SOLVED PROBLEMS MathematicalAnalysis

Tutorial Mathematical Analysis
The exercises can be found at

https://drive.google.com/drive/folders/0B8T_iLf7M-qpdUVEYlFhOVdkRVU
These are tutorial notes which have not undergone systematic proofreading. Readers are asked to
kindly send an email to georgklein53@gmail.com if anything looks like it needs correction or modifi-
cation.
Spis treści
Temat 1 3
Temat 2: School mathematics. Extension. 8
Temat 3: Induction, binomial theorem, inequalities. 25
Temat 4: Weighted mean, bounds of number sets, inequalities between means

(mean values), compound interest, present value, exponential function, logarithm. 38
Temat 5: Sequences I. 55
Temat 6: Sequences II. 74
Academic Year 2015/16, Colloquium I, 20 November, 2015 93
Academic Year 2016/17, Colloquium I, 1 December, 2016 101
Temat 7: Series. 105
Temat 8: The limits of functions. 118
Temat 9: Functions. Continuity of functions. 122
Temat 10: Continuity continued. 127
Temat 11: The derivative of a function 140
Temat 12: Lagrange theorem, derivatives and properties of functions. 146
Temat 13: l’Hôpital’s rule, elasticity, and Taylor’s formula. 152
Questions revisited, using power series and l’Hôpital’s rule 166

Academic Year 2015/16, Colloquium I, 20 November, 2015 . . . . . . . . . . . . . . . 166
Temat 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Academic Year 2015/16, Colloquium II, 15 January, 2016 182
Academic Year 2016/17, Colloquium II, 24 January, 2017 186
Academic Year 2015/16, Examination, 25 January, 2016 189
1
Academic Year 2015/16, Examination, 2 February, 2016 193
Academic Year 2015/16, Examination, 2 March, 2016 199
Academic Year 2016/17, Examination, 1 February, 2017 203
Temat 15: Indefinite integrals. 207
Temat 16: Indefinite, definite and improper integrals 215
Temat 17: Norms and inner products 223
Temat 18: Subsets of Rn . Limits of functions of several variables. Continuity. 232
Temat 19: Functions of several variables 242
Temat 20: Partial derivatives and differentiation of functions of several variables.246
Temat 21: Gradient and growth 262
Academic Year 2014/15, Colloquium I, 17 May, 2015 266
Academic Year 2016/17, Colloquium I, 11 April, 2017, Version A 273
Academic Year 2016/17, Colloquium I, 11 April, 2017, Version B 277
Temat 22: Extrema of functions of several variables. 282
Temat 23: Extrema of functions of several variables II. 289
Temat 24: Additional exercises on the differential calculus of functions of several

variables. 298
Temat 25: Tangent planes and diffeomorphisms. 311
Temat 26: Local invertibility of functions. Implicit function theorem. Extrema

with constraints and Lagrange multipliers. 319
Temat 27: More examples of conditional extrema and the Kuhn-Tucker theorem.332
Academic Year 2015/16, Colloquium II, 3 June, 2016 345
Academic Year 2016/17, Colloquium II, 1 June, 2017 352
Academic Year 2014/15, Examination, 8 September, 2015 359
Academic Year 2015/16, Examination, 6 September, 2016 365
Temat 28: Multiple integrals. Polar substitution.

Gaussian/Euler-Poisson integral. 370
2
Temat 1
Question 1. Let A, B, C be variables with values in the set of students. Every student learns at
least two languages from the set {L1 , L2 , L3 , L4 }. Is the following sentence always true? If A and B
learn at least one language together and B and C learn at least one language together then A and
C learn at least one language together.
Answer: We can show that the sentence is not always true by giving a counter example. Student A
learns L1 and L2 , Student B learns L2 and L3 , and Student C learns L3 and L4 . Then A and C do
not learn a language together.
Question 2. Let a, b, c be natural numbers. Is the following sentence true?

If b|a and c|a then (bc)|a.
Remark: The symbol b|a means that a is divisible by b. In other words, a = b · c for some natural
number c.
Answer: We can show that the sentence is not always true by giving a counter example. 6|12 and
4|12, but 6 · 4 = 24 and 24 6 |12.
Question 3. Is the following sentence a tautology:

1. [(p ∨ q) ∧ ¬p] ⇒ q
2. [(p ⇒ q) ∧ (q ⇒ p)] ⇒ (p ∨ q)
Answer: 1. The question is whether the following statement is true: Given that we have p or q, and
given that we do not have p, we necessarily have q. Intuitively it is clear that the statement is true,
but we have to prove it using a truth table.
p q p ∨ q ¬p (p ∨ q) ∧ ¬p [(p ∨ q) ∧ ¬p] ⇒ q
T T T F F T
T F T F F T
F T T T T T
F F F T F T
Note that the implication in the sixth column could only be false if there was a T in the fifth column
and in the same row an F in the second column. This does not happen, so the last column consists
only of T s, which means that the sentence is a tautology.
2. The question is whether the following statement is true: Given that we have p implies q, and given
that we have q implies p, we necessarily have p or q. We have to construct a truth table.
p q ¬p ¬q p ⇒ q q ⇒ p (p ⇒ q) ∧ (q ⇒ p) p ∨ q [(p ⇒ q) ∧ (q ⇒ p)] ⇒ (p ∨ q)
T T F F T T T T T
T F F T F T F T T
F T T F T F F T T
F F T T T T T F F
There is an F in the last column, so the sentence is not a tautology. It is true in some cases,
corresponding to the first three rows, but it is not always true, so it is not a tautology.
The question came up as to how this exercise relates to the concept of logical equivalence. For
example, one of De Morgan’s laws states that ¬(p ∧ q) is logically equivalent to (¬p) ∨ (¬q). The
proof also uses a truth table.
3
p q ¬p ¬q p ∧ q ¬(p ∧ q) (¬p) ∨ (¬q)
T T F F T F F
T F F T F T T
F T T F F T T
F F T T F T T
The last two columns have the same values in all rows, hence ¬(p ∧ q) and (¬p) ∨ (¬q) are logically
equivalent.
Question 4. Consider the formula Φ(m, x) with two real-valued variables m, x :
Φ(m, x) ⇔ mx2 + mx + 4 > 0.
Which of the sentences are true:

a) ∀m ∀x Φ(m, x).
b) ∀m ∃x Φ(m, x).
c) ∃m ∀x Φ(m, x).
d) ∃m ∃x Φ(m, x).
e) ∀x ∃m Φ(m, x).
f) ∃x ∀m Φ(m, x).
Answer: a) For m = −3 and x = 1, we do not have mx2 + mx + 4 > 0, so the sentence is false.
b) For any value of m, we can chose x = 0, and then mx2 + mx + 4 > 0, so the sentence is true.
c) There exists m, namely m = 0, such that for any value of x, we have mx2 + mx + 4 > 0, so the
sentence is true.
d) Clearly, there exists m, for example m = 0, and there exists x, for example x = 0, such that
mx2 + mx + 4 > 0, so the sentence is true.
e) For any value of x, we can chose m = 0, and then mx2 + mx + 4 > 0, so the sentence is true.
f) There exists x, namely x = 0, such that for any value of m, we have mx2 + mx + 4 > 0, so the
sentence is true.
Question 5. On the real plan, draw the set of points with coordinates (x, y) given by the following
condition: a) x2 + y 2 2
b) |x| + |y| ¬ 2.
√
Answer: a) First one should realize that x2 + y 2 = 2 is the the equation of a circle of radius 2 in the
real plan. This is so because in polar coordinates, x = r cos θ and y = r sin θ and by the Pythagorean
theorem, x2 + y 2 = r2 cos2 θ + r2 sin2 θ = r2 (cos2 θ + sin2 θ) = r2 . So the set of points consists of all
the points on the red circle and all the points outside the red circle.
4
b) One needs to make a distinction of four cases.
Case 1. In the first quadrant (meaning x 0 and y 0) we have |x| = x and |y| = y. Thus
|x| + |y| ¬ 2 is the equivalent to x + y ¬ 2, so y ¬ 2 − x. This gives us the green line and everything
in the first quadrant below the green line.
Case 2. In the second quadrant (meaning x ¬ 0 and y 0) we have |x| = −x and |y| = y. Thus
|x| + |y| ¬ 2 is the equivalent to −x + y ¬ 2, so y ¬ 2 + x. This gives us the blue line and everything
in the second quadrant below the blue line.
Case 3. In the third quadrant (meaning x ¬ 0 and y ¬ 0) we have |x| = −x and |y| = −y. Thus
|x| + |y| ¬ 2 is the equivalent to −x + −y ¬ 2, so −x − 2 ¬ y. This gives us the orange line and
everything in the third quadrant above the orange line.
Case 4. In the fourth quadrant (meaning x 0 and y ¬ 0) we have |x| = x and |y| = −y. Thus
|x| + |y| ¬ 2 is the equivalent to x + −y ¬ 2, so x − 2 ¬ y. This gives us the red line and everything
in the fourth quadrant above the red line.
So the set of points consists of all the points on the rectangle and all the points inside the rectangle
(in four colours).
5
Question 6. Let
y

A = {(x, y) : y < x}, B = (x, y) : <1
x
Find and draw the following sets:
A ∩ B, A ∪ B, A \ B, B \ A.
Answer: We have the line y = x (green). The set A corresponds to everything which is below that
line. For the set B, we have to consider two cases. When x > 0, then y < x and this also corresponds
y
to everything which is below the the green line (in the region x > 0). When x < 0, then < 1
x
is equivalent to y > x. This corresponds to everything which is above the green line (in the region
x < 0).
6
y

A ∩ B = (x, y) : (y < x) ∧ < 1 , this corresponds to everything below the green line in the
x
region x > 0.
y

A ∪ B = (x, y) : (y < x) ∨ < 1 , this corresponds to everything below the green line, and,
x
additionally, in the region x < 0, everything above the green line.
y

A \ B = (x, y) : (y < x) ∧ ¬ < 1 , this corresponds to everything below the green line in the
x
region x ¬ 0.
y

B \ A = (x, y) : < 1 ∧ ¬(y < x) , this corresponds to everything above the green line in the
x
region x < 0.
Question 7. Let
A = {x2 + y 2 ¬ 2}, B = {(x, y) : x + y > 0}
Draw the following sets on the real plane:
A ∩ B, A ∪ B, A \ B, B \ A.
√
We have a circle of radius 2 (red). The set set A corresponds to everything on that circle and
everything inside that circle. We have the line y = −x (blue). The set B corresponds to everything
above that line. The set A ∩ B corresponds to everything which is at the same time above the blue
line and on or inside the red circle. The set A ∪ B corresponds to everything above the blue line in
addition to everything on or inside the red circle. The set A \ B corresponds to everything on or
inside the red circle and at the same time on or below the blue line. The set B \ A corresponds to
everything above the blue line and at the same time outside the red circle.
Question 8. Using mathematical induction, show that for any natural number n:
1. (1 + a)n 1 + na, where a > −1
7
n(n + 1)
2. 1 + 2 + 3 + · · · + n =
2
Answer: 1. For n = 1, we clearly have 1 + a 1 + a. Now we need to show that if the statement
is true for a given n ∈ N, then it is also true for n + 1. This means we have have to show that
(1 + a)n+1 1 + (n + 1)a = 1 + na + a, under the assumption that a > −1 and (1 + a)n 1 + na.
By the induction hypothesis, we have
(1 + a)n+1 = (1 + a)(1 + a)n = (1 + a)n + a(1 + a)n 1 + na + a(1 + a)n .
So now we only need to show that a(1 + a)n a, under the assumption that a > −1. For a = 0,
clearly a(1 + a)n a. For −1 < a < 0, the inequality a(1 + a)n a is equivalent to (1 + a)n ¬ 1,
which holds since for −1 < a < 0, we have 0 < 1 + a < 1. For a > 0, the inequality a(1 + a)n a is
equivalent to (1 + a)n 1, which holds since for a > 0, we have 1 + a > 1.
1(1 + 1)
2. For n = 1, we clearly have 1 = . Now we need to show that if the statement is true for a
2
given n ∈ N, then it is also true for n + 1. This means we have to show that
(n + 1)(n + 1 + 1) (n + 1)(n + 2)
1 + 2 + 3 + · · · + n + (n + 1) = = ,
2 2
n(n + 1)
under the assumption that 1 + 2 + 3 + · · · + n = . Using the induction hypothesis, we get
2
n(n + 1) n(n + 1) + 2(n + 1) (n + 1)(n + 2)
1 + 2 + 3 + · · · + n + (n + 1) = + (n + 1) = = .
2 2 2
Temat 2: School mathematics. Extension.

Question 1. Show that:
a) for any x, y, z ∈ R we have |x + y + z| ¬ |x| + |y| + |z|;
b) for any x, y ∈ R we have ||x| − |y|| ¬ |x − y|;
1
c) for any x, y ∈ R we have max{x, y} = (x + y + |x − y|). Find the analogous formula for
2
min{x, y}.
√
Answer. We will use that for any real number a, we have |a| = a2 .
q √ q √
a) We need to show that (x + y + z)2 ¬ x2 + y 2 + z 2 . Since on both sides of the inequality
we have positive numbers, we can take the square:
q q √
(x + y + z)2 = x2 + y 2 + z 2 + 2xy + 2yz + 2zx ¬ x2 + y 2 + z 2 + 2 x2 y 2 + 2 y 2 z 2 + 2 z 2 x2
√ q √
= x2 + y2 + z2 .
q q
We used the fact that xy ¬ |xy| = (xy)2 = x2 y 2 .
8
s
√ q 2 q
b) We need to show that x2 − y2 ¬
(x − y)2 . Since on both sides of the inequality we
√ q 2
have positive numbers, we can take the square: x − y 2 ¬ (x − y)2 . Now we have
2
√ q 2 q
x2 − y2 = x2 − 2 x2 y 2 + y 2 ¬ x2 − 2xy + y 2 = (x − y)2 .
q q
We used the fact that xy ¬ |xy| = (xy)2 = x2 y 2 .
c) Here we consider three cases. If x = y, then
1 1 1
max{x, y} = (x + y + |x − y|) = (x + x + |x − x|) = (x + x) = x = y.
2 2 2
If x y, then
1 1 1
max{x, y} = (x + y + |x − y|) = (x + y + (x − y)) = (x + x) = x.
2 2 2
If y x, then
1 1 1
max{x, y} = (x + y + |x − y|) = (x + y − (x − y)) = (y + y) = y.
2 2 2
1
The analogous formula for the minimum is min{x, y} = (x+y−|x−y|), the verification is analogous.
2
Question 2. Calculate:
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
(5 5 − 4 4 )(5 5 − 3 3 )(5 5 − 2 2 )(5 5 − 1)(4 4 − 3 3 )(4 4 − 2 2 )(4 4 − 1)(3 3 − 2 2 )(3 3 − 1)(2 2 − 1).
1
1 2 1 1 1
Answer: Since 4 4 = 22 4
= 2 4 = 2 2 , we have 4 4 − 2 2 = 0. Hence the product is equal to zero.
Question 3. Calculate  2
1
− 1 .
 
1
−2

√ 1 −2
2−1
Answer:
 2  2
 2  2
1  1  1 1

1 − 1
 =  √ − 1 = √ − 1 =  √ − 1
−2

2−1 2−1 2−1
−2 √ −2 √ −2
 
√ √
√ 1 −2
2−1 √2−1 −2( 2−1) 1−2 2+2 3−2 2
2−1

√ 2
√ !2
3−2 2 3−2 2
=  √

√ √ − 1
 = √ √ −1
( 2−1)(3−2
√ 2)
− 2(3 − 2 2) 2 − 1 − 2(3 − 2 2)
3−2 2
√ !2 √ !2
3−2 2 3−2 2
= √ √ −1 = √ −1
2 − 1 − 6 + 4 2) 5 2−7
9
√ √ !2 √ √ !2 √ !2
3−2 2 5 2−7 3−2 2−5 2+7 10 − 7 2
= √ − √ = √ = √
5 2−7 5 2−7 5 2−7 5 2−7
√ √ √
(10 − 7 2)2 100 − 140 2 + 2 · 49 2 · (50 − 70 2 + 49)
= √ = √ = √ = 2.
(5 2 − 7)2 2 · 25 − 70 2 + 49 50 − 70 2 + 49
√ 1

√
2 4
Question 4. Let f (x) = 1 + x + x . Find f (−x), f , f ( x), and f (x2 ).
x
q √
Answer: f (−x) = 1 + (−x)2 + (−x)4 = 1 + x2 + x4 = f (x).
s s s
1 1 2 1 4 1 1 x4 + x2 + 1 1√ 1

f = 1+ + = 1+ 2 + 4 = 4
= 2 1 + x2 + x4 = 2 f (x).
x x x x x x x x
√ q √ √
f ( x) = 1 + ( x)2 + ( x)4 .
q √
f (x2 ) = 1 + (x2 )2 + (x2 )4 = 1 + x4 + x8 .
Question 5. Find the maximal domain (the subset of R) on which the functions below are well
defined:
√ √ q √
a) f (x) = x2 ; b) f (x) = ( x)2 ; c) f (x) = 4 − x;
s
q
3
√ x x+1
d) f (x) = 4− x; e) f (x) = ; f) f (x) = .
|x| x−1
√
Answer: a) Since x2 is always greater than or equal to zero, f (x) = x2 is well defined on all of R.
√
b) The square root is defined for non-negative numbers, so f (x) = ( x)2 is well defined on the
interval [0, ∞). Zero is included in the interval, infinity is not.
√ q √
c) For x to be well defined, we need to have x 0. For 4 − x to be well defined we need to
√ q √
have 4 − x 0, which implies x ¬ 16. Hence f (x) = 4 − x is well defined on the interval [0, 16].
Zero and 16 are included in the interval.
√
d) For x to be well defined, we needqto have x 0. The cube root is well defined, for non-negative
3
√
and for negative numbers. So f (x) = 4 − x is well defined on the interval [0, ∞). Zero is included
in the interval, infinity is not.
x
e) The absolute value is defined everywhere, but division by zero is not defined. Hence f (x) =
|x|
is well defined on the real numbers without zero, which we can write as R \ 0.
x+1
f) Division by zero is not defined, so we know that for x = 1, the function is not well defined.
x−1
x+1
The square root is defined for non-negative numbers, so we need to have 0. We will consider
x−1
two cases, first suppose x − 1 > 0. Then we can multiply by x − 1 to get x + 1 0. So in this case
x > 1 and x + 1 0, which can be combined to x > 1. If x − 1 < 0, then when multiplying by x − 1,
we need to reverse the inequality. So we get x + 1 ¬ 0. In this case x < 1 and x ¬ −1, which can be
10
s
x+1
combined to x ¬ −1. So f (x) = is well defined on the union of the intervals (−∞, −1] and
x−1
(1, ∞), which can be written as (−∞, −1] ∪ (1, ∞), square brackets mean the boundary is included,
round brackets mean it is not.
Question 6.
a) Find the equation for the bisector of the line segment between the points (1, −5) and (3, −1).
b) Find the center and the radius of the circle given by the equation 2x2 + 2y 2 + 2x − 2y = 1.
c) Find the vertex of the parabola given by the equation 2y = x2 − 4x + 8.
(1, −5) + (3, −1)
Answer a) We get the midpoint of the segment by taking = (2, −3). The slope of
2
∆y −1 − (−5) 4
the line going through the points (1, −5) and (3, −1) is = = = 2. The slope of the
∆x 3−1 2
1
bisector can be calculate by taking −1 times the reciprocal of the slope of the original line, so is − .
2
1
We need a line through the point (2, −3) with slope − . We know that the line can be expressed as
2
1
y = − x + b. Since it needs to go through the point (x = 2, y = −3), we get −3 = −1 + b, so b = −2
2
1
and the equation for the bisector is y = − x − 2.
2
b) This is an application of the well known formula (a + b)2 = a2 + 2ab + b2 . The application is called
„completing the square”.
1 = 2x2 + 2y 2 + 2x − 2y
1 1 1 1 1
= x2 + y 2 + x − y = x2 + x + − + y 2 − y + −
2 | {z 4} 4 | {z 4} 4
complete square complete square
2 2 2 2
1 1 1 1 1 1 1

= x+ − + y− − = x+ + y− −
2 4 2 4 2 2 2
2 2
1 1

1= x+ + y−
2 2
The equation of a circle is given by r2 = (x − x0 )2 + (y − y0 )2 , where r is the radius and (x0 , y0 ) is
√
1 1

the center of the circle. Hence the radius is 1 = 1, and the center is (x0 , y0 ) = − , .
2 2
c) This is an application of the well known formula (a + b)2 = a2 + 2ab + b2 . The application is called
„completing the square”.
2y = x2 − 4x + 8 = x
|
2
− {z
4x + 4} −4 + 8
complete square
= (x − 2)2 − 4 + 8 = (x − 2)2 + 4
1
y = (x − 2)2 + 2
2
11
The vertex of a parabola given by the equation y = m(x − x0 )2 + y0 is (x0 , y0 ). Hence the vertex of
the parabola is (x0 , y0 ) = (2, 2).
Question 7.
a) Solve the equation |x − 2| + |x − 8| = 1.
b) Solve the inequality ||2x − 3| − 3| > 2.
c) Solve the system of equations:

|x + 1| + |y− 1| = 5,
|x + 1| = 4y − 4.
d) Solve the system of equations: 

 x2
− x − 6 > 0,
2x 15 − x2 .
Answer: a) We need to consider three cases. If x < 2, the equation |x − 2| + |x − 8| = 1 becomes

9
−(x − 2) − (x − 8) = 1, so −2x + 10 = 1 which implies x = and is in contradiction with x < 2, so
2
no solution in this case.
If 2 ¬ x < 8, the equation |x − 2| + |x − 8| = 1 becomes x − 2 − (x − 8) = 1, so 6 = 1, a contradiction,
so no solution in this case.
If x 8, the equation |x − 2| + |x − 8| = 1 becomes x − 2 + x − 8 = 1, so 2x = 11, which implies
11
x= and is in contradiction with x 8, so no solution in this case.
2
b) Case 1.1: 2x − 3 0 and 2x − 3 − 3 0. The inequality ||2x − 3| − 3| > 2 becomes 2x − 3 − 3 > 2,
so 2x > 8 and x > 4. We check that for x > 4, we have 2x − 3 0 and 2x − 3 − 3 0, so x > 4 is a
solution.
Case 1.2. 2x − 3 0 and 2x − 3 − 3 < 0. The inequality ||2x − 3| − 3| > 2 becomes −2x + 3 + 3 > 2,
so −2x > −4 and x < 2. We check that for x < 2, we have 2x − 3 − 3 < 0, but 2x − 3 0 implies
3 3
x . So ¬ x < 2 is a solution
2 2
Case 2.1: 2x − 3 < 0 and −2x + 3 − 3 0. The inequality ||2x − 3| − 3| > 2 becomes −2x + 3 − 3 > 2,
so −2x > 2 and x < −1. We check that for x < −1, we have 2x − 3 < 0 and −2x + 3 − 3 0, so
x < −1 is a solution.
Case 2.2: 2x − 3 < 0 and −2x + 3 − 3 < 0. The inequality ||2x − 3| − 3| > 2 becomes 2x − 3 + 3 > 2,
so 2x > 2 and x > 1. We check that for x > 1, we have −2x + 3 − 3 < 0, but 2x − 3 < 0 implies
3 3
x < , so 1 < x < is a solution.
2 2
Combining Cases 1.1, 1.2, 2.1, and 2.2, we get that the set of all solutions is (−∞, −1)∪(1, 2)∪(4, ∞).
c) Case 1: x + 1 0 and y − 1 0. The system of equations becomes


x + 1 + y − 1 = 5,
x + 1 = 4y − 4.
12
So x + 1 = 6 − y = 4y − 4, hence y = 2 and x = 3. We check that this solutions satisfies x + 1 0
and y − 1 0.
Case 2: x + 1 0 and y − 1 < 0. The system of equations becomes

x + 1 − y+ 1 = 5,
x + 1 = 4y − 4.
8 17 8 5
So x+1 = 4+y = 4y −4, hence y = and x = . But y −1 < 0 and −1 = 0, a contradiction,
3 3 3 3
so this does not give us a solution.
Case 3: x + 1 < 0 and y − 1 0. The system of equations becomes

−x − 1 + y − 1 = 5,
−x − 1 = 4y − 4.
So −x − 1 = 6 − y = 4y − 4, hence y = 2 and x = −5. We check that this solutions satisfies x + 1 < 0

and y − 1 0.
Case 4: x + 1 < 0 and y − 1 < 0. The system of equations becomes

−x − 1 − y + 1 = 5,
−x − 1 = 4y − 4.
8 23 8 5
So −x − 1 = 4 + y = 4y − 4, hence y = and x = − . But y − 1 < 0 and − 1 = 0, a
3 3 3 3
contradiction, so this does not give us a solution.
Combining Cases 1, 2, 3, and 4, we get two solutions, (x, y) = (3, 2) and (x, y) = (−5, 2).
d) It is easiest to first solve the equalities x2 − x − 6 = 0 and 2x = 15 − x2 .

1 1
0 = x2 − x − 6 = x 2 − x + − −6
| {z 4} 4
complete square
2 2
1 1 1 25

= x− − −6= x− −
2 4 2 4
2
25 1

= x−
4 2
5 1
± =x−
2 2
1 5
x = ± = 3 or − 2
2 2
We need x2 − x − 6 > 0 and the parabola x2 − x − 6 is shaped upwards, so we have x < −2 or x > 3.
Now we solve 2x = 15 − x2 , which is equivalent to 0 = x2 + 2x − 15
0 = x2 + 2x − 15 = x
|
2
2x + 1} −1 − 15
+ {z
complete square
13
= (x + 1)2 − 1 − 15 = (x + 1)2 − 16
16 = (x + 1)2
±4 = x + 1
x = −1 ± 4 = 3 or − 5
We need 2x 15 − x2 , and we know that for x = 3 and x = −5, we have 2x = 15 − x2 . This means
that the line 2x intersects with the parabola 15 − x2 at x = 3 and x = −5. Since the parabola 15 − x2
is shaped downwards, we conclude that to have 2x 15 − x2 , we need x < −5 or x > 3.
Since we need simultaneous solutions to x2 − x − 6 > 0 and 2x 15 − x2 , we get x < −5 and x > 3.
Question 8. Calculate:
q √ q √
a) 15 − 6 6 + 15 + 6 6;
q
3 √ q
3 √
b) 20 + 14 2 + 20 − 14 2.
1 1
c) Calculate x3 − 3
under the assumption that x − = 4.
x x
Answer: a)
s
q √ q √ q √ q √ 2
15 − 6 6 + 15 + 6 6 = 15 − 6 6 + 15 + 6 6
r
√ q √ q √ √
= 15 − 6 6 + 2 15 − 6 6 15 + 6 6 + 15 + 6 6
s
√ r
√ √ √
= 15 − 6 6 + 2 15 − 6 6 15 + 6 6 + 15 + 6 6
q √ √ √
= 15 − 6 6 + 2 225 − 36 · 6 + 15 + 6 6
q √ √ √ q √ √ √
= 15 − 6 6 + 2 9 + 15 + 6 6 = 30 + 2 9 = 30 + 6 = 36 = 6
b)
q
3 √ q
3 √
20 + 14 2 + 20 − 14 2 =
s s
3
q
3 √ q
3 √ 3 3
√ 1 √ 1 3
= 20 + 14 2 + 20 − 14 2 = 20 + 14 2 3 + 20 − 14 2 3
s
3
√ √ 2 √ 1 √ 1 √ 2 √
= 20 + 14 2 + 3 20 + 14 2 3 20 − 14 2 3 + 3 20 + 14 2 3 20 − 14 2 3 + 20 − 14 2
s
3
√ 2 √ 1 √ 1 √ 2
= 40 + 3 20 + 14 2 3 20 − 14 2 3 + 3 20 + 14 2 3 20 − 14 2 3
s √ 13 √ 13 √ 1 √ 1
= 3
40 + 3 20 + 14 2 + 20 − 14 2 20 + 14 2 3 20 − 14 2 3
s
3
q
3 √ q
3 √ r √ √
= 40 + 3 20 + 14 2 + 20 − 14 2 3 20 + 14 2 20 − 14 2
14
√ √ √
s q q
3 3
= 3
40 + 3 20 + 14 2 + 20 − 14 2 3 400 − 2 · 196
√ √ √
s q q
3 3 3 3
= 40 + 3 20 + 14 2 + 20 − 14 2 8
√ √
s q q
3 3 3
= 40 + 3 20 + 14 2 + 20 − 14 2 · 2
√ √
s q q
3 3 3
= 40 + 6 20 + 14 2 + 20 − 14 2
√ √ √ √
q q s q q
3 3 3 3 3
So we have 20 + 14 2 + 20 − 14 2 = 40 + 6 20 + 14 2 + 20 − 14 2 and we want to
q
3 √ √ q
3
q
3 √ q
3 √
know what is 20 + 14 2 + 20 − 14 2. We put x = 20 + 14 2 + 20 − 14 2, and then x =
√3
40 + 6x. From this we get x3 = 40 + 6x. Without knowing a general formula for solving a cubic
equation,
q we can √
look for
q a solution by trying small integers for x. We find that x = 4 is a solution.
3 3 √
So x = 20 + 14 2 + 20 − 14 2 = 4.
1 1
c) We know that x − = 4 and we need to calculate x3 − 3 .
x x
3 2
1 1 1 1

43 = 64 = x − = x3 − 3x2 + 3x −
x x x x3
" 2 #
3 21 1 1
=x −3 x −x −
x x x3
 
x 1 1 1 1

= x3 − 3 x − 3
= x3 − 3 − 12

2
− 3
= x − 3 x −  −
3
x x 
| {z x}
 x x
=4
1 1
So x3 − 3
− 12 = 64 and hence x3 − 3 = 76.
x x
Question 9. In Cartesian coordinate system, draw the plot of the function f :

a) y = f (x) = |1 − x| − |x − 2| − |x − 3|;
b) y = f (x) = |x − 1|(|x| − 1) .
Answer: a) Since |1 − x| = | − (1 − x)| = |x − 1|, it is convenient to write the function as

f (x) = |x − 1| − |x − 2| − |x − 3|. There are four cases:
If x 3, the graph of the function is y = (x − 1) − (x − 2) − (x − 3) = −x + 4 (green).
If 2 ¬ x < 3, the graph of the function is y = (x − 1) − (x − 2) + (x − 3) = x − 2 (red).
If 1 ¬ x < 2, the graph of the function is y = (x − 1) + (x − 2) + (x − 3) = 3x − 6 (blue).
If x < 1, the graph of the function is y = −(x − 1) + (x − 2) + (x − 3) = x − 4 (orange).
15
b) There are three cases:
If x < 0, the graph of the function f (x) = |x − 1|(|x| − 1) is
y = −(x − 1)(−x − 1) = (x − 1)(x + 1) = x2 − 1 (green)
If 0 ¬ x < 1, the graph of the function f (x) = |x − 1|(|x| − 1) is y = −(x − 1)(x − 1) = −(x − 1)2
(red)
If x 1, the graph of the function f (x) = |x − 1|(|x| − 1) is y = (x − 1)(x − 1) = (x − 1)2 (blue)
Question 10. Find the value of the parameter m ∈ R under the assumption that the one of the
solutions of the equation 4x2 − 15x + 4m2 = 0 is equal to the square of the other solution.
Answer:
0 = 4x2 − 15x + 4m2
16
15 15 225 225
0 = x2 − x + m2 = x2 − x + − + m2
4 | 4 {z 64} 64
complete square
2
15 225

= x− − + m2
8 64
2
15 225

x− = − m2
8 64
s
15 225
x− =± − m2
8 64
s s s
15 225 15 225 15 225
x= ± − m2 = + − m2 or − − m2
8 64 8 64 8 64
Now we require that one of the solutions is equal to the square of the other solution. This gives
s  s 2
15 225 15 225
+ − m2 =  − − m2 
8 64 8 64
s
225 15 225 225
= − − m2 + − m2
64 4 64 64
s s
120 4 225 225 15 225 225
+ − m2 = − − m2 + − m2
64 4 64 64 4 64 64
s
19 225 330
− m2 = − m2
4 64 64
s
81225 361 2 330
− m = − m2
1024 16 64
2
81225 361 2 165

− m = − m2
1024 16 32
81225 361 2 27225 165 2
− m = − m + m4
1024 16 1024 16
196 2 54000
0 = m4 + m −
16 1024
196 2 36416 38416 54000
0 = m4 + m + − −
| 16 {z 1024} 1024 1024
complete square
2
196 92416

2
= m + −
32 1024
2
196 92416

m2 + =
32 1024
2
49 361

m2 + =
8 4
17
s
49 361 19 76
m2 + =± =± =±
8 4 2 8
49 76 27 125
m2 = − ± = or −
8 8 8 8
s s
27 3 3
Of course m2 cannot be negative, so we get m = ± =± ·
8 2 2
Question 11. Find the value of the parameter m ∈ R in such a way that the solutions x1 , x2 of the
equation x2 + mx + 7 = 0 satisfy |x1 − x2 | = 1.
Answer:
m2 m2
0 = x2 + mx + 7 = x2 + mx + − +7
| {z 4} 4
complete square
m2
2
m

= x+ − +7
2 4
s
m m2
x+ =± −7
2 4
s
m m2
x=− ± −7
2 4
s s
m m2 m m2
So we have x1 = − + − 7 and x2 = − − − 7, therefore
2 4 2 4
v
u s   s 2
q u
m m2 m m2
1 = |x1 − x2 | = (x1 − x2 )2 = t− − 7 − − − − 7
u
+
2 4 2 4
v v
us s 2 u s 2
m2 m2 m2 √ √
u u r
2
−7+ − 7 = − 7 m2 − 28 m2 − 28
u u
= t t2 = =
4 4 4
12 = 1 = m2 − 28
m2 = 29
√
m = ± 29.
1 495
Question 12. Let an be an arithmetic sequence, given by an+1 = an + . We know that a1 a2 a3 = .
2 4
Find a1 + a2 + a3 .
Answer: First, we should make it clear what such a sequence is. For example 1, 1.5, 2, 2.5 . . . or
3.2, 3.7, 4.2, 4.7 . . .. We just need to determine what is a1 , the first term. We know that the product
495
of the first three terms is . To get an approximate idea of what is a1 , we will solve the following
4
18
equation
500 495
a1 a2 a3 ≈ a32 = 125 = ≈
4 4
√
3
√3
Then a2 = 125 = 5 · 5 · 5 = 5. So we try a2 = 5, from which we get a1 = 4.5 and a5 = 5.5. We
check
9 11 9 · 5 · 11 495
a1 a2 a3 = 4.5 · 5 · 5.5 = ·5· = = .
2 2 4 4
Thus
a1 + a2 + a3 = 4.5 + 5 + 5.5 = 15.
Question 13. Show that the polynomial W (x), such that all the coefficients are integers and such
that: W (13) = 3 and W (17) = 5, cannot exist.
Answer: First, we should make it clear what it means to be a polynomial with integer coefficients.
It looks like
W (x) = c0 + c1 x + c2 x2 + c3 x3 + · · · cn xn ,
where all the c’s a are integers. If we assume that there is such a polynomial with W (13) = 3 and
W (17) = 5, then W (17) − W (13) = 5 − 3 = 2. We will establish a contradiction. We know that
17 = 13 + 4, so
2 = W (17) − W (13) = W (13 + 4) − W (13)

c0 + c1 (13 + 4) + c2 (13 + 4)2 + c3 (13 + 4)3 + · · · cn (13 + 4)n
=

c0 + c1 (13) + c2 (13)2 + c3 (13)3 + · · · cn (13)n
−
Of course the two constants, c0 , cancel each other. Now we look at the binomial expansion of (13+4)i ,
for any i with 1 ¬ i ¬ n. We get
! ! ! ! ! !
i i i i i i
(13 + 4)i = 13i 40 + 13i−1 41 + 13i−2 42 + · · · + 132 4i−2 + 131 4i−1 + 130 4i .
0 1 2 i−2 i−1 i
!
i
Since 13i 40 = 13i , it cancels with the corresponding term in W (13). All other terms in the
0
binomial expansion of (13 + 4)i are multiples of 4. So we have expressed 2 as a sum of integers all of
which contain the factor 4. This would imply that 2 is divisible by 4, contradiction.
Question 14. A biker covered a distance of 96 km. He did this in 2 hours less than he planned to.
In every hour, he covered a distance 1 km bigger than he planned to cover in 75 minutes. What was
the speed of the biker? (We assume that the speed was constant)
Answer: We will use v, the actual speed and vp , the planned speed. We will use kilometers per hours
as units for speed. So 75 minutes will be converted to 1.25 hours. From this we get a relation between
the planned speed and the actual speed:
number of km planned for 75 minutes
z }| {
v= 1.25 · vp +1 .
| {z }
(number of km actually covered in one hour)= speed
19
Since he covered 96 km in two hours less than he planned to, using
distance
time = ,
speed
we get
96 96
= − 2.
v vp
Using the expression relating v and vp , we get
96 96
= −2
1.25 · vp + 1 vp
96 · vp = 96(1.25 · vp + 1) − 2(1.25 · vp + 1)vp
5
= 120 · vp + 96 − vp2 − 2 · vp
2
5
0 = − vp2 + 22 · vp + 96
2
44 192
0 = vp2 − vp −
5 5
44 484 484 192 44 484 1444
= vp2 − vp + − − = vp2 − vp + −
| 5 {z 25 } 25 5 | 5 {z 25} 25
2
22 1444

= vp − −
5 25
2
22 1444

vp − =
5 25
s
22 1444 38
vp − =± =±
5 25 5
22 38 16
±
vp = = 12 or − .
5 5 5
Of course only the positive solution for vp makes sense. So we get
v = 1.25 · vp + 1 = 1.25 · 12 + 1 = 16.
Additional explanation: The speed has units km per hour. He planned to do the 96 km in 8 hours,
but he actually did them in only 6 hours.
Question 15. Show that the plot of any polynomial of the degree 2 has an axis of symmetry. Is the
analogous statement true for any polynomial of even degree?
Answer: The plot of a polynomial of degree 2 is a parabola. A parabola clearly has an axis of
symmetry, cutting the parabola in a right and a left half. The general expression for a polynomial of
degree two is
 
!
b c  b b2 b2 c
f (x) = ax2 + bx + c = a x2 + x + = a x2 + x + 2 − 2 + 
 
a a 
| a {z 4a } 4a a 
complete square
20
 !2  !2
b b2 c b b2
= a x + − 2 + =a x+ − +c
2a 4a a 2a 4a
b
If we let x0 = − , we have
2a
!2 !2
b b2 b b b2 b2
f (x0 + x) = a x0 + x + − +c=a − +x+ − + c = ax2 − + c,
2a 4a 2a 2a 4a 4a
and
!2 !2
b b2 b b b2 b2
f (x0 − x) = a x0 − x + − +c=a − −x+ − + c = a(−x)2 − + c.
2a 4a 2a 2a 4a 4a
b
Since x2 = (−x)2 , we have f (x0 + x) = f (x0 − x), which means that x = x0 = − is an axis of
2a
symmetry.
For polynomials of even degree more than two, the analogous statement is not true. For example,
consider f (x) = x4 − x = x3 (x − 1). It intersects the x-axis only at x = 0 and x = 1. So if there was
1
an axis of symmetry, it would need to be at x0 = . But
2
3 1 3

f x0 + =f + = f (2) = 24 − 2 = 14,
2 2 2
3 1 3

f x0 − =f − = f (−1) = (−1)4 − (−1) = 2.
2 2 2
So f (x0 + x) 6= f (x0 − x), a contradiction from which we conclude that the polynomial f (x) = x4 − x
does not have an axis of symmetry.
Question 16. Solve the inequalities:

14 10
a) x3 + 2x2 − 5x − 6 < 0; b) < − 3.
x2 − 5x + 6 2−x
Answer: a) We should first solve the equation x3 + 2x2 − 5x − 6 = 0. We try small integers to find a
solution, and it turns out that x = −1 satisfies the equation. So we need to divide x3 + 2x2 − 5x − 6
by (x − (−1)). We get
x3 + 2x2 − 5x − 6 = (x + 1)(x2 + x − 6),
and we can factor (x2 + x − 6) to get
x3 + 2x2 − 5x − 6 = (x + 1)(x2 + x − 6) = (x + 1)(x − 2)(x + 3).
So we have three solutions to the equation x3 + 2x2 − 5x − 6 = 0, namely x = −1, x = 2, and

x = −3. It is clear that when x is very big, the value of x3 + 2x2 − 5x − 6 is positive, and when x
is very small (big in absolute value, but negative), the value of x3 + 2x2 − 5x − 6 is negative. So we
get that x3 + 2x2 − 5x − 6 > 0 when x > 2 or −3 < x < −1. We can write the set of solutions as
21
a union of intervals, (−3, −1) ∪ (2, ∞), where round brackets mean that the limit is not included in
the interval.
b) We can factor the denominator on the left side of the inequality to get
14 10
< − 3.
(3 − x)(2 − x) 2−x
Now we need to distinguish four cases, keeping in mind that when multiplying by a negative number,
we need to reverse the inequality. Before looking at the four cases, we will solve the equation
14 10
= −3
(3 − x)(2 − x) 2−x
14 = 10(3 − x) − 3(3 − x)(2 − x) = 30 − 10x − 3x2 + 15x − 18
= −3x2 + 5x + 12
0 = −3x2 + 5x − 2
5 2 5 25 25 2 5 2 5 25 1
0 = x2 − + = x2 − + − + = x2 − + = x2 − + −
3 3 | 3
{z 36
} 36 3 3 3 | 3
{z 36
} 36
2
5 1

= x− −
6 36
s
5 1 1
x− =± =±
6 36 6
5 1 2
x= ± = 1 or .
6 6 3
2
So we have equality for x = 1 and x = . Now we look at the different cases.
3
Case 1: we have 3 − x > 0 and 2 − x > 0. Then we are actually solving the inequality
14 10
> −3
(3 − x)(2 − x) 2−x
0 > −3x2 + 5x − 2
0 < 3x2 − 5x + 2
3
The last inequality is about a parabola which is shaped upwards, so it is true when x is between
2
and 1. We check that in this interval, we have 3 − x > 0 and 2 − x > 0.
Case 2: we have 3 − x > 0 and 2 − x < 0. Then we are actually solving the inequality
14 10
> −3
(3 − x)(2 − x) 2−x
0 < −3x2 + 5x − 2
0 > 3x2 − 5x + 2
22
3
The last inequality is about a parabola which is shaped upwards, so it is true when x < or x > 1.
2
Since we have 3 − x > 0 and 2 − x < 0, we only get 2 < x < 3 as a solution.
Case 3: we have 3 − x < 0 and 2 − x > 0. This cannot actually happen.
Case 4: we have 3 − x < 0 and 2 − x < 0. Then we are actually solving the inequality
14 10
> −3
(3 − x)(2 − x) 2−x
0 > −3x2 + 5x − 2
0 < 3x2 − 5x + 2
3
The last inequality is about a parabola which is shaped upwards, so it is true when x is between
2
and 1. But this is in contradiction with 3 − x < 0 and 2 − x < 0. So we don’t get a solution in this
case.
2

Combining the solutions in the four cases, we get , 1 ∪ (2, 3), the round brackets mean that the
3
limit is not included in the interval.
Question 17. Solve the equation

√
x2 + 4x − 16 2x + 20 = 0.
We try small integers and we notice that x = 2 is solution. Now we put 2x = z 2 and make a
substitution
z4
+ 2z 2 − 16z + 20 = 0
4
We will multiply this equation by 4, and we know that z 2 = 4 is a solution, so either z = 2 or
z = −2 is a solution. By trying, we find that z = 2 is a solution, but z = −2 is not. So we can factor
out (z − 2).
0 = z 4 + 8z 2 − 64z + 80
= (z − 2)(z 3 + 2z 2 + 12z − 40)
Now we look at the second factor, and again z = 2 is a solution, so we can take out another factor
(z − 2):
0 = (z − 2)(z 3 + 2z 2 + 12z − 40)

= (z − 2)(z − 2)(z 2 + 4z + 20)
= (z − 2)(z − 2)[(z + 2)2 + 16]
| {z }
never zero
So the only solution we get is z = 2. Since 2x = z 2 , this makes x = 2 the only solution to the
original equation.
Question 18. Show that for any x ∈ R the following inequalities are true:
1
a) x4 − 6x3 + 10x2 − 4x + 4 > 0; b) x2 − x + > 0; c) x4 − x + 1 > 0.
2
23
a)
x4 − 6x3 + 10x2 − 4x + 4 = x4 − 6x3 + 9x2 + x2 − 4x + 4
 
= x2  |
2
{z }

|
2
{z }
2 2 2
 x − 6x + 9  + x − 4x + 4 = x (x − 3) + (x − 2) > 0
| {z } | {z }
complete square complete square 0 0
The summands can be zero, but not for the same x. So the sum is always greater than zero.
1 1 1 1 1 1

b) x − x + = x2 − x + − + = x −
2
+ > 0.
2 | {z 4} 4 2 2 4
complete square
c) It is clear that for x ¬ 0, we have x4 −x+1 > 0. And for x 1, we have x4 x, so x4 −x+1 > 0.
For 0 < x < 1, we have x > x4 , but also x < 1, so x4 − x + 1 > 0.
Question 19. Solve the equation sin x + cos 2x = 0.

Answer: There are two ways of doing this. One may look a the shape of the sin and cos curves,
π π 1

keeping in mind that sin = cos = .
6 3 2
7π 1 14π 2π 1 π 1

By the shape of the curves, sin = − and cos = cos = . Also, sin − =−
6 2 6 6 2 6 2
2π 1

and cos − = . This repeats itself with a period of 2π, so two general solutions for x are given
6 2
7π π
by x = + 2kπ and x = − + 2kπ, for and integer k (positive, zero, or negative).
6 6
π π
Also, sin = 1 and cos(π) = −1, so another general solution for x is given by x = + 2kπ, for
2 2
and integer k (positive, zero, or negative).
Alternatively, one may use the formula cos(2x) = 1 − 2 sin2 x. Then we get
0 = sin x + cos 2x = sin x + 1 − 2 sin2 x
24
1 1 1 1 1 1 1 1 9
0 = sin2 x − sin x − = sin2 x − sin x + − − = sin2 x − sin x + −
2 2 | 2{z 16} 16 2 | 2{z 16} 16
2
1 9

0 = sin x − −
4 16
s
1 9 3
sin x − = ± =±
4 16 4
1 3 1
sin x = ± = 1 or −
4 4 2
From this we get the same solutions for x as above.
Question 20. Solve the system of equations:


π
x + y
 = ;
2
cos2 y − cos2 x = 1.

From the second equation we can immediately deduce that y must be a multiple of π, and x must
π
be an odd multiple of . This is so because since sin and cos only take values between −1 and 1,
2
we need to have cos y = 1 and cos2 x = 0. Then from the first equation we get the general solution
2
π
x = (2k + 1) and y = −kπ, for an integer k (positive, zero, or negative).
2
Temat 3: Induction, binomial theorem, inequalities.

Question 1. Show that for any n 5 we have
2n > n2 .
Answer: Here we do not start with the assumption that n ∈ N. Nevertheless, we will begin by
proving the statement for natural numbers. For n = 5, we have 25 = 32 > 25 = 52 . Now we need to
show that if the statement is true for a given n ∈ N, then it is also true for n + 1. This means that
we have to show that 2n+1 > (n + 1)2 = n2 + 2n + 1, under the assumption that 2n > n2 . Using the
induction hypothesis, we get
2n+1 = 2 · 2n > 2 · n2 = n2 + n2 .
Now we only need to show that n2 2n + 1 if n 5. We use a second induction here. Clearly,
52 10 + 1. So we need to show that (n + 1)2 2(n + 1) + 1 = 2n + 3, under the assumption that
n2 2n + 1 and n 5. By the second induction hypothesis,
(n + 1)2 = n2 + 2n + 1 2n + 1 + 2n + 1 = 4n + 2.
1
2n + 3 and 4n + 2 intersect at n = , and the slope of 4n + 2 is bigger than the slope of 2n + 3. So
2
for n 5, we have 4n + 2 > 2n + 3.
We have shown that for any integer n 5, we have 2n > n2 . Now we would like to generalize this to
any r ∈ R such that r 5. It is possible to do this using the result that we obtained by induction
25
for integers, but this would be unnecessarily complicated. In any case, one needs to use logarithms.
r 2
We have 2r = eln 2 = er ln 2 and r2 = eln r = e2 ln r . So the inequality that we would like to
prove becomes er ln 2 > e2 ln r , and taking logarithms on both sides, it becomes r ln 2 > 2 ln r. Since
2 2
the derivative of 2 ln r is , the slope of r ln 2 is bigger than the slope of 2 ln r for r > ≈ 2.89.
r ln 2
For r = 5, we have 5 ln 2 ≈ 3.47 > 2 ln 5 ≈ 3.22. So for any r 5, the inequality holds.
Question 2. Show that for any n ∈ N \ {1} we have 7|(n7 − n).
Answer: 7|(27 − 2), since 27 − 2 = 128 − 2 = 126 = 18 · 7. So the statement holds for n = 2. Now we
need to show that 7|(n + 1)7 − (n + 1) under the assumption that 7|(n7 − n). We have
! ! ! !
7 7 7 0 7 6 1 7 5 2 7 4 3
(n + 1) − (n + 1) = n1 + n1 + n1 + n1+
0 1 2 3
! ! ! !
7 3 4 7 2 5 7 1 6 7 0 7
+ n1 + n1 + n1 + n 1 − (n − 1)
4 5 6 7
7! 7! 7! 7!
= n7 10 + n6 11 + n5 12 + n4 13 +
0!(7 − 0)! 1!(7 − 1)! 2!(7 − 2)! 3!(7 − 3)!
7! 7! 7! 7!
+ n3 14 + n2 15 + n1 16 + n0 17 − (n − 1)
4!(7 − 4)! 5!(7 − 5)! 6!(7 − 6)! 7!(7 − 7)!
7! 7 0 7! 6 1 7! 5 2 7! 4 3
= n1 + n1 + n1 + n1+
1 · 7! 1 · 6! 2!5! 3!4!
7! 3 4 7! 2 5 7! 1 6 7! 0 7
+ n1 + n1 + n1 + n 1 − (n − 1)
4!3! 5!2! 6! · 1 7! · 1
7·6 5 2 7·6·5 4 3
= n7 10 + 7n6 11 + n1 + n1+
2·1 3·2·1
7·6·5·4 3 4 7·6·5·4·3 2 5 7·6·5·4·3·2 1 6
+ n1 + n1 + n 1 + n0 17 − (n − 1)
4·3·2·1 5·4·3·2·1 6·5·4·3·2·1
= n7 10 + 7n6 11 + 7 · 3 · n5 12 + 7 · 5 · n4 13 +
+ 7 · 5 · n3 14 + 7 · 3 · n2 15 + 7n1 16 + n0 17 − (n − 1)
= n7 + 7n6 + 7 · 3 · n5 + 7 · 5 · n4 + 7 · 5 · n3 + 7 · 3 · n2 + 7n + 1 − (n − 1)
= n7 − n + 7(n6 + 3n5 + 5n4 + 5n3 + 3n2 + n)
Since by the induction hypothesis, 7|(n7 − n), we conclude that

h i
7| n7 − n + 7(n6 + 3n5 + 5n4 + 5n3 + 3n2 + n) .
3. Show that for any n ∈ N \ {1} we have 42|(n7 − n).

7
Answer:
h We check thati for n = 2, we have 2 − n = 126 = 3 · 42. So we need to show that
42| (n + 1)7 − (n + 1) , under the hypothesis that 42|(n7 − n). From the proof of the previous
26
question, we already know that
(n + 1)7 − (n + 1) = n7 − n + 7(n6 + 3n5 + 5n4 + 5n3 + 3n2 + n).
Since we assume that 42|(n7 − n), and since 42 = 7 · 6 and 7 and 6 do not have a common factor, it
suffices to show that 6|(n6 + 3n5 + 5n4 + 5n3 + 3n2 + n). We will use a second induction. Clearly, for
n = 1, we have 18 = n6 + 3n5 + 5n4 + 5n3 + 3n2 + n and 6|18. So we need to show that 6 divides
(n + 1)6 + 3(n + 1)5 + 5(n + 1)4 + 5(n + 1)3 + 3(n + 1)2 + n + 1, under the hypothesis that 6 divides
n6 + 3n5 + 5n4 + 5n3 + 3n2 + n. To save space, we will use Pascal’s triangle to get the binomial
coefficients.
1
1 1
1 2 1
1 3 3 1
1 4 6 4 1
1 5 10 10 5 1
1 6 15 20 15 6 1
(n + 1)6 + 3(n + 1)5 + 5(n + 1)4 + 5(n + 1)3 + 3(n + 1)2 + n + 1 =

= n6 + 6n5 + 15n4 + 20n3 + 15n2 + 6n + 1 + 3(n5 + 5n4 + 10n3 + 10n2 + 5n + 1)+
+ 5(n4 + 4n3 + 6n2 + 4n + 1) + 5(n3 + 3n2 + 3n + 1) + 3(n2 + 2n + 1) + n + 1
= n6 + 6n5 + 15n4 + 20n3 + 15n2 + 6n + 1 + 3n5 + 15n4 + 30n3 + 30n2 + 15n + 3+
+ 5n4 + 20n3 + 30n2 + 20n + 5 + 5n3 + 15n2 + 15n + 5 + 3n2 + 6n + 3 + n + 1
h i
= n6 + 3n5 + 5n4 + 5n3 + 3n2 + n + 6n5 + 15n4 + 20n3 + 15n2 + 6n + 1 + 15n4 + 30n3 +
+ 30n2 + 15n + 3 + 20n3 + 30n2 + 20n + 5 + 15n2 + 15n + 5 + 6n + 3 + 1

h i
= n6 + 3n5 + 5n4 + 5n3 + 3n2 + n + 6n5 + 30n4 + 90n2 + 18 + 66n3 + 60n + 4n3 + 2n
h i
= n6 + 3n5 + 5n4 + 5n3 + 3n2 + n + 6[n5 + 5n4 + 15n2 + 3 + 11n3 + 10n] + 4n3 + 2n
Now all we need to show is that 6 divides 4n3 + 2n. This is clear, by noticing that 4n3 + 2n is an
even number and considering 3 cases:
Case 1: n is divisible by 3. This means that there exists an integer a such that n = 3a. Then 4n3 + 2n =
4(3a)3 + 2(3a) = 3(4 · 32 a3 + 2a). So 4n3 + 2n is divisible by 3.
Case 2: The remainder after dividing n by 3 is 1. This means that there exists an integer a such that
n = 3a + 1. Then 4n3 + 2n = 4(3a + 1)3 + 2(3a + 1) = 4(27a3 + 9a2 + 3a + 1) + 6a + 2 =
3 · [4(9a3 + 3a2 + a) + 2a + 2]. So 4n3 + 2n is divisible by 3.
Case 3: The remainder after dividing n by 3 is 2. This means that there exists an integer a such that
n = 3a + 2. Then 4n3 + 2n = 4(3a + 2)3 + 2(3a + 2) = 4(27a3 + 18a2 + 12a + 8) + 6a + 4 =
3 · [4(9a3 + 6a2 + 4a) + 2a + 12]. So 4n3 + 2n is divisible by 3.
Alternatively, one may prove that 4n3 + 2n is divisible by 6 by induction. For n = 1, we have
4n3 + 2n = 6, so the statement holds in this case. Now we need to show that the statement is true
for n + 1, under the hypothesis that it is true for n. We have 4(n + 1)3 + 2(n + 1) = 4(n3 + 3n2 +
27
3n + 1) + 2n + 2 = (4n3 + 2n) + 6(2n2 + 2n + 1). Since we assume that 4n3 + 2n is divisible by 6, it
follows that (4n3 + 2n) + 6(2n2 + 2n + 1) is divisible by 6.
Question 4. This is Question 8. 2. in Temat 1.
Question 5. Show that for any n ∈ N we have

n(n + 1)(2n + 1)
12 + 22 + 32 + · · · + n2 = .
6
1(1 + 1)(1 + 2)
Answer: Clearly the statement is true for n = 1, since 1 = . Now we need to use the
6
induction hypothesis to show that the statement is true for n + 1. We have
n(n + 1)(2n + 1)
12 + 22 + 32 + · · · + n2 + (n + 1)2 = + (n + 1)2
6
n(n + 1)(2n + 1) + 6(n + 1)2
=
6
(n + 1)[n(2n + 1) + 6(n + 1)]
=
6
(n + 1)(2n2 + 7n + 6)
=
6
(n + 1)(n + 2)(2n + 3)
=
6
(n + 1)(n + 1 + 1)(2(n + 1) + 1)
= .
6
Question 6. Find a general formula for the sum:
3 · 7 + 7 · 11 + 11 · 15 + · · · + (4n − 1)(4n + 3).
Answer: This sum can be witten as

n
X
(4i − 1)(4i + 3) = 3 · 7 + 7 · 11 + 11 · 15 + · · · + (4n − 1)(4n + 3).
i=1
n
X n(n + 1)
Using the results from Exercises 4 and 5, namely that i= and
i=1 2
n n
n(n + 1)(2n + 1)
i2 =
X X
, as well as the trivial fact that 1 = n, we get
i=1 6 i=1
n n
(16i2 − i − 3)
X X
(4i − 1)(4i + 3) =
i=1 i=1
n n n
16i2 −
X X X
= i− 3
i=1 i=1 i=1
28
n n n
i2 −
X X X
= 16 i−3 1
i=1 i=1 i=1
n(n + 1)(2n + 1) n(n + 1)

= 16 − − 3n
6 2

1 1 1 1 n
+ + + ··· + = .
3 · 7 7 · 11 11 · 15 (4n − 1)(4n + 3) 3(4n + 3)
1 1
Answer: Clearly the statement is true for n = 1, since = . Now we need to use the
3·7 3(4 · 1 + 3)
induction hypothesis to show that the statement is true for n + 1. We have
1 1 1 1 1
+ + + ··· + + =
3 · 7 7 · 11 11 · 15 (4n − 1)(4n + 3) (4(n + 1) − 1)(4(n + 1) + 3)
n 1
= +
3(4n + 3) (4(n + 1) − 1)(4(n + 1) + 3)
n 1
= +
3(4n + 3) (4n + 3)(4n + 7)
n(4n + 7) + 3 4n2 + 7n + 3 (n + 1)(4n + 3)
= = =
3(4n + 3)(4n + 7) 3(4n + 3)(4n + 7) 3(4n + 3)(4n + 7)
(n + 1) (n + 1)
= = .
3(4n + 7) 3(4(n + 1) + 3)
Question 8. Using the binomial theorem find a general formulas for:
(1 + 1)4 = · · ·
(2 + 1)4 = · · ·
(3 + 1)4 = · · ·
..
.
4
(n + 1) = · · ·
and next, using the results of Questions 4 and 5 find a general pattern for:
13 + 23 + 33 + · · · + n3 .
Answer: By the binomial theorem, we have

! ! ! ! !
4 4 4 0 4 3 1 4 2 2 4 1 3 4 0 4
(1 + 1) = 11 + 11 + 11 + 11 + 11
0 1 2 3 4
29
! ! ! ! !
44 4 0 4 3 1 4 2 2 4 1 3 4 0 4
(2 + 1) = 21 + 21 + 21 + 21 + 21
0 1 2 3 4
! ! ! ! !
4 4 4 0 4 3 1 4 2 2 4 1 3 4 0 4
(3 + 1) = 31 + 31 + 31 + 31 + 31
0 1 2 3 4
..
.
! ! ! ! !
4 4 0 4 3 1 4 2 2 4 1 3 4 0 4
(n + 1)4 = n1 + n1 + n1 + n1 + n1
0 1 2 3 4
!
4 4! 4!
We will use = = = 1, but the remaining coefficients we first leave them as they
0 0!(4 − 0)! 1 · 4!
are, since this will lead to the answer of Question 9. We put the first summand on the right to the
other side of the equation, and the barred terms will cancel when we add all the equations.
! ! ! !
4 3 1 4 2 2 4 1 3 4 0 4
24 − 14 10 = 11 + 11 + 11 + 11
1 2 3 4
! ! ! !
4 3 1 4 2 2 4 1 3 4 0 4
34 −
24
10 = 21 + 21 + 21 + 21
1 2 3 4
! ! ! !
4 4 3 1
40 4 2 2 4 1 3 4 0 4
4 −
31 =
31 + 31 + 31 + 31
1 2 3 4
..
.
! ! ! !
4 4 3 1 4 2 2
40 4 1 3 4 0 4
(n + 1) −
n1 = n1 + n1 + n1 + n1

1 2 3 4
Now we add all the equations
! !
4 4 4
(n + 1) − 1 = (13 + 23 + 33 + · · · + n3 ) + (12 + 22 + 32 + · · · + n2 )+
1 2
! !
4 4
+ (1 + 2 + 3 + · · · + n) + (10 + 20 + 30 + · · · + n0 )
3 4
Using the results from Questions 4 and 5, and the trivial fact that
1| 0 + 20 + 3{z
0
+ · · · + n0} = |1 + 1 + 1{z+ · · · + 1} = n,
n times n times
we get
! ! ! !
4 4 4 n(n + 1)(2n + 1) 4 n(n + 1) 4
(n + 1) − 1 = (13 + 23 + 33 + · · · + n3 ) + + + n
1 2 6 3 2 4
We isolate the sum which we are interested in and divide by it’s coefficient to get

4 n(n+1)(2n+1) 4 n(n+1) 4
3 3 3 3
(n + 1)4 − 1 − 2 6
− 3 2
− 4
n
1 + 2 + 3 + ··· + n =
4
(1)
1
Only now we convert the choose functions to numbers, to get
3 3 3 (n + 1)4 − 1 − 6 n(n+1)(2n+1)
3 6
− 4 n(n+1)
2
−1·n
1 + 2 + 3 + ··· + n =
4
30
(n + 1)4 − 1 − n(n + 1)(2n + 1) − 2n(n + 1) − n
=
4
Question 9. Assuming that we already know the general paterns for the following sums:
def
S1 (n) = 11 + 21 + 31 + · · · + n1
def
S2 (n) = 12 + 22 + 32 + · · · + n2
def
S3 (n) = 13 + 23 + 33 + · · · + n3
..
.
def
Sk (n) = 1k + 2k + 3k + · · · + nk
find a general formula for the sum:

def
Sk+1 (n) = 1k+1 + 2k+1 + 3k+1 + · · · + nk+1 .
Answer: This is a straightforward generalization of what we did in Question 8, where we used

n(n + 1) n(n + 1)(2n + 1)
S0 = 1| + 1 + 1{z+ · · · + 1} = n, S1 = , and S2 = to get a formula for S3 .
2 6
n times
So all we need to do is adapt equation (1). We do this by realizing that in the choose functions of
equation (1), the 4 corresponds to k + 2 in the present situation. This is so because in Question 8,
we used the known formulas up until k = 2. So we get

k+2 k+2 k+2 k+2 k+2
(n + 1)k+2 − 1 − 2
Sk (n) − 3
Sk−1 (n) − · · · − k
S2 (n) − k+1
S1 (n) − k+2
n
Sk+1 (n) =
k+2
1

k+2 k+2 k+2 k+2
(n + 1)k+2 − 1 − 2
Sk (n) − 3
Sk−1 (n) − · · · − k
S2 (n) − k+1
S1 (n) − n
=
k+2

n
!
n
= 2n .
X
k=0 k
Answer: If we realize that 2 = 1 + 1, this is a simple application of the binomial theorem
2n = (1 + 1)n =
! ! ! ! ! !
n n 0 n n−1 1 n n−2 2 n n n 0 n
= 1 1 + 1 1 + 1 1 + ··· + 12 1n−2 + 11 1n−1 + 11
0 1 2 n−2 n−1 n
n
! ! ! ! ! ! !
n n n n n n X n
= + + + ··· + + + =
0 1 2 n−2 n−1 n k=0 k
31
n
!
X
k n
(−1) = 0.
k=0 k
Answer: If we realize that 0 = 1 − 1, this is a simple application of the binomial theorem
0 = 0n = (1 − 1)n =
! ! ! ! !
n n n n−1 n n−2 n n 0
= 1 (−1)0 + 1 (−1)1 + 1 (−1)2 + · · · + 11 (−1)n−1 + 1 (−1)n
0 1 2 n−1 n
n
! ! ! ! ! !
n n n n n n
− · · · + (−1)n−1 + (−1)n (−1)k
X
= − + =
0 1 2 n−1 n k=0 k
Question 12. For any n ∈ N calculate:

! ! ! ! !
n n n n n
−1 +2 −3 + · · · + (−1)n−1 (n − 1) + (−1)n (n) .
1 2 3 n−1 n
Answer: Since for an integer i with 1 ¬ i ¬ n we have

!
n n! n(n − 1) · · · 1
i =i =i
i i!(n − i)! [i(i − 1) · · · 1][(n − i)(n − i − 1) · · · 1]
(n − 1) · · · 1
=n
[(i − 1) · · · 1][(n − i)(n − i − 1) · · · 1]
!
(n − 1)! n−1
=n =n ,
(i − 1)!(n − 1 − (i − 1))! i−1
we can rewrite the sum as

! ! ! ! !
n n n n n
S = −1 +2 −3 + · · · + (−1)n−1 (n − 1) + (−1)n (n)
1 2 3 n−1 n
! ! ! ! !
n−1 n−1 n−1 n−1 n−1
= −n +n −n + · · · + (−1)n−1 n + (−1)n (n)
0 1 2 n−2 n−1
" ! ! ! ! !#
n−1 n−1 n−1 n−1 n−1
=n − + − + · · · + (−1)n−1 + (−1)n ,
0 1 2 n−2 n−1
and by Question 11, we get S = n(−0) = 0. There is however a special case in which we cannot use
the result of Question 11, namely when n = 1. Of course for n = 1, we get S = −1.
Question 13. Using the binomial theorem show that for any n ∈ N the Bernoulli inequality with n
as a power and x 0, is true.
32
Answer: The Bernoulli inequality states that for any real number x −1 and any integer n 0,
we have (1 + x)n 1 + nx. In this question we have the special case of a real number x 0 and an
integer n 1. By the binomial theorem we have
(1 + x)n =
! ! ! ! ! !
n n 0 n n−1 1 n n−2 2 n n n 0 n
= 1 x + 1 x + 1 x + ··· + 12 xn−2 + 11 xn−1 + 1x
0 1 2 n−2 n−1 n
! ! ! !
n n−2 2 n n n 0 n
= 1 + nx + 1 x + ··· + 12 xn−2 + 11 xn−1 + 1x
2 n−2 n−1 n
Since x 0, the terms inside the bracket sum to a number greater than or equal to zero. Hence
(1 + x)n 1 + nx.
Question 14. Using the binomial theorem show that for any n ∈ N the Bernoulli inequality with n
as a power and x > −1, is true.
Remark: First we will answer this question, and then we will give an alternative proof without using
the binomial theorem which is much simpler.
Answer: Before doing this question, it is essential to have understood the answer to Question 13.
Now we will show that the terms inside the bracket will sum to a number greater than or equal to
zero also when x > −1 (and not necessarily x 0 as in Question 13). We will do this by induction
!
2 0 2
on n. For n = 1, the Bernoulli inequality is trivially true, and for n = 2 we have 1 x = x2 0.
2
Now we need to show that
0 ¬ Sn+1 =
! ! ! !
n + 1 (n+1)−2 2 n + 1 (n+1)−3 3 n+1 n + 1 0 n+1
= 1 x + 1 x + ··· + 11 x(n+1)−1 + 1x
2 3 (n + 1) − 1 n+1
under the hypothesis that

! ! ! !
n n−2 2 n n−3 3 n n 0 n
0 ¬ Sn = 1 x + 1 x + ··· + 11 xn−1 + 1x .
2 3 n−1 n
! ! !
n n n+1
For all integers i with 0 ¬ i ¬ n − 1, we have + = . In Question 13 we already
i i+1 i+1
proved the statement for x 0, so here we can assume that −1 < x < 0. Therefore we obtain
0 ¬ (1 + x)Sn = Sn + xSn =
! ! ! !
n n−2 2 n n−3 3 n n 0 n
= 1 x + 1 x + ··· + 11 xn−1 + 1x +
2 3 n−1 n
! ! ! !
n n−2 3 n n−3 4 n n 0 n+1
+ 1 x + 1 x + ··· + 11 xn + 1x
2 3 n−1 n
! ! ! ! !
n n−2 2 n + 1 (n+1)−3 3 n + 1 (n+1)−4 4 n+1 1 n n + 1 0 n+1
= 1 x + 1 x + 1 x + ··· + 1x + 1x
2 3 4 n n+1
33
" ! ! #
n n−2 2 n + 1 (n+1)−2 2
= 1 x − 1 x + Sn+1
2 2
" ! !# " ! !#
n n+1 2 2 n n+1
So we have 0 ¬ − x + Sn+1 . Since x > 0 and − < 0, it follows
2 2 2 2
that Sn+1 > 0.
Now we will give an alternative proof without using the binomial theorem. We want to show that
for any integer n 1 and for any real number x > −1 we have (1 + x)n 1 + nx. For n = 1, the
statement is clearly true. Now we need to show that (1 + x)n+1 1 + (n + 1)x, under the hypothesis
that (1 + x)n 1 + nx. Since x > −1, we have the positive number 1 + x, and we can multiply both
sides of the inequality by this number
(1 + x)n+1 =
= (1 + x)n (1 + x) (1 + nx)(1 + x) = 1 + nx + x + nx2 = 1 + (n + 1)x + nx2 1 + (n + 1)x.
Question 15. Show that for any nonnegative real numbers a and b we have
a+b √
ab .
2
Answer: We have (a−b)2 = a2 −2ab+b2 0. Hence a2 +b2 2ab. Then (a+b)2 = a2 +2ab+b2 4ab
and by taking square roots and dividing by 2 we get
a+b √
ab .
2
Question 16. Prove the inequality between an „arithmetic mean” and a „geometric mean” for any
number n of nonnegative real numbers .
Answer: The first thing we need to do is to state the inequality which we want to prove:
x1 + x2 + · · · + xn √
n x1 x2 · · · xn .
n
We will prove an equivalent statement, namely
n
x1 + x2 + · · · + xn

x1 x2 · · · xn .
n
For n = 1 the statement is trivially true. So we need to show that
n+1
x1 + x2 + · · · + xn + xn+1

x1 x2 · · · xn xn+1 ,
n+1
under the hypothesis that n
x1 + x2 + · · · + xn

x1 x2 · · · xn .
n
If x1 = x2 = · · · = xn = xn+1 , we clearly have
n+1 !n+1
x1 + x2 + · · · + xn + xn+1 (n + 1)x1

= = xn+1
1 = x1 x2 · · · xn xn+1 ,
n+1 n+1
34
so we may assume that amongst the x’s, at least one is bigger than the arithmetic mean and at least
one is smaller than the arithmetic mean. Without loss of generality, we can assume that these are
xn and xn+1 respectively. Then
x1 + x2 + · · · + xn + xn+1
y = xn + xn+1 −
n+1
is a positive number and by the induction hypothesis we have
n
x1 + x2 + · · · + xn−1 + y

x1 x2 · · · xn−1 y.
n
Since xn is bigger than the arithmetic mean and xn+1 is smaller than the arithmetic mean, we have
x1 + x2 + · · · + xn + xn+1 x1 + x2 + · · · + xn + xn+1

xn − − xn+1 > 0.
n+1 n+1
It follows that
x1 + x2 + · · · + xn + xn+1
y· − xn xn+1 > 0,
n+1
and hence
x1 + x2 + · · · + xn + xn+1
y· > xn xn+1 .
n+1
Thus
x1 + x2 + · · · + xn + xn+1 n+1

n+1
x1 + x2 + · · · + xn + xn+1 n x1 + x2 + · · · + xn + xn+1

=
n+1 n+1
 n
x1 +x2 +···+xn +xn+1
x1 + x2 + · · · + xn + xn+1 − n+1

x1 + x2 + · · · + xn + xn+1

= 
n n+1
x1 + x2 + · · · + xn−1 + y n x1 + x2 + · · · + xn + xn+1

=
n n+1
x1 + x2 + · · · + xn + xn+1

x1 x2 · · · xn−1 y > x1 x2 · · · xn−1 xn xn+1 .
n+1
Question 17. Show that for any x > 0 and y > 0 the inequality
(x + y)n ¬ 2n−1 (xn + y n )
holds true for any n ∈ N.

Answer: We will do this by induction. Clearly, the statement holds for n = 1. Now we need to show
that (x + y)n+1 ¬ 2n (xn+1 + y n+1 ) under the hypothesis that (x + y)n ¬ 2n−1 (xn + y n ). We can
multiply both sides of the inequality by the positive number (x + y) and we get
(x + y)n+1 = (x + y)n (x + y) ¬ 2n−1 (xn + y n )(x + y) = 2n−1 (xn+1 + y n+1 + xy n + yxn ).
Now all we need to show is that xy n + yxn ¬ xn+1 + y n+1 . If x = y, then clearly this is true. So we
assume x 6= y, without loss of generality we can assume x > y. Then y n < xn , and we can multiply
35
both sides of this inequality by the positive number x − y. We get (x − y)y n < (x − y)xn , hence
xy n − y n+1 < xn+1 − yxn , hence xy n + yxn < xn+1 + y n+1 . So
(x + y)n+1 ¬ 2n−1 (xn+1 + y n+1 + xy n + yxn ) < 2n−1 (xn+1 + y n+1 + xn+1 + y n+1 ) =
= 2 · 2n−1 (xn+1 + y n+1 ) = 2n (xn+1 + y n+1 ).

1 1 1 1
+ + + ··· + < 1.
1·2 2·3 3·4 n · (n + 1)
Answer: If we realize that

1 1 1 1 1 1 1 1 1 1 1 1
= − , = − , = − , and so on until = − ,
1·2 1 2 2·3 2 3 3·4 3 4 n · (n + 1) n n+1
then it is clear that

1 1 1 1
+ + + ··· + =
1·2 2·3 3·4 n · (n + 1)
! ! ! !
1 1 1 1 1 1 1 1 1 1 1
= − + − + − + ··· + − + − =1− .
1 2 2 3 3 4 n−1 n n
n+1 n+1
1
For any n ∈ N, we have S = 1 − < 1.
n+1

1 1 1 1
2
+ 2 + 2 + · · · + 2 < 2.
1 2 3 n
Answer: Since
1 1 1 1 1 1 1 1
< , < , < , and so on until < ,
22 1·2 32 2·3 42 3·4 n 2 (n − 1) · n
using the result of Question 18 with n replaced by n − 1 we get

1 1 1 1 1 1 1
2
+ 2 + 2 + ··· + 2 = 1 + 2 + 2 + ··· + 2
1 2 3 n 2 3 n
1 1 1 1
<1+ + + + ··· +
1·2 2·3 3·4 (n − 1) · n
1

=1+ 1− < 1 + 1 = 2.
n−1+1
There is of course the special case of n = 1 in which we cannot use the result of Question 18 with
n replaced by n − 1, but in this case the sum is 1 which is smaller than 2.
36
1 1 1 1 √
√ + √ + √ + · · · + √ n.
1 2 3 n
1 √
Answer: We will do this by induction. For n = 1, we have √ 1. Now we need to show that
1
1 1 1 1 √ 1 1 1 1 √
√ + √ + √ +· · ·+ √ n + 1 under the assumption that √ + √ + √ +· · ·+ √ n.
1 2 3 n+1 1 2 3 n
We get
1 1 1 1 1 1 1 1 1
√ + √ + √ + ··· + √ = √ + √ + √ + ··· + √ + √
1 2 3 n+1 1 2 3 n n+1
√ 1
n+ √
n+1
√ √ √
√ n+1 1 n n+1+1
= n√ +√ = √
n+1 n+1 n+1
q √
n(n + 1) + 1 n2 + n + 1
= √ = √
n+1 n+1
√ √ √
n2 + 1 n+1 n+1 n+1 √
> √ =√ = √ = n + 1.
n+1 n+1 n+1

1 3 5 2n − 1 1
· · · ··· · <√ .
2 4 6 2n 2n
1
Answer: Induction will not work here, in fact √ is an overestimate. So this requires a trick.
2n
Compare the product
1 3 5 2n − 3 2n − 1
A = · · · ··· · ·
2 4 6 2n − 2 2n
with the product
2 4 6 2n − 2
B = · · · ··· · .
3 5 7 2n − 1
The first thing to notice is that A has n terms and B has only n − 1 terms. This means we need to
1 1
assume n 2, which is not a problem since clearly for n = 1, we have < √ . Comparing the first
2 2
n − 1 terms of A with the terms of B, we notice that the corresponding term in B is bigger, i.e.
2 1 4 3 6 5 2n − 2 2n − 3
> , > , > , and so on until > .
3 2 5 4 7 6 2n − 1 2n − 2
2n − 1
Furthermore, the last term of A is smaller than 1, since < 1. It follows that A2 is smaller
2n
37
than AB. When multiplying A and B, almost all the terms cancel:
1 3 5 2n − 3 2n − 1 2 4 6 2n − 2 1
AB = · · · ··· · · · · · · ··· · = .
2 4 6 2n − 2 2n 3 5 7 2n − 1 2n
1 1
So A2 < AB = , which implies A < √ .
2n 2n
Homework Question: Use mathematical induction to show that for any m, n ∈ N with n m
n
2i = 2n+1 − 2m .
X
i=m
Answer: Here the base case is n = m. In this case 2m = (2 − 1)2m = 2m+1 − 2m . Now we need to
n+1 n
2i = 2n+2 − 2m under the assumption that 2i = 2n+1 − 2m .
X X
show that
i=m i=m
n+1
2i = · · ·
X
i=m
= · · · = 2n+2 − 2m .
Temat 4: Weighted mean, bounds of number sets, inequ-

alities between means (mean values), compound interest,
present value, exponential function, logarithm.
Question 1. When two electric motors M1 and M2 are powered by an emergency power generator,
we know that if we run M2 only, the fuel for the generator will last for 10 hours. If we run both
motors, the fuel will last for 4 hours. How long will the emergency power last if we use M1 only.
Answer: We use the equation
F = C · T,
where F is the amount of fuel used, C is the consumption, and T is the time. We get
F = C2 · 10 F = (C1 + C2 ) · 4
So
F F 4F 4F 4 40 40
T = = F −4C2 = = F = 4 = = ,
C1 4
F − 4C2 F − 4 10 1 − 10 10 − 4 6
40
which means that if we use M1 only, the emergency power will last hours. In minutes, this is
6
40
· 60 = 400. Converting back into hours, this is 6 hours and 40 minutes.
6
Question 2. Assume we can regulate the power usage of the first motor M1 , but we can’t regulate
the power usage of the second motor. We know that if we run M2 only, then the fuel will last for 10
38
hours. If we run both of them with M1 at maximum usage, then the fuel lasts for 4 hours. By how
much should we reduce the power usage of M1 so that both can run together for 6 hours?
Answer: It is essential to have understood the previous question before doing this. From the previous
question, we have
F 6F F
C1 = 40 = C2 =
6
40 10
where C1 is the consumption of M1 at maximum usage and C2 is the consumption of M2 . Now we

multiply C1 by a factor r, such that both motors can run together for 6 hours:
F = (rC1 + C2 ) · 6
6F F

F = r + ·6
40 10
6 1

1= r + ·6
40 10
36 6
1=r +
40 10
40 = 36 · r + 24
40 − 24 16 4
r= = =
36 36 9
4 5
So we should reduce the power usage of M1 by 1 − = , which corresponds approximately to
9 9
55.56%.
Question 3.
a) Find an upper bound of the set of all numbers which have the form:
( )
a b c
∆= + + ,
b+c c+a a+b
where the numbers a, b, c satisfy the system of inequalities: a < b + c , b < c + a , c < a + b.
(Equivalent formulation: Among all triangles with side lengths a, b, c, find an upper bound of the
set ∆).
b) Find a lower bound of the set:
( )
a b c
∆= + + ,
b+c c+a a+b
where a, b, c are any positive numbers.

Answer: We will use the answer from (b) to do (a), so we start with (b). The first thing to notice
is that lower bound should be read as „greatest lower bound”. Indeed, since a, b, c are any positive
numbers, it is trivial to give the lower bound zero. To find the greatest lower bound, we notice that
a b c
if we multiply all of a, b, and c by the same positive number, the value of + +
b+c c+a a+b
does not change. So we may assume that c = 1. We define the function
39
a b 1 a b 1
f (a, b) = + + = + + .
b+1 1+a a+b 1+b 1+a a+b
To get a possible extremum of this function, the partial derivatives of f (a, b) with respect to a and
b must be zero. We get
∂ 1 b 1
f (a, b) = − 2
− = 0,
∂a 1 + b (1 + a) (a + b)2
∂ a 1 1
f (a, b) = − + − = 0.
∂b (1 + b)2 1 + a (a + b)2
Thus
1 b a 1
− = − +
b + 1 (1 + a)2 (1 + b)2 1 + a
(1 + a)2 (1 + b) − b(1 + b)2 = −a(1 + a)2 + (1 + a)(1 + b)2
(1 + 2ab + a2 )(1 + b) − b(1 + 2b + b2 ) = −a(1 + 2a + a2 ) + (1 + a)(1 + 2b + b2 )
1 + 2ab + a2 + b + 2ab2 + ba2 − b − 2b2 − b3 = −a − 2a2 − a3 + 1 + 2b + b2 + a + 2ab + ab2
Hence
0 = 1 + 2ab + a2 + b + 2ab2 + ba2 − b − 2b2 − b3 − (−a − 2a2 − a3 + 1 + 2b + b2 + a + 2ab + ab2 )

= a3 − b3 + ba2 − ab2 + 3a2 − 3b2
= (a − b)(a2 + b2 + 2ba + 3a + 3b).
Clearly, a = b is a solution, so we put this back into the partial derivative.

∂ 1 a 1
f (a, b) = − 2
− =0
∂a 1 + a (1 + a) (a + a)2
0 = (2a)2 (1 + a) − a(2a)2 − (1 + a)2
= 4a2 + 4a3 − 4a3 − 1 − 2a − a2
= 3a2 − 2a − 1
2 1
0 = a2 − a −
3 3
2 1 1 1 2 1 4
= a2 − a + − − = a2 − a+ −
| 3{z 9} 9 3 | 3{z 9} 9
2
1 4

= a− −
3 9
1 2
a− =±
3 3
1 2 1
a = ± = 1 or −
3 3 3
40
We are dealing with positive numbers, so the only solution here is a = 1. We look for a possible
second solution
0 = (a − b)(a2 + b2 + 2ba + 3a + 3b)

 
!2 !2
2b + 3 2b + 3
 
 2 2

= (a − b) 
a + (2b + 3)a + − +b + 3b
 2 2 

| {z }
complete square
 !2 !2 
2b + 3 2b + 3
= (a − b)  a + − + b2 + 3b
2 2
!2 !2
2b + 3 2b + 3
0= a+ − + b2 + 3b
2 2
!2
2b + 3 4b2 + 12b + 9
= a+ − + b2 + 3b
2 4
!2
2b + 3 9
= a+ −
2 4
s
2b + 3 9 3
a+ =± =±
2 4 2
3 3
a = −b − ± = −b or − b − 3
2 2
Since we are dealing with positive numbers, the only solution is b = a = 1. Since
1 1 1 3
f (1, 1) = + + = ,
1+1 1+1 1+1 2
and using other numbers for a or b increases the value of the function, we conclude that at b = a = 1,
the function f (a, b) has a local minimum. So the greatest lower bound for
( )
a b c
∆= + + ,
b+c c+a a+b
3
where a, b, c are any positive numbers, is .
2
a) The first thing to notice is that upper bound should be read as „least upper bound”. Indeed,
1 1 1
since a < b + c , b < c + a , c < a + b, it is trivial to give the upper bound 3 = + + . From
1 1 1
the answer to (b) above, we know that the function
a b 1
f (a, b) = + + ,
1+b 1+a a+b
where a, b are any positive numbers, only has one local extremum which is a minimum. As before, we
can scale the triangle and assume that c = 1. So when deforming the triangle with side lengths a, b, c,
a b c
we cannot possibly reach a local or absolute maximum. So the greatest value for + +
b+c c+a a+b
41
is obtained when we cannot deform the triangle any further, namely when we are approaching the
situation where two sides become equal and the third side becomes zero. In this situation, the limit
we are approaching is
1 0 1
+ + = 2.
0+1 1+1 1+0
So the least upper bound is 2.
Question 4. a) For two consecutive days the value of a certain stock was growing 10% a day, and
for the next two days was dropping 10% a day. What was the percentage change of the value of this
stock after the forth day?
b) For two consecutive days the value of a stock was dropping 10% a day, and for the next two days
was growing 10% a day. What was the percentage change of the value of this stock after the fourth
day?
c) On the first day the value of a stock grew by 10%, on the second day the value dropped 10%, on
the third day it again grew 10%, on the fourth day it again declined 10%. What was the percentage
change of the value of this stock after the forth day.
d) For ten consecutive days an index was growing 10% per day. What was the percentage change of
the value of this index after the tenth day.
e) For ten consecutive days an index was dropping 10% per day. What was the percentage change
of the value of this index after the tenth day.
Answer: a) Let V0 be the initial value of the stock, and let V4 be the value of the stock after the
10
fourth day. Growing by 10% means that we need to multiply by 1 + = 1.1, and dropping by
100
10
10% means that we need to multiply by 1 − = 0.9. Thus
100
1.99

V4 = V0 · (1.1) · (1.1) · (0.9) · (0.9) = 0.9801 · V0 = (1 − 0.0199) · V0 = 1 − · V0 .
100
So after the fourth day, the value of the stock has dropped by 1.99%.
b) Let V0 be the initial value of the stock, and let V4 be the value of the stock after the fourth day.
10
Growing by 10% means that we need to multiply by 1 + = 1.1, and dropping by 10% means
100
10
that we need to multiply by 1 − = 0.9. Thus
100
1.99

V4 = V0 · (0.9) · (0.9) · (1.1) · (1.1) = 0.9801 · V0 = (1 − 0.0199) · V0 = 1 − · V0 .
100
So after the fourth day, the the value of the stock has dropped by 1.99%.
c) Let V0 be the initial value of the stock, and let V4 be the value of the stock after the fourth day.
10
Growing by 10% means that we need to multiply by 1 + = 1.1, and dropping by 10% means
100
10
100
1.99

V4 = V0 · (1.1) · (0.9) · (1.1) · (0.9) = 0.9801 · V0 = (1 − 0.0199) · V0 = 1 − · V0 .
100
42
So after the fourth day, the the value of the stock has dropped by 1.99%.
d) Let V0 be the initial value of the index, and let V10 be the value of the index after the tenth day.
10
Growing by 10% means that we need to multiply by 1 + = 1.1. Thus
100
159.37

V10 = V0 · (1.1)10 = 2.5937 · V0 = (1 + 1.5937) · V0 = 1 + · V0 .
100
So after the tenth day, the value of the index has grown by 159.37%.
e) Let V0 be the initial value of the index, and let V10 be the value of the index after the tenth day.
10
Dropping by 10% means that we need to multiply by 1 − = 0.9. Thus
100
65.13

V10 = V0 · (0.9)10 = 0.3487 · V0 = (1 − 0.6513) · V0 = 1 − · V0 .
100
So after the tenth day, the value of the index has dropped by 65.13%.
Question 5. The economy of some state was developing with a stable rate of 2% growth in GDP per
year for the last 100 years. During the same period, the economy of a neighbour state was developing
with a stable rate of 2% growth in GDP per year for the first 50 years, but then had 11 years of
deep crises where GDP was dropping 4% per year, followed by 21 years of stagnation without change
in GDP, and then 18 years of dynamic development with every year GDP growth 8%. If at the
beginning both sates were at the same level of development, which one is currently more wealthy?
Answer: Let WS be the current wealth of the state and let WN be the current wealth of the neighbour
state. Let W be the wealth of both states individually 100 years ago. Then
100
2

WS = W · 1 + = W · (1 + 0.02)100 = W · (1.02)100 = 7.2446 · W
100
50 11 21 18
2 4 0 8

WN = W · 1 + 1− 1+ 1+
100 100 100 100
= W · (1 + 0.02)50 (1 − 0.04)11 (1 + 0)21 (1 + 0.08)18
= W · (1.02)50 (0.96)11 (1)21 (1.08)18 = W · (1.02)50 (0.96)11 (1.08)18 = 6.8647 · W
Since WS > WN , the state is currently more wealthy than the neighbour state.
Question 6. There are 7 billion inhabitants on some planet who consider themselves reasonable.
Assuming that their natural population growth per year is 1.2%, calculate the population size after:
a) 5 years, b) 10 years and c) 20 years.
How many years does it take for the population size to double? When will the population size exceed
20 billion.
1.2

Answer: Growing by 1.2% means we need to multiply by 1 + = (1 + 0.012) = 1.012.
100
a) Let P5 be the population size after 5 years.
P5 = 7000000000 · (1.012)5 = 7430201687.
43
b) Let P10 be the population size after 10 years.
P10 = 7000000000 · (1.012)10 = 7886842445.
c) Let P20 be the population size after 20 years.
P20 = 7000000000 · (1.012)20 = 8886040537.
To find out after how many years the population size will have doubled, we need to solve the equation
7000000000 · (1.012)x = 14000000000,
where the variable x stands for the number of years. We get
7000000000 · (1.012)x = 14000000000

(1.012)x = 2
x
eln 1.012 =2
ex ln 1.012 = 2
x ln 1.012 = ln 2
ln 2
x= = 58.11
ln 1.012
So after 58.11 years the population size will have doubled.
To find out after how many years the population size will exceed 20 billion, we need to solve the
equation
7000000000 · (1.012)y = 20000000000,
where the variable y stands for the number of years. We get
7000000000 · (1.012)y = 20000000000

20
(1.012)y =
7
y 20
eln 1.012 =
7
20
ey ln 1.012 =
7
20
y ln 1.012 = ln
7
ln 20
7
y= = 88.01
ln 1.012
So after 88.01 years the population size will exceed 20 billion.
Recall that the nominal rate of interest on deposits or credit is given by percentage per annum without
taking into account capitalization of interest (i.e compounding is not included). The effective rate of
44
interest (also given by percentage per annum) - takes into account compounding of interest over 1
year. If compounding is continuous, then the amount At with nominal interest rate r, after t years,
can be calculated from the equation At = A0 ert , where A0 is the initial deposit.
Question 7. a) A fund offers a nominal interest rate of 8% per annum compounded every quarter
(three months). What is the effective interest rate?
b) A fund uses monthly compounding of interest and advertises that the effective interest rate is
10%. If deposits are taken only for 3 months, how much percentage profit can a client earn in 3
months.
c) Do the above examples a) and b) with the additional assumption that each time interest is
compounded, the interest is taxed at a rate of 19%.
Answer: a) The nominal interest rate is 8% per year. To get the quarterly interest rate, we have to
divide this by 4. Since compounding occurs at the end of each quarter, if an amount A0 is deposited,
one year later it has become
4 4
1 8 2

A1 = A0 1+ · = A0 1+ = A0 (1 + 0.02)4 = A0 (1.02)4 = 1.0824 · A0 .
4 100 100
So the effective interest rate is
A1 8.24
− 1 = 1.0824 − 1 = 0.0824 = ,
A0 100
which means 8.24%.
b) It is essential to have understood part a). First we need to calculate the nominal interest rate.
We let x be a variable whose value is the nominal interest rate. Then
10

A1 = A0 1 + = A0 (1 + 0.10) = 1.1 · A0
100
12
x

= A0 1 +
12
12
x

1+ = 1.1
12
x 1
1+ = (1.1) 12
12
x 1
= (1.1) 12 − 1
12
1
9.569
x = 12 · (1.1) 12 − 1 = 0.09569 =
100
So the nominal interest rate is 9.569%. What we actually need to continue the calculations is
x
= 0.007974. After three months, the deposited amount has become
12
3
x

A 1 = A0 1 + = A0 (1 + 0.007974)3 = A0 (1.007974)3 = 1.0241 · A0
4 12
45
So the percentage profit earned after 3 months is
A1 2.41
4
− 1 = 1.0241 − 1 = 0.0241 = ,
A0 100
which means to 2.41%.
c) When the interest at the time of compounding is taxed at a rate of 19%, the interest rate needs
19

to be multiplied by 1 − (1 − 0.19) = 0.81 in the calculations for a) and b). For a), we get
100
4 4
1 8 2

A1 = A0 1+ · · 0.81 = A0 1+ · 0.81
4 100 100
= A0 (1 + 0.02 · 0.81)4 = A0 (1.0162)4 = 1.0664 · A0 .
So the effective interest rate is

A1 6.64
− 1 = 1.0664 − 1 = 0.0664 = ,
A0 100
which means 6.64%.
For b), when the interest at the time of compounding is taxed at a rate of 19%, we get that after
three months, the deposited amount has become
3
x

A 1 = A0 1 + · 0.81 = A0 (1 + 0.007974 · 0.81)3 = A0 (1.006459)3 = 1.0195 · A0
4 12
So the percentage profit earned after 3 months is
A1 1.95
4
− 1 = 1.0195 − 1 = 0.0195 = ,
A0 100
which means to 1.95%.
Question 8. a) What nominal interest rate with yearly compounding of interest is required so that
a deposit will double after 10 years.
b) What nominal interest rate with monthly compounding of interest is required so that a deposit
will double after 10 years.
c) What nominal interest rate with compounding of interest daily (assume every year is 365 days,
don’t take into account leap years), is required so that a deposit will double after 10 years.
Answer a) Let the nominal interest rate be x. With yearly compounding of interest, after 10 years,
an initially deposited amount of A0 will have become
A10 = (1 + x)10 A0 .
The amount A0 is required to have double after 10 years, so we get A10 = (1 + x)10 A0 = 2 · A0 . Hence
(1 + x)10 = 2
1
1 + x = 2 10
46
1 7.18
x = 2 10 − 1 = 1.0718 − 1 = 0.0718 = .
100
So the required nominal interest rate is 7.18%.
b) Let the nominal interest rate be x. Ten years corresponds to 120 months. With monthly compo-
unding of interest, after 10 years, an initially deposited amount of A0 will have become
120
x

A10 = 1+ A0 .
12
120
x

The amount A0 is required to have double after 10 years, so we get A10 = 1 + A0 = 2 · A0 .
12
Hence
120
x

1+ =2
12
x 1
1+ = 2 120
12
x 1
= 2 120 − 1 = 1.005793 − 1 = 0.005793
12
6.95
x = 12 · 0.005793 = 0.0695 = .
100
c) Let the nominal interest rate be x. Ten years corresponds to 3650 days. With daily compounding
of interest, after 10 years, an initially deposited amount of A0 will have become
3650
x

A10 = 1+ A0 .
365
365
x

The amount A0 is required to have double after 10 years, so we get A10 = 1+ A0 = 2 · A0 .
3650
Hence
3650
x

1+ =2
365
x 1
1+ = 2 3650
365
x 1
= 2 3650 − 1 = 1.00018992 − 1 = 0.00018992
365
6.93
x = 365 · 0.00018992 = 0.0695 = .
100
Question 9. a) A bank advertises “interest rates on deposits up to 12%”, hiding the fact that they
use the following progressive system: at the end of the first month 1% interest is paid, after the
second month 2% interest is paid, and so on until after the twelfth month 12% is paid. What is
effective interest rate?
b) Do a) taking into account Belka tax (problem 7c).
47
Answer a) We assume that the 1% interest rate for the first month is a yearly interest rate, and the
same goes for the interest rates of the subsequent months. So after the first month, an amount A0
has become
1 1 1

A 1 = A0 · 1 + · = A0 · 1 + 0.01 · = A0 · (1 + 0.00083333) = 1.00083333 · A0 ,
12 100 12 12
after the second month, we get
2 1 1

A2 =A1 · 1+ · = A 1 · 1 + 0.02 · = A 1 · (1 + 0.001666666) = 1.001666666 · A 1 ,
12 12 100 12 12 12 12 12
and so on, until after the the twelfth month we get

12 1 1

A1 = A 12 = A 11 · 1 + · = A 11 · 1 + 0.12 · = A 11 · (1 + 0.01) = 1.01 · A 11 .
12 12 100 12 12 12 12 12
So we have
1 1 2 1 11 1 12 1

A1 = A 12 = A0 · 1 + · 1+ · · ··· · 1 + · 1+ ·
12 100 12 100 12 100 12 100 12
1200 + 1 1200 + 2 1200 + 11 1200 + 12

= A0 · · · ··· · ·
1200 1200 1200 1200
1201 1202 1211 1212

= A0 · · · ··· · ·
1200 1200 1200 1200
1
= A0 · · (1201) · (1202) · · · · · (1211) · (1212)
120012
1 1212!
= A0 · ·
120012 1200!
!
1 1212 6.69

= A0 · · · (12!) = 1.0669 · A0 = (1 + 0.0669) · A 0 = 1 + · A0
120012 12 100
So
A1 6.69

= 1+ ,
A0 100
which means that the the effective interest rate is 6.69%.
b) If the interest is taxed at a rate of 19%, the interest rate for every individual month needs to be
multiplied by
19 100 19 81

1− = − =
100 100 100 100
to find out how much an amount A0 has become after 1 year. From the calulation in part a), since
1 81 81
· = , we get
12 100 1200
1 81 2 81 11 81 12 81

A1 = A 12 = A0 · 1 + · 1+ · · ··· · 1 + · 1+ ·
12 100 1200 100 1200 100 1200 100 1200
120000 + 1 · 81 120000 + 2 · 81 120000 + 11 · 81 120000 + 12 · 81

= A0 · · · ··· · ·
120000 120000 120000 120000
1
= A0 · (120000 + 1 · 81) · (120000 + 2 · 81) · · · · · (120000 + 11 · 81) · (120000 + 12 · 81)
12000012
48
1
= A0 · (120081) · (120162) · · · · · (120891) · (120972)
12000012
9.39672 · 1060
= · A0 = 1.0539 · A0
12000012
So
A1 5.39

= 1+ ,
A0 100
which means that the the effective interest rate is 5.39%.
Question 10. A bank is giving credit with nominal interest rate 8%. The repayments are to be made
in 12 equal monthly payments over one year. The interest charged is calculated from the value of
the debt in the given month and added to the debt at the end of each month before the monthly
payment is subtracted. If a client took a credit of 100000 zlotys, what will be the total interest paid?
First we need to know to which multiplicative factor the nominal interest rate of 8% corresponds to
on a monthly basis. We get
8 1 8
1+ · =1+
100 12 1200
Let x be the amount which is payed at the end of every month. At the beginning of the second
month, the debt is
8

D = 100000 1 + − x,
1200
at the beginning of the third month, the debt is
8 8

D = 100000 1 + −x 1+ − x,
1200 1200
at the beginning of the fourth month, the debt is
8 8 8

D= 100000 1 + −x 1+ −x 1+ − x,
1200 1200 1200
and so on until at the beginning of the thirteenth month, everything has been paid and the debt is
zero. So we get
12 11 2
8 8 8 8

0 = 100000 1 + −x 1+ − ··· − x 1 + −x 1+ −x
1200 1200 1200 1200
 
12 11 2
8 8 8 8
 
= 100000 1 + − x 1 + + ··· + 1 + + 1+ + 1 .
 
1200 
|
1200 {z
1200 1200 }

S
We let 11 10 2

8 8 8 8

S = 1+ + 1+ + ··· + 1 + + 1+ + 1,
1200 1200 1200 1200
then
8 8 8

1+ S−S = 1+ −1 S = S=
1200 1200 1200
12 11
8 2

8 8 8

= 1+ + 1+
+ ··· + 1 + + 1 + −
1200 1200
1200 1200

49
 
11 10 2

8 8 8 8

− 1+
+ 1+
+ ··· + 1 +
+ 1+ + 1
1200 1200 1200
1200
12
8

= 1+ − 1,
1200
because all other terms cancel, and we get
12 12
8 8
1+ 1200
−1 1+ 1200
−1
S= 8 = 1200 · .
1200
8
Thus
12
8

0 = 100000 1 + − Sx
1200
12
8

8
12 1+ 1200
−1
= 100000 1 + − 1200 · ·x
1200 8
12
8
100000 1 + 1200
x= 8 12 = 8698.8437
(1+ 1200 ) −1
1200 · 8
Since
12 · 8698.8437 − 100000 = 4386.12,
the total interest payed is 4386.12 zlotys.
Question 11. 10000 zlotych are deposited with nominal interest rate 8%, and the interest rates are
calculated in a continuous way. What will be the average annual income if the deposit is made for 5
years, and what if the deposit is made for 20 years?
Answer: If compounding is continuous, then the amount At with nominal interest rate r, after t
years, can be calculated from the equation At = A0 ert , where A0 is the initial deposit.
After 5 years, we get
8 40
A5 = 10000 e 100 ·5 = 10000 e 100 = 10000 e0.4 = 14918.25
14918.25 − 10000
Since = 983.64, the average annual income is 983.64 zlotych.
5
After 20 years, we get
8 160
A20 = 10000 e 100 ·20 = 10000 e 100 = 10000 e1.6 = 49530.32
49530.32 − 10000
Since = 1976.52, the average annual income is 983.64 zlotych.
20
Question 12. Suppose we wish to make a deposit so that after 2 years the deposit will be worth
2 million. What initial deposit would be required in the following cases: A nominal interest rate of
10%, compounded:
a) every year
b) every half a year
50
c) monthly
d) continuously
Answer: a) Let A0 be the initial deposit. With yearly compounding, we get
2
10

A2 = 2000000 = A0 · 1 + = A0 · (1.1)2
100
2000000
So A0 = = 1652892.56
(1.1)2
b) Let A0 be the initial deposit. With bi-annual compounding, we get
4
10 1

A2 = 2000000 = A0 · 1 + · = A0 · (1.05)4
100 2
2000000
So A0 = = 1645404.95
(1.05)2
c) Let A0 be the initial deposit. With monthly compounding, we get
24
10 1

A2 = 2000000 = A0 · 1 + · = A0 · (1.008333333)24
100 12
2000000
So A0 = = 1638819.10
(1.008333333)24
d) If compounding is continuous, then the amount At with nominal interest rate r, after t years,
can be calculated from the equation At = A0 ert , where A0 is the initial deposit. So we get
10
A2 = 2000000 = A0 e 100 ·2 = A0 e0.2
2000000
So A0 = = 1637461.51
e0.2
In each case of to the previous question, the solution is an example of what economists call current
value.
Question 13. For some investment, the current value of an amount V1 , which will be paid in t1 years,
is exactly the same as the current value of an amount V2 , which will be paid in t2 years. Show that
the current value of V1 paid in t1 + k years is the same as the current value of the amount V2 paid
in t2 + k years.
We consider two cases. First we consider compounding periods which are a fraction of a year, such
as in parts a), b), and c) of the previous exercise. For example, if compounding is monthly, we divide
the year by n = 12.
The current value V1 at time t1 is
t1 ·n
1

V1 (t1 ) = D1 · 1 + r · ,
n
where D1 is the initial deposit.
t2 ·n
1

V2 (t2 ) = D2 · 1 + r · ,
n
51
The current value V1 at time t1 + k is
(t1 +k)·n
1

V1 (t1 + k) = D1 · 1 + r · .
n
(t2 +k)·n
1

V2 (t2 + k) = D2 · 1 + r · .
n
We need to show that V1 (t1 + k) = V2 (t2 + k), given that V1 (t1 ) = V2 (t2 ). We get
(t1 +k)·n
1

V1 (t1 + k) = D1 · 1 + r ·
n
t1 ·n+k·n
1

= D1 · 1 + r ·
n
t1 ·n k·n
1 1

= D1 · 1 + r · · 1+r·
n n
k·n
1

= V1 (t1 ) · 1 + r ·
n
k·n
1

= V2 (t2 ) · 1 + r ·
n
t2 ·n k·n
1 1

= D2 · 1 + r · · 1+r·
n n
t2 ·n+k·n
1

= D2 · 1 + r ·
n
(t2 +k)·n
1

= D2 · 1 + r · = V2 (t2 + k).
n
Now we consider the case where compounding is continous, such as in part d) of the previous
exercise.
V1 (t1 ) = D1 · er·t1 ,
V2 (t2 ) = D2 · er·t2 ,
V1 (t1 + k) = D1 · er·(t1 +k) .
V2 (t2 + k) = D2 · er·(t2 +k) .
52
We need to show that V1 (t1 + k) = V2 (t2 + k), given that V1 (t1 ) = V2 (t2 ). We get
V1 (t1 + k) = D1 · er·(t1 +k)
= D1 · er·t1 +r·k
= D1 · er·t1 · er·k
= V1 · er·k
= V2 · er·k
= D2 · er·t2 · er·k
= D2 · er·t2 +r·k
= D2 · er·(t2 +k) = V2 (t2 + k).
Question 14. Market research shows that when advertising a given product ends, the decline of its
sale follows an equation of the form S(t) = S0 e−λt where λ is some coefficient characterized by the
given product and S0 the sales in the last month with advertising. A company stops advertising one
of its products with the plan to start advertising again when the sale of this product has dropped by
25%. If after the first week the sales dropped by 5%, at what time in the future should the company
expect to start advertising again?
Answer: First we need to determine the constant λ. After the first week, the sales dropped by 5%,
5

so S(1) = S0 1 − = S0 e−λ·1 . Thus
100
5

S0 1 − = S0 e−λ·1
100
100 5

− = e−λ
100 100
95
= e−λ
100
95

ln = −λ
100
−1 !
95 95 100

λ = − ln = ln = ln = 0.051293
100 100 95
Now we need to use the formula with λ = 0.051293 to determine when the sales will have dropped
by 25%. We get
25

S(t) = S0 1 − = S0 e−0.051293·t
100
100 25

− = e−0.051293·t
100 100
75
= e−0.051293·t
100
53
75

ln = −0.051293 · t
100
75

− ln = 0.051293 · t
100
−1 !
75

ln = 0.051293 · t
100
100

ln = 0.051293 · t
75

100
ln 75
t= = 5.61
0.051293
So after 5.61 weeks, the company needs to start advertising again.
Question 15. Don Giovanni is lending 100000 e for one year with nominal interest rate 100% per
year. Can Don Giovanni construct a compounding scheme with compounding periods of equal length
so that after one year his return is:
a) 2.5 mln e ; b) 2.7 mln e ; c) 3 mln e .
In the cases which have a solution, give an equation that Don Giovanni can use to quickly established
the schemes ?
Answer: No compounding scheme can give more increase than continuous compounding. If compo-
unding is continuous, then the amount At with nominal interest rate r, after t years, can be calculated
from the equation At = A0 ert , where A0 is the initial deposit. So for one year at a nominal interest
of 100%, we get
100
A1 = 1000000 e 100 ·1 = 1000000 e1 = 1000000 e = 2718281.82
Consequently, a) and b) have solutions, but c) does not. To find these solutions, we look for an integer
n, such that when dividing the year in n equal compounding periods, we get the desired result. If
n = 1, we get
100
!1 1
1

100
A1 = 1000000 · 1 + = 1000000 · 1 + = 1000000 · (1 + 1) = 1000000 · 2 = 2000000
1 1
This is less than 2500000, so we need to increase n. We try compounding periods of 6 months, so
n = 2. We get
100
!2 2 2
1 3

100
A1 = 1000000 · 1 + = 1000000 · 1 + = 1000000 · = 2250000
2 2 2
n = 3. We get
100
!3 3 3
1 4

100
A1 = 1000000 · 1 + = 1000000 · 1 + = 1000000 · = 2370370.37
3 3 3
n = 4. We get
100
!4 4 4
1 5

100
A1 = 1000000 · 1 + = 1000000 · 1 + = 1000000 · = 2441406.25
4 4 4
54
This is less than 2500000, so we need to increase n. One year cannot nicely be divided into 5 equal
parts. We try compounding periods of 2 months, so n = 6. We get
100
!6 6 6
1 7

100
A1 = 1000000 · 1 + = 1000000 · 1 + = 1000000 · = 2521626.37
6 6 6
So compounding periods of 2 months work for a).

For part b), we notice that 2.7 mln is quite close to 2718281.82, so what we need is near continuous
compounding. We try compounding perio ds of 1 day, so n = 365. We get
100
!365 365 365
1 366

100
A1 = 1000000 · 1 + = 1000000 · 1 + = 1000000 · = 2714567.47
365 365 365
This is bigger than 2.7 mln, so we try to see if we can decrease n. The integer 365 is not divisible by
365
2, 3, or 4. So we try compounding periods of 5 days, which means that n = = 73. We get
5
100
!73 73 73
1 74

100
A1 = 1000000 · 1 + = 1000000 · 1 + = 1000000 · = 2699894.23
73 73 73
This is close enough to 2.7 mln.
Question 16. The moment a baby is born, we open a savings account with P zlotys so as to build
a fund for the child’s education. The nominal interest rate is 6%. We will compound with a high
frequency, so practically the interest compounds continuously. Suppose we plan that at 18 years of
age the child will have 30000 zlotys in this account. What should the value of P be?
Answer: If compounding is continuous, then the amount At with nominal interest rate r, after t years,
can be calculated from the equation At = A0 ert , where A0 is the initial deposit. So we have
6 108
30000 = P · e 100 ·18 = P · e 100
30000 30000
P = 108 = = 10187.87
e 100 e1.08
So P should be 10187.87 zlotys.
Temat 5: Sequences I.
Question 1. Find the limit of the following sequences if they converge, otherwise prove that they
diverge:
π
a) an = sin ,
n
b) an = sin n,
1−n
c) an = √ ,
1−2 n
√
1 + 99 n
d) an = √ .
1 + 100 n
55
π π

Answer: a) Since n→∞
lim = 0 and sin 0 = 0, we have n→∞
lim sin = 0.
n n
b) Regardless on how big n is, sin n can have any value between −1 and 1. So the limit does not
exist.
c)    √ 
√1 − √n √1 − n
!
1−n n n n
lim √ = lim  √  = lim  1
 =∞
n→∞ 1−2 n n→∞ √ − 2 √n
1 n→∞ √ −2
n
n n
d)
1
 
√ !  1
 1
+ n 99
1+ n 99
1 + n 99 1
 n 100 n 100
1 
lim √ = lim  1
 = lim 
1

n→∞ 1 + 100 n n→∞ 1 + n 100 n→∞  1
+ n 100

1 1
n 100 n 100
 1 1
  1

1
1 + n 99 − 100 1
1 + n 9900
n 100 n 100
= lim  = lim  =∞
   
1  1 
n→∞ 1 +1 n→∞ 1 +1
n 100 n 100
Question 2. Find the limit of

n3 + 4n2 + n − 1
an = .
n4 + n3 + 2n2 − 2n − 1
Answer:
 
n3 +4n2 +n−1 1 4 1 1
! !
n3 + 4n2 + n − 1 n4 n
+ n2
+ n3
− n4
lim = lim n4 +n3 +2n
 2 −2n−1
 = lim 1 2 2 1 =0
n→∞ n4 + n3 + 2n2 − 2n − 1 n→∞
n4
n→∞ 1+ n
+ n2
− n3
− n4
Question 3. Give a lower bound for each of the following sequences:

a) an = n2 − 5n + 6,
b) an = |n2 − 5n + 6| ,
c) an = |n2 − 5n + 1|,
d) an = |n2 − 13| − 2000n,
√
e) an = n2 − 3n + 6 .
2
5 25 5

2
Answer: a) The function f (x) = x − 5x + 6 = x − −
+ 6 is a parabola with vertex x = .
2 4 2
5
The smallest value of the sequence an = n2 − 5n + 6 occurs at the integers n which are closest to ,
2
thus 2 and 3. We have a2 = a3 = 0, and 0 is the lower bound.
b) The expression inside the absolute value is the same as in part a), and for integer values of n it
is always non-negative. So as in part a), the lower bound is 0.
c) Here we also have a parabola, but the negative part becomes positive because of the absolute
value. Calculating an for small integers, we will find that soon the values will increase more an more
56
rapidly, and then we are past the lower bound:
a1 = 3 a2 = 5 a3 = 5 a4 = 3 a5 = 1 a6 = 7 a7 = 15
So the lower bound is 3.

d) The absolute value sign only makes a difference if n2 < 13, so for n equal to 1, 2, or 3. Given
that we subtract 2000n, it is clear that the lower bound has not been reached with these small values
of n. So we consider the parabola x2 − 13 − 2000x = (x − 1000)2 − 1000000 − 13. The vertex is at
x = 1000, so the lower bound is reached for n = 1000. Since a1000 = −1000013, the lower bound is
1000013.
√
e) This is the same expression as in part a), only with a square root. The lower bound is 0 = 0.
Question 4. Determine the smallest n0 such that for all n n0 , either an+1 > an or an+1 < an , if:
a) an = n2 − 8n + 7
1
b) an = −
3n + 17
5 n

c) an = 1 −
n
√
d) an = n n
Answer: a) The function x2 − 8n + 7 = (x − 4)2 − 16 + 7 is a parabola with vertex at x = 4. So
n0 = 4.
b) As n increases, the denominator gets bigger. So an is monotone increasing. It is decreasing in
absolute value, but because of the minus sign it is increasing. So n0 = 1.
c) We calculate the first 10 terms.
9 8 1 16 27 38 49 510
a1 = −4 a2 = a3 = − a4 = a5 = 0 a6 = 6 a7 = 7 a8 = 8 a9 = 9 a10 = 10
4 27 256 6 7 8 9 10
We claim that n0 = 5. This seems like a reasonable guess, but it needs a proof. We need to show
that for n 5, we have
(n − 4)n+1 (n − 5)n
> .
(n + 1)n+1 nn
For n = 5, this means
16
> 0.
66
For n = 6, it means
27 16 27 77
> which is equivalent to > .
77 66 16 66
For n = 7, it means
38 27 38 88
> which is equivalent to > .
88 77 27 77
And so on. This is obtained by applying five times the following fact: For positive integers x and y
such that x ¬ y, we have
xy+1 (x + 1)y+1
> .
(x − 1)y xy
57
This fact can be proved by induction, starting with y = 1. We need to show that
xy+1 xy > (x + 1)y+1 (x − 1)y .
For y = 1, we have
(x + 1)2 (x − 1) = (x2 + 2x + 1)(x − 1) = x3 + 2x2 + x − x2 − 2x − 1

= x3 + |x2 −{zx − 1} < x3 = x2 x for x ¬ 1
tail
Now we want to show that x2+1 x2 > (x + 1)2+1 (x − 1)2 . We use the result for y = 1 and multiply
by (x − 1)(x + 1) = x2 − 1.

(x + 1)2+1 (x − 1)2 = (x + 1)2 (x − 1)(x2 − 1) = x3 + x2 − x − 1 (x2 − 1)

= x5 + x4 − x3 − x2 − x3 + x2 − x − 1
= x5 + x
|
4
− 2x3 −{z2x2 + x + 1} < x5 = x3 x2 for x ¬ 2
tail
Now we want to show that x3+1 x3 > (x + 1)3+1 (x − 1)3 . We use the result for y = 2 and multiply by
(x − 1)(x + 1) = x2 − 1.

(x + 1)3+1 (x − 1)3 = (x + 1)2+1 (x − 1)2 (x2 − 1) = x5 + x4 − 2x3 − 2x2 + x + 1 (x2 − 1)

= x7 + x6 − 2x5 − 2x4 + x3 + x2 − x5 + x4 − 2x3 − 2x2 + x + 1
= x7 + x
|
6
− 3x5 − 3x4 +{z3x3 + 3x2 − x − 1} < x7 = x4 x3 for x ¬ 3
tail
Now there is an apparent pattern of how this is going to look like for any y. The tail is always a
polynomial of degree 2y, which is negative valued for x ¬ y. Going to y + 1, one needs to use this
together with the fact that one has multiplied by x2 − 1 to settle the cases for x ¬ y. Then the only
new case to check is x = y + 1, which can be verified using the pattern.
√ 1
d) We calculate the first four terms of an = n n = n n .
1 1 √ 1 √ 1
1
1 1 √
a1 = 1 1 = 11 = 1
3 2
a2 = 2 2 = 2 a3 = 3 3 = 3 a4 = 4 4 = 4 2 = 22 = 2
√ √ √ 3 √ 2 √ √
We know that 2 > 1. We know that 2 < 1.5, since 1.52 = 2.25. Since 2 = 2 2=2 2
√ 3 √ √ 3 √ 3 √ √
3 3 3
and 3 = 3, and 3 = 2 · 1.5 > 2 2, we know that 3 > 2 which implies 3 > 2. So
a3 > a2 and a4 < a3 , which makes n0 = 3 a reasonable guess. But it needs a proof. We consider the
1
function f (x) = x x , and we calculate the derivative.
d 1 d ln x x1 d 1 ln x
f 0 (x) = xx = e = ex
dx dx dx
d 1

1
ln x
=e x ln x we used the chain rule
dx x
d 1

1
= xx ln x
dx x
58
" #
d 1 1 d

1
=x x ln x + ln x we used the product rule
dx x x dx
1
−1 1 1 xx

1 1
=x x ln x 2
+ · = 2 (1 − ln x) = x x −2 (1 − ln x)
x x x x
1
For the function f (x) = x x to reach an extremum, the derivative must be zero. We know that
1
x x −2 is never zero, so (1 − ln x) must be zero, which occurs when ln x = 1, thus when x = e.
We know that e is strictly between 2 and 3, so the only extremum is reached before x = 3, which
confirms that n0 = 3.
Question 5. Determine if the following sequences are monotone:

22 + 3n+1
a) an =
2n+1 + 3n
1
b) an = 1, an+1 = − a2n + 2an
2
1
c) an = 3, an+1 = − a2n + 2an
2
1
d) an = 5, an+1 = − a2n + 2an
2
Answer: a) It is clear that when n gets big, then the numerator grows faster than the denominator,
which makes the terms of the sequence get bigger. But it is not clear for which n this starts to
22 + 3x+1
happen. So it is useful to calculate the derivative of the function f (x) = x+1 . We will use the
2 + 3x
quotient rule and the chain rule.
2x + 3x+1
!
d
dx 2x+1 + 3x
d d
(2x+1 + 3x ) dx (2x + 3x+1 ) − (2x + 3x+1 ) dx (2x+1 + 3x )
=
(2x+1 + 3x )2
x x+1 x+1 x
x+1 x d ln 2 ln 3 x x+1 d ln 2 ln 3
(2 +3 ) dx
e + e − (2 + 3 ) dx
e + e
=
(2x+1 + 3x )2

d d
(2x+1 + 3x ) dx ex ln 2 + e(x+1) ln 3 − (2x + 3x+1 ) dx e(x+1) ln 2 + ex ln 3
=
(2x+1 + 3x )2

d d
(2x+1 + 3x ) dx ex ln 2 + e(x+1) ln 3 (2x + 3x+1 ) dx e(x+1) ln 2 + ex ln 3
= −
(2x+1 + 3x )2 (2x+1 + 3x )2

d d
(2x+1 + 3x ) ex ln 2 dx [x ln 2] + e(x+1) ln 3 dx [(x + 1) ln 3]
=
(2x+1 + 3x )2

d d
(2x + 3x+1 ) e(x+1) ln 2 dx [(x + 1) ln 2] + ex ln 3 dx [x ln 3]
−
(2x+1 + 3x )2
59

d d
(2x+1 + 3x ) 2x dx [x ln 2] + 3x+1 dx [(x + 1) ln 3]
=
(2x+1 + 3x )2

d d
(2x + 3x+1 ) 2x+1 dx [(x + 1) ln 2] + 3x dx [x ln 3]
−
(2x+1 + 3x )2
(2x+1 + 3x ) (2x ln 2 + 3x+1 ln 3)
=
(2x+1 + 3x )2
(2x + 3x+1 ) (2x+1 ln 2 + 3x ln 3)
−
(2x+1 + 3x )2
2x+1 2x ln 2 + 2x+1 3x+1 ln 3 + 3x 2x ln 2 + 3x 3x+1 ln 3
=
(2x+1 + 3x )2
2x 2x+1 ln 2 + 2x 3x ln 3 + 3x+1 2x+1 ln 2 + 3x+1 3x ln 3
−
(2x+1 + 3x )2

2 + 2x+1 3x+1 ln 3 + 3x 2x ln 2 +

22x+1
ln 32x+1
ln 3
=
(2x+1 + 3x )2
ln 2 + 2x 3x ln 3 + 3x+1 2x+1 ln 2 +

22x+1 32x+1
ln 3
−

(2x+1 + 3x )2
2x+1 3x+1 ln 3 + 3x 2x ln 2 − (2x 3x ln 3 + 3x+1 2x+1 ln 2)
=
(2x+1 + 3x )2
2x+1 3x+1 ln 3 + 3x 2x ln 2 − 2x 3x ln 3 − 3x+1 2x+1 ln 2
=
(2x+1 + 3x )2
ln 3 (2x+1 3x+1 − 2x 3x ) + ln 2 (3x 2x − 3x+1 2x+1 )
=
(2x+1 + 3x )2
ln 3 (2x+1 3x+1 − 2x 3x ) − ln 2 (3x+1 2x+1 − 3x 2x )
=
(2x+1 + 3x )2
(ln 3 − ln 2) (2x+1 3x+1 − 2x 3x )
=
(2x+1 + 3x )2
Since ln 3 > ln 2 and for x > 0 we have 2x+1 3x+1 > 2x 3x , the numerator is always greater than
zero. So the derivative is positive valued, which implies that the sequence an is monotone increasing.
1
b) We need to determine whether the sequence with the recurrence relation an+1 = − a2n + 2an
2
and an = 1 is monotone. We calculate a few terms
a1 = 1
1 1 3
a2 = − a21 + 2a1 = − · 12 + 2 · 1 =
2 2 2
2
1 1 3 3 9 6 9 24 15
a3 = − a22 + 2a2 = − · +2· =− + =− + =
2 2 2 2 8 2 8 8 8
60
It seems that the sequence is monotone increasing and approaches 2. This is a guess and needs a
proof. First we will show that for an strictly between 0 and 2, we have an+1 > an . When an is
strictly between 0 and 2, then
1

1 < 2 − an < 2,
2
which implies that
1 2 1

an+1 = − an + 2an = an 2 − an > an .
2 2
Now we will show that when an 6= 2, we have an+1 < 2. We have
1 1

an+1 = − a2n + 2an = 2 − a2n + an
2 4
  
1 1 

=2 − a2n − 4an = 2  − a2n − 4an + 4 −4
 
4 4 | {z }
complete square
 
1
h i
=2 − (an − 2)2 − 4 = 2 1 − (an − 2)2 
 
4 | {z }
< 1 when an 6= 2
1
This confirms that the sequence with the recurrence relation an+1 = − a2n + 2an and an = 1 is
2
monotone increasing.
1
c) We need to determine whether the sequence with the recurrence relation an+1 = − a2n + 2an
2
and an = 3 is monotone. We have
1 1 9 3
a2 = − a21 + 2a1 = − 32 + 2 · 3 = − + 6 = .
2 2 2 2
So a2 < a1 , but from part a) we know that from a2 onwards, the sequence will be increasing. So
the sequence is not monotone.
1
d) We need to determine whether the sequence with the recurrence relation an+1 = − a2n + 2an
2
and an = 5 is monotone. We have
1 1 25 5
a2 = − a21 + 2a1 = − 52 + 2 · 5 = − + 10 = − .
2 2 2 2
So a2 < a1 , and it is clear that for an < 0, we have an+1 < an . So the sequence is monotone
decreasing.
Question 6. Prove that the sequence

√
a1 = 5,
q √
a2 = 5+ 5,
r q √
a3 = 5+ 5+ 5,
..
.
61
r q √
an = 5+ 5+ 5 + ···
| {z }
n times
is increasing and bounded.

Answer: We call this sequence S. A sequence which is increasing and bounded must reach a limit,
we will calculate that limit and call it sup S. First we prove that S is increasing. Clearly, a2 > a1 ,
q √ √
since 5 + 5 > 5. Now we argue by induction. For positive numbers β and γ, β > γ if and only
if β 2 > γ 2 . We have √ 2
a2n+1 = 5 + an = 5 + an > 5 + an−1 = a2n
To find sup S, we first assume that sup S exists, which in the context of an increasing sequence means
that the limit of an for n tending towards infinity exists and is finite. If this limit exists, we let it be
x, and x must satisfy the recurrence relation. Thus
√
x= 5+x
x2 = 5 + x
x2 − x = 5
1 1
x2 − x + − =5
4 4
2
1 1

x− − =5
2 4
2
1 1

x− =5+
2 4
s
1 1
x− =± 5+
2 4
s
1 1
x= ± 5+
2 4
s
1 1
We only keep the solution which is positive, so x = + 5 + . So far we only know that the
2 4 s
1 1
sequence is monotone increasing, and that if the limit exists, it is equal to x = + 5 + . Now we
2 4
s s
1 1 1 1
assume that there exists n0 such that an0 > + 5 + , say an0 = + 5 + + for some > 0.
2 4 2 4
Then s
q 2 1 1
a2n0 +1 = 5 + an0 = 5 + an0 =5+ + 5+ +
2 4
and
 s 2
1 1
a2n0 = + 5 + + 
2 4
62
s s 2
1 1 1
= + 5 + + +  5 + + 
4 4 4
s s 2 s 
1 1 1 1
= + 5 + + +  5 +  + 2  5 +  + 2
4 4 4 4
s s
1 1 1 1
= + 5 + + + 5 + + 2 · 5+ + 2
4 4 4 4
s s
1 1 1
=5+ + 5 + + + 2 · 5+ + 2
2 4 4
s
1
= a2n0 +1 + 2 · 5+ + 2 > a2n0 +1
4
So a2n0 > a2n0 +1 , which implies an0 > an0 +1 , in contradiction with the fact that the sequence is
increasing. Hence the sequence is bounded above and
s
1 1
sup S = x = + 5+
2 4
Question 7. The Fibonacci sequence, (Fn ), is defined by the following recurrence relation: F1 = 1,
F2 = 1, Fn = Fn−1 + Fn−2 for n = 3, 4, 5, . . .. Use induction to show that:
" √ !n √ !n #
1 1+ 5 1− 5
Fn = √ − .
5 2 2
Fn+1
Is the sequence an = monotone?
Fn
F2n F2n+1
Are bn = and cn = monotone?
F2n−1 F2n
√ √
1+ 5 1− 5
Answer: We put α = and β = . We have
2 2
√ !2 √ √ √ √
2 1+ 5 1+2 5+5 4 + (2 + 2 5) 4 (2 + 2 5) 1+ 5
α = = = = + =1+ =1+α
2 4 4 4 4 2
and
√ !2 √ √ √ √
2 1− 5 1−2 5+5 4 + (2 − 2 5) 4 (2 − 2 5) 1− 5
β = = = = + =1+ = 1 + β.
2 4 4 4 4 2
The formula we need to prove by induction is
" √ !n √ !n #
1 1+ 5 1− 5 1
Fn = √ − = √ [αn − β n ] .
5 2 2 5
We have
 √ !1 √ !1  " √ ! √ !# √
1  1+ 5 1− 5  1 1+ 5 1− 5 1 2 5
F1 = √ − =√ − =√ · =1
5 2 2 5 2 2 5 2
63
and
 √ !2 √ !2 
1  1+ 5 1− 5 
F2 = √ −
5 2 2
" √ ! √ !# √
1 1+2 5+5 1−2 5+5 1 4 5
=√ − =√ · =1
5 4 4 5 4
We will use the fact that α2 = 1 + α and β 2 = 1 + β. We need to show that Fn = Fn−1 + Fn−2 .
1 h i 1 h i
Fn−1 + Fn−2 = √ αn−1 − β n−1 + √ αn−2 − β n−2
5 5
1 h n−1 i
=√ α + αn−2 − β n−1 + β n−2
5
1 h n−2 i
=√ α + αn−1 − β n−2 + β n−1
5
1 h n−2 i
=√ α + αn−2 α − β n−2 + β n−2 β
5
1 h i
= √ αn−2 (1 + α) − β n−2 (1 + β)
5
1 h i 1
= √ αn−2 α2 − β n−2 β 2 = √ [αn − β n ] = Fn
5 5
Fn+1
Now we want to see whether the sequence an = is monotone. We need some Fibonacci
Fn
numbers for this.
F1 = 1 F2 = 1 F3 = 1 + 1 = 2 F4 = 2 + 1 = 3 F5 = 3 + 2 = 5
F6 = 5 + 3 = 8 F7 = 8 + 5 = 13 F8 = 13 + 8 = 21 F9 = 21 + 13 = 34 F10 = 34 + 21 = 55
We have
1 2 3 5
a1 = a2 = =2 a3 = a4 =
1 1 2 3
Since a2 > a1 but a3 < a2 , the sequence is not monotone.
F2n
Now we want to see whether the sequence bn = is monotone. We have
F2n−1
F2 1 F4 3 F6 8 F8 21
b1 = = b2 = = b3 = = b4 = =
F1 1 F3 2 F5 5 F7 13
It seems that the sequence bn is monotone increasing, this needs a proof. We have
F2(n+1) F2n+2 √1 [α2n+2 − β 2n+2 ] α2n+2 − β 2n+2 α2n+2 − β 2n β 2

5
bn+1 = = = = =
F2(n+1)−1 F2n+1 √1 [α2n+1 − β 2n+1 ] α2n+1 − β 2n+1 α2n+1 − β 2n−1 β 2
5
and
F2n √1 [α2n − β 2n ] α2n − β 2n
5
bn = = =
F2n−1 √1 [α2n−1 − β 2n−1 ] α2n−1 − β 2n−1
5
64
α2n − β 2n α2 α2n α2 − β 2n α2 α2n+2 − β 2n α2
= · = =
α2n−1 − β 2n−1 α2 α2n−1 α2 − β 2n−1 α2 α2n+1 − β 2n−1 α2
We want to know whether we have the following inequality
α2n+2 − β 2n β 2 α2n+2 − β 2n α2
bn+1 = > = bn
α2n+1 − β 2n−1 β 2 α2n+1 − β 2n−1 α2
The only difference in the two expressions is a factor of α2 which gets replaced by a factor of β 2 . Since
β 2 < α2 , and β 2n > 0 whereas β 2n−1 < 0, this makes the numerator on the left side of the inequality
bigger than the numerator on the right side of the inequality, and it makes the denominator on the
left side of the inequality smaller than the denominator on the right side of the inequality, which
implies that the inequality holds. Thus the sequence bn is monotone increasing.
F2n+1
Now we want to see whether the sequence cn = is monotone. We have
F2n
F3 2 F5 5 F7 13 F9 34
c1 = = c2 = = c3 = = c4 = =
F2 1 F4 3 F6 8 F8 21
It seems that the sequence cn is monotone decreasing, this needs a proof. We have
F2(n+1)+1 F2n+3 √1 [α2n+3 − β 2n+3 ] α2n+3 − β 2n+3 α2n+3 − β 2n+1 β 2

5
cn+1 = = = = =
F2(n+1) F2n+2 √1 [α2n+2 − β 2n+2 ] α2n+2 − β 2n+2 α2n+2 − β 2n β 2
5
and
F2n+1 √1 [α2n+1 − β 2n+1 ] α2n+1 − β 2n+1

5
cn = = =
F2n √1 [α2n − β 2n ] α2n − β 2n
5
α2n+1 − β 2n+1 α2 α2n+1 α2 − β 2n+1 α2 α2n+3 − β 2n+1 α2

= · = =
α2n − β 2n α2 α2n α2 − β 2n α2 α2n+2 − β 2n α2
We want to know whether we have the following inequality
α2n+3 − β 2n+1 β 2 α2n+3 − β 2n+1 α2

cn+1 = < = cn
α2n+2 − β 2n β 2 α2n+2 − β 2n α2
The only difference in the two expressions is a factor of α2 which gets replaced by a factor of β 2 . Since
β 2 < α2 , and β 2n+1 < 0 whereas β 2n > 0, this makes the numerator on the left side of the inequality
smaller than the numerator on the right side of the inequality, and it makes the denominator on
the left side of the inequality bigger than the denominator on the right side of the inequality, which
implies that the inequality holds. Thus the sequence cn is monotone decreasing.
Question 8. Find a function f such that an = f (n) for the following recurrence relations:
a) an+2 = an+1 + an + n2 , a1 = 1, a2 = 2;
b) an+2 = 4an+1 + 4an , a1 = 1, a2 = 2;
c) an+2 = 4an+1 + 4an + 2n , a1 = 1, a2 = 2;
d) an+2 = 2an+1 − an , a1 = 1, a2 = 2;
e) an+2 = 2an+1 − an + n, a1 = 1, a2 = 2;
65
f) an+2 = an , a1 = 1, a2 = 2;
g) an+2 = an + n, a1 = 1, a2 = 2;
h) an+2 = an + (−1)n , a1 = 1, a2 = 2;
1 1
i) an+2 = an+1 + an , a1 = 0, a2 = 1.
2 2
Answer: a) We need to calculate some terms to find a pattern. We use the expression Fn for the
nth Fibbonaci number, as in the previous exercise. We have an+2 = an+1 + an + n2 , a1 = 1, a2 = 2.
a1 = 1 = F2
a2 = 2 = F3
a3 = a2 + a1 + 12 = F2 + F3 + 12 = F4 + 12
a4 = a3 + a2 + 22 = F4 + 12 + F3 + 22 = (F4 + F3 ) + 12 + 22 = F5 + 12 + 22
a5 = a4 + a3 + 32 = F5 + 12 + 22 + F4 + 12 + 32 = F6 + 2 · 12 + 22 + 32
a6 = a5 + a4 + 42 = F6 + 2 · 12 + 22 + 32 + F5 + 12 + 22 + 42
= F7 + (2 + 1)12 + 2 · 22 + 32 + 42 = F7 + F4 · 12 + F3 · 22 + F2 · 32 + F1 · 42
a7 = F8 + F5 · 12 + F4 · 22 + F3 · 32 + F2 · 42 + F1 · 52
From this we get

n−2
Fn−1−i · i2 .
X
an = f (n) = Fn+1 +
i=1
b) We have an+2 = 4an+1 + 4an , a1 = 1, a2 = 2. This looks somewhat similar to the Fibonacci
sequence, with different coefficients in the recurrence relation. Just like we used Fn = f (n) for the
Fibonacci sequence, here we will use Bn = f (n). The only reason for choosing the letter B is because
this is part b). The following is a standard method to solve this kind of recurrence relation. We
introduce a function s(x), such that the power series expansion of s(x) is
∞
Bn xn−1 = B1 + B2 x + B3 x2 + B4 x3 + · · ·
X
s(x) =
n=1
It needs to be said that typically the first term in such a power series expansion would be called B0 .
But in this exercise we are given B1 = a1 = 1 and B2 = a2 = 2. We can express s(x) as follows.
∞ ∞
Bn xn−1 = B1 + B2 x + Bn xn−1
X X
s(x) =
n=1 n=3
∞ ∞ ∞
4(Bn−1 + Bn−2 )xn−1 = B1 + B2 x + 4Bn−1 xn−1 + 4Bn−2 xn−1
X X X
= B1 + B2 x +
n=3 n=3 n=3
∞ ∞ ∞ ∞
Bn−1 xn−1 + 4 Bn−2 xn−1 = B1 + B2 x + 4 Bn xn + 4 Bn xn+1
X X X X
= B1 + B2 x + 4
n=3 n=3 n=2 n=1
∞ ∞ ∞ ∞
xBn xn−1 + 4 x2 Bn xn−1 = B1 + B2 x + 4x Bn xn−1 + 4x2 Bn xn−1
X X X X
= B1 + B2 x + 4
n=2 n=1 n=2 n=1
66
∞ ∞
Bn xn−1 + 4x2 Bn xn−1
X X
= B1 + B2 x − 4B1 x + 4B1 x + 4x
n=2 n=1
∞ ∞
X X
= B1 + B2 x − 4B1 x + 4xB1 + 4x
n=2 n=1
∞ ∞
X X
= B1 + B2 x − 4B1 x + 4x
n=1 n=1
= B1 + B2 x − 4B1 x + 4xs(x) + 4x2 s(x)

= 1 + 2x − 4 · 1 · x + 4xs(x) + 4x2 s(x) = 1 − 2x + 4xs(x) + 4x2 s(x)
Hence
s(x) = 1 − 2x + 4xs(x) + 4x2 s(x) = 1 − 2x + (4x + 4x2 )s(x)

(1 − 4x − 4x2 )s(x) = 1 − 2x
1 − 2x 2x − 1 2x − 1 2x − 1
s(x) = = 2 = 2 = 2
1 − 4x − 4x 2 4x + 4x − 1 4x +{z4x + 1} −1 − 1 4x +{z4x + 1} −2
| |
2x − 1 2x − 1
= = √ √
(2x + 1) − 2
2
((2x + 1) − 2)((2x + 1) + 2)
x
2x − 1 −1
= √ √ = √2 4 √
(2x + (1 − 2))(2x + (1 + 2)) x+ 1− 2
2
x + 1+ 2
2
Now we use partial fractions and we write

√ √
x 1+ 2 1− 2
−1 P Q P x+ 2
+ Q x + 2
s(x) = √2 4 √ = √ + √ = √ √
1− 2 1+ 2 1− 2 1+ 2 1− 2 1+ 2
x+ 2
x + 2
x+ 2
x+ 2
x + 2 x+ 2
So √ ! √ !
x 1 1+ 2 1− 2
− =P x+ +Q x+
2 4 2 2
which by comparing the coefficients gives the system of equations

1


2
 = P + Q,
√ ! √ !
1 1+ 2 1− 2
−


 =P +Q .
4 2 2
1
From the first equation we get Q = − P , and putting this in the second equation gives
2
√ ! √ ! " √ ! √ !# √ !
1 1+ 2 1 1− 2 1+ 2 1− 2 1 1− 2

− =P + −P =P − + ·
4 2 2 2 2 2 2 2
√ √ ! √ ! √ !
1−1+ 2+ 2 1− 2 √ 1− 2
=P + =P 2+
2 4 4
67
√ ! √ √
√ 1− 2 1 1+1− 2 2− 2
−P 2 = + = =
4 4 4 4
√ √ √ √ √ √ √
−1 2 − 2 2−2 2 2−2 2 2−2 2 1− 2
P =√ · = √ = √ √ = =
2 4 4 2 4 2 2 4·2 4
√ √ √ √
1 1 1− 2 2 1− 2 2 − (1 − 2) 1+ 2
Q= −P = − = − = =
2 2 4 4 4 4 4
Thus
√ √
1− 2 1+ 2
P Q 4 √ 4 √
s(x) = √
1− 2
+ √
1+ 2
= +
x+ 2
x+ 2
x + 1−2 2 x + 1+2 2
√ √
1− 2 2√ 1+ 2 2√
4
· 1− 2 4
· 1+ 2
1
2
1
2
= 2√
+ 2√
= 2√
+ 2√

x· 1− 2
+1 x· 1+ 2
+1 x· 1− 2
+1 x· 1+ 2
+1
1 1 1 1
2 2 2 2
= + −2
= + −2

−x · √2 +1 −x · √ +1 1− √2 x 1− √ x
2−1 1+ 2 2−1 1+ 2
1 1 1 1
2 2 2 2 √
= √ + √ = √ +
2( 2+1) −2(1− 2) 2 2+2 −2+2 2
1− √ √ x 1− √ √ x 1− 2−1
x 1− 1−2
x
( 2−1)( 2+1) (1+ 2)(1− 2)
1 1 1 1
2 2 √ 2 2 √
= √
2 2+2
+ −2+2 2
= √
2 2+2
+ 2−2 2
1− 1
x 1− −1
x 1− 1
x 1− 1
x
 
1 1 1
=  √ + √ 
2 1− 2+2 2 x 1− 2−2 2 x
We need to expand the two functions of type

1
1 − cx
as a power series. For 0 < cx < 1, we put
Sk = 1 + cx + (cx)2 + (cx)3 + (cx)4 + · · · + (cx)k−1 + (cx)k .
Then
cxSk = cx + (cx)2 + (cx)3 + (cx)4 + (cx)5 + · · · + (cx)k + (cx)k+1 .
Hence
(1 − cx)Sk = Sk − cxSk = 1 + cx + (cx)2 + (cx)3 + (cx)4 + · · · + (cx)k−1 + (cx)k

h i
− cx + (cx)2 + (cx)3 + (cx)4 + (cx)5 + · · · + (cx)k + (cx)k+1
= 1 − (cx)k+1
Thus
1 − (cx)k+1
Sk =
1 − cx
68
and since 0 < cx < 1,
S = 1 + cx + (cx)2 + (cx)3 + (cx)4 + · · ·
1 − (cx)k+1 1
= lim Sk = lim =
k→∞ k→∞ 1 − cx 1 − cx
So we have the required power series expansion
1
= 1 + cx + (cx)2 + (cx)3 + (cx)4 + · · ·
1 − cx
= 1 + cx + c2 x2 + c3 x3 + c4 x4 + · · ·
∞ ∞
cn x n = cn−1 xn−1
X X
=
n=0 n=1
We have
 
∞
X
n−1 1 1 1
s(x) = Bn x =  √ + √ 
n=1 2 1− 2+2 2 x 1− 2−2 2 x
∞ √ n−1 n−1 X ∞ √ n−1 n−1
" #
1 X
= 2+2 2 x + 2−2 2 x
2 n=1 n=1
∞
X 1
√ n−1 n−1 √ n−1 n−1
= 2+2 2 x + 2−2 2 x
n=1 2
Thus
1
√ n−1 √ n−1
f (n) = Bn = 2+2 2 + 2−2 2 .
2
c) We have an+2 = 4an+1 + 4an + 2n , a1 = 1, a2 = 2. We use the sequence Bn from part b). By
analogy with part a), one obtains
n−2
Fn−1−i · 2i .
X
an = f (n) = Bn +
i=1
Since we argue by analogy, we should comment on the fact that here we have Bn whereas in part a)
we have Fn+1 . This is because in the sequence of part a), the first term is F2 and the second term is
F3 , whereas here the first term is B1 and the second term is B2 .
d) We need to calculate some terms to find a pattern. We have an+2 = 2an+1 − an , a1 = 1, a2 = 2.
a1 = 1
a2 = 2
a3 = 2a2 − a1 = 2 · 2 − 1 = 3
a4 = 2a3 − a2 = 2 · 3 − 2 = 4
a5 = 2a4 − a3 = 2 · 4 − 3 = 5
69
So we get
an = f (n) = n.
e) We have an+2 = 2an+1 − an + n, a1 = 1, a2 = 2. Using the result from part d), and by analogy
with part a), we get
n−2
X
an = f (n) = n + Fn−1−i · i.
i=1
f) We have an+2 = an , a1 = 1, a2 = 2. It is clear that this sequence is 1, 2, 1, 2, 1, 2, . . .. So

3 (−1)n
an = f (n) = + .
2 2
g) We have an+2 = an + n, a1 = 1, a2 = 2. Using the result from part f), and by analogy with
part a), we get
3 (−1)n n−2 X
an = f (n) = + + Fn−1−i · i.
2 2 i=1
h) We need to calculate some terms to find a pattern. We have an+2 = an + (−1)n , a1 = 1, a2 = 2.
a1 = 1 a2 = 2
a3 = a1 + (−1)1 = 1 − 1 = 0 a4 = a2 + (−1)2 = 2 + 1 = 3
a5 = a3 + (−1)3 = 0 − 1 = −1 a6 = a4 + (−1)4 = 3 + 1 = 4
a7 = a5 + (−1)5 = −1 − 1 = −2 a8 = a6 + (−1)6 = 4 + 1 = 5
So we get
   
1 1 n−1 1 1 n

 + (−1)n+1  1 −
an = f (n) =   + (−1)n  1 +
+ .
 
2 2  2 2 2  2
| {z } | {z }
0 when n even 0 when n odd
1 1
i) We have an+2 = an+1 + an , a1 = 0, a2 = 1. This looks somewhat similar to the Fibonacci
2 2
sequence, with different coefficients in the recurrence relation. Just like we used Fn = f (n) for the
Fibonacci sequence, here we will use In = f (n). The only reason for choosing the letter I is because
this is part i). The following is a standard method to solve this kind of recurrence relation. We
introduce a function s(x), such that the power series expansion of s(x) is
∞
In xn−1 = I1 + I2 x + I3 x2 + I4 x3 + · · ·
X
s(x) =
n=1
It needs to be said that typically the first term in such a power series expansion would be called I0 .
But in this exercise we are given I1 = a1 = 0 and I2 = a2 = 1. We can express s(x) as follows.
∞ ∞
In xn−1 = I1 + I2 x + In xn−1
X X
s(x) =
n=1 n=3
∞ ∞ ∞
1 1 1
(In−1 + In−2 )xn−1 = I1 + I2 x + In−1 xn−1 + In−2 xn−1
X X X
= I1 + I2 x +
n=3 2 n=3 2 n=3 2
70
∞ ∞ ∞ ∞
1X 1X 1X 1X
= I1 + I2 x + In−1 xn−1 + In−2 xn−1 = I1 + I2 x + In xn + In xn+1
2 n=3 2 n=3 2 n=2 2 n=1
∞ ∞ ∞ ∞
1X 1X 1 X 1 X
= I1 + I2 x + xIn xn−1 + x2 In xn−1 = I1 + I2 x + x In xn−1 + x2 In xn−1
2 n=2 2 n=1 2 n=2 2 n=1
∞ ∞
1 1 1 X 1 X
= I1 + I2 x − I1 x + I1 x + x In xn−1 + x2 In xn−1
2 2 2 n=2 2 n=1
∞ ∞
1 1 1 X 1 X
= I1 + I2 x − I1 x + xI1 + x In xn−1 + x2 In xn−1
2 2 2 n=2 2 n=1
∞ ∞
1 1 X 1 X
= I1 + I2 x − I1 x + x In xn−1 + x2 In xn−1
2 2 n=1 2 n=1
1 1 1
= I1 + I2 x − I1 x + xs(x) + x2 s(x)
2 2 2
1 1 1 1 1
= 0 + x − · 0 · x + xs(x) + x2 s(x) = x + xs(x) + x2 s(x)
2 2 2 2 2
Hence
1 1 1
s(x) = x + xs(x) + x2 s(x) = x + x + x2 s(x)
2 2 2
1

1− x + x2 s(x) = x
2

2 − x − x2 s(x) = 2x

x2 + x − 2 s(x) = −2x
−2x −2x −2x
s(x) = = =
x2 + x − 2 1 1
x2 + x + − 14 − 2 x2 + x + − 49
| {z 4} | {z 4}
−2x −2x
= 2 = 1

3

1

3

x+ 1
− 9 x+ 2
− 2
x+ 2
+ 2
2 4
−2x −2x
= =
x+ 1
− 3
x+ 1
+ 3 (x − 1)(x + 2)
2 2 2 2
Now we use partial fractions and we write

−2x P Q P (x + 2) + Q(x − 1)
s(x) = = + =
(x − 1)(x + 2) x−1 x+2 (x − 1)(x + 2)
So
−2x = P (x + 2) + Q(x − 1)
which by comparing the coefficients gives the system of equations

−2
 = P + Q,
0 = 2P − Q.

71
From the first equation we get Q = −2 − P , and putting this in the second equation gives
0 = 2P − (−2 − P ) = 2P + 2 + P = 3P + 2
−2
P =
3
−6 −2 4
Q = −2 − P = − =−
3 3 3
Thus
P Q − 23 − 43 − 23 − 43 2
3
− 32 2
3
− 23
s(x) = + = + = + = + = +
1 − x 1 + 12 x

x−1 x+2 x−1 x+2 −1 + x 2 + x 1 − x 1 − −1x
2
 
2 1 1
=  − 
3 1 − x 1 − −1x
2
We need to expand the two functions of type

1
1 − cx
as a power series. For 0 < cx < 1, we put
Sk = 1 + cx + (cx)2 + (cx)3 + (cx)4 + · · · + (cx)k−1 + (cx)k .
Then
cxSk = cx + (cx)2 + (cx)3 + (cx)4 + (cx)5 + · · · + (cx)k + (cx)k+1 .
Hence
(1 − cx)Sk = Sk − cxSk = 1 + cx + (cx)2 + (cx)3 + (cx)4 + · · · + (cx)k−1 + (cx)k

h i
− cx + (cx)2 + (cx)3 + (cx)4 + (cx)5 + · · · + (cx)k + (cx)k+1
= 1 − (cx)k+1
Thus
1 − (cx)k+1
Sk =
1 − cx
and since 0 < cx < 1,
S = 1 + cx + (cx)2 + (cx)3 + (cx)4 + · · ·
1 − (cx)k+1 1
= lim Sk = lim =
k→∞ k→∞ 1 − cx 1 − cx
So we have the required power series expansion
1
= 1 + cx + (cx)2 + (cx)3 + (cx)4 + · · ·
1 − cx
= 1 + cx + c2 x2 + c3 x3 + c4 x4 + · · ·
72
∞ ∞
cn x n = cn−1 xn−1
X X
=
n=0 n=1
We have
 
∞
2 1 1
In xn−1
X
s(x) = =  − 
n=1 3 1 − x 1 − −1x
2
∞ ∞
1 n−1 n−1
" #
2 X

(1)n−1 xn−1 −
X
= − x
3 n=1 n=1 2
∞ n−1
" #
2 1

(1)n−1 xn−1 − − n−1
X
= x
n=1 3 2
Thus " n−1 # " n−1 #

2 1 2 1

f (n) = In = (1)n−1 − − = 1− − .
3 2 3 2
Question 9. Using the results of questions 4, 5, 8 and 9 from Temat 3, calculate the following limits:
1 + 2 + 3 + ··· + n
a) lim
n→∞ n2
12 + 22 + 32 + · · · + n2
b) n→∞
lim
n3
13 + 23 + 33 + · · · + n3
c) lim
n→∞ n4
1k + 2k + 3k + · · · + nk
d) lim
n→∞ nk+1
Answer: a) From the result of question 4 from Temat 3, we know that
n(n + 1)
1 + 2 + 3 + ··· + n =
2
Thus
n(n+1) 1
1 + 2 + 3 + ··· + n 2 n(n + 1) n2 + n 1+ n 1
lim = n→∞
lim = n→∞
lim = lim = n→∞
lim =
n→∞ n2 n2 2n 2 n→∞ 2n 2 2 2
b) From the result of question 5 from Temat 3, we know that
n(n + 1)(2n + 1)
12 + 22 + 32 + · · · + n2 =
6
Thus
n(n+1)(2n+1)
12 + 22 + 32 + · · · + n2 6 n(n + 1)(2n + 1) 2 1
lim = n→∞
lim = n→∞
lim = =
n→∞ n3 n3 6n 3 6 3
73
We used the fact that the highest power of n, namely n3 , has a coefficient of 2 in the numerator and
a coefficient of 6 in the denominator.
c) From the result of question 8 from Temat 3, we know that
(n + 1)4 − 1 − n(n + 1)(2n + 1) − 2n(n + 1) − n

13 + 23 + 33 + · · · + n3 =
4
Thus
(n+1)4 −1−n(n+1)(2n+1)−2n(n+1)−n
13 + 23 + 33 + · · · + n3 4
lim = lim
n→∞ n4 n→∞ n4
(n + 1)4 − 1 − n(n + 1)(2n + 1) − 2n(n + 1) − n 1
= lim 4
=
n→∞ 4n 4
We used the fact that the highest power of n, namely n4 , has a coefficient of 1 in the numerator and
a coefficient of 4 in the denominator.
d) From the result of question 9 from Temat 3, we know that
Sk (n) = 1k + 2k + 3k + · · · + nk

k+1 k+1 k+1 k+1
(n + 1)k+1 − 1 − 2
Sk−1 (n) − 3
Sk−2 (n) − · · · − k−1
S2 (n) − k
S1 (n) − n
=
k+1
Thus
1k + 2k + 3k + · · · + nk
lim =
n→∞ nk+1
(n+1)k+1 −1−(k+1
2 ) k−1
S (n)−(k+1
3 ) k−2
S k+1
(n)−···−(k−1 )S2 (n)−(k+1
k ) 1
S (n)−n
k+1
= lim
n→∞ nk+1

k+1 k+1 k+1 k+1
(n + 1)k+1 − 1 − 2
Sk−1 (n) − 3
Sk−2 (n) − · · · − k−1
S2 (n) − k
S1 (n) − n
= lim
n→∞ (k + 1)nk+1
1
=
k+1
We used the fact that the highest power of n, namely nk+1 , has a coefficient of 1 in the numerator
and a coefficient of k + 1 in the denominator. It is clear that the highest power of n which appears
in Si (n) with i < k is equal to ni+1 , so nk+1 is higher.
Temat 6: Sequences II.

Question 1. For each of the following sequences, calculate the limit if the limit exists. If the limit
does not exist, explain why it does not exist.
n2 + (−1)n · n
a) an = √ 2
n+ 3
74
√ 4
q
n+8 n
ą) an = √ 2
(n + 3 n)
√ 2
n 3n − 7
b) an = √ 4
n+2+2
v √
u n + (−1)n + 4 n
u
c) an = u
t √ 2
3n + 2n
n+2 2

n+1
ć) an =
2 + 4 + 6 + · · · + 2n
√ √
d) an = n + 1 − n
√ √
e) an = n2 + 5n − n2 − n
√ √
ę) an = 5n2 − 3n − n 5 + 8
√ √
f) an = n4 + 3n2 + 5 − n4 − n2 + n
q √ q √
n2 + n + 1 − n2 − n − 1
g) an = √ √
n+1− n
1 − 2 + 3 − 4 + · · · + (2n − 1) − 2n
h) an = √
n2 + 3
12 32 52 (2n − 1)2
i) an = + + + · · · +
n3 n3 n3 n3
√
3
j) an = n n3 + n + 2 − n
q q
3 3
j) bis) an = n(n + 1)2 − n(n − 1)2
(3n)!
k) an =
(n!)3
cos(n!)
l) an = √
n
√
n
3−1
ł) an =
2 + (−1)n
√n
n−1
m) an =
3 + (−1)n
v v
u u !
n
u
n
u u
n) an = t1 + t
2
√
n
ń) an = 2n + n2
√
n
o) an = 3n + 5n + 7n
75
s
1 n
n n
n 1 1
ó) an = + +
2 3 4
1 1 1

p) an = n 2 + 2 + ··· + 2
n +1 n +2 n +n
1 2 n
q) an = 2 + 2 + ··· + 2
n +1 n +2 n +n
14 + 24 + 34 + · · · + n4
r) an = 4
1 + 24 + 34 + · · · + n4 + (n + 1)4
1! + 3! + 5! + · · · + (2n − 1)!
s) an =
2! + 4! + 6! + · · · + (2n)!
n
1

ś) an = 1 −
n
2n−1
1

t) an = 1 −
n2
2n+1
1

u) an = 1 +
2n
2n+1
n−1

v) an =
n+3
!(−1)n n
(−1)n
w) an = 1 +
n
!2n2 −n
n2 − 1
x) an =
n2
√ !n
1+ n2
y) an =
2
Fn+1
z) an = , where Fn is the n-th Fibonacci number
Fn
1 5

ź) a1 = 1, an+1 = an +
2 an
1
ż) a1 = 1, an+1 = 1 +
an
Answer: a)
n2 + (−1)n · n n2 + (−1)n · n 1 + (−1)n · n1

lim √ 2 = lim √ = lim √ =1
n→∞ n2 + 2n 3 + 3 n→∞ 1 + 2 3 + 3

n→∞
n+ 3 n n 2
ą)
√ 4
q
n+8 n √ 2 √
(n + 8 n) n2 + 16n n + 64n 1 + √16n + 64
n
lim √ 2 = n→∞
lim √ 2 = n→∞
lim √ = lim =1
n→∞ (n + 3 n) (n + 3 n) 2
n + 6n n + 9n n→∞ 1 + n + n9
√6
76
b)
√ 2 √ √
n
n 3n − 7 n 3n − 14 3n + 49 n2
3n − 14 3n + 49
lim 4 = lim 2 = lim
n→∞ √ √ √
2
n→∞ n→∞ 1
n+2+2 n+2+4 n+2+4 n2
n+2+4 n+2+4
√
3n
3 − 14 n
+ 49
n
= lim √ 2 =3
n→∞ 2 n+2
1+ n
+ 4 n + n4
c)
√
v v
u 1
u n + (−1)n + 4 n n + (−1)n + n 4
u u
lim u
√ 2 = n→∞
lim t √ =0
n→∞ 9n2 + 6 2n + 2n
t
3n + 2n
The highest power of n in the denominator is higher than the highest power of n in the numerator,
thus the limit is zero.
ć)
2
n+2 2 (n+2)! 2

(n+2)!
n+1 (n+1)!·[(n+2)−(n+1)]! (n+1)!·1!
lim = lim = lim
n→∞ 2 + 4 + 6 + · · · + 2n n→∞ 2(1 + 2 + 3 + · · · + n) n→∞ 2(1 + 2 + 3 + · · · + n)
2
(n+2)(n+1)(n)(n−1)···1 2

(n+2)!
(n+1)! (n+1)n(n−1)(n−2)···1
= n→∞
lim = n→∞
lim
2(1 + 2 + 3 + · · · + n) 2(1 + 2 + 3 + · · · + n)
(n + 2)2 n2 + 2n + 4
= n→∞
lim = n→∞
lim
2(1 + 2 + 3 + · · · + n) 2(1 + 2 + 3 + · · · + n)
n2 + 2n + 4 n2 + 2n + 4 1 + n2 + 4
n2
= lim = lim = lim =1
n→∞ 2· n(n+1)
2
n→∞ n2 + n n→∞ 1 + n1
d)
√ √ √ √ √
√
√ √ n+1− n n+1− n n+1+ n
lim n + 1 − n = n→∞
lim = n→∞
lim √ √
n→∞ 1 n+1+ n
√ 2 √ 2
n + 1 − ( n) n+1−n 1
= n→∞
lim √ √ lim √
= n→∞ lim √
√ = n→∞ √ =0
n+1+ n n+1+ n n+1+ n
e)
 
√ √ v v
25 25 1 1
u u 
u 2 u 2
lim n2 + 5n − n2 − n = lim un + 5n + − −u
un −n+ − 
n→∞ n→∞  u
t| {z 4} 4 t| {z 4} 4 

s s 
2 2
5 25 1 1

= lim  n+ − − n− − 
n→∞ 2 4 2 4
77
s s 
2 2
5 1

= lim  n+ − n− 
n→∞ 2 2
5 1 6

= n→∞
lim n+ − n− = n→∞
lim =3
2 2 2
ę)
s 
√ √
3
√
lim 5n − 3n − n 5 + 8 = lim
2  5 n − n − n 5 + 8
2
n→∞ n→∞ 5
 s 
√ 3 √
= lim  5 · n2 − n − n 5 + 8
n→∞ 5
 
√ v √
3 9 9
u 
 u 2 
= lim  5 · un − n− + − n 5 + 8
n→∞  u
t| 5{z 100} 100 

complete square
 
v
√ u 2 √
3 9

 u 
= lim  u n−
5·u + −n 5 + 8
n→∞ 
t 10 |100
{z }


small in proportion
 s 
√
3
2 √
= lim  5 · n− − n 5 + 8
n→∞ 10
√ √
3

= n→∞
lim 5· n− −n 5+8
10
√
√ √
!
3 5
= lim n 5− −n 5+8
n→∞ 10
√ ! √ √
3 5 3 5 3 5 3
= lim − +8 =− +8=8− √ √ =8− √
n→∞ 10 10 2 5 5 2 5
f)
√ √
lim n4 + 3n2 + 5 − n4 − n2 + n
n→∞
√ √
n4 + 3n2 + 5 − n4 − n2 + n
= lim
n→∞ 1
√ √ √ √
n4 + 3n2 + 5 − n4 − n2 + n n4 + 3n2 + 5 + n4 − n2 + n
= n→∞
lim √ √
n4 + 3n2 + 5 + n4 − n2 + n
√ 2 √ 2
n4 + 3n2 + 5 − n4 − n2 + n n4 + 3n2 + 5 − (n4 − n2 + n)
= lim √ √ = lim √ √
n→∞ n4 + 3n2 + 5 + n4 − n2 + n n→∞ n4 + 3n2 + 5 + n4 − n2 + n
78
4n2 − n + 5 4 − n1 + n52
= lim √ √ = lim 1
√
4 + 3n2 + 5 + 1
√
n→∞ n4 + 3n2 + 5 + n4 − n2 + n n→∞ n2 n n2 n4 − n2 + n
1 5 1 5
4− n
+ 2 4− + 2
= n→∞
lim q qn = n→∞
lim q n qn
1 1 3 5 1 1
n4
(n4 + 3n2 + 5) + n4
(n4 − n2 + n) 1+ n2
+ n4
+ 1− n2
+ n3
4 4 4
=√ √ = = =2
1+ 1 1+1 2
g)
q √ q √
n2 + n + 1 − n2 − n−1
lim √ √
n→∞ n+1− n
q
√ q √ q
√ q √
n2 + n + 1 − n2 − n − 1 n2 + n + 1 + n2 − n − 1
= n→∞
lim √ √ q 2 √
q √
n+1− n n + n+1+ n − n−1
2
q
√ 2 q
√ 2
n2 + n+1 − n2 − n−1
= lim √ √ q √ q √
n→∞
n+1− n n2 + n+1+ n − n−1
2
√ √
n2 + n + 1 − n2 − n − 1
= n→∞
lim √ √ q 2 √
q √
n+1− n n + n + 1 + n2 − n − 1
√ √
n+1+ n−1
= n→∞
lim √ √ q 2 √
q √
n+1− n n + n+1+ n − n−1
2
√ √ √ √
n+1+ n n+1+ n−1
= n→∞
lim √ √ √ √ q 2 √
q √
n+1+ n n+1− n n + n+1+ n − n−1
2
√
√ √ √
n+1+ n n+1+ n−1
= n→∞
lim √ 2 √ 2 q 2 √
q √
n + 1 − ( n) n + n + 1 + n2 − n − 1
√ √ √ √
n+1+
n n+1+ n−1
= lim q
√ q √
n→∞
[n + 1 − n] n + n+1+ n − n−1
2 2
√ √ √ √
n+1+ n n+1+ n−1
= lim q √ q √
n→∞
n2 + n + 1 + n2 − n − 1
√ √ √ √
1
n
n + 1 + n n + 1 + n − 1
= n→∞
lim q
√ q √
1
n
n + n+1+ n − n−1
2 2
79
√ √ √ √
√1 n + 1 + n √1n n+1+ n−1
n
= lim q
√ q √
n→∞ 1
n
n + n+1+ n − n−1
2 2
q
n 1
q q
n n 1
q
n 1
q
1
√ q 1
q
1

n
+ n
+ n n
+ n
+ n
− n 1+ n
+ 1 1+ n
+ 1− n
= n→∞
lim q √ q √ = n→∞
lim q √ q √
n2 n+1 n2 n−1 n+1 n−1
+ + n2
− n2 n2
1+ n2 n2
+ 1− n2
√ √ √ √
1+ 1 1+ 1 (1 + 1)(1 + 1) 2·2 4
= √ √ = = = =2
1+ 1 1+1 2 2
h)
-1 -1 -1
z }| { z }| { z }| {
1 − 2 + 3 − 4 + · · · + (2n − 1) − 2n 1 − 2 + 3 − 4 + · · · + (2n − 1) − 2n
lim √ = lim √
n→∞ 2
n +3 n→∞ n2 + 3
(−1) · n −n −n n1
= lim √ = lim √ = lim √
n→∞ n2 + 3 n→∞ n2 + 3 n→∞ n1 n2 + 3
−1 −1
= lim q 2 = lim q = −1
n→∞ n +3 n→∞ 3
n2
1+ n2
i)
!
12 32 52 (2n − 1)2 12 + 32 + 52 + · · · + (2n − 1)2
lim + + + · · · + = n→∞
lim
n→∞ n3 n3 n3 n3 n3
We need to simplify the numerator.

n n n n n
12 + 32 + 52 + · · · + (2n − 1)2 = (2i − 1)2 = 4i2 − 4i + 1 = 4i2 +
X X X X X
(−4i) + 1
i=1 i=1 i=1 i=1 i=1
n n n
n(n + 1)(2n + 1) n(n + 1)
i2 − 4
X X X
=4 i+ 1=4 −4 +n
i=1 i=1 i=1 6 2
2n(n + 1)(2n + 1)
= − 2n(n + 1) + n
3
Thus
!
12 32 52 (2n − 1)2 12 + 32 + 52 + · · · + (2n − 1)2
lim + + + · · · + = lim
n→∞ n3 n3 n3 n3 n→∞ n3
2n(n+1)(2n+1)
3
− 2n(n + 1) + n
= lim
n→∞ n3
4
The highest power of n in the numerator is n3 with a coefficient of , and the highest power of n in
3
4
4
the denominator is n3 with a coefficient of 1, thus limit is 3 = .
1 3
80
√
3
j) We are interested in the limit of the sequence an = n n3 + n + 2 − n . We assume that the
√
3
lim n
limit exists and that it is finite, let’s say n→∞ n3 + n + 2 − n = x. We will use the binomial
theorem.
√ 3 √ 3
3
lim n3
3 3
x = lim n
n→∞
n3 +n+2−n = n→∞ n3 + n + 2 − n
1 3
= lim n3 n3 + n + 2 3
−n
n→∞
3 2 1
3 3 3 3 3 2 3 3 3
= lim n n +n+2 − 3n n + n + 2 + 3n n +n+2 −n
n→∞
2 1
3 3 3 3 2 3 3 3
= lim n n + n + 2 − 3n n + n + 2 + 3n n +n+2 −n
n→∞
2 1
= lim n3 n + 2 − 3n n3 + n + 2 3
+ 3n2 n3 + n + 2 3
n→∞
1 1
3 3 3 3 3
= lim n n + 2 − 3n n +n+2 −n n +n+2
n→∞
√ 1
3 3 3 3
= lim n n + 2 − 3n n3 +n+2−n n +n+2
n→∞
√ 1
= lim n3 (n + 2) − lim n3 3n n3 + n + 2
3 3
n3 + n + 2 − n
n→∞ n→∞
√ 1
3 3 3 3 3
= lim n (n + 2) − lim n n3 +n+2−n · lim 3n n +n+2
n→∞ n→∞ n→∞
1
3 3 3 3
= lim n (n + 2) − x lim 3n n +n+2
n→∞ n→∞
Keeping in mind that x is a finite number which we would like to determine, we get
1
3 3 3 3 3
x = lim n (n + 2) − x lim 3n n +n+2
n→∞ n→∞
1
3 3
x lim 3n n +n+2 3
= lim n3 (n + 2) − x3
n→∞ n→∞
limn→∞ (n3 (n + 2)) − x3

x= 1
limn→∞ 3n3 (n3 + n + 2) 3
limn→∞ (n3 (n + 2) − x3 )
= 1
limn→∞ 3n3 (n3 + n + 2) 3
n3 (n + 2) − x3 (n + 2) − x3
= n→∞
lim 1 = n→∞
lim 1
3n3 (n3 + n + 2) 3 3 (n3 + n + 2) 3

3 2 x3
1
(n + 2) − x3 1+ n
− n 1 1
n
= lim 1 = n→∞
lim 1 = 1 =
n→∞ 3
(n3 + n + 2) 3 1 2 3 3 (1) 3 3
n 3 1+ n2
+ n3
If our original assumption, that the limit exists and that it is finite, was wrong, then such a procedure
81
would not allow us to determine x but lead to a contradiction. Thus
√ 1
3 3+n+2−n = .
lim n n
n→∞ 3
q q
3 3
j) bis) We are interested in the limit of the sequence an = n(n + 1)2 − n(n − 1)2 We assume
q q
3 3
that the limit exists and that it is finite, let’s say lim n(n + 1)2 − n(n − 1)2 = x. We will use
n→∞
the binomial theorem.
q q 3 q q 3
3 3 3 3 3
x = lim n(n + 1)2 − n(n − 1)2 = lim n(n + 1)2 − n(n − 1)2
n→∞ n→∞
1 1 3
2 3 2 3
= n→∞
lim n(n + 1) − n(n − 1)
3 2 1 1 2
2 3 2 3 2 3 2 3 2 3
= n→∞
lim n(n + 1) − 3 n(n + 1) n(n − 1) + 3 n(n + 1) n(n − 1) −
3
− n(n − 1)2 3
2 1 1 2
lim n n2 + 2n + 1 − 3 n(n + 1)2
= n→∞ 3
n(n − 1)2 3
+ 3 n(n + 1)2 3
n(n − 1)2 3
−

−n n2 − 2n + 1
2 1 1 2
= lim
n→∞
n3 + 2n2 +
n − 3 n(n + 1)2 3
n(n − 1)2 3
+ 3 n(n + 1)2 3
n(n − 1)2 3
−

− n3 − 2n2 +
n
2 1 1 2
2 2 3 2 3 2 3 2 3
= lim 4n − 3 n(n + 1) n(n − 1) + 3 n(n + 1) n(n − 1)
n→∞
1 1 1 1
2 2 3 2 3 2 3 2 3
= lim 4n − n(n + 1) − n(n − 1) · 3 n(n + 1) n(n − 1)
n→∞
q q 1 1
= lim 4n2 − 3
n(n + 1)2 − 3
n(n − 1)2 · 3 n n2 + 2n + 1 3
n n2 − 2n + 1 3
n→∞
q q 1 1
2
= lim 4n − lim 3
n(n + 1)2 − 3
n(n − 1)2 · lim 3 n n2 + 2n + 1 3
n n2 − 2n + 1 3
n→∞ n→∞ n→∞
1 1
= lim 4n2 − x · lim 3 n n2 + 2n + 1 3
n n2 − 2n + 1 3
n→∞ n→∞
1 1
lim 4n2 − x · n→∞
= n→∞ lim 3 n3 + 2n2 + n 3
n3 − 2n2 + n 3
Keeping in mind that x is a finite number which we would like to determine, we get
1 1
x3 = n→∞
lim 4n2 − x · n→∞
lim 3 n3 + 2n2 + n 3
n3 − 2n2 + n 3
1 1
3 2 3 2
x · lim 3 n + 2n + n 3
n − 2n + n 3
= lim 4n2 − x3
n→∞ n→∞
limn→∞ (4n2 ) − x3
x= 1 1
limn→∞ 3 (n3 + 2n2 + n) 3 (n3 − 2n2 + n) 3
82
limn→∞ (4n2 − x3 )
= 1 1
limn→∞ 3 (n3 + 2n2 + n) 3 (n3 − 2n2 + n) 3
4n2 − x3
= lim 1 1
n→∞
3 (n3 + 2n2 + n) 3 (n3 − 2n2 + n) 3
x3
4− n2
= lim 1 1
n→∞ 3
n2
(n3 + 2n2 + n) 3 (n3 − 2n2 + n) 3
x3
4− n2
= n→∞
lim 1 1
3 1
n
(n3 + 2n2 + n) 3 n
(n3 − 2n2 + n) 3
x3
4− n2 4 4
= n→∞
lim 1 1 = 1 1 =
2 1 3 2 1 3 3 (1) (1) 3 3 3
3 1+ n
+ n3
1− n
+ n3
If our original assumption, that the limit exists and that it is finite, was wrong, then such a procedure
would not allow us to determine x but lead to a contradiction. Thus
4
q q
3 3
lim n(n + 1) − n(n − 1) = .
2 2
n→∞ 3
k) Here we need to construct an argument based on the definition of the factorial. In the denominator
one has the product of the first n integers to the power three. But in the denominator one has the
product of the first 3n integers. So
(3n)!
(3n)! 2n!
· (2n)!
n!
· n! (3n)!
2n!
· (2n)!
n!
lim = lim = lim
n→∞ (n!)3 n→∞ (n!)3 n→∞ (n!)2
3n(3n−1)(3n−2)···2·1 2n(2n−1)(2n−2)···2·1
2n(2n−1)(2n−2)···2·1
· n(n−1)(n−2)···2·1
= lim
n→∞ [n(n − 1)(n − 2) · · · 2 · 1]2
[3n(3n − 1)(3n − 2) · · · (2n + 2)(2n + 1)] · [2n(2n − 1)(2n − 2) · · · (n + 2)(n + 1)]
= n→∞
lim
[n(n − 1)(n − 2) · · · 2 · 1]2
(2n)n · nn (2n)n 2n nn
> lim 2 = lim n
= lim n
= lim 2n = ∞
n→∞ n
(n ) n→∞ n n→∞ n n→∞
√
l) Since the value of cos is always between −1 and 1, but n tends towards infinity, we have
cos(n!)
lim √ = 0.
n→∞ n
√n
ł) Since 3 tends towards one, the numerator tends towards zero. But the denominator is always
either 1 or 3, depending on whether n is odd or even. So
√
n
3−1
lim = 0.
n→∞ 2 + (−1)n
√
m) We first need to determine the limit of n n.
√ √
n
1 1
lim n
n = lim eln n = lim eln n n = lim e n ln n
n→∞ n→∞ n→∞ n→∞
83
1
We need to find the limit of the exponent ln n. Since ln n grows more slowly than n, we have
n
1
lim ln n = 0
n→∞ n
Thus
√ 1
lim n
n = lim e n ln n = e0 = 1
n→∞ n→∞
√
n
Since n tends towards one, the numerator tends towards zero. But the denominator is always
√
n
n−1
lim
n→∞ 3 + (−1)n
= 0.
n)
v v v v
u u ! u s u s
n n! n!
u u u
n
u u n n
lim t1 + t = lim t1 + = lim t1 +
n→∞ 2 n→∞ 2!(n − 2)! n→∞ 2(n − 2)!
v v v
u u s
n(n − 1)(n − 2) · · · 2 · 1 n(n − 1)
u u u
n
u u n
= n→∞
lim t1 + t = lim 1+
t
2(n − 2)(n − 3)(n − 4) · · · 2 · 1 n→∞ 2
1  q  1 q
 s n
n n(n−1) 1 n(n−1)
n(n − 1)  ln 1+ 2 n
ln 1+ 2
= n→∞
lim 1 + lim e
= n→∞ lim e
= n→∞
 
2

 s 
1 n(n − 1) 
We will determine the limit of the exponent ln 1 + . When n tends towards infinity,
n 2
we have s s s
n(n − 1) n2 − n n2 √
1+ =1+ <1+ < 1 + n2 < 1 + n
2 2 2
The number 1 is negligibly small in proportion, and ln n grows more slowly than n. Thus
 s 
1 n(n − 1) 
lim ln 1 +
n→∞ n
=0
2
So we get
v v q
u n(n−1)
1
!
ln 1+
u
n
u
n n 2
= e0 = 1
u u
lim t1 + t = lim e
n→∞ 2 n→∞
ń)
√ 1 1
n 1
lim eln(2 +n ) lim e n ln(2 +n )
n 2 n 2

lim 2n + n2 n = n→∞
n
lim
n→∞
2n + n2 = n→∞ = n→∞
84
1 n
We will find the limit of the exponent ln 2 + n2 . We will use the fact that when n is large, 2n
n
n2
grows faster than n2 , and thus lim = 0.
n→∞ 2n
1 n 1 √ n n √ √
ln 2 + n2 = n→∞
n n n
lim lim ln 2n + n2 lim ln
= n→∞ 2n + n2 = n→∞
lim ln 2n + n2
n→∞ n n n
 s 
1√ 1 n

n n
= lim ln 2 · 2n + n2 = lim ln 2 · (2 + n2 )
n→∞ 2 n→∞ 2n
 s   s 
n 2n n2  n n2
lim ln 2 ·
= n→∞ + lim ln 2 ·
= n→∞ 1 + n  = ln(2 · 1) = ln 2
2n 2n 2
So we get
√ 1
2n + n2 = lim e n ln(2 +n ) = eln 2 = 2
n n 2
lim
n→∞ n→∞
o)
 s 
√
n

1√

n 1 n
lim 3n + 5n + 7n = n→∞
lim 7 · n 3n + 5n + 7n = n→∞
lim 7 · (3 + 5n + 7n )
n→∞ 7 7n
 
 s  v
3 n
n n u n n
5 7 u 3 5
 
n
 
= lim 7 · + +  = lim 7 · u + +1
n→∞ 7 7 7 n→∞  t| 7
n
u
{z
7 } 

0 when n goes to ∞
=7·1=7
ó)
s  s 
1 n
n n n n n
n 1 1 1 1 1 1
lim + + lim  · 2 n
= n→∞ + + 
n→∞ 2 3 4 2 2 3 4
 s   s 
n n n n n n
1 n 1 1 1 1 n 2 2 2
lim  ·
= n→∞ 2n + +  = lim  · + + 
2 2 3 4 n→∞ 2 2 3 4
 
v
n n 
1 2 2  1 1
 u
· ·1=
u
= lim  u1 + +  =
n→∞ 2 n
3 4 2 2
u 
t | {z }
p) We are interested in finding

1 1 1

lim n + + · · · + .
n→∞ n2 + 1 n2 + 2 n2 + n
85
We put this on a common denominator. The smallest common denominator is the product of all
n
the denominators. The highest power of n in this common denominator will be n2 = n2n with
coefficient
n−1
1. In the numerator one gets a sum of n terms, the highest power of n in each of the terms
2
is n = n2n−2 with coefficient 1. And there is a factor of n outside the brackets. So the highest
power of n in the numerator is n · n · n2n−2 = n2n with coefficient 1. Dividing the numerator and the
denominator by n2n , everything except the coefficient 1 in the numerator and the coefficient 1 in the
1
denominator becomes zero. Thus the limit is equal to = 1.
1
q) We are interested in finding
1 2 n

lim 2
+ 2 + ··· + 2 .
n→∞ n +1 n +2 n +n
We put this on a common denominator. The smallest common denominator is the product of all
n
the denominators. The highest power of n in this common denominator will be n2 = n2n with
coefficient
n−1
1. In the numerator one gets a sum of n terms, the highest power of n in each of the terms
2
is n = n2n−2 with coefficient n. We know that
n(n + 1) n2 + n
1 + 2 + 3 + ··· + n = = .
2 2
1
So in the numerator the highest power of n is n2 · n2n−2 = n2n with coefficient . Dividing the
2
2n 1
numerator and the denominator by n , everything except the coefficient in the numerator and
2
1
2 1
the coefficient 1 in the denominator becomes zero. Thus the limit is equal to = .
1 2
r) From questions 8 and 9 from Temat 3 we know that
(n + 1)4 − 1 − n(n + 1)(2n + 1) − 2n(n + 1) − n

13 + 23 + 33 + · · · + n3 =
4
and that in a similar expression for
14 + 24 + 34 + · · · + n4
1
the highest power of n would be n5 with coefficient . Replacing n by n + 1 does not change the
5
highest power of n or its coefficient. Thus
1
14 + 24 + 34 + · · · + n4 5
lim = 1 = 1.
n→∞ 14 + 24 + 34 + · · · + n4 + (n + 1)4 5
s) We are interested in finding
1! + 3! + 5! + · · · + (2n − 1)!
lim
n→∞ 2! + 4! + 6! + · · · + (2n)!
There is a one to one correspondence between the terms in the numerator and the terms in the
denominator. We compare 1! and 2!, then 3! and 4!, and so on up until (2n − 1)! and 2n!. The
86
corresponding term in the denominator is bigger, by a factor 2, 4, and so on up until 2n. So on
average the factor is
2 n(n+1)
Pn
2 ni=1 i
P
i=1 2i 2
= = = n + 1.
n n n
Moreover, the bigger the term, the bigger the factor. Thus
1! + 3! + 5! + · · · + (2n − 1)! 1
lim ¬ lim = 0.
n→∞ 2! + 4! + 6! + · · · + (2n)! n→∞ n + 1
Thus
1! + 3! + 5! + · · · + (2n − 1)!
lim =0
n→∞ 2! + 4! + 6! + · · · + (2n)!
n
1

ś) We are interested in determining lim 1− . We will use the fact that
n→∞ n
n
x

lim 1+ = ex
n→∞ n
With x = −1, we get n n
1 −1 1

lim 1− = lim 1+ = e−1 =
n→∞ n n→∞ n e
2n−1
1

t) We are interested in determining lim 1− . We will use the fact that
n→∞ n2
n
x

lim 1+ = ex
n→∞ n
Then we also have 2
x n

lim 1 + 2 = ex
n→∞ n
The −1 after the 2n in the exponent is negligibly small in proportion. With x = −1, we get
2n−1 2n 2n 2 n2
1 1 −1 −1

n
lim 1− 2 = n→∞
lim 1− 2 = n→∞
lim 1+ 2 = n→∞
lim 1+ 2
n→∞ n n n n
n2 ! n2 1
−1 2 1 1

n
−1 n −2 n
= lim 1+ 2 = lim e = lim e = lim =1
n→∞ n n→∞ n→∞ n→∞ e2
2n+1
1

u) We are interested in determining 1 + n . We will use the fact that
2
n
x

lim 1 + = ex
n→∞ n
Then we also have 2n
x

lim 1+ n = ex
n→∞ 2
With x = 1, we get
2n+1 2·2n 2n !2 2n !2
1 1 1 1
2
lim 1+ n = lim 1+ n = lim 1+ n = lim 1+ n = e1 = e2
n→∞ 2 n→∞ 2 n→∞ 2 n→∞ 2
87
2n+1
n−1

v) We are interested in determining n→∞
lim . We will use the fact that
n+3
n
x

lim 1 + = ex
n→∞ n
The +3 in the denominator becomes negligibly small in proportion after taking the n out of the
numerator. The +1 in the exponent is negligibly small in proportion. With x = −4, we get
2n+1 2n+1 2n
n−1 −4 −4

lim = n→∞
lim 1+ = n→∞
lim 1+
n→∞ n+3 n+3 n
n 2 n 2
−4 −4 1
2
= n→∞
lim 1+ = lim 1 + = e−4 = e−8 =
n n→∞ n e8
w) We are interested in finding the limit of the sequence

!(−1)n n
(−1)n
an = 1 + .
n
We will use the fact that n
x

lim 1+ = ex .
n→∞ n
With x = 1, we get n
1

lim 1 + = e.
n→∞ n
We need to check whether the limit of the sequence
−n
−1

an = 1 +
n
is also e.
With x = −1, we get n
−1 1

lim 1+ = .
n→∞ n e
Thus −n n −1 n −1 −1
−1 −1 −1 1

lim 1+ = lim 1+ = lim 1+ = = e.
n→∞ n n→∞ n n→∞ n e
So we have !(−1)n n
(−1)n
lim 1+ = e.
n→∞ n
!2n2 −n
n2 − 1
x) We are interested in determining lim . We will use the fact that
n→∞ n2
n
x

lim 1 + = ex
n→∞ n
Then we also have n2
x

lim 1+ = ex
n→∞ n2
88
With x = −1, we get
!2n2 −n 2n2 2n2
2n2 −n −1 −1
n2 − 1

−1 1+ n2
1+ n2
lim = n→∞
lim 1+ 2 = n→∞
lim n = n→∞
lim 1 n2
n→∞ n2 n 1+ −1 −1 n
n2 1+ n2
n2 2
−1
1+ n2 (e−1 )
2
(e−1 ) e−2
2
1
= lim n2 n1
= lim 1 = −1 0 = = e−2 = 2
n→∞
−1
n→∞
(e−1 ) n (e ) 1 e
1+ n2
y) Will will use the binomial theorem.

√ !n  1
n  1
n Pn n i 1 n−i
i=0 i 1 2
n
1+ 2 1+2 
n 1+2 
n n
lim = n→∞
lim  = n→∞
lim  = n→∞
lim Pn n
n→∞ 2 2 1+1 i=0 1i 1n−i i
1
n−0 n
1
n− n 1
n−n
n n n
0
10 2 n + ··· + n 1 2 2n 2
+ ··· + n
1n 2 n
2
= n→∞
lim
n

n n n

n
0
10 1n−0 + · · · + n 1 2 1n− 2 + · · · + n
1n 1n−n
2
1
n 1
n 1
0
n n 2 n
0
2n + ··· + n 2n + ··· + n
2n
2
= lim
n

n

n
n→∞
0
+ ··· + n + ··· + n
2
n
n
0
1

n n n n n n
0
2n + · · · + n 2 2n + · · · + n
2n 0
21 + · · · + n 22 + · · · + n
20
2 2
= lim
n

n

n
= n→∞
lim
n

n

n
n→∞
0
+ ··· + n + ··· + n 0
+ ··· + n + ··· + n
2 2
√
n n n
0
21 + · · · + n 2 + ··· + n
20 √
2
= lim
n

n

n
= 2
n→∞
0
+ ··· + n + ··· + n
2
To see what happens in the last expression, we will look at an example with finite numbers, then
one should make abstraction from the finiteness and get the picture of what would happen if n was
infinite. We put n = 10 and consider the sequence
! ! 4 6 ! 3 7 ! 2 8 ! 1 9 ! 0 10
10 5 10 2 10 + 2 10 10 2 10 + 2 10 10 2 10 + 2 10 10 2 10 + 2 10 10 2 10 + 2 10
2 10 , , , , ,
5 6 2 7 2 8 2 9 2 10 2
! !
10 10
In five steps, the choose function moves from = 252 to = 1. The factor thereafter moves
5 10
0 10
5 1 √ 2 10 + 2 10 1+2
in five steps from 2 10 = 2 2 = 2 to = = 1.5, which is a move of less than 0.1. Now
2 2
we consider the binomial distribution as in the following picture, but for much larger n.
89
!
n
Just like in the example for n = 10, for n infinite there will be a change in the coefficient and
k
k n−k
2n + 2 n
a corresponding change in the function . As n gets bigger, the ratio of hight divided by
2
base in the graph gets bigger, and the curve will get infinitely steep. So the dominant factor is the
√
!
n
one with coefficient n , which is 2.
2
Fn+1
z) We are interested in finding the limit of the sequence an = , where Fn is the n-th Fibonacci
Fn
number By question 7 from Temat 5, we know that
" √ !n √ !n #
1 1+ 5 1− 5
Fn = √ − .
5 2 2
√
1− 5
Since < 1 , we have
2
√ n+1 √ n+1 √ n+1 √ n+1 √ n+1
√1 1+ 5
− 1− 5
1+ 5 1− 5 1+ 5
5 2 2 2
− 2 2
lim
n→∞
h √ n √ n i = n→∞
lim √ n √ n = n→∞
lim √ n
√1 1+ 5
− 1− 5 1+ 5
− 1− 5 1+ 5
5 2 2 2 2 2
√ !n+1−n √ √
1+ 5 1+ 5 1+ 5
= n→∞
lim = n→∞
lim =
2 2 2
1 5

ź) We are interested in finding the limit of the sequence a1 = 1, an+1 = an + . If the
2 an
sequence has a limit, this limit needs to satisfy the recurrence relation. We first assume that the
sequence has a limit, and call this limit x. Then
1 5

x= x+
2 x
5
2x = x +
x
2x2 = x2 + 5
x2 = 5
√
x=± 5
√
We only√keep the positive solution, so x = 5. So far, all we know is that the limit, if it√exists,
must be 5. Now we will show that starting with n = 2, the sequence is bounded below by 5. We
90
1+5 √ √ √
have a2 = = 3 > 5. Now we will show that if an > 5, then an+1 > 5. We assume that
√ 2
an = 5 + for some > 0.
 √ 2   √ 2 
√ 5 + 5 + + 5
!
1 5 1 5  1
an+1 = 5++ √ =  √ +√ =  √ 
2 2 2

5+ 5+ 5+ 5+
√ ! √ ! √ 2
15 + 2 5 + 2 + 5 1 10 + 2 5 + 2 5 + 5 + 2
= √ = √ = √
2 5+ 2 5+ 5+
√ 2  
√ 5 + + 2√5 √ 2 √
= 5· √ = 5  1 + √ √  > 5
5+ 2 5 5+
Furthermore,
 
√ 2 √ 2 √
5  1 + √ √  = 5 + √ = 5 + · √
2 5 5+ 2 5+ 2 5+
√ √ 1 √
= 5+· √ = 5+· √
2 5
< 5+
2 5 + 2
+2
√ √
So for n 2, we have an+1 > 5 and an+1 < an . Hence the limit exists and is equal to 5.
1
ż) We are interested in finding the limit of the sequence a1 = 1, an+1 = 1 + . If the sequence
an
has a limit, this limit needs to satisfy the recurrence relation. We first assume that the sequence has
a limit, and call this limit x. Then
1
x=1+
x
x2 = x + 1
x2 − x = 1
1 1
x2 − x + − =1
| {z 4} 4
complete square
2
1 1

x− − =1
2 4
2
1 5

x− =
2 4
√
1 5
x− =±
2 2
√ √
1 5 1± 5
x= ± =
2 2 2
√
1+ 5
We only keep the positive solution, so x = . So far, all we know is that the limit, if it exists,
√ 2
1+ 5
must be .
2
91
Now we will show that |an+2 − an+1 | < |an+1 − an |, which means that consecutive terms get closer
3
and closer together. We have a1 = 1, a2 = 2, and a3 = . So we may assume that 1 < an < 2.
2
1 1 1

?
|an+2 − an+1 | = 1 + 1 − 1+ < 1+ − an = |an+1 − an |
1 + an an an
1 1 1 1

?
1+ 1 − 1+ < 1+ − an we will multiply by 1 +
1 + an an an an
2 2
1 1 1 1

?
1+ +1− 1+ < 1+ − an 1+
an an an an
! !
1 2 1 ? 2 1
1+ +1− 1+ + 2 < 1+ + 2 − an − 1
an an an an an
1 1 ? 2 1
1− − 2 < −an + + 2 we will multiply by a2n
an an an an
?
a2n − an − 1 < −a3n + 2a2n + 1
We assume that 1 < an < 2. It is straightforward to verify that the inequality holds in that interval.
1
Indeed, a2 − a − 1 is a parabola with vertex at a = and it has value 1 for a = 2 and value −1
2
3 2
for a = 1. The other function to consider is −a + 2a + 1. For 1 < a < 2, we have a < 2, and thus
a3 < 2a2 . Hence −a3 + 2a2 > 0 and −a3 + 2a2 + 1 > 1.
So consecutive terms get closer and closer together. This does not yet guarantee that the sequence
√
1+ 5
converges to to the limit . It may have subsequences converging to different limits and conse-
2
cutive terms get closer and closer together. We will show that applying the recurrence relation twice
92
gives rise to same equation to solve to get the limit.
1 1
1+ 1 =x we will multiply by 1 +
1+ x
x
1
1+ +1=x+1
x
1
1+ =x
x
If the recurrence relation is applied three times or more, this also works. If it is applied three times,
the equation to solve to get the limit is
1 1
1+ =x we will multiply by 1 +
1 + 1+1 1 1+ 1
x
x
!
1 1
1+ 1 +1=x 1+ 1
1+ x
1+ x
!
1 1 1
1 +2=x 1+ 1 we will multiply by 1 +
1+ x
1+ x
x
2 1

1+2+ =x 1+ +1
x x
2 1

+3=x +2
x x
2
+ 3 = 1 + 2x
x
2
+ 2 = 2x note that 2 is the Fibonacci number F3
x
1
1+ =x
x
We use the Fibonacci numbers (see question 7 from Temat 5). Applying the recurrence relation m
times and solving the equation for the limit, as in the second last line in the previous calculation,
one obtains
Fm
+ Fm = Fm x
x
which is equivalent to
1
1 + = x.
x
√
1+ 5
Hence the limit is .
2
Academic Year 2015/16, Colloquium I, 20 November, 2015

Question 1. Find a simple expression for the following sum:
Sn = 2 · 6 + 7 · 11 + · · · + (5n − 3) · (5n + 1).
93
Answer:
Sn = 2 · 6 + 7 · 11 + · · · + (5n − 3) · (5n + 1)
n n
(25i2 − 10i − 3)
X X
= (5i − 3) · (5i + 1) =
i=1 i=1
n n n n n n
2 2
X X X X X X
= 25i + (−10i) + (−3) = 25 i − 10 i−3 1
i=1 i=1 i=1 i=1 i=1 i=1
n(n + 1)(2n + 1) n(n + 1)

= 25 · − 10 · −3·n
6 2
25n(n + 1)(2n + 1) − 30n(n + 1) − 18n
=
6
n[25(n + 1)(2n + 1) − 30(n + 1) − 18]
=
6
n[(n + 1)[25(2n + 1) − 30] − 18]
=
6
n[(n + 1)(50n − 5) − 18]
=
6
Question 2. Find a simple expression for the following sum:

1 1 1
Sn = + + ··· + .
2 · 7 7 · 12 (5n − 3) · (5n + 2)
Answer:
5 5 5
5 · Sn = + + ··· +
2 · 7 7 · 12 (5n − 3) · (5n + 2)
7 − 2 12 − 7 5n + 2 − (5n − 3)
= + + ··· +
2·7 7 · 12 (5n − 3) · (5n + 2)
1 1 1 1 1 1

= − + − + ··· + −
2 7 7 12 5n − 3 5n + 2
! ! !
1 1 1 1 1 1
= − + − + ··· + −
2 7 7 12 5n − 3 5n + 2

1 1
= −
2 5n + 2
1 1 1 1 5n + 2 − 2 n

Sn = − = =
5 2 5n + 2 5 10n + 4 10n + 4
! ! !
51 51 51
Question 3. Calculate the value of the following expression: + + ··· +
0 1 25
94
Answer: Let
! ! ! !
51 51 51 51
S= + + ··· + +
0 1 50 51
! ! ! !
51 51 51 51
= 1·1+ 1 · 1 + ··· + 1·1+ 1·1
0 1 50 51
! ! ! !
51 51 0 51 50 1 51 1 50 51 0 51
= 1 1 + 1 1 + ··· + 11 + 11
0 1 50 51
= (1 + 1)51 = 251
! ! ! ! ! !
51 51 51 51 51 51
Since = , = , = , and so on, we have
0 51 1 50 2 49
! ! !
51 51 51 1 1
+ + ··· + = S = · 251 = 250 .
0 1 25 2 2
Question 4. In a class of 31 students, each passed at least two of the three exams in microeconomics,
macroeconomics and linear algebra. It is known that 25 students passed microeconomics, 28 students
passed macroeconomics and 29 students passed linear algebra. How many students passed all three
exams?
Answer: If we take the sum of 25, 28, and 29, we are counting all students, but we are over-counting.
We are counting every student at least twice, and we are counting those students who passed all
three exams three times. Let x be the number of students who passed all three exams. From
25 + 28 + 29 − x
= 31,
2
we get x = 20.
1 1
Question 5. Calculate the value of the expression x4 + 4
given that x − = 3.
x x
Answer: We have
2
1 1 1 1

32 = 9 = x − = x2 − 2x · + 2 = x2 − 2 + 2
x x x x
2
1 1 1

2 2 2 2
9 = 81 = x − 2 + 2 = x −2+ 2 x −2+ 2
x x x
x2 2 x2 2 1
= x4 − 2x2 + 2
− 2x 2
+ 4 − 2
+ 2
− 2+ 4
x x x x x
2 2 1
= x4 − 2x2 + 1 − 2x2 + 4 − 2 + 1 − 2 + 4
x x x
4 1
= x4 − 4x2 + 8 − 2 − 2 + 4
x x
 
1 1
= x4 − 4  2

x − 2 +
−2+
2
x}  x4
| {z
=9
95
1
= x4 − 36 − 2 +
x4
1
x4 + = 81 + 36 + 2 = 119
x4
Question 6. Solve the following inequality:
|x2 − 6x + 7| ¬ (x − 3)2 .
Answer: It is useful to rewrite the inequality as
(x − 3)2 − 2 ¬ (x − 3)2 .
So we have twice the same parabola, but on the left side of the inequality it is lowered by 2 and we
take the absolute value. It is useful to make a sketch
Clearly, the intersection occurs when

(x − 3)2 − 2 = − (x − 3)2 − 2 = −(x − 3)2 + 2
So we need to solve
−(x − 3)2 + 2 = (x − 3)2

2(x − 3)2 = 2
(x − 3)2 = 1
x − 3 = ±1
x = 3 ± 1 = 2 or 4
The red curve needs to be below the blue curve or on the blue curve. So the solution is
(−∞, 2] ∪ [4, ∞).
96
Round brackets mean that the limit is excluded, square brackets mean that the limit is included.
Question 7. Solve the following inequality:
cos2 x < sin2 x.
Answer: We have sin2 + cos2 = 1 . So the inequality becomes
1 − sin2 < sin2 x

1 < 2 sin2 x
1
< sin2 x
2
1 1 π 3π 1 5π 7π 1
So either sin x > √ or sin x < − √ . Since sin = sin = √ , and sin = sin = −√ ,
2 2 4 4 2 4 4 2
from the shape of the sin curve, we get the solution
π 3π

x ∈ ∪k∈Z + kπ, + kπ .
4 4
Round brackets mean that the limit is excluded from the interval.
Question 8. On the first day the value of a stock grew by 10%, on the second day the value dropped
10%, on the third day it again grew 10%, on the fourth day it again declined 10%. What was the
percentage change of the value of this stock after the forth day.
Answer: Let V0 be the initial value of the stock, and let V4 be the value of the stock after the fourth
10
day. Growing by 10% means that we need to multiply by 1 + = 1.1, and dropping by 10% means
100
10
100
1.99

V4 = V0 · (1.1) · (0.9) · (1.1) · (0.9) = 0.9801 · V0 = (1 − 0.0199) · V0 = 1 − · V0 .
100
So after the fourth day, the the value of the stock has dropped by 1.99%. The percentage change is
−1.99%.
Question 9. Solve the following equation for x: log3 (3x − 8) = 2 − x.

Answer: By definition, log3 (3x − 8) is the number we need to put in the exponent of 3 to get 3x − 8.
So
log3 (3x − 8) = 2 − x
x −8)
3log3 (3 = 32−x
3x − 8 = 32−x
3x − 32−x = 8.
97
We try small integers for x, and doing so, we notice that
32 − 32−2 = 9 − 30 = 9 − 1 = 8.
So x = 2.
Question 10. Solve the inequality: x3 + 2x2 − 5x − 6 > 0.

Answer: It is useful to first solve the equation x3 + 2x2 − 5x − 6 = 0. When looking for the roots
of a polynomial of degree 3, it is useful to try small integers for x. We find that x = 2 is a solution.
So x3 + 2x2 − 5x − 6 has a factor x − 2. We need to take out this factor, this is done by dividing
x3 + 2x2 − 5x − 6 by x − 2. We get
x3 + 2x2 − 5x − 6 = (x − 2)(x2 + 4x + 3) = 0.
To find the two remaining roots, we have to solve x2 + 4x + 3 = 0. We get
x2 + 4x + 3 = 0
(x + 2)2 − 4 + 3 = 0
(x + 2)2 = 1
x + 2 = ±1
x = −2 ± 1 = −1 or − 3
So the original inequality becomes
x3 + 2x2 − 5x − 6 = (x − 2)(x2 + 4x + 3) = (x − 2)(x − (−1))(x − (−3)) = (x − 2)(x + 1)(x + 3) > 0.
When x tends to −∞, then x3 + 2x2 − 5x − 6 tends to −∞, and when x tends to +∞, then
x3 + 2x2 − 5x − 6 tends to +∞. So the solution is
(−3, −1) ∪ (2, +∞).
Round brackets mean that the limit is not included in the interval.
98
q √ √
Question 11. Let an = n + 3 n − n. Calculate lim an .
n→∞
Answer:
q √ √
q
√ √ n+3 n− n
lim n + 3 n − n = lim
n→∞ n→∞ 1
√ √ √ √
q q
n+3 n− n n−3 n+ n
= n→∞
lim q √ √
n+3 n+ n
√ √ 2
q 2
n + 3 n − ( n) √
n+3 n−n
= n→∞
lim q √ √ = n→∞
lim q √ √
n+3 n+ n n+3 n+ n
√
3 n 3
= lim q √ √ = lim q √
n→∞
n + 3 n + n n→∞ √1n n + 3 n + 1
3 3 3 3
= lim q √ = lim q =√ =
n→∞ 1
(n + 3 n) + 1 n→∞
1+ √3 +1 1+1 2
n n
√
q 4
1

nπ
n+8 n
Question 12. Let an = sin + lim an .
√ 2 . Calculate n→∞
n 6 (n + 3 n)
Answer: The first thing to notice is that since the sin function oscillates between 1 and −1, but the
1 1 nπ

limit of when n tends to infinity is zero, lim sin = 0. Thus
n n→∞ n 6
√ 4 √ 4
 q   q 
1 nπ
n+8 n 
1

nπ
 n+8 n 
lim  sin + √ = lim sin + lim √ 2 
   
n→∞  n 2  
6 (n + 3 n) n→∞ n 6 n→∞  (n + 3 n) 
√
 q 4 
 n+8 n 
= 0 + lim  √ 2
 

n→∞  (n + 3 n) 
 √ 2
(n + 8 n)
= lim  √ 
n→∞ (n + 3 n)2
In the last expression, the highest power of n in the numerator is n2 with coefficient 1, and the
highest power of n in the denominator is n2 with coefficient 1. Dividing the numerator and the
denominator by the highest power of n, which is n2 , gives
 √ 2
(n + 8 n)  1
lim  √ 2 = lim =1
n→∞ (n + 3 n) n→∞ 1
99
√ √
Question 15. Let α > 0, a1 = α and an = α + an−1 when n ∈ N and n 2. If S = {an : n ∈ N}
then find sup S and inf S.
√ √q
Answer: S is monotone increasing. Clearly, a2 > a1 , since α + α > α. Now we argue by
induction. For positive numbers β and γ, β > γ if and only if β 2 > γ 2 . We have
√ 2
a2n+1 = α + an = α + an > α + an−1 = a2n
√ √
So the smallest element in S is a1 = α, which means that inf S = α.
To find the least upper bound, we first assume that a least upper bound exists, which in the context
of an increasing sequence means that the limit of an for n tending towards infinity exists and is finite.
If this limit exists, we let it be x, and x must satisfy the recurrence relation. Thus
√
x= α+x
x2 = α + x
x2 − x = α
1 1
x2 − x + − =α
4 4
2
1 1

x− − =α
2 4
2
1 1

x− =α+
2 4
s
1 1
x− =± α+
2 4
s
1 1
x= ± α+
2 4
s
1 1
We only keep the solution which is positive, so x = + α + . So far we only know that the
2 4 s
1 1
sequence is monotone increasing, and that if the limit exists, it is equal to x = + α + . Now
2 4
s s
1 1 1 1
we assume that there exists n0 such that an0 > + α + , say an0 = + α+ + for some
2 4 2 4
> 0. Then s
√ 2 1 1
a2n0 +1 = ( α + an0 ) = α + an0 = α + + α+ +
2 4
and
 s 2
1 1
a2n0 = + α + + 
2 4
s s 2
1 1 1
= + α + + +  α + + 
4 4 4
100
s s 2 s 
1 1 1 1
= + α + + +  α +  + 2  α +  + 2
4 4 4 4
s s
1 1 1 1
= + α + + + α + + 2 · α+ + 2
4 4 4 4
s s
1 1 1
=α+ + α + + + 2 · α+ + 2
2 4 4
s
1
= a2n0 +1 + 2 · α+ + 2 > a2n0 +1
4
So a2n0 > a2n0 +1 , which implies an0 > an0 +1 , in contradiction with the fact that the sequence is
increasing. Hence the sequence is bounded above and
s
1 1
sup S = x = + α+
2 4
Academic Year 2016/17, Colloquium I, 1 December, 2016

Question 1. Give the formal definitions for the following expressions:
(i) ex where x ∈ R
(ii) π x where x ∈ R
(iii) loga (x) where a and x are positive real numbers.
Answer:
n
x

x
(i) e = n→∞
lim 1+
n
x
(ii) π x = eln π = ex ln π
(iii) loga (x) is the number y such that ay = x. So
ln (ay ) = y ln a = ln x
and hence
ln x
loga (x) = y =
ln a
Question 2. Find a rational expression (fraction) for the following sum:
n
X 1
.
k=1 (6k − 5)(6k + 1)
Answer:
n
X 1 1 1 1 1
S= = + + + ··· +
k=1 (6k − 5)(6k + 1) (1)(7) (7)(13) (13)(19) (6n − 5)(6n + 1)
101
6 6 6 6
6S = + + + ··· +
(1)(7) (7)(13) (13)(19) (6n − 5)(6n + 1)
7−1 13 − 7 19 − 13 6n + 1 − (6n − 5)
= + + + ··· +
(1)(7) (7)(13) (13)(19) (6n − 5)(6n + 1)
1 1 1 1 1 1 1 1

= − + − + − + ··· + −
1 7 7 13 13 19 6n − 5 6n + 1
! ! ! !
1 1 1 1 1 1 1 1
= − + − + − + ··· + −
1 7 7 13 13 19 6n − 5 6n + 1

1 1 6n + 1 − 1 6n
= − = =
1 6n + 1 6n + 1 6n + 1
6S 6n n
S= = =
6 6(6n + 1) 6n + 1
Question 3. Find the maximum value of −9x2 +6x+20 and the value/values of x where the maximum
is attained.
Answer:
complete square
z }| {
−9x2 + 6x + 20 = −(9x2 − 6x − 20) = −(9x2 − 6x + 1 −1 − 20)
h i
2
= −(9x
|
−{z6x + 1} −21) = − (3x − 1)2 − 21 = −(3x − 1)2 + 21
complete square
1
So the maximum value is 21 and it is attained at x = .
3
Question 4. Find all values x such that cos2 (x) − sin2 (x) > 0.
Answer:

cos2 (x) − sin2 (x) = cos2 (x) − 1 − cos2 (x) = 2 cos2 (x) − 1 > 0
2 cos2 (x) > 1

1
cos2 x >
2
1 −1
cos x > √ or cos x < √
2 2
102
π −π 1 3π 5π −1
We know that cos = cos = √ , and we also have cos = cos = √ . So the solution is
4 4 2 4 4 2
π π

x ∈ ∪k∈Z − + kπ, + kπ
4 4
Round brackets mean that the limit is excluded from the interval.
Question 5. A fund offers a nominal interest rate of 6% per annum compounded every four months.
What is the effective interest rate?
Answer: Let A0 be the original amount deposited and A1 be what this amount has become after
one year.
6
!3 3
0.06

A1 = 1 + 100
A0 = 1 + A0 = (1 + 0.02)3 A0 = (1.02)3 A0
3 3
6.1208

= 1.061208 · A0 = (1 + 0.061208) A0 = 1 + A0
100
So the effective interest rate is 6.1208%.
Question 6. Calculate the limit (possibly improper)

−n
2016

lim 1−
n→∞ n
n
x

Answer: We know that lim 1+ = ex . We will use this with x = −2016.
n→∞ n
−n n −1 n −1
2016 2016 −2016

lim 1− = lim 1− = lim 1+
n→∞ n n→∞ n n→∞ n
n −1
−2016
−1
= lim 1+ = e−2016 = e(−1)(−2016) = e2016
n→∞ n

!
1 2n + 1
lim
n→∞ n 2n
Answer:
!
1 2n + 1 1 (2n + 1)! 1 (2n + 1)! 1 (2n + 1)!
lim = lim · = lim · = lim ·
n→∞ n 2n n→∞ n 2n!(2n + 1 − 2n)! n→∞ n 2n!1! n→∞ n 2n!
1 (2n + 1)(2n)(2n − 1) · · · 2 · 1 2n + 1 1

= lim · = lim = lim 2 + =2
n→∞ n (2n)(2n − 1)(2n − 2) · · · 2 · 1 n→∞ n n→∞ n
103
1 + sin((your age)n )
lim
n→∞ ln n
Answer: The sin function has always a value between 1 and −1, so the numerator is always between
0 and 2. But the logarithm in the denominator grows towards infinity. So the limit is zero.
Question 9. Calculate the following limits (possibly improper) for the two cases α = n and α = n2
!α
n2 + 1
lim
n→∞ n2 − 1
2
x n x n

Answer: We know that n→∞ lim 1 + = ex . So we also have n→∞
lim 1 + 2 = ex . We will use
n n
this with x = 2. The −1 in the denominator becomes negligibly small after taking the n2 in the
numerator out of the fraction. For α = n, we get
!n ! 1 n2  !n2  n1
n2 + 1 n2 + 1 n
n2 + 1
lim = lim = lim  2 
n→∞ n2 − 1 n→∞ n2 − 1 n→∞ n −1
n2 ! n1 n2 ! n1
2 2 1

= lim 1+ 2 = lim 1+ 2 = lim e2 n
=1
n→∞ n −1 n→∞ n n→∞
For α = n2 , we get
!n2 n2 n2
n2 + 1 2 2

lim = lim 1+ 2 = lim 1+ 2 = e2
n→∞ n2 − 1 n→∞ n −1 n→∞ n
Question 10. Find an explicit numerical value of x such that

n
!!
1 X n
ln(x) = ln
n k=0 k
Answer: By the binomial theorem, we know that

n n
! !
n n
X n k n−k X n
2 = (1 + 1) = 1 1 =
k=0 k k=0 k
Thus
n
!!
1 n 1 1 n 1 n ln 2
ln (2n ) = ln eln 2 = ln en ln 2 =
X
ln(x) = ln = = ln 2
n k=0 k n n n n
and hence
x = eln x = eln 2 = 2
104
Temat 7: Series.
X
Question 1. Prove that if the series an converges to a real limit, then lim an = 0.
X
Answer: If the series an converges to a real limit, we have
∞
X k
X
an = lim an = L
k→∞
n=1 n=1
for a real number L. Since we also have

∞
X k+1
X
an = lim an = L
k→∞
n=1 n=1
we get
k+1
X k
X
lim an = lim an = lim an − lim an = L − L = 0
n→∞ k→∞ k→∞
n=1 n=1
Question 2. Calculate, assuming that |x| < 1, a 6= 0:

∞ ∞ ∞
axn naxn n2 axn
X X X
a) b) c)
n=1 n=1 n=1
∞ k
axn = lim axn and we put
X X
Answer: a) We have
k→∞
n=1 n=1
k
axn = ax + ax2 + ax3 + · · · + axk−1 + axk .
X
Sk =
n=1
Then
h i
(1 − x)Sk = (1 − x) ax + ax2 + ax3 + · · · + axk−1 + axk
h i h i
= ax + ax2 + ax3 + · · · + axk−1 + axk − ax2 + ax3 + ax4 + · · · + axk + axk+1
h i h i
k−1 k k
ax2 +
= ax + ax3 + · · · +
ax ax
+ ax2 +
− ax3 +
ax4 + · · · +
ax + axk+1

= ax − axk+1
(1 − x)Sk ax − axk+1
Thus Sk = = , and because |x| < 1, we get
1−x 1−x
∞ k
ax − axk+1 ax
axn = lim axn = lim Sk = lim
X X
= .
n=1
k→∞
n=1
k→∞ k→∞ 1−x 1−x
∞
X ∞
X
b) We will use the Cauchy product. Let αi and βj be two infinite series. The Cauchy product
i=1 j=1
of these two infinite series is defined as follows:
∞
! ∞  
∞

k
X X X X
α i ·  βj  = γk  where γk = αl βk−l+1
i=1 j=1 j=k l=1
105
If we put αi = xi and βj = axj , we get
1
xl axk−l+1 = x · ax = ax2
X
γ1 =
l=1
2
xl axk−l+1 = x · ax2 + x2 · ax = 2x3
X
γ2 =
l=1
3
xl axk−l+1 = x · ax3 + x2 · ax2 + x3 · ax = 3ax4
X
γ3 =
l=1
..
.
n
xl axk−l+1 = x · axn + x2 · axn−1 + · · · + xn−1 · ax2 + xn · ax = naxn+1
X
γn =
l=1
∞
ax
axn =
X
From part a) we know that . So
n=1 1−x
∞ ∞ ∞
!  
1 1 x ax ax
naxn = · xi ·  axj  = ·
X X X
· =
n=1 x i=1 j=1 x 1−x 1−x (1 − x)2
∞
X ∞
X
c) We will use the Cauchy product. Let αi and βj be two infinite series. The Cauchy product
i=1 j=1
of these two infinite series is defined as follows:
∞
! ∞  
∞

k
X X X X
α i ·  βj  = γk  where γk = αl βk−l+1
i=1 j=1 j=k l=1
If we put αi = xi and βj = naxj , we get

1
xl naxk−l+1 = x · nax = nax2
X
γ1 =
l=1
2
xl naxk−l+1 = x · nax2 + x2 · nax = 2nx3
X
γ2 =
l=1
3
xl naxk−l+1 = x · nax3 + x2 · nax2 + x3 · nax = 3nax4
X
γ3 =
l=1
..
.
n
xl naxk−l+1 = x · naxn + x2 · naxn−1 + · · · + xn−1 · nax2 + xn · nax = n2 axn+1
X
γn =
l=1
∞ ∞
ax ax
axn = naxn =
X X
From part a) we know that and from part b) we know that .
n=1 1−x n=1 (1 − x)2
So !  
∞ ∞ ∞
1 1 x ax ax
n2 axn = · xi ·  naxj  = ·
X X X
· =
n=1 x i=1 j=1 x 1 − x (1 − x)2 (1 − x)3
106
X
Question 3. Examine the convergence of the series an , where :
√
3n + 2 n−1 1
a) an = 2 b) an = √ 2 c) an = √
5n − n + 1 2 n +1 n n2 + n − n
√ √
n+2− n n10 1000n
d) an = √ e) an = n f) an =
n+1 10 n!
(n!)2 (n!)2 · 3n 2n · n!
g) an = h) an = i) an =
(2n)! (2n)! nn
n2
n n+1 !n
2n + 2 2 + (−1)n

4
n
j) an = k) an = l) an =
3n + 2 4n π
Answer: a) We are examining the convergence of the series

∞
X 3n + 2
n=1 5n2−n+1
We will use the limit comparison test. We know that the series
∞
X 1 1 1 1
= 1 + + + ···
n=1 n 2 3 4
diverges. We have
3n+2
5n2 −n+1 3n2 + 2n 3 + n2 3
lim 1 = lim = lim =
n→∞ n→∞ 5n2 − n + 1 n→∞ 5 − 1 + 1
5
n n n2
∞
X 1
Since the limit exists and is different from zero, from the the fact that diverges it follows that
n=1 n
∞
X 3n + 2
diverges as well.
n=1 5n2−n+1
b) We are examining the convergence of the series
∞ √
X n−1
√
2
n=1 2 n + 1
∞
X 1 1 1 1
= 1 + + + ···
n=1 n 2 3 4
√
diverges. Since for all n, we have n ¬ n, it follows that the series
∞
X 1 1 1 1
√ = 1 + √ + √ + √ ···
n=1 n 2 3 4
diverges as well. We have
√
√n−1
√
2 n− n 1 − √1n 1 − √1n 1 − √1n 1
lim 2 n1 +1 = lim √ 2 = lim 2 √ 2 = lim q = lim q =
n→∞ √
n
n→∞ 2 n + 1 n→∞
n
n + 1 n→∞ 2 12 (n2 + 1) n→∞ 2 1 + 12 2
n n
107
∞
X 1
Since the limit exists and is different from zero, from the the fact that √ diverges it follows
n
∞ √ n=1
X n−1
that √ diverges as well.
2
n=1 2 n + 1
c) We are examining the convergence of the series

∞ ∞
√
X 1 X n2 + n + n
√ = √ √
n=1 n n2 + n − n n=1 n n2 + n − n n2 + n + n
∞
√ ∞
√ ∞
√
X n2 + n + n X n2 + n + n X n2 + n + n
= √ 2 = =
n=1 n (n + n − n )
2 2 n2
n=1 n n2 + n − n2 n=1
∞
X 1 1 1 1
= 1 + + + ···
n=1 n 2 3 4
diverges. We have
√
n2 +n+n
√ s  s 
n2 n2 + n + n 1 1
lim 1 = lim = lim  (n2 + n) + 1 = lim  1 + + 1 = 2
n→∞
n
n→∞ n n→∞ n2 n→∞ n
∞
X 1
Since the limit exists and is different from zero, from the the fact that diverges it follows that
√ n=1 n
∞
X n2 + n + n
diverges as well.
n=1 n2
d) We are examining the convergence of the series
√ √ √ √ √ √ √ 2
√ 2
∞
X n+2− n X ∞ n+2− n n+2+ n X∞ n + 2 − ( n)
√ = √ √ √ = q q
n=1 n+1 n=1 n+1 n+2+ n n=1 (n + 1)(n + 2) + (n + 1)n
∞ ∞
X n+2−n X 2
= √ √ = √ √
n=1 n2 + 3n + 2 + n2 + n n=1 n2 + 3n + 2 + n2 + n
∞
X 1 1 1 1
= 1 + + + ···
n=1 n 2 3 4
diverges. We have
√ 2 √
n2 +3n+2+ n2 +n 2n 2
lim 1 lim √ 2
= n→∞ √ = lim 1
√ √
n→∞
n n + 3n + 2 + n2 + n n→∞ n
n2 + 3n + 2 + n2 + n
2 2
= n→∞
lim q q = lim q q
1
n2
(n2 + 3n + 2) + 1
n2
(n2 + n) n→∞ 1 + 3
n
+ 2
n2
+ 1+ 1
n
2
=√ √ =1
1+ 1
108
∞
X 1
Since the limit exists and is different from zero, from the the fact that diverges it follows
n=1 n
∞
X 2
that √ √ diverges as well.
n=1 n2 + 3n + 2 + n2 + n
e) We are examining the convergence of the series
∞
X n10
n
n=1 10
We have
(n+1)10 10 10
an+1 1 (n + 1)10 1 1 1 1 1

10n+1
lim
n→∞ a
= n→∞
lim n10
= n→∞
lim · 10
= n→∞
lim · (n + 1) = n→∞
lim · 1+ =
n 10n
10 n 10 n 10 n 10
∞
1 X n10
Since < 1, by the ratio test, the series n
converges.
10 n=1 10
f) We are examining the convergence of the series

∞
X 1000n
n=1 n!
We have
1000n+1
an+1 (n+1)! n! n(n − 1) · · · 1 1000
lim = lim 1000n = lim 1000 · = lim 1000 · = lim =0
n→∞ a
n
n→∞
n!
n→∞ (n + 1)! n→∞ (n + 1)n · · · 1 n→∞ n+1
∞
X 1000n
Since 0 < 1, by the ratio test, the series converges.
n=1 n!
g) We are examining the convergence of the series
∞
X (n!)2
n=1 (2n)!
We have
((n+1)!)2 ((n+1)!)2 ((n+1)n···1)2
an+1 (2(n+1))! (2n+2)! (2n+2)(2n+1)···1 (n + 1)2
lim = lim (n!)2
= lim (n!)2 = lim = lim
n→∞ a n→∞ n→∞ n→∞ (n(n−1)···1)2 n→∞ (2n + 2)(2n + 1)
n
(2n)! (2n)! (2n)(2n−1)···1
2 1
n2 + 2n + 1 1+ n
+ n2 1
= n→∞
lim 2
= n→∞
lim 6 2 =
4n + 6n + 2 4+ n
+ n2
4
∞
1 X (n!)2
Since < 1, by the ratio test, the series converges.
4 n=1 (2n)!
h) We are examining the convergence of the series

∞
X (n!)2 · 3n
n=1 (2n)!
an+1 3
In comparison with part g), in n→∞
lim , the only difference will be a factor of 3. Since < 1, by
an 4
109
∞
X (n!)2 · 3n
the ratio test, the series converges.
n=1 (2n)!
i) We are examining the convergence of the series
∞
X 2n · n!
n=1 nn
We have
s s s s
n 2n · n! n n! n n(n − 1) · · · 2 · 1 n n n−1 2 1
lim n
= n→∞
lim 2 n
= n→∞
lim 2 = n→∞
lim 2 · · ·
n→∞ n n n · n···n · n n n n n
We know from Question 16 in Temat 3 that the „arithmetic mean” and a „geometric mean” inequality
states
x1 + x2 + · · · + xn √
n x1 x2 · · · xn ,
n
in fact this is a strict inequality unless x1 = x2 = · · · = xn . Thus
s s
n n−1 2 1
n 2n · n! n n n−1 2 1 n
+ n
+ ··· + n
+ n
lim n
= n→∞
lim 2 · · · < n→∞
lim 2 ·
n→∞ n n n n n n
n + (n − 1) + · · · + 2 + 1 n(n + 1)
= lim 2 · 2
= lim 2 ·
n→∞ n n→∞ 2n2
n+1 1

= lim = lim 1 + =1
n→∞ n n→∞ n
s
∞
n 2n · n! X 2n · n!
Since lim < 1, by the the nth root test we know that the series converges.
n→∞ nn n=1 nn
j) We are examining the convergence of the series
∞ n
2n + 2 4
X
n=1 3n + 2
We have
s
n 1 ·n 1 2
!1 1
2n + 2 2n + 2 2n + 2 2+ 4
2

n 4 n 4 4 4
n
lim = n→∞
lim = n→∞
lim = n→∞
lim 2 = <1
n→∞ 3n + 2 3n + 2 3n + 2 3+ n
3
s
n ∞ n
2n + 2 2n + 2 4

n 4 X
Since lim < 1, by the the nth root test we know that the series
n→∞ 3n + 2 n=1 3n + 2
converges.
k) We are examining the convergence of the series
n2
∞ n+1
X n
n=1 4n
We have
v
u n2 1 ·n2 n n
n+1 n+1 n n+1 1
u
n
t n n n
1+ n e
lim = lim 1 = lim = lim = <1
n→∞ 4n n→∞ 4 n
·n n→∞ 4 n→∞ 4 4
110
v
u n2 n2
u
n
n+1 ∞ n+1
t n X n
Since n→∞
lim < 1, by the the nth root test we know that the series converges.
4n n=1 4n
l) We are examining the convergence of the series

∞
!n
X 2 + (−1)n
n=1 π
As in parts i), j) and k) one could use the nth root test. Only here the limit does not exist and one
would need the lim sup, whereas in parts i), j) and k) the limit exists and is consequently equal to
the lim sup. But we might as well just compute the sum
∞
!n 2 3 4 5 6
X 2 + (−1)n 1 3 1 3 1 3
S= = + + + + + + ···
n=1 π π π π π π π
 
1 3 5  3 2 4 6
1 1 3 3
= 
 + + + · · ·
+ + + + ···
π π π  π π {z π
| {z } | }
S1 S2
2
1 3
π π
= 2 + 2
1 3
1− π
1− π
Since we computed the sum and it is a finite number, the series clearly converges. We will show how
we got S1 , the sum S2 was obtained similarly.
" 3 5 2n−1 #
1 1 1 1
S1 = lim + + + ··· +
n→∞ π π π π
2 ! " 3 5 2n−1 " 3 5 2n+1 ##
1 1 1 1 1 1 1 1
1− S1 = lim + + + ··· + − + + ··· +
π n→∞ π π π π π π π
  
3 5 2n−1
1  1 3
5 2n+1
1 1 1 1 1

lim  +
= n→∞ + + ··· + − + + ··· + 
π π π π π π π
" 2n+1 #
1 1 1
= n→∞
lim − =
π π π
1
π
S1 = 2
1
1− π
∞
X xn
Question 4. Show that ∀x > 0 the following series is convergent .
n=0 n!
Answer:
xn+1
an+1 (n+1)! n(n − 1) · · · 2 · 1 x
lim = xn = lim x = lim =0
n→∞ a
n n!
n→∞ (n + 1)n · · · 2 · 1 n→∞ n + 1
∞
X xn
Hence by the the comparison test, the series converges ∀x > 0.
n=0 n!
111
X X X
Question 5. Find a series an with all an > 0 such that an < +∞ and an ln n = +∞.
We use the series ∞ ∞
X X 1
an = 2
n=1 n=1 n (ln(n + 1))
We have
∞ ∞ ∞
2k 1
2k a2k =
X X X
k k 2 = k 2
k=1 k=1 2 (ln(2 + 1)) k=1 (ln(2 + 1))
∞ ∞ ∞ ∞
X 1 X 1 X 1 1 X 1 π2
< k 2
= 2 = 2 = = < +∞
k=1 (ln(2 )) k=1 (k ln 2)
2
k=1 k (ln 2) (ln 2)2 k=1 k 2 6 (ln 2)2
Hence by the Cauchy condensation test,

∞
X 1
2 < +∞
n=1 n (ln(n + 1))
And we have

∞ ∞ 2k ln 2k ∞ ln 2k ∞ ln 2k ∞ ln 2k
2k a2k ln 2k =
X X X X X
= > =
k=1 k=1 2k (ln(2k + 1))2 k=1 (ln(2k + 1))2 k=1 (ln(2 · 2k ))2 k=1 (ln(2k+1 ))2
∞ ∞ ∞ ∞
X k ln 2 X k ln 2 X k 1 X k
= 2 = 2 2 = 2 =
k=1 ((k + 1) ln 2) k=1 (k + 1) (ln 2) k=1 (k + 1) ln 2 ln 2 k=1 (k + 1)2
∞ ∞ ∞
1 X k 1 X k 1 X 1
> 2 = 2
= = +∞
ln 2 k=1 (2k) ln 2 k=1 4k 4 ln 2 k=1 k
Hence by the Cauchy condensation test,

∞
X ln n
2 = +∞
n=1 n (ln(n + 1))
∞
X X X an
Question 6: Find a series an with all an > 0 such that an = +∞ and < +∞.
n=2 ln(ln n)
We use the series ∞ ∞

X X 1
an =
n=1 n=1 n ln (ln (n + 2)) ln (n + 1)
We have
∞ ∞ k ∞
X
k 2k
X 2k 22 X 2k
2 2 a22k = =
2k 2k + 2 ln 22k + 1 2k + 2 ln 22k + 1
k=1 k=1 2 ln ln 2 k=1 ln ln 2
∞ ∞
X 2k X 2k
> = k+1 ln 2) 2k+1 ln 2
k=1 ln ln 22k+1 ln 22k+1 k=1 ln (2
∞ ∞ ∞ ∞
X 1 X 1 X 1 X 1
= k+1
> k+2
= >
k=1 ln (2 ln 2) 2 ln 2 k=1 2 ln (2 ) k=1 2(k + 2) ln 2 k=1 2(k + 2)
112
∞ ∞
X 1 1X 1
> = = +∞
k=1 6k 6 k=1 k
Hence by double application of the Cauchy condensation test,
∞
X 1
= +∞
n=1 n ln (ln (n + 2)) ln (n + 1)
And we have
∞ ∞ k
X
k 2k a22k X 2k 22
2 2 =
k=1 ln ln 22k k=1 22k ln ln 22k + 2 ln 22k + 1 ln ln 22k
∞
X 2k
=
k=1 ln ln 22k + 2 ln 22k + 1 ln ln 22k
∞
X 2k
<
k=1 ln ln 22k ln 22k ln ln 22k
∞ ∞
X 2k X 1
= k k k
= k k
k=1 ln (2 ln 2) 2 ln 2 ln (2 ln 2) k=1 ln (2 ln 2) ln 2 ln (2 ln 2)
∞
X 1
= k k
k=1 [ln (2 ) + ln (ln 2)] ln (2) [ln (2 ) + ln (ln 2)]
∞
X 1
=
k=1 [k ln 2 + ln (ln 2)] ln (2) [k ln 2 + ln (ln 2)]
∞
1 X 1
= 3 < +∞
2) 2

(ln 2) k=1 k + ln(ln
ln 2
∞
X 1
To see why the last inequality holds, one may use the limit comparison test with the series 2
.
k=1 k
By double application of the Cauchy condensation test,
∞
X 1
< +∞
n=2 n ln (ln (n + 2)) ln (n + 1) ln(ln n)
∞
X
Question 7. Examine the convergence of an :
n=2
n 1 1
a) an = b) an = c) an =
10ln n ln(n!) n · 10ln(ln n)
Answer a)
∞ ∞ ∞ ∞ ∞ ∞
X n X n X n X n X n X 1
ln n
= ln n = ln 10 ln n
= ln 10 = ln 10
= ln 10−1
< +∞
n=2 10 n=2 e n=2 n n=2 n
ln 10 ) ln n )
n=2 (e n=2 (e
The last inequality holds because 2 < ln 10, and thus 1 < ln 10 − 1. To see why we have 2 < ln 10, it
∞
n
suffices to realize that 32 = 9 < 10 and e < 3. Hence the series
X
ln n
is convergent.
n=2 10
113
b)
∞ ∞ ∞ ∞
X 1 X 1 X 1 X 1
= > n
= = +∞
n=2 ln(n!) n=2 ln[n(n − 1) · · · 2 · 1] n=2 ln (n ) n=2 n ln n
To get the last equality, we applied the Cauchy condensation test as follows:
∞ ∞ ∞ ∞
X 2k X 1 X 1 1 X 1
k k
= k
= = = +∞
k=1 2 ln (2 ) k=1 ln (2 ) k=1 k ln 2 ln 2 k=1 k
∞
X 1
Hence the series is divergent.
n=2 ln(n!)
c)
∞ ∞ ∞ ∞ ∞
X 1 X 1 X 1 X 1 X 1
n)
= ln(ln n)
= n)
= ln 10 = ln 10
n=2 n · 10 n=2 n · e
ln(ln ln 10 ln(ln
n=2 n · (ln n)
ln(ln n) )
n=2 n · (e n=2 n · (e
ln 10 )
∞ ∞
1 X 1 1 X 1
= ln 10 + ln 10 < ln 10 + 2 < +∞
2 · (ln 2) n=3 n · (ln n) 2 · (ln 2) n=3 n · (ln n)
We again used the fact that ln 10 > 2. We applied the Cauchy condensation test as follows:
∞ ∞ ∞ ∞
X 2k X 1 X 1 1 X 1
2 = 2 = 2 = 2 < +∞
k k
k=2 2 (ln (2 ))
k
k=2 (ln (2 )) k=2 (k ln 2) (ln 2) k=2 k 2
∞
X 1
Hence the series ln(ln n)
is convergent.
n=2 n · 10
Question 8. Examine the convergence of the following series:

1 1 1 1 1
a) − + − + − + ···
2 6 10 14 18
1 1 1 1 1 1
b) − − + + − − + · · ·
2 3 4 5 6 7
1 1 1 1 1 1
c) − + + − + + − ···
2 4 8 16 32 64
1 1 1 1 1 1 1 1 1 1 1 1
d) − + + − + + − + + − + + − ···
2 5 6 4 9 10 8 17 18 16 33 34
1 1 1 1 1 1 1 1
e) 1 − − + − − + − − + · · ·
2 3 4 5 6 7 8 9
1 1 1 1 1 1 1 1 1 1 1 1 1 1
f) 1 + − − − + + − − − + + − − − + ···
2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 1 1 1 1 1 1 1 1 1 1
g) 1 − − + − − + − − + − − + ···
2 4 3 6 8 5 10 12 7 14 16
1 1 1 1 1 1 1 1 1 1 1
h) 1 + − + + − + + − + + − + ···
3 2 5 7 4 9 11 6 13 15 8
1 2 3 4 5 6 7 8 9
i) 1 + 2 − 3 + 4 + 5 − 6 + 7 + 8 − 9 + · · ·
2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1
j) 2 + 2 − 2 + 2 + 2 − 2 + 2 + 2 − 2 + · · ·
1 3 2 5 7 4 9 11 6
114
1 1 1 1 1
Answer a) We are examining the convergence of the series − + − + − + · · · . We know
2 6 10 14 18
1 1 1 1
that + + + · · · is divergent. Since
2 3 4 5
1 1 1 1 1 1 1 1 1 1

+ + + + + ··· > + + + + · · · = +∞,
2 6 10 14 18 4 2 3 4 5
1 1 1 1 1
the series − + − + − + · · · does not converge absolutely. But since the terms are
2 6 10 14 18
monotone decreasing in absolute value, and have limit zero at infinity, by the alternating series test,
it converges.
1 1 1 1 1 1
b) We are examining the convergence of the series − − + + − − + · · · . We know that
2 3 4 5 6 7
1 1 1 1 1 1 1 1 1 1
+ + + · · · is divergent. Hence the series − − + + − − + · · · does not converge
2 3 4 5 2 3 4 5 6 7
1 1 1 1 1 1

absolutely. We can write the series as − + + + − + + · · · . Now the terms are
2 3 4 5 6 7
monotone decreasing in absolute value, and have limit zero at infinity. Hence by the alternating series
test, the series converges.
1 1 1 1 1 1
c) We are examining the convergence of the series − + + − + + − · · · . We put
2 4 8 16 32 64
1 1 1 1
Sk = + + + ··· + k.
2 4 8 2
Then
1 1 1 1 1 1 1 1 1 1 1

1− Sk = + + + · · · + k − + + + · · · + k+1 = − k+1
2 2 4 8 2 4 8 16 2 2 2
Thus
1 1 1
1 1 1 − k+1
+ + + · · · = lim Sk = lim 2 2 1 = 2
1 =1
2 4 8 k→∞ k→∞ 1 −
2 2
1 1 1 1 1 1
So the series − + + − + + − · · · converges absolutely, and hence converges.
2 4 8 16 32 64
d) We are examining the convergence of the series
1 1 1 1 1 1 1 1 1 1 1 1
− + + − + + − + + − + + − ···
2 5 6 4 9 10 8 17 18 16 33 34
We have
1 1 1 1 1 1 1 1 1 1 1 1
+ + + + + + + + + + + + ···
2 5 6 4 9 10 8 17 18 16 33 34
1 1 1 1 1 1 1 1 1 1 1 1 1
= + + + + + + + + + + + + +···
2 |4 {z
5 6} |8 9{z 10} |16 17{z 18} |32 33 {z 34}
<3· 41 <3· 18 1
<3· 16 1
<3· 32
∞
X 1
<3 k
=3·1=3
k=1 2
So the series
1 1 1 1 1 1 1 1 1 1 1 1
− + + − + + − + + − + + − ···
2 5 6 4 9 10 8 17 18 16 33 34
115
∞
X 1
converges absolutely and hence converges. We used k
= 1 which we proved in part c).
k=1 2
e) We are examining the convergence of the series

1 1 1 1 1 1 1 1
1− − + − − + − − + ···
2 3 4 5 6 7 8 9
We know that the series does not converge absolutely. We have
1 1 1 1 1 1 1 1
1− − + − − + − − + ···
2 3 4 5 6 7 8 9
   
1 1 1 1 1  1 1 1
  

= 1 − + − + − + · · ·
−  = −∞
+ + + · · ·


| 2 4 5{z 7 8 }
 3

| 6 9 }
P {z ∞
converges by alternating series test = 13 1
k=1 k
=∞
So the series
1 1 1 1 1 1 1 1
1− − + − − + − − + ···
2 3 4 5 6 7 8 9
diverges.
f) We are examining the convergence of the series
1 1 1 1 1 1 1 1 1 1 1 1 1 1
1+ − − − + + − − − + + − − − + ···
2 3 4 5 6 7 8 9 10 11 12 13 14 15
We know that the series does not converge absolutely. We have
1 1 1 1 1 1 1 1 1 1 1 1 1 1
1+ − − − + + − − − + + − − − + ···
2 3 4 5 6 7 8 9 10 11 12 13 14 15
 
 1 1 1 1 1 1 1 1 1 1 1

=
 1+ − + + + − + + + − − −
+ · · ·

|
2 3 4 6 7 8 {z 9 11 12 13 14 }

converges by alternating series test
 
1 1 1
 

− + + + · · ·
 = −∞
|5 10 {z 15

P∞ }
= 15 1
k=1 k
=∞
So the series
1 1 1 1 1 1 1 1 1 1 1 1 1 1
1+ − − − + + − − − + + − − − + ···
2 3 4 5 6 7 8 9 10 11 12 13 14 15
diverges.
g) We are examining the convergence of the series
1 1 1 1 1 1 1 1 1 1 1
1− − + − − + − − + − − + ···
2 4 3 6 8 5 10 12 7 14 16
We have
1 1 1 1 1 1 1 1 1 1 1
1+ + + + + + + + + + + + ···
2 4 3 6 8 5 10 12 7 14 16
116
∞
1 1 1 1 1 1 1 1 1X 1
> + + + + + + + + ··· = =∞
2 4 6 8 10 12 14 16 2 k=1 k
Hence the series does not converge absolutely. We have

1 1 1 1 1 1 1 1 1 1 1
1−− + − − + − − + − − + ···
2 4 3 6 8 5 10 12 7 14 16
1 1 1 1 1 1 1 1 1 1 1

=1− + + − + + − + + − + + ···
2 4 3 6 8 5 10 12 7 14 16
and
1 1 1 1 1 1 1 1 1 1 1

1> + > > + > > + > > + > ···
2 4 3 6 8 5 10 12 7 14 16
Hence by the alternating series test, the series
1 1 1 1 1 1 1 1 1 1 1
1− − + − − + − − + − − + ···
2 4 3 6 8 5 10 12 7 14 16
converges.
h) We are examining the convergence of the series
1 1 1 1 1 1 1 1 1 1 1
1+ − + + − + + − + + − + ···
3 2 5 7 4 9 11 6 13 15 8
We have
1 1 1 1 1 1 1 1 1 1 1
1+ + + + + + + + + + + + ···
3 2 5 7 4 9 11 6 13 15 8
∞
1 1 1 1 1X 1
> + + + ··· = =∞
2 4 6 8 2 k=1 k
Hence the series does not converge absolutely. We have

1 1 1 1 1 1 1 1 1 1 1
1+− + + − + + − + + − + ···
3 2 5 7 4 9 11 6 13 15 8
1 1 1 1 1 1 1 1 1 1 1

= 1+ − + + − + + − + + − + ···
3 2 5 7 4 9 11 6 13 15 8
and
1 1 1 1 1 1 1 1 1 1 1

1+ > > + > > + > > + > > ···
3 2 5 7 4 9 11 6 13 15 8
Hence by the alternating series test, the series
1 1 1 1 1 1 1 1 1 1 1
1+ − + + − + + − + + − + ···
3 2 5 7 4 9 11 6 13 15 8
converges.
i) We are examining the convergence of the series
1 2 3 4 5 6 7 8 9
1
+ 2 − 3 + 4 + 5 − 6 + 7 + 8 − 9 + ···
2 2 2 2 2 2 2 2 2
k+1 1
2k+1 (k + 1)2k k+1 1+ k 1
We have lim = lim = lim = lim =
k→∞ kk k→∞ k · 2k+1 k→∞ 2k k→∞ 2 2
2
117
Thus by the ratio test, the series converges absolutely and hence converges.
j) We are examining the convergence of the series
1 1 1 1 1 1 1 1 1
2
+ 2 − 2 + 2 + 2 − 2 + 2 + 2 − 2 + ···
1 3 2 5 7 4 9 11 6
∞
X 1 π2
Since 2
= , the series converges absolutely and hence converges.
k=1 k 6
Temat 8: The limits of functions.

Find the limit (finite or infinite) or prove that the limit does not exist:
3x2 − 5x − 2
Question 1. lim
x→2 5x2 − 20
Answer: x = 2 is a root of the numerator and of the denominator. We need to take out the factor
x − 2.
3x2 − 5x − 2 (x−2)
(3x + 1)
7
lim = lim =
x→2 5x − 20
2 (x− 2)
x→2 5(x + 2)
20
2x3 + 250
Question 2. lim
x→−5 x2 + 4x − 5
Answer: x = −5 is a root of the numerator and of the denominator. We need to take out the factor
x − (−5) = x + 5.
2x3 + 250 (2x2 − 10x + 50)
(x+5)

150
lim = lim = = 25
x→−5 x2 + 4x − 5 x→−5 (x − 1)
(x+ 5)

6
xn − 1
Question 3. lim
x→1 x − 1
Answer: x = 1 is a root of the numerator and of the denominator. We need to take out the factor
x − 1.
xn − 1 (xn−1 + xn−2 + · · · + x + 1)
(x−1)

lim = lim =n
x→1 x − 1 x→1 1 ·
(x−1)
√ √
Question 4. lim ex + 1 − ex − 1
x→+∞
Answer:
√
√ √
√ ex + 1 − ex − 1
lim ex + 1 − ex − 1 = lim
x→+∞ x→+∞ 1
√ √ √ √
ex + 1 − ex − 1 ex + 1 + ex − 1
= lim √ x √
x→+∞ e + 1 + ex − 1
√ 2 √ 2
ex + 1 − ex − 1 ex + 1 − (ex − 1)
= lim √ x √ x = lim √ x √
x→+∞ e +1+ e −1 x→+∞ e + 1 + ex − 1
118
2
= lim √ √ =0
x→+∞ ex + 1 + ex − 1
√
x2 + 1 − 1
Question 5. lim √ 2
x→0 x + 25 − 5
Answer:
√ √ √ √
x2 + 1 − 1 x2 + 1 + 1 x2 + 1 − 1 x2 + 25 + 5
lim √ 2 = lim √ √ √
x→0 x + 25 − 5 x→0 x2 + 1 + 1 x2 + 25 − 5 x2 + 25 + 5
√ √
2 √
− 12

x2 + 1 x2 + 25 + 5 [x2 + 1 − 1] x2 + 25 + 5
= lim √ √ 2 = lim √
x→0 x→0 x 2 + 1 + 1 [x2 + 25 − 25]
2
x +1+1 x + 25 − 5
2 2
√ √
x2 x2 + 25 + 5 25 + 5 10
= lim √ = √ = =5
x→0 x2 + 1 + 1 x2 1+1 2
1
Question 6. lim
x2
x→0
Answer: Because of the square in the denominator, we get +∞ when approaching zero from the left.
We also get +∞ when approaching zero from the right. So
1 1 1
lim− 2
= lim+ 2 = lim 2 = +∞
x→0 x x→0 x x→0 x
1
Question 7. lim
x→0 x3
Answer: Because of the third power in the denominator, we get −∞ when approaching zero from
the left. We get +∞ when approaching zero from the right. So
1 1
lim− 3
= −∞ =
6 lim+ 3 = +∞
x→0 x x→0 x
and the limit does not exist.
1
Question 8. lim e x
x→0
Answer: We get e−∞ = 0 when approaching zero from the left. We get e+∞ = +∞ when approaching
zero from the right. So
1 1
lim− e x = 0 6= lim+ e x = +∞
x→0 x→0
and the limit does not exist.
1
Question 9. lim e− x2
x→0
119
Answer: Because of the square in the denominator of the exponent, we get e−∞ = 0 when approaching
zero from the left. We also get e−∞ = 0 when approaching zero from the right. So
1 1 1
lim− e− x2 = lim+ e− x2 = lim e− x2 = 0
x→0 x→0 x→0
sin(ax)
Question 10. lim , b 6= 0.
x→0 bx
Answer: When α is very close to zero, then sin α ≈ α. This is because the sin curve is like a line
with slope 1 in the proximity of the origin. So we use sin(ax) ≈ ax and we get
sin(ax) a
x a
lim = lim =
x→0 bx x→0 b
x b
cx
Question 11. lim , d 6= 0.
x→0 tg(dx)
with slope 1 in the proximity of the origin. So we use sin(dx) ≈ dx and we get
cx cx cx c
x c
lim = lim sin(dx)
= lim = lim =
x→0 tg(dx) x→0 x→0 sin(dx) x→0 d
x d
cos(dx) 1
sin(ax)
Question 12. lim , b 6= 0.
x→0 tg(bx)
with slope 1 in the proximity of the origin. So we use sin(ax) ≈ ax and sin(bx) ≈ bx and we get
sin(ax) sin(ax) sin(ax) a

x a
lim = lim sin(bx) = lim sin(bx) = lim =
x→0 tg(bx) x→0
cos(bx)
x→0
1
x→0 b
x b
sin x
Question 13. lim .
x→−∞ x
Answer: The sin function always has a value between 1 and −1. When deviding by ∞ or −∞ this
becomes zero. So we get
sin x
lim =0
x→−∞ x
cos x
Question 14. limπ
x→ 2 x − π2
Answer:
120
π π π
In the proximity of , the cos function looks like the line − x. So we use cos x ≈ − x and we get
2 2 2

(−1) x−π2
π

cos x 2
−x
limπ π = lim π = lim
π = −1
x→ 2 x − x→ π2 x − x→ π2 x −
2 2
2
1 + cos x
Question 15. lim
x→π sin2 x
Answer:
1 + cos x 1 + cos x 1+
( (( cos
((x 1
lim
x→π sin2 x
= x→π
lim = x→π
lim =
1 − cos x
2
(1+cos x)(1 − cos x)

2
!x−x2
2x2 + 1
Question 16. lim
x→+∞ 3x2 + 1
Answer:
!x−x2 1
!x−x2 1
!−x(x−1) −∞ +∞
2x2 + 1 2+ x2
2+ x2 2 3
lim = lim 1 = lim 1 = = = +∞
x→+∞ 3x2 + 1 x→+∞ 3+ x2
x→+∞ 3+ x2
3 2
!−x2 +x3
2x2 + 1
Question 17. lim
x→0 3x2 + 1
Answer: !−x2 +x3

2x2 + 1
lim = 10 = 1
x→0 3x2 + 1
!
1
Question 18. limπ tg x − π
x→ 2
2
−x
Answer:
121
π π π
In the proximity of , the cos function looks like the line − x. So we use cos x ≈ − x. In the
2 2 2
π
proximity of , the sin function looks like the line with constatant value 1. So we use sin x ≈ 1 and
2
we get
! ! !
1 sin x 1 1 1
lim tg x − π = limπ − π = limπ π − π = limπ 0 = 0
x→ π2
2
−x x→ 2 cos x 2
−x x→ 2
2
−x 2
−x x→ 2
Temat 9: Functions. Continuity of functions.

Question 0. Find the „left–side” and „right–side” limits of the functions:
1
a) f (x) = 2 x−1 , at the point x = 1
1
2x + 3
b) f (x) = 1 , at the point x = 0
3x + 2
1
Answer: a) When approaching 1 from the right, the exponent becomes +∞, so we get
x−1
1
lim+ 2 x−1 = 2∞ = ∞
x→1
| {z }
„right–side” limit
1
When approaching 1 from the left, the exponent becomes −∞, so we get
x−1
1 1
lim− 2 x−1 = 2(−∞) = =0
x→1
| {z }
2∞
„left–side” limit
1
b) When approaching 0 from the right, the exponent becomes +∞, so we get
x
1 1 ∞ ∞
1 2x 3 2 x 3 2 3 2
2 +3x 1 + 1 3
+ 1
3
+ 3∞ 3
+0 ∞
2
3x 3x 3x
lim 1 = lim+ 2 = lim+ 2 = 2 = = =0
x→1+ 3 x + 2} x→1 1+ 1 x→1 1+ 1 1+ 3∞
1+0 3
| {z 3x 3x
„right–side” limit
122
1
When approaching 0 from the left, the exponent becomes −∞, so we get
x
1
2x + 3 2(−∞) + 3 0+3 3
lim− 1 = (−∞)
= =
x→1 3 x + 2} 3 +2 0+2 2
| {z
„left–side” limit
Question 1. Find the functions f (g(x)) and g(f (x)), where

a) f (x) = 1 − x2 , g(x) = 2x + 3
b) f (x) = −17, g(x) = |x|
√
c) f (x) = x2 − 3, g(x) = x2 + 3
1
d) f (x) = x2 + 1, g(x) =
x2 + 1
√
e) f (x) = x3 − 4, g(x) = 3 x + 4
Answer: a)
f (g(x)) = 1 − (2x + 3)2 g(f (x)) = 2(1 − x2 ) + 3 = −2x2 + 5
b)
f (g(x)) = −17 g(f (x)) = | − 17| = 17
c) q √ 2
f (g(x)) = (x2 + 3)2 − 3 g(f (x)) = x2 − 3 +3
d)
2
1 1

f (g(x)) = +1 g(f (x)) =
x2 + 1 (x2 + 1)2 + 1
e)
√
3
3 √
3
√
3
f (g(x)) = x+4 −4=x+4−4=x g(f (x)) = x3 − 4 + 4 = x3 = x
Question 2. Find a function of the form f (x) = xk (k does not have to be an integer) and a function
g(x) in such a way that f (g(x)) = h(x), where:
1
a) h(x) =
1 + x2
1
b) h(x) = √
x + 10
1
c) h(x) =
(1 + x + x2 )3
1
Answer: a) We put k = −1 and g(x) = 1 + x2 . Then f (x) = x−1 = and
x
1
h(x) = f (g(x)) = .
1 + x2
123
1 1 1 1
b) We put k = − and g(x) = x + 10. Then f (x) = x− 2 = 1 = √ and
2 x2 x
1
h(x) = f (g(x)) = √ .
x + 10
1
c) We put k = −3 and g(x) = 1 + x + x2 . Then f (x) = x−3 = and
x3
1
h(x) = f (g(x)) = .
(1 + x + x2 )3
Question 3. Let f (x) = 1 + x2 . Find a function g(x), so as to have
f (g(x)) = 1 + x2 − 2x3 + x4 .
Answer: We put g(x) = x − x2 . Then

2 2
f (g(x)) = 1 + (g(x))2 = 1 + x − x2 = 1 + x2 − 2x · x2 + x2 = 1 + x2 − 2x3 + x4 .
√
Question 4. Let g(x) = 1 + x. Find a function f (x), so as to have
√
f (g(x)) = 3 + 2 x + x.
Answer: We put f (x) = 2 + x2 . Then

√ 2 √ √ 2 √
f (g(x)) = 2 + 1 + x =2+1+2 x+ x = 3 + 2 x + x.
Question 5. Find a function g(x) to have f (g(x)) = h(x), where
f (x) = x2 , h(x) = x4 + 1.
√
Answer: We put g(x) = x4 + 1. Then
√ 2
f (g(x)) = (g(x))2 = x4 + 1 = x4 + 1 = h(x).
Question 6. a) Is this possible to choose the value f (1), in such a way that the function defined for
x 6= 1 by the formula
|x − 1|
f (x) =
(x − 1)2
is continuous on R?
b) Is this possible to choose the values f (−2), f (3), in such a way that the function defined for
x ∈ R \ {−2, 3} as
x+1
f (x) = 2
x −x−6
124
is continuous on R?
c) Is this possible to choose the values f (−1), f (1), in such a way that the function defined for
x ∈ R \ {−1, 1} as
|x2 − 1|
f (x) = 2
x −1
is continuous on R?
Answer: a) Both the numerator and the denominator are always non-negative. So we have
|x − 1| |x − 1| 1
lim f (x) = lim = lim = lim = ∞.
x→1 x→1 (x − 1)2 x→1 |x − 1||x − 1| x→1 |x − 1|
Hence it is not possible to choose the value f (1) so as to make f (x) continuous on R.
b) We have
x+1 x+1
f (x) = = .
x2 −x−6 (x + 2)(x − 3)
When approaching −2 from the left, we get the „left–side” limit
x+1 x+1
lim − f (x) = lim − = lim = −∞.
x→−2 x→−2 x2 − x − 6 x→−2− (x + 2)(x − 3)
Hence it is not possible to choose the value f (−2) so as to make f (x) continuous on R.
c) We have 
|x2 − 1| 1 if |x| 1
f (x) = 2 =
x −1 −1 if |x| < 1
Hence it is not possible to choose the value f (−1) so as to make f (x) continuous on R.
Question 7. a) Show that equation x4 + 2x − 1 = 0 is satisfied for some x ∈ [0, 1].

b) Show that equation x5 − 5x3 + 3 = 0 is satisfied for some x ∈ [−3, 2].
c) Show that equation x3 − 4x + 1 = 0 has three different real valued solutions.
π
d) Show that there is an x between and π such that tg x = −x.
2
Answer: a) We put f (x) = x4 + 2x − 1. Then f (0) = −1 and f (1) = 2. The function f (x) is
continuous on R. Thus for some x ∈ [0, 1], we must have f (x) = 0.
b) We put f (x) = x5 − 5x3 + 3. Then f (−2) = 11 and f (2) = −5. The function f (x) is continuous
on R. Thus for some x ∈ [−2, 2], and consequently for some x ∈ [−3, 2], we must have f (x) = 0.
c) We put f (x) = x3 − 4x + 1. Then f (−3) = −14, f (−1) = 4, f (1) = −2, and f (3) = 16. The
function f (x) is continuous on R. Thus for some x1 ∈ [−3, −1], for some x2 ∈ [−1, 1], and for some
x3 ∈ [1, 3], we must have f (x1 ) = f (x2 ) = f (x3 ) = 0.
π
d) We put f (x) = tg x + x. Then the „right–side” limit at x = is
2
sin x

lim + f (x) = lim + (tg x + x) = lim + + x = −∞,
x→( π2 ) x→( π2 ) x→( π2 ) cos x
125
π

and f (π) = π. The function f (x) is continuous on the interval , π . The round bracket means
2
π

that the limit is not included in the interval. Hence there exists an x ∈ , π such that f (x) = 0,
2
which means that tg x = −x.
Question 8. At which points is the function


0 if x ∈ Q
f (x) = 
x2 if x ∈
/Q
continuous?
Answer: We will show that the function f (x) is nowhere continuous. For this we need some pro-
perties of Q and R. We denote by Q the set of all fractions with integers in the numerator √ and the
denominator. The reals, denoted by R, include all rationals, but also numbers such as 2 and π
which are not rational and therefore called irrational. To show that the function f (x) is nowhere
continuous, it suffices to show that between any two rationals there is an irrational, and between any
two irrationals there is a rational.
First we will show that between any two irrationals, there is a rational. We start with two irrationals
a and b. There must be a first decimal in which they differ, in the following example it is the 6th
decimal, but it could be a much higher numbered decimal.
a = 1.23456432993 . . .
b = 1.23456840932 . . .
Then
1234568
c = 1.234568 =
1000000
is a rational strictly between a and b.
Now we will show that between any two rationals, there is an irrational. We need the fact that the
rationals are countable, they can be counted as in the following diagram.
Now we consider two rationals p and q, and we will show that there exists an irrational in the interval
[p, q]. There must be a first decimal in which p and q differ, in the following example it is the 8th
decimal, but it could be a much higher numbered decimal.
p = 2.498377323492871 . . .
126
q = 2.498377376390248 . . .
Now assume the numbers between p and q were all rational. This would imply that they are countable.
If they were countable, we could put them in one-to-one correspondence with the natural numbers.
Assume we did this, this would lead to a list like the following:
1. 2.498377339248935 . . .
2. 2.498377353859389 . . .
3. 2.498377343957290 . . .
4. 2.498377350345093 . . .
5. 2.498377364928043 . . .
..
.
If we have such a list, we can create a number between p and q which is not on the list as follows.
We take one decimal, starting from the 9th, from each number in the list, in the order in which they
are listed. This gives
2.498377333948 . . .
Now we replace these decimals by 4 if they are greater than or equal to 5, and by 5 if they are less
than or equal to 4. This gives
2.498377355454 . . .
This number is different from all the numbers on the list. Hence the numbers between p and q are
not countable. Any subset of the rationals is countable. Hence between a and b there must be an
irrational.
Temat 10: Continuity continued.

Question 1. Where possible, find the upper and lower bounds of the following functions. Indicate
which functions actually attain the bounds.
1
a) f (x) = ; Df = R.
1 + x2
2
b) f (x) = ; Df = R.
ex + e−x
ex − e−x
c) f (x) = ; Df = R.
ex + e−x
√
d) f (x) = x 1 − x2 ; Df = (−1, 1).
2 + x − x2
e) f (x) = ; Df = (1, +∞).
(x − 1)2
x2 + x − 1
f) f (x) = ; Df = (1, +∞).
x−1
x2 + x − 1
g) f (x) = ; Df = R \ {1}.
x−1
127
1
h) f (x) = cos ; Df = R \ {0}.
x
i) f (x) = e−x cos x; Df = (0, +∞).
j) f (x) = e−x cos x; Df = [0, +∞).
1
Answer: a) We have f (x) = and Df = R. The denominator is always positive, so the lower
1 + x2
bound is zero. The function gets very close to zero when x gets big, but never reaches the bound.
The upper bound, 1, is reached when x = 0.
2
b) We have f (x) = and Df = R. This is an even function, which means that f (−x) = f (x).
ex+ e−x
So it is enough to look at non-negative x. Here is what ex and e−x look like
When starting at x = 0 and moving to the right, ex (green) increases and e−x (red) decreases, but
2 2
less rapidly. So the upper bound of f (x) = x −x
is reached at x = 0 and is = 1. When x gets
e +e 2
2
large, the function f (x) = x gets very close to zero, so the lower bound is zero, but it is never
e + e−x
reached.
128
ex − e−x
c) We have f (x) = and Df = R. This is an odd function, which means that
ex + e−x
f (−x) = −f (x). So it is enough to look at non-negative x and then make use of the symmetry.
We have f (0) = 0, and when moving to the right, the numerator grows, since ex increases and e−x
decreases. So for positive x, both the numerator and the denominator are always positive. And the
numerator is always smaller than the denominator. If x gets large, the dominant term in both the
numerator and the denominator is ex . So the upper bound is 1, but it is never reached. By symmetry,
the lower bound is −1, but it is never reached.
√
d) We have f (x) = x 1 − x2 and Df = (−1, 1). We know that y = x is a line with slope 1. We
√
also need to know the shape of y = 1 − x2 . This is a half-circle, since x2 + y 2 = 1 is the equation
of a circle with radius 1.
Since we multiply the function corresponding to the green line with the function corresponding to
the red semicircle, the product will be an odd function which means that f (−x) = −f (x). We have
f (0) = 0. When moving to the right, the line increases and the semicircle decreases. Until they cross,
the line increases more rapidly than the semicircle decreases, both absolutely and in proportion to it’s
1
current value. After they cross, the situation is reversed. So the upper bound is reached at x = √
2
1 1 1
and is . By symmetry, the lower bound is reached at x = − √ and is − .
2 2 2
129
2 + x − x2
e) We have f (x) = and Df = (1, +∞). We divide the numerator by (x−1)2 = x2 −2x+1.
(x − 1)2
This gives
2 + x − x2 −x + 3
f (x) = = −1 +
(x − 1)2 (x − 1)2
When approaching x = 1 from the right, f (x) goes towards ∞. So there is no upper bound. When
x gets very big, the function gets closer and closer to −1. But what is the lower bound? We need to
look at y = −x + 3 and y = (x − 1)2 .
1 0
We have f (2) = −1 + = 0 and f (3) = −1 + = −1. When x is increasing and we are past x = 3,
1 4
we need to look at the proportional growth of the numerator and the denominator, when they are
equal, the minimum is reached. The absolute growth of −x + 3 is constant and equal to −1. The
absolute growth of (x − 1)2 is 2(x − 1). Now we equate the proportional growth, which means the
absolute growth divided by the current value, of the numerator and the denominator.
−1 2(x − 1)
=
−x + 3 (x − 1)2
130
−(x − 1) = 2(−x + 3)
x=5
−5 + 3 9
So the lower bound is reached at x = 5 and is f (5) = −1 + =−
(5 − 1)2 8
x2 + x − 1
f) We have f (x) = and Df = (1, +∞). We divide the numerator by the denominator.
x−1
This gives
x2 + x − 1 1
f (x) = =x+2+
x−1 x−1
So when x gets big, we are asymptotically approaching the line x + 2. When x is approaching 1 from
the right, f (x) goes towards infinity. So there is no upper bound. We need to look at y = x and
1 1
y= . The vertical asymptote of y = also appears in red, but it is not part of the curve.
x−1 x−1
1
The slope od y = x is constant and equal to 1. The slope of y = is equal to −1 at x = 2.
x−1
131
Between x = 1 and x = 2, it is smaller than −1, so f (x) is decreasing, and for x > 2, it is bigger
than −1, so f (x) is increasing. Thus the lower bound is reached at x = 2 and is equal to f (2) = 5
x2 + x − 1
g) We have f (x) = and Df = R \ {1}. When x goes to infinity, f (x) goes to infinity, so
x−1
there is no upper bound. When x goes to minus infinity, f (x) goes to minus infinity, so there is no
lower bound.
1 1
h) We have f (x) = cos and Df = R \ {0}. On Df = R \ {0}, the function takes on every value
x x
1
in R with the exception of zero. So f (x) = cos takes on all the values that the cos function takes
x
on, which means that the upper bound 1 is reached and the lower bound −1 is reached.
i) We have f (x) = e−x cos x and Df = (0, +∞). We need to know what the functions y = e−x
(green) and y = cos x (red) look like
If we multiply these two functions, for Df = R \ {0}, the upper bound is 1, but it it is not reached
since zero is excluded from the domain of definition. The particularity of the exponential function
132
y = ex , is that the absolute growth of the function is equal to the value of the function at any x.
Hence its proportional growth is 1 and the proportional growth of y = e−x is −1. The particularity
of the trigonometric function y = cos x is that its absolute growth is − sin x. Hence its proportional
− sin x
growth is . To find the lower bound, we equate the absolute value of the proportional growth
cos x
of y = cos x and the proportional growth of y = e−x and search for solutions in the region where
sin x
y = cos x is negative in the previous picture. From = −1 we get cos x = − sin x and the
cos x
3π 3π 3π 1 3π

solution we are looking for is x = . So the lower bound is cos e 4 = − √ e 4 and is reached
4 4 2
3π
at x =
4
j) We have f (x) = e−x cos x and Df = [0, +∞). This is as part i), only that here the upper bound
is reached since zero is included in the domain of definition.
Question 2. Determine if the following functions have an inverse. In the cases that they do, give the
domain of the inverse function and determine if it is continuous.
a) f (x) = x2 ; Df = [−2, 1].
b) f (x) = x2 ; Df = [−2, −1].
c) f (x) = x3 ; Df = R.
3
d) f (x) = 2x ; Df = R.
e) f (x) = ctg x; Df = (−2π, −π).
( )
ax + b d
f) f (x) = , c 6= 0; Df = R \ − .
cx + d c
ex − e−x
g) f (x) = ; Df = R.
2
ex + e−x
h) f (x) = ; Df = R \ {0}.
ex − e−x
Answer: a) We have f (x) = x2 and Df = [−2, 1]. The function f (x) is not one-to-one on the
domain of definition, so it does not have an inverse.
b) We have f (x) = x2 and Df = [−2, −1]. The function f (x) is one-to-one on Df = [−2, −1].
√
Its inverse is f (y) = − y and the domain of the inverse function is [1, 4]. The inverse function is
continuous on its domain.
133
c) We have f (x) = x3 and Df = R. The function f (x) is one-to-one on Df = R. Its inverse is
1
f (y) = y 3 and the domain of the inverse function is R. The inverse function is continuous.
d) We have f (x) = 2(x ) and Df = R. The function f (x) is one-to-one on Df = R. Its inverse
3
1
is f (y) = (log2 y) 3 and the domain of the inverse function is R>0 = {y ∈ R | y > 0}. The inverse
function is continuous.
1
e) We have f (x) = ctg x = and Df = (−2π, −π). The function f (x) is one-to-one on
tg x
!
1 3π
Df = (−2π, −π). Its inverse is arc tg − and the domain of the inverse function is R \ {0}.
y 2
3π π π

The additive term − is due to the fact that the range of the arc tg function is − , . The
2 2 2
inverse function is continuous on its domain. There is a jump at y = 0, but zero is not part of the
domain.
134
b − da
( )
ax + b a c d
f) We have f (x) = = + , c 6= 0 and Df = R \ − . This is the equation of a
cx + d c cx + d c
1
hyperbola. If we put a = d = 0 and b = c = 1 we get the well known function y = .
x
b − da d a

c
The inverse is f (y) = − , and the domain of the inverse function is R \ . The inverse
cy − a c c
a a
function is continuous on its domain. There is a jump at y = , but is not part of the domain.
c c
x −x
e −e
g) We have f (x) = and Df = R. This is the arithmetic mean (red) of two continuous
2
and steadily increasing functions y = ex (blue) and y = −e−x (green).
135
Hence the function corresponding to the red curve is invertible, the domain of the inverse is R, and
the inverse is continuous.
ex + e−x
h) We have f (x) = x and Df = R \ {0}. This is an odd function, meaning that
e − e−x
f (−x) = −f (x). So it is sufficient to look at non-negative x and then make use of the symmetry.
When approaching 0 from the right, the function goes towards ∞. This is so because in the numerator
we have ex − e−x , and for x positive, ex > e−x . When moving away from zero and towards the right,
both the numerator and the denominator of f (x) grow. The proportional change in the denominator
is bigger than the proportional change in the numerator, so f (x) is always decreasing. When x is
very big, the dominant term in the numerator as well as in the denominator is ex , so f (x) approaches
1. It follows that f (x) looks as in the following picture.
The vertical asymptote is shown in red as well, but it is not part of the curve. The function f (x) is
one-to-one. It is invertible and the domain of the inverse function is (−∞, −1) ∪ (1, ∞). The inverse
function is continuous on its domain.
136
1
Question 3. a) Show that the function f (x) = satisfies a Lipschitz condition on (, +∞), > 0,
x
i.e., there is a positive constant L (f ) such
|f (x) − f (y)| ¬ L (f )|x − y|
for all x, y ∈ (, +∞). Conversely, show that no Lipschitz condition is possible for f on (0, +∞).
b) Show that f (x) = x2 satisfies a Lipschitz condition on any bounded interval but does not on the
open intervals (−∞, 0), (0, ∞) and (−∞, ∞).
c) Show that a Lipschitz function with bounded domain is a bounded function.
1
Answer: a) We know what the function f (x) = looks like, it is a hyperbola. We are only interested
x
in the part which is in first quadrant (x 0 and y 0). At every point of the hyperbola, we can
draw a tangent line as in the following picture
1
This tangent line has a slope which we can determine. In the picture, the tangent line has slope − .
4
1 1

In general, at the point , , the slope of the tangent line is − 2 . So

1
|f (x) − f (y)| ¬ |x − y|
2
But when approching 0, the slope of the tangent line becomes −∞. So no Lipschitz condition is
possible for f on (0, +∞).
b) This is very similar to what we did in part a), only now we have f (x) = x2 . Again we draw a
tangent line as in the following picture
137
This tangent line has a slope

which we can determine. In the picture, the tangent line has slope −4.

2
In general, at the point , , the slope of the tangent line is 2. For a bounded interval (α, β), we
put = α if |α| |β| and = β otherwise. Then
|f (x) − f (y)| ¬ |2||x − y|
But when approaching ∞, the slope of the tangent line becomes ∞, and when approaching −∞,
the slope of the tangent line becomes −∞. So no Lipschitz condition is possible for f on the open
intervals (−∞, 0), (0, ∞) and (−∞, ∞).
c) Let f (x) be a Lipschitz function with bounded domain. The domain is bounded below by a real
number α and bounded above by a real number β. For a Lipschitz function, we have a constant L
such that |f (x) − f (y)| ¬ L|x − y| for all x, y in Df . So |f (x) − f (y)| ¬ L|α − β| for all x, y in
Df . We pick an x0 in Df . Then f (x) is bounded below by f (x0 ) − L|α − β| and bounded above by
f (x0 ) + L|α − β|.
A note about terminology: in some exercises, „lower bound” means „greatest lower bound”, and
„upper bound” means „least upper bound”. In the present exercise, the lower bound we have found
is just „a lower bound” and the upper bound we have found is just „an upper bound”.
Question 4. Give an example of a continuous function f (x), with domain Df = [0, 1), such that the
range, f (Df ), is the interval:
a) (0, 1], b) [0, 1], c) (0, 1),
d) [0, +∞), e) (0, +∞), f) (−∞, ∞).
Answer: a) f (x) = 1 − x
1 2

b) f (x) = 4 x −
2
1 x 1

c) f (x) = + sin
2 2 1−x
1 1
d) f (x) = − −1= −1
x−1 1−x
138
2
1 1

e) f (x) = sin +1−x
1−x 1−x
1 1

f) f (x) = sin
1−x 1−x
Question 5. a) Explain why among the triangles inscribed in a circle of radius R there is one with
maximal area.
b) A mountain climber climbs a mountain with a stop watch timing the ascent. At the top, the
climber immediately turns around, resets the timer to zero, and begins to descend along the same
path. Show that there is a point on the path where the travel time up is the same as the travel time
down.
Answer: a)
The area of a triangle inscribed in a circle of radius R is determined uniquely by the side length
c and the angle α. We have 0 < c ¬ 2R and 0 < α < 180◦ . So the area is a continuous bounded
function of two variables with bounded domain and must reach a maximum.
b)
The picture corresponds to a path of length 8 km. The climber does the ascent in 4 hours (red line)
and the descent in 2 hours (blue line). Regardless on what is the actual distance, time of ascent, time
of descent, and shape of the curves, it is clear that the red and the blue curve must intersect.
139
Temat 11: The derivative of a function
Question 1. Calculate the derivative of the function f wherever it exists.
1 1 x+1
a) f (x) = b) f (x) = c) f (x) =
x3 sin x x−1
√ √
d) f (x) = sin3 x
3
e) f (x) = 3 x f) f (x) = 1 + x3
2
g) f (x) = e−x h) f (x) = ex i) f (x) = x ln x
j) f (x) = log2 x k) f (x) = logx 2 l) f (x) = xx
2
m) f (x) = xx n) f (x) = (xx )2
1
Answer: a) We have f (x) = = x−3 . Thus
x3
3
f 0 (x) = −3x(−3−1) = −3x−4 = −
x4
1
b) We have f (x) = = (sin x)−1 . Using the chain rule, we get
sin x
−1 −1 cos x
f 0 (x) = −1(sin x)(−1−1) (sin x)0 = −1(sin x)−2 (sin x)0 = 2
(sin x)0 = 2
cos x = −
(sin x) (sin x) (sin x)2
x+1
c) We have f (x) = . Using the quotient rule, we get
x−1
0 (x + 1)0 (x − 1) − (x + 1)(x − 1)0 1(x − 1) − (x + 1) · 1 −2

f (x) = = =
(x − 1) 2 (x − 1) 2 (x − 1)2
d) We have f (x) = sin3 x = (sin x)3 . Using the chain rule, we get
f 0 (x) = 3(sin x)(3−1) (sin x)0 = 3(sin x)2 (sin x)0 = 3 sin2 x(sin x)0 = 3 sin2 x cos x
√
3
1
e) We have f (x) = x = x 3 . Thus
1 1 1 1 3 1 2 1 1 1 1
f 0 (x) = x( 3 −1) = x( 3 − 3 ) = x− 3 = 2 = 1
= 1 = √
3x( 3 )
3
3 3 3 3x 3 2·
3 (x2 ) 3 3 x2
√ 1
1 + x3 = 1 + x3 3 . Using the chain rule, we get
3
f) We have f (x) =
1 ( 1 −1) 0 1 ( 1 − 3 ) 0 1 − 2 0
f 0 (x) = 1 + x3 3 1 + x3 = 1 + x3 3 3 1 + x3 = 1 + x3 3 1 + x3
3 3 3
1 − 2 x2 x2 x2 x2
= 1 + x3 3 · 3x2 = 2 = 1
= 1 = q
3 (1 + x3 )(2· 3 ) (1 + x3 )2

(1 + x3 ) 3 (1 + x3 )2 3
3
140
g) We have f (x) = e−x . Using the chain rule, we get
f 0 (x) = e−x (−x)0 = e−x (−1) = −e−x
h) We have f (x) = ex = e(x ) . Using the chain rule, we get

2 2
0
f 0 (x) = e(x ) x2 = e(x ) · 2x(2−1) = e(x ) · 2x = 2xex
2 2 2
2
i) We have f (x) = x ln x. Using the product rule, we get

1
f 0 (x) = (x)0 ln x + x(ln x)0 = 1 · ln x + x · = ln x + 1
x
j) We have f (x) = log2 x. We need to express this in terms of natural logarithms in order to take the
derivative. We have y = f (x) = log2 x. By the definition of the logarithm, this means 2y = x. Since
y
eln 2 = 2 , we have eln 2 = ey ln 2 = x. Taking the natural logarithm on both sides of the equation
ln x
ey ln 2 = x, we get y ln 2 = ln x, hence y = . We get
ln 2
!0
0 ln x 1 1 1 1
f 0 (x) = (log2 x) = = (ln x)0 = · =
ln 2 ln 2 ln 2 x x ln 2
k) We have f (x) = logx 2. We need to express this in terms of natural logarithms in order to take
the derivative. We have y = f (x) = logx 2. By the definition of the logarithm, this means xy = 2.
y
Since eln x = x , we have eln x = ey ln x = 2. Taking the natural logarithm on both sides of the
ln 2
equation ey ln x = 2, we get y ln x = ln 2, hence y = . Using the chain rule, we get
ln x
!0 0
ln 2 1

0 0
f (x) = (logx 2) = = ln 2 = ln 2(ln x)−1 = ln 2(−1)(ln x)−1−1 (ln x)0
ln x ln x
1 − ln 2 − ln 2 1 − ln 2
= − ln 2(ln x)−2 (ln x)0 = − ln 2 · 2
(ln x)0 = 2
(ln x)0 = 2
· =
(ln x) (ln x) (ln x) x x(ln x)2
l) We have f (x) = xx . We need to express this as a power of e in order to take the derivative. Since
x
x = eln x , we have xx = eln x = ex ln x . Using the chain rule and the product rule, we get
0
f 0 (x) = (xx )0 = ex ln x = ex ln x (x ln x)0 = xx (x ln x)0
1

= xx ((x)0 ln x + x(ln x)0 ) = xx 1 · ln x + x · = xx (ln x + 1)
x
m) We have f (x) = xx = x(x ) . We need to express this as a power of e in order to take the
2 2
( x2 )
derivative. Since x = eln x , we have xx = x(x ) = eln x
2
2
2
= ex ln x . Using the chain rule and the
141
product rule, we get
0
ln x 0
2
2
2
0 2
0
f 0 (x) = xx = ex = ex ln x
x2 ln x = xx x2 ln x
1

2 0

x2 0 x2 2
=x x 2
ln x + x (ln x) =x 2x 2−1
ln x + x · = xx (2x ln x + x)
2
x
2 2 +1
= xx · x (2 ln x + 1) = xx (2 ln x + 1)
n) We have f (x) = (xx )2 . From part l), we know that (xx )0 = xx (ln x + 1). Using the chain rule, we
get 0
f 0 (x) = (xx )2 = 2 (xx )2−1 (xx )0 = 2xx (xx )0 = 2xx xx (ln x + 1) = 2 (xx )2 (ln x + 1)
Question 2. Examine if the function is differentiable at the point x0 = 0.

1

x sin

for x 6= 0
a) f (x) = x|x| d) f (x) =  x
0 for x = 0
1

x2 sin

for x 6= 0
b) f (x) = |x|3 e) f (x) = x

0 for x = 0
 1
e− x
 for x > 0
c) f (x) = | sin3 (x)| f) f (x) = 
0 for x ¬ 0
Answer: a) We have f (x) = x|x|. We will use the definition of the derivative. Approaching zero
from the right, we get
f (0 + h) − f (0) (0 + h)|0 + h| − 0 · |0| h|h|

lim+ = lim+ = lim+ = lim+ |h| = 0
h→0 h h→0 h h→0 h h→0
Approaching zero from the left, we get
f (0 + h) − f (0) (0 + h)|0 + h| − 0 · |0| h|h|

lim− = lim− = lim− = lim− |h| = 0
h→0 h h→0 h h→0 h h→0
The „left–side” and „right–side” limits are the same, so the function is differentiable at x0 = 0 and
f 0 (0) = 0.
b) We have f (x) = |x|3 . We will use the definition of the derivative. Approaching zero from the
right, we get
f (0 + h) − f (0) |0 + h|3 − |0|3 |h|3 |h| 2

lim+ = lim+ = lim+ = lim+ |h| = lim+ |h|2 = 0
h→0 h h→0 h h→0 h h→0 h h→0
f (0 + h) − f (0) |0 + h|3 − |0|3 |h|3 |h| 2

lim− = lim− = lim− = lim− |h| = lim− −|h|2 = 0
h→0 h h→0 h h→0 h h→0 h h→0
142
f 0 (0) = 0.
Answer: c) We have f (x) = | sin3 (x)| = (sin x)3 . We will use the definition of the derivative.
Approaching zero from the right, keeping in mind that in the proximity of zero we have sin α ≈ α,
we get
f (0 + h) − f (0) | sin3 (0 + h)| − | sin3 (0)| | sin3 (h)|

lim+ = lim+ = lim+
h→0 h h→0 h h→0 h
|h3 | |h| 2
= lim+ = lim+ |h | = lim+ |h2 | = 0
h→0 h h→0 h h→0
Approaching zero from the left, keeping in mind that in the proximity of zero we have sin α ≈ α,
we get
f (0 + h) − f (0) | sin3 (0 + h)| − | sin3 (0)| | sin3 (h)|

lim− = lim− = lim−
h→0 h h→0 h h→0 h
|h3 | |h| 2
= lim− = lim− |h | = lim− −|h2 | = 0
h→0 h h→0 h h→0
f 0 (0) = 0.
1

x sin

for x 6= 0
d) We have f (x) =  x
0 for x = 0
We will use the definition of the derivative. Approaching zero from the right, we get
1
f (0 + h) − f (0) (0 + h) sin 0+h −0 h sin h1 1
lim+ = lim+ = lim+ = lim+ sin
h→0 h h→0 h h→0 h h→0 h
1
As h tends to zero, the argument of sin goes to infinity. So the limit does not exist and the
h
function is not differentiable at x0 = 0.
1

x2 sin

for x 6= 0
e) We have f (x) = x

0 for x = 0
1
f (0 + h) − f (0) (0 + h)2 sin 0+h −0 h2 sin h1 1
lim+ = lim+ = lim+ = lim+ h sin = 0
h→0 h h→0 h h→0 h h→0 h
1
f (0 + h) − f (0) (0 + h)2 sin 0+h −0 h2 sin h1 1
lim− = lim− = lim− = lim− h sin = 0
h→0 h h→0 h h→0 h h→0 h
f 0 (0) = 0.
143
 1
e − x
 for x > 0
f) We have f (x) =
0 for x ¬ 0

1 1
f (0 + h) − f (0) e− 0+h − 0 e− h 1 1 1 1 1 1
lim+ = lim+ = lim+ = lim+ · e− h = lim+ eln h e− h = lim+ eln h − h
h→0 h h→0 h h→0 h h→0 h h→0 h→0
1
Now we put g = . When h approaches zero from the right, g approaches +∞. When g is large, g
h
grows much faster than ln g, and we have
lim (ln g − g) = −∞.

g→+∞
Thus
f (0 + h) − f (0) 1 1
lim+ = lim+ eln h − h = lim eln g−g = e−∞ = 0
h→0 h h→0 g→+∞
f (0 + h) − f (0) 0−0 0
lim− = lim− = lim− = lim− 0 = 0
h→0 h h→0 h h→0 h h→0
f 0 (0) = 0.
Question 3. Find the equation of the line which is tangent to the graph of function f (x) = xx at
point (2, 4).
Answer: From Question 1 l) (Temat 11) we know that f 0 (x) = xx (ln x + 1). So
f 0 (2) = 22 (ln 2 + 1) = 4 + 4 ln 2
We need the equation of a line which has slope 4 + 4 ln 2 and which goes through the point (2, 4).
y = mx + b
y = (4 + 4 ln 2)x + b
4 = (4 + 4 ln 2)2 + b
b = −8 ln 2 − 4
y = (4 + 4 ln 2)x − 8 ln 2 − 4
So the equation of the tangent line is
y = (4 + 4 ln 2)x − 8 ln 2 − 4
Question 4. Calculate the angles under which the graphs of functions f (x) = x2 and g(x) = x3
intersect themselves.
Answer: The graphs of the functions f (x) and g(x) touch each other at (0, 0) and they intersect at
144
(1, 1). We have f 0 (x) = 2x, thus f 0 (1) = 2. We have g 0 (x) = 3x2 , thus g 0 (1) = 3. So the slope of f (x)
at the point (1, 1) is 2, and the slope of g(x) a at the point (1, 1) is three. To get the angle γ, we use
vectors with the same slope, a = [1, 2] and b = [1, 3], respectively.
a · b = |a||b| cos γ
! ! !
a·b [1, 2] · [1, 3] 1+6
γ = arc cos = arc cos = arc cos √ √
|a||b| |[1, 2]||[1, 3]| 1+4 1+9
!
7 7 7 7
= arc cos √ √ = arc cos √ = arc cos √ = arc cos √ ≈ 8.13◦
5 10 50 25 · 2 5 2
√
Question 5. Calculate the angles under which the graphs of functions f (x) = 3
x and g(x) = x3
intersect themselves.
Answer: The graphs of the functions f (x) and g(x) intersect each other at (−1, −1), (0, 0) and (1, 1).
√ 1
For the intersection at (0, 0), because the slope of x3 at that point is zero, the slope of 3 x = x 3 is ∞
and the angle of intersection is 90◦ . By symmetry, the angle of intersection at (−1, −1) is the same as
1 2 1
the angle of intersection at (1, 1). We have f 0 (x) = x− 3 , thus f 0 (1) = . We have g 0 (x) = 3x2 , thus
3 3
1
g 0 (1) = 3. So the slope of f (x) at the point (1, 1) is , and the slope of g(x) a at the point (1, 1) is
3
three. To get the angle γ, we use vectors with the same slope, a = [3, 1] and b = [1, 3], respectively.
a · b = |a||b| cos γ
! ! !
a·b [3, 1] · [1, 3] 3+3
γ = arc cos = arc cos = arc cos √ √
|a||b| |[3, 1]||[1, 3]| 9+1 1+9
!
6 6
= arc cos √ √ = arc cos ≈ 53.13◦
10 10 10
Question 6. Using the theorem about the derivative of an inverse function calculate (f −1 )0 (0), where
f (x) = x + sin x.
df dy
Answer: The derivative of f (x) is f 0 (x) = = .
dx dx
df −1 dx
The derivative of the inverse function is (f −1 )0 (y) = = .
dy dy
So
dx 1 1 1
(f −1 )0 (y) = = dy = 0 = 0 −1
dy dx
f (x) f [f (y)]
Now we need to know what is f −1 (0). There is only one x value for which f (x) = 0, namely x = 0.
So f −1 (0) = 0. Thus
1 1 1 1
(f −1 )0 (0) = 0 = = =
f (0) 1 + cos 0 1+1 2
145
Question 7. Let f : R → R be a differentiable odd function (i.e. f (−x) = −f (x) for any x). Show
that f 0 (x) is an even function.
Answer: We need to show that f 0 (x) = f 0 (−x). By the definition of the derivative,
f (x + h) − f (x) f (x + h) − f (x) f (x + h) − f (x)

f 0 (x) = lim = lim+ = lim−
h→0 h h→0 h h→0 h
and
f (−x + h) − f (−x) f (−x + h) − f (−x) f (−x + h) − f (−x)
f 0 (−x) = lim = lim+ = lim−
h→0 h h→0 h h→0 h
Since f (x) = −f (−x), we have
f (x + h) − f (x) f (x − h) − f (x) −f (−x + h) − (−f (−x))

f 0 (x) = lim+ = lim− = lim−
h→0 h h→0 −h h→0 −h
−f (−x + h) + f (−x) f (−x + h) − f (−x)
= lim− = lim− = f 0 (−x)
h→0 −h h→0 h
Question 8. Let f : R → R be an even function, differentiable at point x0 = 0. Show that f 0 (0) = 0.

Answer: By the definition of the derivative,
f (0 + h) − f (0) f (0 + h) − f (0) f (0 + h) − f (0)

f 0 (0) = lim = lim+ = lim−
h→0 h h→0 h h→0 h
Since f is an even function, we have f (x) = f (−x), thus
f (0 + h) − f (0) f (0 − h) − f (0) f (0 + h) − f (0) f (0 + h) − f (0)

lim+ = lim+ = lim− = − lim−
h→0 h h→0 h h→0 −h h→0 h
So f 0 (0) = −f 0 (0), which means that 2f 0 (0) = 0 and therefore f 0 (0) = 0.
Question 9. Let f : R → R be a differentiable function. Is this true that for any x0 ∈ R there exists
f (b) − f (a)
a pair of points a < x0 < b which is such that f (x0 ) = ?
b−a
Answer: It is not true. For example, we can take f (x) = x3 . Then f 0 (0) = 0, but for a and b such
f (b) − f (a)
that a < 0 < b, we always have > 0.
b−a
Temat 12: Lagrange theorem, derivatives and properties of

functions.
Question 1. Use The Mean Value Theorem of Lagrange to show that:
x
a) < ln(1 + x) < x, for x > 0.
x+1
b) | arc tg x − arc tg y| ¬ |x − y|, where x, y ∈ R.
146
Answer:
The mean value theorem states that for any function that is continuous on [a, b] and differentiable
on (a, b) there exists some c in the interval (a, b) such that the secant joining the endpoints of the
interval [a, b] is parallel to the tangent a c.
a) First we will show that for x > 0, we have ln(1 + x) < x. Assume that there was an x0 > 0 such
that ln(1 + x0 ) x0 . Then by the mean value theorem, there would be an x ∈ (0, x0 ) for which
1 ln(1 + x0 ) x0
= = 1. This is impossible, so we conclude that ln(1 + x) < x, for x > 0.
1+x x0 − 0 x0 − 0
x
Now we will show that for x > 0, we have < ln(1 + x). Assume that there was an x0 > 0 such
x+1
x0
that ln(1 + x0 ) ¬ . Then by the mean value theorem, there would be an x ∈ (0, x0 ) for which
x0 + 1
x0
1 ln(1 + x0 ) 1 x
= ¬ x0 +1 = . This is impossible, so we conclude that < ln(1 + x),
1+x x0 − 0 x0 − 0 x0 + 1 x+1
for x > 0.
b) Answer: If we can show that the absolute value of the derivative of arc tg(x) is always smaller
than or equal to 1, the result follows by the mean value theorem. So we calculate the derivative of
arc tg(x). Let x = tg(y). Then y = arc tg(x) and we have
!0
dx sin y (sin y)0 cos y − sin y(cos y)0 cos y cos y − sin y(− sin y)
= (tg(y))0 = = =
dy cos y cos2 y cos2 y
cos2 y + sin2 y 1
= =
cos2 y cos2 y
From the last line of this calculation, we also get
dx cos2 y + sin2 y cos2 y sin2 y
= 2
= 2
+ 2
= 1 + tg2 y
dy cos y cos y cos y
Since y = arc tg(x), we get
dy 1 1 1 1 1
(arc tg x)0 = = dx = 2
= 2
= 2
=
dx dy
1 + tg y 1 + tg (arc tg x) 1 + [tg(arc tg x)] 1 + x2
1
Since ¬ 1 for all x, we have |(arc tg x)0 | ¬ 1 for all x, which is what we need.
1 + x2
Question 2. Find the intervals on which the function is monotonic:
147
x3
a) f (x) =
x−2
b) f (x) = x · ln x, x > 0
c) f (x) = ex cos x
3x2 (x − 2) − x3 3x3 − 6x2 − x3 2x3 − 6x2 2x2 (x − 3)
Answer: a) f 0 (x) = = = =
(x − 2)2 (x − 2)2 (x − 2)2 (x − 2)2
The derivative is positive for x > 3, it is zero for x = 3, and it is negative for x < 3. The function
f (x) is not defined for x = 2. So the function f (x) is monotonically decreasing on (−∞, 2) and (2, 3),
and it is monotonically increasing on (3, ∞).
1
b) f 0 (x) = 1 · ln x + x · = ln x + 1
x
1 1 1
We have ln = − ln e = −1. Hence f 0 (x) < 0 for 0 < x < , and f 0 (x) > 0 for x > . So the
e e e
1 1

function f (x) is monotonically decreasing on 0, , and it is monotonically increasing on ,∞ .
e e
c) f 0 (x) = ex cos x + ex (− sin x) = ex (cos x − sin x)
The derivative is positive when cos x > sin x, and it is negative when sin x > cos x. So the func-
3π π

tion f (x) is monotonically increasing on all intervals − + 2kπ, + 2kπ with k ∈ Z, and it is
4 4
π 5π

monotonically decreasing on all intervals + 2kπ, + 2kπ with k ∈ Z.
4 4
Question 3. a) Show that f (x) = x + sin x is strictly increasing on R.

b) Use the above to show the existence of the inverse f −1 : R → R.
c) Examine for which arguments y ∈ R there exists the derivative of the inverse, (f −1 )0 (y) (finite or
infinite).
Answer: a) The derivative is f 0 (x) = 1 + cos x.
The derivative is positive, except for isolated points, namely odd multiples of π, where the derivative
is zero. It follows that the function f (x) is strictly increasing on R.
b) A function which is strictly increasing on its domain must be one-to-one. Hence f (x) is invertible.
Since x covers all of R, and cos x just oscillates between −1 and 1, it is clear that f maps R onto R,
and f −1 maps R onto R.
c) From Question 6 in Temat 11, we know that
dx 1 1 1
(f −1 )0 (y) = = dy = 0 = 0 −1
dy dx
f (x) f [f (y)]
So the derivative of the inverse, (f −1 )0 (y), (finite or infinite), exists, except at points y0 such that
h i h i h i h i
f 0 f −1 (y0 ) = 0 and f 0 f −1 (y) changes sign at y0 . But f 0 f −1 (y) = 1 + cos f −1 (y) is always
non-negative. So the derivative of the inverse, (f −1 )0 (y) (finite or infinite), exists for all arguments
y ∈ R.
148
Question 4. Find the local extrema of the function f (x) = x + sin 2x.
The derivative is f 0 (x) = 1 + 2 cos 2x.
We have local extrema at points where the derivative is zero and changes sign. It is clear that
f 0 (x) = 1 + 2 cos 2x changes sign whenever it is zero. So we need to find the solutions to
1 2π
1 + 2 cos 2x = 0. We get cos 2x = − which happens when 2x = + 2kπ with k ∈ Z or
2 3
4π
2x = + 2kπ with k ∈ Z. So the extrema are
3
√ !
π π 3
+ kπ, + kπ + with k ∈ Z
3 3 2
and √ !
2π 2π 3
+ kπ, + kπ − with k ∈ Z
3 3 2
Question 5. Show that among all rectangles with fixed perimeter the square has the biggest area.
We take a rectangle with sides a, b, a, b.
The perimeter is p = a + b + a + b = 2a + 2b and the area is A = ab. We would like to have A as a

function of one variable
p − 2a p
A(a) = ab = a · = −a2 + · a
2 2
p
The derivative is A0 (a) = −2a + . We put the derivative equal to zero and solve for a, which gives
2
p p
a = . We then use p = 2a + 2b and get b = . Hence an extremum is only possible when a = b,
4 4
this must be a maximum.
Question 6. Determine the edge length of a rectangular container without a lid, with a square base
and a capacity of 0.5 cubic meter, so as to build it using as little material as possible.
The base is square, say its edge length is a. Let the hight be h.
149
We have 0.5 = a2 h, and the surface area is A = a2 + 4ah. We want to minimize the surface area so
as to use as little material as possible. We would like to have A as a function of one variable
0.5 2
A(a) = a2 + 4ah = a2 + 4a · 2
= a2 +
a a
2
The derivative is A0 (a) = 2a − . We put the derivative equal to zero and solve for a, which gives
a2
a = 1. We then use 0.5 = a2 h and get h = 0.5. Hence an extremum is only possible when a = 1
and h = 0.5 (dimensions are meters), this must be a minimum.
Question 7. Prove that a light beam which is moving at a constant speed c from point A to point B,
with reflection at a mirror surface at a point so that the travel time is minimal, fulfills the condition
of „angle of incidence equals the angle of reflection”. Compare the result obtained with the law of
refraction when passing from one medium to another.
Answer:
The speed is constant, so we want to minimize a + b and show that this implies α = β. The points
A and B are given, so f , g, and d + e are constants. We put l = d + e. The angles α and β are equal
d e
if = , so we need to determine d or e. We use e = l − d and will determine d. The function we
f g
150
want to minimize is
q q q q
t(d) = a + b = f2 + d2 + g2 + e2 = f2 + d2 + g 2 + (l − d)2
We take the derivative with respect to the variable d, keeping in mind that f , g and l are constants.
1 2d 1 −2(l − d) d −(l − d)
t0 (d) = ·√ 2 + · q = √ + q
2 f + d2 2 g 2 + (l − d)2 f 2 + d2 g 2 + (l − d)2
We put the derivative equal to zero and solve for d.
d −(l − d)
√ +q =0
f2 +d 2
g 2 + (l − d)2
d l−d
√ 2 2
=q
f +d g 2 + (l − d)2
q q
d g2 + (l − d)2 = (l − d) f 2 + d2

d2 g 2 + (l − d)2 = (l − d)2 f 2 + d2

d2 g 2 + l2 − 2ld + d2 = l2 − 2ld + d2 f 2 + d2
d2 g 2 + d2 l2 − 2ld3 + d4 = l2 f 2 − 2ldf 2 + d2 f 2 + l2 d2 − 2ld3 + d4

d2 g 2 = l2 f 2 − 2ldf 2 + d2 f 2

d2 g 2 − f 2 + 2ldf 2 − l2 f 2 = 0
l2 f 4 l2 f 4
d2 g 2 − f 2 + 2ldf 2 + − − l2 f 2 = 0
g2 − f 2 g2 − f 2
| {z }
complete square
l2 f 4 l2 f 4 + l2 f 2 (g 2 − f 2 )
d2 g 2 − f 2 + 2ldf 2 + − =0
g2 − f 2 g2 − f 2
| {z }
complete square
l2 f 4 l2 f 2 g 2
d2 g 2 − f 2 + 2ldf 2 + − =0
g2 − f 2 g2 − f 2
| {z }
complete square
!2
q lf 2 l2 f 2 g 2
d g2 − f2 +√ 2 − =0
g − f2 g2 − f 2
!2
q lf 2 l2 f 2 g 2
d g2 − f 2 + √ 2 =
g − f2 g2 − f 2
q lf 2 lf g
d g2 − f2 +√ = ±√ 2
g −f
2 2 g − f2
lf 2 lf g
d+ =± 2
g −f
2 2 g − f2
151
lf 2 lf g
d=− ±
g2 − f 2 g2 − f 2
According to the picture, g > f , so we get one positive solution
lf 2 lf g fg − f2
d=− 2 + =l· 2
g − f 2 g2 − f 2 g − f2
This must correspond to a minimum of the length of the trajectory t(d).

d e l−d
Now we need to show that = = .
f g g
2
d l · fg2g−f
−f 2 g−f
= =l· 2
f f g − f2
2 2 g 2 −f 2 −f g+f 2 g 2 −f g
e l−d l − l · fg2g−f
−f 2
1 − fg2g−f
−f 2 g 2 −f 2 g 2 −f 2 g−f
= = =l· =l· =l· =l·
g g g g g g g2 − f 2
We will compare the result obtained with the law of refraction when passing from one medium to
another. There too, the path is so that the time of travel is minimal.
Refraction of light at the interface between two media of different refractive indices, with n2 > n1 .
Since the velocity is lower in the second medium (v2 < v1 ), the angle of refraction θ2 is less than the
angle of incidence θ1 ; that is, the ray in the higher-index medium is closer to the normal.
Temat 13: l’Hôpital’s rule, elasticity, and Taylor’s formula.

Question 1. Using l’Hôpital’s rule (if necessary), calculate the following limits:
ex − 1 1 − x + ln x sin x
a) lim b) lim c) lim
x→0 sin(2x) x→1 1 + cos(πx) x→0 x + x2
ex ln x 1 1

d) lim e) lim √ f) lim −
x→+∞ x + x2 x→+∞ x x→0 x sin x
152
√
!
1 1 1

g) lim − ctg x h) lim − i) lim x2 + 3x − x
x→0 x x→2 x − 2 ln(x − 1) x→+∞
x
1

1
j) lim x ln x k) lim (cos x) x2 l) lim x 1+ −e
x→0+ x→0 x→+∞ x
ex − 1
Answer: a) We want to find lim . For x = 0, the numerator and the denominator are both
x→0 sin(2x)
zero, so we will use l’Hôpital’s rule.
ex − 1 (ex − 1)0 ex 1
lim = lim 0 = lim =
x→0 sin(2x) x→0 (sin(2x)) x→0 2 cos(2x) 2
1 − x + ln x
b) We want to find lim . For x = 1, the numerator and the denominator are both zero.
x→1 1 + cos(πx)
We will use l’Hôpital’s rule twice.
1 − x + ln x (1 − x + ln x)0 −1 + x1
lim = lim = lim
x→1 1 + cos(πx) x→1 (1 + cos(πx))0 x→1 −π sin(πx)
0
1
−1 + x − x12 −1 1
= lim = lim = = −
x→1 (−π sin(πx))0 x→1 −π 2 cos(πx) −π 2 (−1) π2
sin x
c) We want to find lim . For x = 0, the numerator and the denominator are both zero, so
x→0 x + x2
we will use l’Hôpital’s rule.
sin x (sin x)0 cos x

lim = lim = lim =1
x→0 x + x2 x→0 (x + x2 )0 x→0 1 + 2x
d) Using the well known fact that the exponential function grows much faster than x or x2 when x
is large, we get
ex
lim = +∞
x→+∞ x + x2
One can check this applying l’Hôpital’s rule twice
ex (ex )0 ex (ex )0 ex
lim = lim = lim = lim = lim = +∞
x→+∞ x + x2 x→+∞ (x + x2 )0 x→+∞ 1 + 2x x→+∞ (1 + 2x)0 x→+∞ 2
ln x
e) We want to find lim √ . For x approaching infinity, the numerator and the denominator are
x→+∞ x
both going towards infinity, so we will use l’Hôpital’s rule.
√
ln x (ln x)0 x
1
2 x 2
lim √ = lim √ 0 = lim 1 = lim = lim √ = 0
x→+∞ x x→+∞ ( x) x→+∞ √
2 x
x→+∞ x x→+∞ x
f) It is possible to get the answer without calculation. When x is very small, sin x ≈ x, so the limit
153
0
we will calculate will be zero. We use a common denominator, this will be of the form and we can
0
apply l’Hôpital’s rule. We actually need to apply it twice.
1 1 sin x − x (sin x − x)0 (cos x − 1) cos x − 1

lim − = lim = lim 0
= lim 0 0
= lim
x→0 x sin x x→0 x sin x x→0 (x sin x) x→0 (x) sin x + x(sin x) x→0 sin x + x cos x
(cos x − 1)0 − sin x

= lim = lim
x→0 (sin x + x cos x)0 x→0 cos x + (x)0 cos x + x(cos x)0
− sin x 0 0
= lim = = =0
x→0 cos x + cos x + x(− sin x) 1+1+0 2
g) Using the well know facts that in the proximity of 0, we have sin x ≈ 0 and cos x ≈ 1, we get
1 1 cos x

lim − ctg x = lim − =0
x→0 x x→0 x sin x
One can check this applying l’Hôpital’s rule twice
1 1 cos x sin x − x cos x (sin x − x cos x)0

lim − ctg x = lim − = lim = lim
x→0 x x→0 x sin x x→0 x sin x x→0 (x sin x)0
cos x − cos x − x(− sin x) x sin x (x sin x)0
= lim = lim = lim
x→0 sin x + x cos x x→0 sin x + x cos x x→0 (sin x + x cos x)0
sin x + x cos x 0
= lim = =0
x→0 cos x + cos x + x(− sin x) 2
h) We will put this on a common denominator and apply l’Hôpital’s rule twice.
[ln(x − 1) − (x − 2)]0
!
1 1 ln(x − 1) − (x − 2)
lim − = lim = lim
x→2 x − 2 ln(x − 1) x→2 (x − 2) ln(x − 1) x→2 [(x − 2) ln(x − 1)]0
1
−1
x−1 1 − (x − 1)
= lim x−2 = lim
x→2 ln(x − 1) + x→2 (x − 1) ln(x − 1) + x − 2
x−1
2−x [2 − x]0
= lim = lim
x→2 (x − 1) ln(x − 1) + x − 2 x→2 [(x − 1) ln(x − 1) + x − 2]0
−1 1
= lim x−1 =−
x→2 ln(x − 1) + +1 2
x−1
i)
√ √
√ x2 + 3x − x x2 + 3x + x x2 + 3x − x2
lim x2 + 3x − x = lim √ = lim √
x→+∞ x→+∞ x2 + 3x + x x→+∞ x2 + 3x + x
3x 3 3
= lim √ = lim 1
√ = lim q
x→+∞ x2 + 3x + x x→+∞ x2 + 3x + x x→+∞ 1
(x2 + 3x) + 1
x x2
154
3 3 3
= lim q =√ =
x→+∞
1+ 3
+1 1+1 2
x
j) We will use l’Hôpital’s rule

1
ln x (ln x)0 x
lim x ln x = lim+ 1 = lim+ 0 = lim+ = lim+ (−x) = 0
x→0+ x→0
x
x→0 1 x→0 − x12 x→0
x
1
k) We want to find lim (cos x) x2 . We have
x→0
1 ln(cos x)
1
ln(cos x) x2
lim (cos x) x2 = lim e = lim e x2
x→0 x→0 x→0
Now we use l’Hôpital’s rule twice to find the limit of the exponent.
− sin x
ln(cos x) [ln(cos x)]0 cos x − sin x
lim = lim 0 = lim = lim
x→0 x2 x→0 [x2 ] x→0 2x x→0 2x cos x
(− sin x)0 − cos x −1 1

= lim 0
= lim = =−
x→0 (2x cos x) x→0 2 cos x + 2x(− sin x) 2+0 2
So we get
1 ln(cos x) 1 11
lim (cos x) x2 = lim e x2 = e− 2 = =√ 1
x→0 x→0 e e 2
x x
1 1

l) We want to find lim x 1+ − e . We know that lim 1+ = e. So we can put this
x→+∞ x x→+∞ x
0
in the form and then use l’Hôpital’s rule twice
0
x 0
x h x i0 ln(1+ x1 )

1
x 1+ 1
−e 1+ 1
−e e −e
x x
lim x 1+ − e = lim 1 = lim h i0 = lim h i0
x→+∞ x x→+∞
x
x→+∞ 1 x→+∞ 1
x x
0
1 1
ex ln(1+ x ) − e ex ln(1+ x ) ln 1 +

1 x
x
+ 1+ x1
· − x12
= lim h i0 = lim
x→+∞ 1 x→+∞ − x12
x

1 x
ln 1 + x
+
1+ x1
· − x12
1+ x1
= lim ex ln( ) · lim
x→+∞ x→+∞ − x12

1 x

1
x ln 1 + x
+ 1+ x1
· − x12
= lim 1+ · lim
x→+∞ x x→+∞ − x12

1 x
· − x12

ln 1 + + 1 1
x 1+ x1 ln 1 + x
− x+1
= e · lim = e · lim
x→+∞ − x12 x→+∞ − x12
i0
1 −1
· − x12 −
h
1 1
ln 1 + x
− x+1 1+ x1 (x+1)2
= e · lim h i0 = e · lim 2
x→+∞ x→+∞
− x12 x3
155

x3 −x3 x3
1+ x1
· − x12 − (x+1)2
− 1+x 1 + (x+1)2
= e · lim = e · lim x
x→+∞ 2 x→+∞ 2
x 2 x3 x3 −x2 (x+1) x3 −x3 −x2
− x+1 + (x+1)2 (x+1)2 (x+1)2
= e · lim = e · lim = e · lim
x→+∞ 2 x→+∞ 2 x→+∞ 2
−x2 −1
−1
x2 +2x+1 1+ x2 + 1
x2 1 e
= e · lim = e · lim =e· =−
x→+∞ 2 x→+∞ 2 2 2
Question 2. Elasticity.
a) Let D be a demand function varying inversely proportional to the price p. If p increases by one
percent, what is the percentage change in demand.
b) Assume that, as a function of the price p, the demand for chocolate from the rural population
is three times smaller than the demand from the urban population with elasticity −0.8 in the rural
case, and elasticity 0.3 in the urban case. Approximately what percentage change in global demand
for chocolate will result from an increase in the price by 1 % ?
C(q)
c) Let C(q) denote the cost of producing q widgets, and c(q) = the average cost of producing
q
one widget. Assume that C extends to a differentiable function of (0, ∞) and show that
C 0 (q) = c(q)(1 + Eq (c)).
Answer: a) Since the demand varies inversely proportional to the price, we have a function
k
D(p) =
p
for some constant k. Let the initial price be p0 , and then it increases by one percent, p1 = (1.01) · p0 .
The percentage change in demand is
k k
D1 D(p1 ) p1 p0 (1.01) 1 1
= = k = k = ≈ 0.990099 ≈ 0.9901 = 1 − 0.0099 = (100 − 0.99)
D0 D(p0 ) p0 p0
1.01 100
So the demand decreases by 0.99%.

b) Price elasticity of demand is a measure used in economics to show the responsiveness, or elasticity,
of the quantity demanded of a good or service to a change in its price. More precisely, it gives the
percentage change in quantity demanded in response to a one percent change in price.
Let Dr be the demand for chocolate from the rural population, Du the demand of chocolate from the
the urban population, and D = Dr + Du the global demand. Since the demand for chocolate from
the rural population is three times smaller than the demand from the urban population, we know
D 3D
that Dr = and Du = . When the price increases by one percent, the demand from the rural
4 4
population decreases by −0.8% and the demand from the urban population increases by 0.3%. So
the percentage change in global demand is
D 3D
(−0.008) · 4
+ (0.003) · 4 1 3 −0.008 + 0.009 0.001
= (−0.008) · + (0.003) · = = = 0.00025
D 4 4 4 4
156
which means an increase of 0.25%.
c) One has to make sure to distinguish c and C. We have the percentage change in c resulting from
one percent increase of q given by the elasticity
∆c ∆c q
∆c ∆q
· ∆q ∆q
· 100 q ∆c q dc q d q
Eq (c) = · 100 = · 100 = · 100 = · = · = · c(q) = · c0 (q)
c c c c ∆q c dq c(q) dq c(q)
c(q) C(q)
So c0 (q) = · Eq (c). Since c(q) = , we have C(q) = q · c(q), and using the product rule for
q q
differentiation, we get
c(q)
C 0 (q) = q 0 ·c(q)+q·c0 (q) = 1·c(q)+q·c0 (q) = 1·c(q)+q· ·Eq (c) = 1·c(q)+c(q)·Eq (c) = c(q)(1+Eq (c))
q
Question 3. Calculate the Taylor series around the point x0 and determine the radius of convergence.
1 1
a) f (x) = , x0 = 0 b) f (x) = , x0 = 0
1+x 2 + x2
1+x 1
c) f (x) = , x0 = 0 d) f (x) = , x0 = 3
2 + 3x2 (x − 2)(4 − x)
1 5 1
e) f (x) = , x0 = f) f (x) = , x0 = 5
(x − 2)(4 − x) 2 (x − 4)2
1
g) f (x) = (1 + x)2 , x0 = 2 h) f (x) = (1 + x) 2 , x0 = 0
1 1
i) f (x) = (1 − x) 3 , x0 = 0 j) f (x) = (1 − x)− 3 , x0 = 0

k) f (x) = sin x2 , x0 = 0 l) f (x) = sin2 (x), x0 = 0
Answer:
∞
cn xn , which says that the radius
X
Note: There is a formula for the convergence of a power series
n=0
of convergence is
1
R= ,
s
where q
n
s = lim sup |cn |
and lim sup denotes the supremum limit.
There are cases where it is more straightforward to use the ratio test to find the radius of convergence.
Using known series, such as the geometric series and the binomial series, allows us to get the radius
of convergence without using convergence tests or formulas every time.
1
a) We want to find the Taylor series of the function f (x) = around x0 = 0. The Taylor
1+x
157
series is a power series, since x0 = 0 it will contain only powers of (x − 0) = x. It is not necessary to
actually use Taylor’s method to find the series, we can use the geometric series
1
= 1 + α + α2 + α3 + α4 + · · ·
1−α
which converges for |α| < 1.
∞
1 1
(−x)n
X
f (x) = = =
1+x 1 − (−x) n=0
= (−x)0 + (−x)1 + (−x)2 + (−x)3 + (−x)4 − · · ·

= 1 − x + x2 − x3 + x4 − · · ·
Since α = −x, we need to have | − x| = |x| = |x − 0| < 1. So the radius of convergence is 1.

1
b) We want to find the Taylor series of the function f (x) = around x0 = 0. The Taylor
2 + x2
1
= 1 + α + α2 + α3 + α4 + · · ·
1−α
1 1 1 1 1
f (x) = = · 2 = ·
2 + x2 2 1 + x2 2 1 − − x2
2
∞
!n
1X x2
= −
2 n=0 2
 !0 !1 !2 !3 !4 
1 x2 x2 x2 x2 x2
=  − + − + − + − + − − ···
2 2 2 2 2 2
" #
1 x2 x4 x6 x8
= 1− + − + − ···
2 2 4 8 16
1 x2 x4 x6 x8
= − + − + − ···
2 4 8 16 32
x2 x2 x2 (x − 0)2 √
Since α = − , we need to have − = = < 1. So |x − 0| < 2 and the radius of
2 2 2 2
√
convergence is 2.
1+x
c) We want to find the Taylor series of the function f (x) = around x0 = 0. The Taylor
2 + 3x2
1
= 1 + α + α2 + α3 + α4 + · · ·
1−α
158
" #  
1+x 1 x 1 1 x 1 1 x
f (x) = = + = 3x 2 + 2 = + 
2 + 3x 2 2 + 3x 2 2 + 3x 2 2 1+ 2 1 + 3x2 2 1− − 3x
2
2
1 − − 3x2
2
∞
" !n ∞
!n #
1 X 3x2 X 3x2
= − +x −
2 n=0 2 n=0 2
1 x 3x2 3x3 9x4 9x5 27x6 27x7 81x8 81x9
= + − − + + − − + + − ···
2 2 4 4 8 8 16 16 32 32
s
3x2 3x2 3x2 3(x − 0)2 2
Since α = − , we need to have − = = < 1. So |x − 0| < and the
2 2 2 2 3
s
2
radius of convergence is .
3
1
d) We want to find the Taylor series of the function f (x) = around x0 = 3. The
(x − 2)(4 − x)
Taylor series is a power series, since x0 = 3 it will contain only powers of x − 3. It is not necessary
to actually use Taylor’s method to find the series, we can use the geometric series
1
= 1 + α + α2 + α3 + α4 + · · ·
1−α
which converges for |α| < 1. In order to make use of this we need to decompose f (x) as follows
1 A B A(4 − x) + B(x − 2)
f (x) = = + =
(x − 2)(4 − x) x−2 4−x (x − 2)(4 − x)
4A − 2B = 1
−A + B = 0
1
A=B=
2
1 1
2 2 1 1 1 1
f (x) = + = · + ·
x−2 4−x 2 x−2 2 4−x
Now we can make use of the geometric series formula.
1 1 1 1 1 1 1 1 1
f (x) = = · + · = · + ·
(x − 2)(4 − x) 2 x−2 2 4−x 2 1 − [−(x − 3)] 2 1 − (x − 3)
∞ ∞ ∞
1X 1X 1X
= [−(x − 3)]n + (x − 3)n = [1 + (−1)n ](x − 3)n
2 n=0 2 n=0 2 n=0
= 1 + (x − 3)2 + (x − 3)4 + (x − 3)6 + (x − 3)8 + · · ·

∞
(x − 3)2n
X
=
n=0
Since |α| = |x − 3|, we need to have |x − 3| < 1. So 2 < x < 4 and the radius of convergence is 1.
1 5
e) We want to find the Taylor series of the function f (x) = around x0 = . The
(x − 2)(4 − x) 2
159
5 5
Taylor series is a power series, since x0 = it will contain only powers of x − . It is not necessary
2 2
to actually use Taylor’s method to find the series, we can use the geometric series
1
= 1 + α + α2 + α3 + α4 + · · ·
1−α
which converges for |α| < 1. From part d) we know that
1 1
1 2 2
f (x) = = +
(x − 2)(4 − x) x−2 4−x
It is clear that
1
2 1 1
= = h i
x−2 2x − 4 1 − −2 x − 52
but for the other term, a calculation is needed to get it in a usable form.
1
2 P
=
4−x 1−Q x− 5
2
1 2P
=
8 − 2x 2 − 2Qx + 5Q
2 − 2Qx + 5Q = 16P − 4P x
2 + 5Q = 16P
−2Q = −4P
2 1
P = =
6 3
2
Q = 2P =
3
1 1
2 3
=
4−x 1− 2
x− 5
3 2
Now we can make use of the geometric series formula.

1 1 1
1 1
f (x) = = 2 + 2 = h i + 3
(x − 2)(4 − x) x−2 4−x 1 − −2 x − 52 1 − 32 x − 52
∞ n ∞ n
5 1X 2 5
X
= −2 x − + x−
n=0 2 3 n=0 3 2
∞ n n ∞ n
1 2 5 3 · (−6)n + 2n 5

n
X X
= (−2) + · x− = n
x−
n=0 3 3 2 n=0 3·3 2
2
4 16 5 112 5

= − x− + x− − ···
3 9 2 27 2
We have combined two series into one, so we need to take the smaller of the radii of convergence.
5

−2 x − <1
2
160
1
gives a radius of convergence of , and
2
2 5

x− <1
3 2
3
gives a radius of convergence of , so taking the smaller of the radii of convergence gives a radius of
2
1
convergence of .
2
1
f) We want to find the Taylor series of the function f (x) = around x0 = 5. The Taylor
(x − 4)2
series is a power series, since x0 = 5 it will contain only powers of x − 5. It is not necessary to actually
use Taylor’s method to find the series, we can use the geometric series
1
= 1 + α + α2 + α3 + α4 + · · ·
1−α
2 !2
1 1 1

f (x) = = =
(x − 4)2 x−4 1 − [−(x − 5)]
∞
!2 ∞
!2
n n n
X X
= [−(x − 5)] = (−1) (x − 5)
n=0 n=0
2
= 1 − (x − 5) + (x − 5)2 − (x − 5)3 + (x − 5)4 − (x − 5)5 + · · ·
∞ n ∞ n
" ! # " ! #
k k−n n 2k −n n
X X X X
= (−1) (−1) (x − 5) = (−1) (−1) (x − 5)
n=0 k=0 n=0 k=0
∞ n n ! ∞ n
" # " ! #
X X
2
k 1 n
X X
n n
= (−1) (x − 5) = (−1) (x − 5)
n=0 k=0 −1 n=0 k=0
= 1 − 2(x − 5) + 3(x − 5)2 − 4(x − 5)3 + 5(x − 5)4 − 6(x − 5)5 + · · ·
Since α = x − 5, we need to have |α| = |x − 5| < 1. So 4 < x < 6 and the radius of convergence is 1.
g) We want to find the Taylor series of the function f (x) = f (x) = (1 + x)2 around x0 = 2. The
Taylor series is a power series, since x0 = 2 it will contain only powers of x − 2. It is not necessary
to actually use Taylor’s method to find the series. Since f (x) is a polynomial of degree two, only the
the first three coefficients will be different form zero.
f (x) = (1 + x)2 = 1 + 2x + x2 = 9 + 6(x − 2) + (x − 2)2
This converges for all x, so the radius of convergence is ∞.

1
h) We want to find the Taylor series of the function f (x) = (1 + x) 2 around x0 = 0. The Taylor
series is a power series, since x0 = 0 it will contain only powers of (x − 0) = x. We will use Taylor’s
method to calculate the power series expansion of f (x) = (1 + x)α for any α.
We need to calculate the first few derivatives to get the pattern
f (x) = (1 + x)α f (0) = (1 + 0)α = 1
161
f 0 (x) = α(1 + x)α−1 f 0 (0) = α(1 + 0)α−1 = α
f 00 (x) = α(α − 1)(1 + x)α−2 f 00 (0) = α(α − 1)(1 + 0)α−2 = α(α − 1)
f 000 (x) = α(α − 1)(α − 2)(1 + x)α−3 f 000 (0) = α(α − 1)(α − 2)(1 + 0)α−3 = α(α − 1)(α − 2)
From this one can infer that f 0000 (0) = f (4) (0) = α(α − 1)(α − 2)(α − 3),
f 00000 (0) = f (5) (0) = α(α − 1)(α − 2)(α − 3)(α − 4), and so on. The Taylor series of f (x) = (1 + x)α
around x = 0, also called the Maclaurin series of f (x), is
f (0) f 0 (0) f 00 (0) f 000 (0)
f (x) = · (x − 0)0 + · (x − 0)1 + · (x − 0)2 + · (x − 0)3 + · · ·
0! 1! 2! 3!
f 00 (0) 2 f 000 (0) 3
= f (0) + f 0 (0) · x + ·x + · x + ···
2! 3!
∞ ∞
!
Xf (n) (0) n X α n
= ·x = x
n=0 n! n=0 n
where f (n) denotes the nth derivative of f and

!
α α(α − 1)(α − 2) · · · (α − n + 1)
=
n n!
!
α
with the convention that = 1. The Maclaurin series of f (x) = (1 + x)α is called the binomial
0
series and the radius of convergence is 1. To see why the radius of convergence is 1, we use the ratio
test.
α(α−1)(α−2)···(α−(n+1)+1) α(α−1)(α−2)···(α−n)
an+1 (n+1)!
· xn+1 (n+1)!
· xn+1
lim = lim α(α−1)(α−2)···(α−n+1)
= lim
n→∞ a n→∞ · xn n→∞ α(α−1)(α−2)···(α−n+1) · xn
n
n! n!
α(α−1)(α−2)···(α−n)
(n+1)n(n−1)···2·1
· xn+1 α−n
= lim
n→∞ α(α−1)(α−2)···(α−n+1) · xn
= n→∞
lim ·x
n(n−1)(n−2)···2·1
n+1
α−n
= lim · lim |x| = 1 · lim |x| = |x| < 1
n→∞ n + 1 n→∞ n→∞
For α a non-negative integer, the series only has finitely many nonzero coefficients and the radius of
convergence is infinite. But we are interested in the case where α is not an integer, in this part and
1
in parts i) and j), so the radius of convergence will be 1. Now we put α = . Then
2
∞ 1
!
1
xn
X
f (x) = (1 + x) = 2 2
n=0 n
1 1
· −1 1
· −1
· −3 1
· −1
· −3 · −5
=1+ ·x+ 2 2
· x2 + 2 2 2
· x3 + 2 2 2 2
· x4
2 2! 3! 4!
(−1)1 2! (−1)2 4! (−1)3 6!
1 22 ·2·(1!) 2 23 ·22 ·(2!) 3 24 ·23 ·(3!)
=1+ ·x+ ·x + ·x + · x4
2 2! 3! 4!
∞ n ∞
1 (−1) (2n)! n+1 (−1)n (2n)!
xn+1
X X
=1+ ·x+ n+1 n
x = 1 + 2n+1 2
2 n=1 2 2 (n!)[(n + 1)!] n=0 2 (n!) (n + 1)
162
1
i) We want to find the Taylor series of the function f (x) = (1 − x) 3 around x0 = 0. We use the
1
binomial series from part h). We put α = and replace x by −x. We get
3
∞ 1
!
1
(−x)n
X
f (x) = (1 − x) = 3 3
n=0 n
1 1
· −2 1
· −2
· −5 1
· −2
· −5 · −8
=1+ · (−x) + 3 3
· (−x)2 + 3 3 3
· (−x)3 + 3 3 3 3
· (−x)4 + · · ·
3 2! 3! 4!
1 2 1
1 · · 32 · 5 1
· 32 · 53 · 8
=1− · x − 3 3 · x2 − 3 3
· x3 − 3 3
· x4 − · · ·
3 2! 3! 4!
The radius of convergence is 1.
1
j) We want to find the Taylor series of the function f (x) = (1 − x)− 3 around x0 = 0. We use the
1
binomial series from part h). We put α = − and replace x by −x. We get
3
∞
− 13
!
− 13
(−x)n
X
f (x) = (1 − x) =
n=0 n
−1
1 · −4 −1
· −4
· −7 −1
· −4
· −7 · −10
=1− · (−x) + 3 3
· (−x)2 + 3 3 3
· (−x)3 + 3 3 3 3
· (−x)4 + · · ·
3 2! 3! 4!
1 4 1
1 · · 43 · 7 1
· 34 · 73 · 10
=1+ · x + 3 3 · x2 + 3 3
· x3 + 3 3
· x4 + · · ·
3 2! 3! 4!
The radius of convergence is 1.

k) We want to find the Taylor series of the function sin x2 around x0 = 0. The Taylor series is a
power series, since x0 = 0 it will contain only powers of (x − 0) = x. We will use Taylor’s method to
calculate the power series expansion of f (x) = sin x.
We need to calculate the first few derivatives to get the pattern
f (x) = sin x f (0) = sin 0 = 0

f 0 (x) = cos x f 0 (0) = cos 0 = 1
f 00 (x) = − sin x f 00 (0) = − sin 0 = 0
f 000 (x) = − cos x f 000 (0) = − cos 0 = −1
From this one can infer that f 0000 (0) = f (4) (0) = 0, f 00000 (0) = f (5) (0) = 1, and in general
1 1

n−1
f (n) (0) = (−1) 2 + (−1)(n+1)
2 2
where f (n) denotes the nth derivative of f .
The Taylor series of f (x) = sin x around x = 0, also called the Maclaurin series of f (x), is
f (0) f 0 (0) f 00 (0) f 000 (0)
f (x) = · (x − 0)0 + · (x − 0)1 + · (x − 0)2 + · (x − 0)3 + · · ·
0! 1! 2! 3!
163
f 00 (0) 2 f 000 (0) 3
= f (0) + f 0 (0) · x + ·x + · x + ···
2! 3!
n−1

1
∞
f (n) (0) n X ∞ (−1) 2
2
+ (−1)(n+1) 12
· xn
X
= ·x =
n=0 n! n=0 n!
1 1 1
=x− · x3 + · x5 − · x7 + · · ·
3! 5! 7!
It is useful to write the series in the following way
1 1 1
f (x) = x − · x3 + · x5 − · x7 + · · ·
3! 5! 7!
∞
(−1)k
· x2k+1
X
=
k=0 (2k + 1)!
The radius of convergence is infinite. To see this, we use the ratio test
(−1)k+1 (−1)k+1 (−1)k+1
ak+1 (2(k+1)+1)!
· x2(k+1)+1 (2k+3)!
· x2k+3 (2k+3)(2k+2)(2k+1)···2·1
· x2k+3
lim = lim (−1)k
= lim (−1)k = lim (−1)k
k→∞ ak k→∞ · x2k+1 k→∞ · x2k+1 k→∞ · x2k+1
(2k+1)! (2k+1)! (2k+1)2k(2k−1)···2·1
−1 −1
= lim · x2 = lim · lim x2 = 0 · x2 < 1
k→∞ (2k + 3)(2k + 2) k→∞ (2k + 3)(2k + 2) k→∞
Clearly, 0 · x2 < 1 holds for any x, so the radius of convergence is infinite. We know the Taylor
series for sin x around x0 = 0. It has infinite

radius of convergence and we will use it to get the Taylor
series around x0 = 0 of the function sin x just by replacing x by x2
2
1 2 3 1 2 5 1 2 7
sin x2 = x2 − x + x − x + ···
3! 5! 7!
1 1 1
= x2 − x6 + x10 − x14 + · · ·
3! 5! 7!
The radius of convergence is infinite.
l) We want to find the Taylor series of the function (sin x)2 around x0 = 0. The Taylor series is
a power series, since x0 = 0 it will contain only powers of (x − 0) = x. from part k), we know the
Taylor series of sin x around x0 = 0 and we will use it here in the following way
1 1 1 1 1 1

2
(sin x) = x − x3 + x5 − x7 + · · · · x − x3 + x5 − x7 + · · ·
3! 5! 7! 3! 5! 7!
" # " #
2 2 1 2 2 2 1 2

= x − x4 +
2
+ x6 − + x8 + + + x10 − · · ·
3! 5! (3!)2 7! 3! · 5! 9! (5!)2 3! · 7!
∞
" n #
1 1
(−1)n+1 x2n
X X
= ·
n=1 k=1 (2k − 1)! [2(n − k) + 1]!
Question 4. Use Taylor’s formula together with Lagrange’s formula for the error to calculate:
164
a) The number e with 5 decimal places of accuracy.
√
b) The number 105 with 3 decimal places of accuracy.
Answer: a) We will use the Taylor series of f (x) = ex around x0 = 0. Since ex is equal to its own
derivative, we have
ex = f (x) = f 0 (x) = f 00 (x) = f 000 (x) = f 0000 (x) = · · ·
So we get the Taylor series of f (x) = ex around x0 = 0
f (0) 0 f 0 (0) 1 f 00 (0) 2 f 000 (0)

f (x) = · (x − 0) + · (x − 0) + · (x − 0) + · (x − 0)3 + · · ·
0! 1! 2! 3!
f 00 (0) 2 f 000 (0) 3 f 0000 (0) 4
= f (0) + f 0 (0) · x + ·x + ·x + · x + ···
2! 3! 4!
e0 2 e0 3 e0 4
= e0 + e0 · x + · x + · x + · x + ···
2! 3! 4!
1 1 1
= 1 + x + · x2 + · x3 + · x4 + · · ·
2! 3! 4!
Hence
1 1 1
e = e1 = f (1) = 1 + 1 + + + + ···
2! 3! 4!
By Lagrange’s formula for the remainder,
f (n+1) (c)
Rn (x) = (x − x0 )n+1 ,
(n + 1)!
where c is between x0 = 0 and x = 1. We want 5 decimal places of accuracy. We have
f (9) (1) 9 e1 2.7 2.7

·1 = ≈ = = 0.0000074 < 0.00001
9! 9! 9! 362880
1
If we use the terms before , by the choice of c = 1, the error will be less than 0.0000074
9!
1 1 1 1 1 1 1
e≈1+1+ + + + + + + = 2.71827877 = 2.71828 rounded
2! 3! 4! 5! 6! 7! 8!
which is accurate up until the 5th decimal in comparison with the actual value of e = 2.718281828 . . .
√
b) We will use the Taylor series of f (x) = 100 + x around x0 = 0. The first few derivatives are
√ 1 √
f (x) = 100 + x = (100 + x) 2 f (0) = 100 = 10
1 1 1 1
f 0 (x) = · (100 + x)− 2 f 0 (0) = √ =
2 2· 100 20
1 −1 3 −1 −1
f 00 (x) = · · (100 + x)− 2 f 00 (0) = √ 3 =
2 2 4· 100 4000
1 −1 −3 5 3 3
f 000 (x) = · · · (100 + x)− 2 f 000 (0) = √ 5 =
2 2 2 8· 100 800000
165
√
So we get the Taylor series of f (x) = 100 + x around x0 = 0
f (0) f 0 (0) f 00 (0) f 000 (0)
f (x) = · (x − 0)0 + · (x − 0)1 + · (x − 0)2 + · (x − 0)3 + · · ·
0! 1! 2! 3!
f 00 (0) 2 f 000 (0) 3 f 0000 (0) 4
= f (0) + f 0 (0) · x + ·x + ·x + · x + ···
2! 3! 4!
1 1 3
= 10 + ·x− · x2 + · x3 − · · ·
20 4000 · 2! 800000 · 3!
Hence
√ √ 1 1 3
105 = 100 + 5 = f (5) = 10 + ·5− · 52 + · 53 − · · ·
20 4000 · 2! 800000 · 3!
f (n+1) (c)
Rn (x) = (x − x0 )n+1 ,
(n + 1)!
where c is between x0 = 0 and x = 5. We want 3 decimal places of accuracy. We have
f 000 (0) 3 3
·5 = · 53 = 0.000078 < 0.001
3! 800000 · 3!
3
If we use the terms before · 53 , by the choice of c = 0, the error will be less than 0.000078
800000 · 3!
√ 1 1
105 ≈ 10 + ·5− · 52 = 10.246875
20 4000 · 2!
which is accurate up until the 3rd decimal in comparison with the actual value of
√
105 = 10.24695077 . . .
Questions revisited, using power series and l’Hôpital’s rule
Academic Year 2015/16, Colloquium I, 20 November, 2015

q √ √
Question 11. Let an = n + 3 n − n. Calculate n→∞
lim an .
q √
Answer: We are going to expand n + 3 n as a power series. To do this, let us determine the
√
Maclaurin series (the Taylor series around x = 0) of the function 1 + x. We need to calculate the
first few derivatives to get the pattern
√ 1 √
f (x) = 1 + x = (1 + x) 2 f (0) = 1 + 0 = 1
1 1 1 1 1
f 0 (x) =
· (1 + x)− 2 f 0 (0) = · (1 + 0)− 2 =
2 2 2
1 1 1 1 1

3 3
00
f (x) = · − · (1 + x)− 2 f 00 (0) = · − · (1 + 0)− 2 = −
2 2 2 2 4
1 1 3 1 1 3 3

5 5
f 000 (x) = · − − · (1 + x)− 2 f 000 (0) = · − − · (1 + 0)− 2 =
2 2 2 2 2 2 8
166
15 7 · 15
From this one can infer that f 0000 (0) = f (4) (0) = − , f 00000 (0) = f (5) (0) = , and so on. So the
√ 16 32
Maclaurin series of 1 + x is
√ f (0) 0 f 0 (0) 1 f 00 (0) 2 f 000 (0) 3
f (x) = 1+x= ·x + ·x + ·x + · x + ···
0! 1! 2! 3!
f (0) 0 f 0 (0) 1 f 00 (0) 2 f 000 (0) 3
= ·x + ·x + ·x + · x + ···
1 1 2·1 3·2
f 00 (0) 2 f 000 (0) 3
= f (0) + f 0 (0) · x + ·x + · x + ···
2 6
1 − 14 2 38 3
=1+ ·x+ · x + · x + ···
2 2 6
1 1 2 1
=1+ ·x− ·x + · x3 − · · ·
2 8 16
3
Now we use this to calculate the limit. We let x = √ . Since we are interested in the limit of n
n
3
tending towards infinity, x = √ is close to zero and using the Maclaurin series is appropriate. We
n
get
q
√ √
lim an = lim n+3 n− n
n→∞ n→∞

√ q √
= n→∞
lim − n+ n+3 n
s
√ √
!
3
= n→∞
lim − n+ n· 1+ √
n
  !2 !3 
√ √
!
1 3 1 3 1 3
lim − n + n · 1 + · √
= n→∞ − · √ + · √ − · · · 
2 n 8 n 16 n
√ √
" !#
3 1 32 1 33
lim − n +
= n→∞ n+ − · √ + · √ − ···
2 8 n 16 ( n)2
!
3 1 32 1 33 3
= n→∞
lim − ·√ + · √ 2 − ··· =
2 8 n 16 ( n) 2
It is possible to find the limit by different means.
One can calculate the limit simply by completing the square.
q
√ √
lim n+3 n− n
n→∞
s 

√ 3
2
9 √
= n→∞
lim  n+ − − n
2 4
s 

√ 3
2
√ 9
= lim  n+ − n because is negligibly small when n tends towards infinity
n→∞ 2 4
167

√ 3 √ 3

3

= n→∞
lim n + − n = n→∞
lim =
2 2 2
Yet another way to find the limit is the following.

√ q√
q
√ √ n+3 n− n
lim n + 3 n − n = lim
n→∞ n→∞ 1
√ √ √ √
q q
n+3 n− n n−3 n+ n
= lim q √ √
n→∞
n+3 n+ n
√ 2 √
q 2
n + 3 n − ( n) √
n+3 n−n
= lim q √ √ = lim q √ √
n→∞ n→∞
n+3 n+ n n+3 n+ n
√
3 n 3
= n→∞
lim q √ √ = n→∞
lim q √
n+3 n+ n √1 n+3 n+1
n
3 3 3 3
= n→∞
lim q = n→∞
√ lim q =√ =
1
(n + 3 n) + 1 1+ √3 +1 1+1 2
n n
!n3 (n−2)
n4 − 1
Question 13. Let an = . Calculate lim an .
n4 + 1 n→∞
n
x

Answer: We will use the fact that lim = ex . Here we will actually use n4 instead of n, in
1+
n n→∞
the limit of n going to ∞ this gives the same result. The number 1 is negligibly small in proportion.
!n3 (n−2) n4
n3 (n−2) n4 −2n3 −2
4
n −1

−2

−2 1+ n4 +1
lim lim 1 +
= n→∞ lim 1 +
= n→∞ = n→∞
lim 2n3
n→∞ n4 + 1 n4 +1 n4 +1 1+ −2
n4 +1
n4 n 4
−2 −2
limn→∞ 1 + n4 +1
limn→∞ 1 + n4 +1
= 2n3 = 2 n4
−2 −2 n
limn→∞ 1 + n4 +1 limn→∞ 1 + n4 +1
n4
−2
limn→∞ 1 + n4 +1 e−2 e−2 e−2 −2 1
= n4 n2
= 2 = 0 = = 2 =

−2 limn→∞ (e−2 ) n (e−2 ) 1 e2
limn→∞ 1+ n4 +1
1
Question 14. Let an = (n + 2015) n . Calculate n→∞
lim an .
Answer: When n goes to infinity, the number 2015 is negligibly small in proportion. So we get
1
1
1 1

ln n
lim (n + 2015)
n→∞
n = n→∞
lim (n) n = n→∞
lim e n
lim e n ln n
= n→∞
168
1

Now we use the fact that n→∞
lim ln n = 0, because ln n grows more slowly than n. Thus
n
1 1

lim (n + 2015) n = lim e n ln n = e0 = 1
n→∞ n→∞
!n3 (n−2)
n4 − 1
Question 13. Let an = . Calculate lim an .
n4 + 1 n→∞
Answer: We will use l’Hôpital’s rule. But first we need to put the expression in a suitable form.
!n3 (n−2) n3 (n−2)
n4 − 1 n4 −1 n4 −1

ln n3 (n−2) ln
= e n4 +1 =e n4 +1
n4 + 1
Now we consider the exponent. When n tends to infinity, n3 (n − 2) tends to infinity, and since
n4 − 1
ln(1) = 0, the expression ln 4 tends to zero. So we can apply l’Hôpital’s rule.
n +1
!
3 n4 − 1
lim n (n − 2) ln 4
n→∞ n +1
4
 
ln nn4 −1
+1 0
= n→∞
lim  1
 of type
n3 (n−2)
0

4
0 
n −1
 ln n4 +1 
= lim  0  take the derivatives of the numerator and the denominator
n→∞ 1
n3 (n−2)
 0 
n4 +1 n4 −1
 n4 −1
· n4 +1 
= n→∞
lim 
  we applied the chain rule
−1 0

(n3 (n−2))2
(n3 (n − 2))
 

n4 +1 ( ) ( )
4n3 n4 +1 − n4 −1 4n3
 n4 −1
· (n4 +1)2 
= lim   we applied the quotient rule and the product rule
n→∞  −1

(n3 (n−2))2
(3n2 (n − 2) + n3 · 1)
 4n3 (n4 +1)−(n4 −1)4n3 
1
 n4 −1 · n4 +1
= lim

−1
 
n→∞ 2 (n − 2) + n3 )
6
n (n−2) 2 (3n
4n3 (n4 +1)−(n4 −1)4n3

 
(n4 −1)(n4 +1)
= lim 

−1


n→∞
n4 (n−2)2
(3(n − 2) + n)
(n4 +1)−(n4 −1)

 
3
 4n · (n4 −1)(n4 +1)
= n→∞
lim  
−1

n4 (n−2)2
(4n − 6)
 
2
4n3 · (n4 −1)(n4 +1) 
= lim  6−4n
n→∞
n4 (n−2)2
!
8n3 · n4 (n − 2)2
= lim
n→∞ (6 − 4n) (n4 − 1) (n4 + 1)
169
In the last expression, the highest power of n in the numerator is n9 with coefficient 8, and the
highest power of n in the denominator is n9 with coefficient −4. Dividing the numerator and the
denominator by the highest power of n, which is n9 , gives
!
8n3 · n4 (n − 2)2 8
lim = lim = −2
n→∞ (6 − 4n) (n4 − 1) (n4 + 1) n→∞ (−4)
We use the limit which we have obtained and put it in the exponent.
 !n3 (n−2) 
n4 − 1 n4 −1 1

n3 (n−2) ln
lim an = lim  4  = lim e n4 +1 = e−2 =
n→∞ n→∞ n +1 n→∞ e2
1
Question 14. Let an = (n + 2015) n . Calculate n→∞
lim an .
Answer: We will use l’Hôpital’s rule. But first we need to put the expression in a suitable form.
1
1 1
(n + 2015) n = eln(n+2015) n
= e n ·ln(n+2015)
Now we consider the exponent. It is of type zero times infinity.

1

lim · ln(n + 2015)
n→∞ n
!
ln(n + 2015) ∞
= lim of type
n→∞ n ∞
(ln(n + 2015))0
!
= n→∞
lim take the derivatives of the numerator and the denominator
(n)0
1
· (n + 2015)0
!
n+2015
= n→∞
lim we applied the chain rule
1
1 !
1

n+2015
= lim = lim =0
n→∞ 1 n→∞ n + 2015
So we get
1 1

lim e n ·ln(n+2015) = e0 = 1
lim (n + 2015) n = n→∞
n→∞
Temat 6
Question 1. For each of the following sequences, calculate the limit if the limit exists. If the limit
does not exist, explain why it does not exist.
√ √
d) an = n + 1 − n
√ √
f) an = n4 + 3n2 + 5 − n4 − n2 + n
q √ q √
n + n + 1 − n2 − n − 1
2
g) an = √ √
n+1− n
170
√
3
j) an = n n3 + n + 2 − n
q q
j) bis) an = 3 n(n + 1)2 − 3
n(n − 1)2
√n
n−1
m) an =
3 + (−1)n
v v
u u !
n
u
n
u u
n) an = t1 + t
2
√
n
ń) an = 2n + n2
√
n
o) an = 3n + 5n + 7n
s
1 n
n n
n 1 1
ó) an = + +
2 3 4
n
1

ś) an = 1 −
n
2n−1
1

t) an = 1 − 2
n
2n+1
1

u) an = 1 + n
2
2n+1
n−1

v) an =
n+3
!(−1)n n
(−1)n
w) an = 1 +
n
!2n2 −n
n2 − 1
x) an =
n2
√ !n
1+ n2
y) an =
2
Answer:
d)
√ √
lim
n→∞
n+1− n =0
√One may argue as follows: When

√ n gets large, 1 is negligibly small in proportion, so the value of
n + 1 approaches the value n, which makes the difference get closer and closer to zero.
But such a simple argument doesn’t work in all cases, so it is appropriate to mention here the power
√ 1
series expansion of the function 1 + x = (1 + x) 2 . We will find the Maclaurin series (the Taylor
√
series around x = 0) of the function 1 + x. We need to calculate the first few derivatives to get the
pattern
√ 1 √
f (x) = 1 + x = (1 + x) 2 f (0) = 1 + 0 = 1
1 1 1 1 1
f 0 (x) = · (1 + x)− 2 f 0 (0) = · (1 + 0)− 2 =
2 2 2
171
1 1 1 1 1

3 3
00
f (x) = · − · (1 + x)− 2 00
f (0) = · − · (1 + 0)− 2 = −
2 2 2 2 4
1 1 3 1 1 3 3

5 5
000
f (x) = · − − · (1 + x)− 2 000
f (0) = · − − · (1 + 0)− 2 =
2 2 2 2 2 2 8
15 7 · 15
From this one can infer that f 0000 (0) = f (4) (0) = − , f 00000 (0) = f (5) (0) = , and so on. So the
√ 16 32
Maclaurin series of 1 + x is
√ f (0) 0 f 0 (0) 1 f 00 (0) 2 f 000 (0) 3
f (x) = 1+x= ·x + ·x + ·x + · x + ···
0! 1! 2! 3!
f (0) 0 f 0 (0) 1 f 00 (0) 2 f 000 (0) 3
= ·x + ·x + ·x + · x + ···
1 1 2·1 3·2
f 00 (0) 2 f 000 (0) 3
= f (0) + f 0 (0) · x + ·x + · x + ···
2 6
1 − 14 2 38 3
=1+ ·x+ · x + · x + ···
2 2 6
1 1 2 1
=1+ ·x− ·x + · x3 − · · ·
2 8 16
√ 1
Now we use the power series expansion for 1 + x with x = to calculate the limit as follows:
n
s   s 
√ √
1

√ √ 1 √
lim n + 1 − n = lim  n 1+ − n = lim  n · 1 + − n
n→∞ n→∞ n n→∞ n
√

1 1 1 1 1 1

√
= n→∞
lim n 1+ · − · 2 + · + ··· − n
2 n 8 n 16 n3
√ √ √
√ √
" # !
1 n 1 n 1 n
= lim n+ · − · 2 + · + ··· − n
n→∞ 2 n 8 n 16 n3
  
√ 1 1 1 1 1 1  √ 
  
=  n+
lim  ·√ − · 3 + · 5 + · · ·
 − n

n→∞  2 n 8 n2 16 n 2  
| {z }
√ √
= lim n− n =0
n→∞
√
When using the Maclaurin series of 1 + x, it is necessary for x to be small, such as for example
1
x = with n going to infinity. It would not work if one chose x = n for large n, so it was indeed
n √
necessary to take the factor of n outside before doing the expansion.
f)
√ √
lim
n→∞
n4 + 3n2 + 5 − n4 − n2 + n =
172
 
v
v !
n2 − n
u
9 9
u
u 4 2
u 
= lim un + 3n + − + 5 − tn4 1− 
n→∞  u
t| {z 4} 4 n4 

complete square
 
v s
2
3 9 n2 − n 
u 
− + 5 − n2 1 −
u
2
= u n
lim u + 
n→∞  2 4 n4 
t | {z } 
negligible
s s 
2
3 n − 1

= lim  n2 + − n2 1−
n→∞ 2 n3
 s 
3 1 − n
= lim n + − n2 1 +
2
n→∞ 2 n3
  
2 3
3 1 1−n 1 1−n 1 1−n
  
lim n2 +
= n→∞ − n2 1 + − + − · · ·
  
 2 
|
2 n3 8 n3{z 16 n3 }

power series expansion
" 2 3 #!
3 1 1−n n2 1−n n2 1 − n

= n→∞
lim n2 + − n2 + − + − ···
2 2 n 8 n3 16 n3
" 2 3 #!
3 n 1 n2 1−n n2 1 − n

= lim n + − n2 −
2
+ − + − ···
n→∞ 2 2n 2n 8 n3 16 n3
  
2 3
3  1 1 n2 1−n n2 1 − n
 
 2  2
= lim n + − n − + − + − · · ·

n→∞  2  2 |2n 8 n3 {z
16 n3 }

3 1 4

= n→∞
lim 2
n + − n2 − = n→∞
lim =2
2 2 2
It is explained in part d) how to obtain the power series expansion.
g) We are interested in finding
q √ q √
n2 + n + 1 − n2 − n−1
lim √ √ .
n→∞ n+1− n
From part d), we know the appropriate power series expansion. Thus
r √ r √
√ √

n+1 n−1
q q
n + n + 1 − n2 − n − 1
2 n2 1+ n2
− n2 1 − n2
lim √ √ = lim =
√
r
n→∞ n+1− n n→∞
1
n n+ n
− n
q √ q √ q √ q √
n+1 n−1 n+1 n−1
n 1+ −n 1− n2 n2
n 1+ −n 1+
n2 −n2
= lim √ q √ = lim √ √
n→∞
n · 1 + n1 − n n→∞ n+1− n
173
√ √ √ √
n+1 2 n−1 2

1 n+1 1 1 n−1 1
n 1+ 2 n2
− 8 n 2 + ··· − n 1 + −2 −n2 8 −n2 + ···
= n→∞
lim √ h √ i
n 1 + 12 · 1
n
− 18 · 1
n2
+ ··· − n
√ √ √ √
n+1 2 n−1 2

n+1 n n−1
n+ 1
2 n
− 8 n2
+ ··· − n + 1
2
− n8
−n −n2
+ ···
= n→∞
lim h√ √
n
√
n
i √
n + 12 · n
− 81 · n2
+ ··· − n
√ √ √ √
n+1 2 n−1 2

n+1 n 1 n−1
1
2 n
− 8 n 2 + ··· − 2 −n
− n8 −n 2 + ···
= n→∞
lim 1
√
n
√
n
2
· n
− 18 · n2
+ ···
h √ √ √ √
n−1 2 n+1 2
i
n+1 n−1 n
1
2 n
+ 1
2 n
+ 8 −n2
− n8 n2
+ ···
= lim 1
√
n 1
√
n
n→∞
2
· n
− 8
· n2
+ ···
√ √ √ √
n−1 2 n+1 2

1 n+1+ n−1 n n
2 n
+ 8 −n2
− 8 n2
+ ···
= lim √
n
√
n→∞ 1
2
· n
− 81 · n2n + · · ·
√ √ √ √
n−1 2 n+1 2

n+1+ n−1 √n + √n n
1
2 n
· n n 8 −n2
− n8 n2
+ ···
= lim √ √
n→∞ 1
2
· nn · √n
n
− √n · 1 · 2n +
n 8 n
···
√ √ √ √
n−1 2 n+1 2

1 n+1+
√ n−1 √n n n
2 n
+ n 8 −n2 − 8 n2 + ···
= lim 1
n→∞
2
− 18 · 1
n
+ ···
√ √ √ √
n−1 2 n+1 2

1 √ n+1 1 √ n−1 √n n n
2 n
+ 2 n
+ n 8 −n2 − 8 n2 + ···
= lim 1
n→∞
2
− 18 · 1
n
+ ···
√ √
n−1 2 n+1 2

1
+ 12 + √n n
− n
+ ···
2 n 8 −n2 8 n2
= lim 1
n→∞
2
− 81 · 1
n
+ ···
0 when n tends to ∞
√ √
z  }| {
!2 !2 
n n n−1 n n+1 
1+ √  − + ···
n 8 −n2 8 n2 1
= lim = lim =2
n→∞ 1 1 1 n→∞ 1
2
− · + ··· 2
| 8 n
{z }
j)
s   s 
√
n+2

n+2
3 3 3
lim n n3 + n + 2 − n = lim n  n3 1+ 3
− n = lim n n 1 + − n
n→∞ n→∞ n n→∞ n3
We are going to use a power series expansion. We will find the Maclaurin series (the Taylor series
174
√
around x = 0) of the function 3 1 + x. We need to calculate the first few derivatives to get the
pattern
√ 1 √
f (x) = 3 1 + x = (1 + x) 3 f (0) = 3 1 + 0 = 1
1 2 1 2 1
f 0 (x) = · (1 + x)− 3 f 0 (0) =· (1 + 0)− 3 =
3 3 3
1 2 1 2 2

5 5
f 00 (x) = · − · (1 + x)− 3 f 00 (0) = · − · (1 + 0)− 3 = −
3 3 3 3 9
1 2 5 1 2 5 10

8 8
000
f (x) = · − − · (1 + x)− 3 000
f (0) = · − − · (1 + 0)− 3 =
3 3 3 3 3 3 27
10 · 8 10 · 8 · 11
From this one can infer that f 0000 (0) = f (4) (0) = − 4 , f 00000 (0) = f (5) (0) = , and so on.
√ 3 35
So the Maclaurin series of 3 1 + x is
√
3 f (0) 0 f 0 (0) 1 f 00 (0) 2 f 000 (0) 3
f (x) = 1+x= ·x + ·x + ·x + · x + ···
0! 1! 2! 3!
f (0) 0 f 0 (0) 1 f 00 (0) 2 f 000 (0) 3
= ·x + ·x + ·x + · x + ···
1 1 2·1 3·2
f 00 (0) 2 f 000 (0) 3
= f (0) + f 0 (0) · x + ·x + · x + ···
2 6
1 − 92 2 10
=1+ ·x+ · x + 27 · x3 + · · ·
3 2 6
1 1 2 5
=1+ ·x− ·x + · x3 − · · ·
3 9 81
√
3 n+2
Now we use the power series expansion for 1 + x with x = to calculate the limit as follows:
n3
s   s 
√
n+2

n+2
3 3
lim n n3 + n + 2 − n = lim n  3 n3 1+ 3
− n = lim n n 1 + − n =
n→∞ n→∞ n n→∞ n3
" 2 3 # !
1 n+2 1 n+2 5 n+2

= lim n n 1 + 3
− + − ··· − n
n→∞ 3 n 9 n3 81 n3
" 2 3 # !
1 n+2 1 n+2 5 n+2

2 2
= lim n 1+ 3
− + − ··· − n
n→∞ 3 n 9 n3 81 n3
" 2 3 # !
n2 n+2 n2 n+2 5n2 n+2

2 2
= n→∞
lim n + − + − ··· − n
3 n3 9 n3 81 n3
2 3 !
n2 n+2 n2 n+2 5n2 n+2

= n→∞
lim − + − ···
3 n3 9 n3 81 n3
2 3 !
n3 + 2n2 n2 n+2 5n2 n+2

= lim − + − ···
n→∞ 3n3 9 n3 81 n3
175
 
1 + 2 n2

n+2
2
5n2

n+2
3  1
n
= n→∞
lim  − + − · · · =
 
 3
|
9 n3 81
{z
n3 }
 3
√
When using the Maclaurin series of 3 1 + x, it is necessary for x to be small, such as for example
n+2
x= with n going to infinity.
n3
j) bis) We will use power series expansions, it is shown in part j) how this power series can be
computed.
q q q q
3 3 3 3
lim n(n + 1)2 − n(n − 1)2 = lim n (n2 + 2n + 1) − n (n2 − 2n + 1) =
n→∞ n→∞
s s   s s 
2n + 1 1 − 2n  2n + 1 1 − 2n 

= lim  3 n3 1+ − 3
n3 1+ = lim n 3 1 + − n
3
1 +
n→∞ n2 n2 n→∞ n2 n2
" 2 # " 2 #!
n 2n + 1 n 2n + 1 n 1 − 2n n 1 − 2n

= lim n+ 2
− + ··· − n + − + ···
n→∞ 3 n 9 n2 3 n2 9 n2
" 2 # " 2 #!
n 2n + 1 n 2n + 1 n 1 − 2n n 1 − 2n

= n→∞
lim 2
− + ··· − − + ···
3 n 9 n2 3 n2 9 n2
" ! 2 # " ! 2 #!
2n2 + n n 2n + 1 n − 2n2 n 1 − 2n

= n→∞
lim 2
− + ··· − − + ···
3n 9 n2 3n2 9 n2
   
1 2 1 2
! !
2+ n 2n + 1 −2 n 1 − 2n 2 −2 4
   
n n
= lim 
− −
+ · · · 
− + · · ·
 =

− =
n2{z n2{z

n→∞  3 |
9 }
  3 |
9 }
 3 3 3
0 when n tends to ∞ 0 when n tends to ∞
√
n
m) We first need to determine the limit of n.
√ √
n
1 1
lim n
n = lim eln n = lim eln n n
= lim e n ln n
n→∞ n→∞ n→∞ n→∞
1
Now we will use l’Hôpital’s rule to find the limit of the exponent ln n.
n
1
1 ln n 1

lim ln n = lim = lim n = lim = 0
n→∞ n n→∞ n n→∞ 1 n→∞ n
Thus
√ 1
lim
n→∞
n
lim e n ln n = e0 = 1
n = n→∞
√
Since n n tends towards one, the numerator tends towards zero. But the denominator is always
√
n
n−1
lim = 0.
n→∞ 3 + (−1)n
176
n)
v v v v
u u ! u s u s
n n! n!
u u u
n
u u n n
lim t 1+t = lim t 1+ = lim t 1+
n→∞ 2 n→∞ 2!(n − 2)! n→∞ 2(n − 2)!
v v v
u u s
n(n − 1)(n − 2) · · · 2 · 1 n(n − 1)
u u u
n
u u n
= lim 1+t = lim 1+
t t
n→∞ 2(n − 2)(n − 3)(n − 4) · · · 2 · 1 n→∞ 2
1  q  1 q
 s n
n n(n−1) 1 n(n−1)
n(n − 1)  ln 1+ 2 n
ln 1+ 2
= lim 1 + = lim e = lim e
 
2

n→∞ n→∞ n→∞
 s 
1 n(n − 1) 
Now we will use l’Hôpital’s rule to find the limit of the exponent ln 1 + .
n 2
q q
n(n−1) n2 −n
ln 1 + ln 1 +
 s 
1  n(n − 1)  2 2
lim ln 1 + = n→∞
lim lim
n→∞ n 2 n n→∞ n
− 1
q1 n2 −n
· 12 · 2
2
n− 1
2
n2 −n
1+ 2
= lim
n→∞ 1
!− 1
1 1 n2 − n 2
1

= lim q · · n−
n→∞
1+ n2 −n 2 2 2
2

1 1 1 1
2
n− 2 2
n− 2
= n→∞
lim q = n→∞
lim q =0
n2 −n n2 −n
q
n2 −n n2 −n +
1+ 2 2 2 2
So we get
v v q
u n(n−1)
1
!
ln 1+
u
n
u
n n 2
= e0 = 1
u u
lim t1 + t = lim e
n→∞ 2 n→∞
ń)
√ 1 1
n 1
lim eln(2 +n ) lim e n ln(2 +n )
n 2 n 2

lim 2n + n2 n = n→∞
n
lim
n→∞
2n + n2 = n→∞ = n→∞
1 n
Now we will use l’Hôpital’s rule to find the limit of the exponent ln 2 + n2 .
n
1
1 n 2
ln (2n + n2 ) 2n +n2
· (2n ln 2 + 2n) 2n ln 2 + 2n
lim ln 2 + n = lim = n→∞
lim = n→∞
lim
n→∞ n n→∞ n 1 2n + n2
2n (ln 2)2 + 2 2n (ln 2)3 2n (ln 2)4 (ln 2)4
= lim = lim = lim = lim
n→∞ 2n ln 2 + 2n n→∞ 2n (ln 2)2 + 2 n→∞ 2n (ln 2)3 n→∞ (ln 2)3
177
(ln 2)4
= = ln 2
(ln 2)3
So we get
√ 1
2n + n2 = lim e n ln(2 +n ) = eln 2 = 2
n n 2
lim
n→∞ n→∞
o)
√ 1 n +5n +7n )
1 1 n +5n +7n )
lim n
3n + 5n + 7n = lim (3n + 5n + 7n ) n = lim eln(3 n
= lim e n ln(3
n→∞ n→∞ n→∞ n→∞
1
Now we will use l’Hôpital’s rule to find the limit of the exponent ln (3n + 5n + 7n ).
n
1
1 ln (3n + 5n + 7n ) 3n +5n +7n
(3n ln 3 + 5n ln 5 + 7n ln 7)
lim ln (3n + 5n + 7n ) = lim = lim
n→∞ n n→∞ n n→∞ 1
3n n
3n ln 3 + 5n ln 5 + 7n ln 7 7n
ln 3 + 57n ln 5 + ln 7
= n→∞
lim = n→∞
lim 3n n
3n + 5n + 7n 7n
+ 57n + 1
n n
3 5
7
ln 3 + 7
ln 5 + ln 7 ln 7
= n→∞
lim n n = = ln 7
3
+ 5
+1 1
7 7
So we get
√ 1 n n n
lim n
3n + 5n + 7n = lim e n ln(3 +5 +7 ) = eln 7 = 7
n→∞ n→∞
ó)
s
n n n n n n n1
n 1 1 1 1 1 1
lim + + = n→∞
lim + +
n→∞ 2 3 4 2 3 4
n n n
1 n n n
n
ln(( 21 ) +( 13 ) +( 14 ) ) 1 1 1 1
= lim e = lim e n ln(( 2 ) +( 3 ) +( 4 ) )
n→∞ n→∞
n n n
1 1 1 1
Now we will use l’Hôpital’s rule to find the limit of the exponent ln + + .
n 2 3 4
n n n
1 1 1
1
n
1
n
1
n
1 ln 2
+ 3
+ 4
lim ln + + = n→∞
lim
n→∞ n 2 3 4 n
n n n
1 1
1 n n
+ 13 + 41
n ln 21 + 1
ln 13 + 1
ln 41
( ) ( ) ( )
2
2 3 4
= lim
n→∞ 1
n n n
1
2
ln 12 + 1
3
ln 31 + 1
4
ln 14
= lim n n n
n→∞ 1 1 1
2
+ 3
+ 4
n n n
2
2
ln 12 + 2
3
ln 31 + 2
4
ln 14
= lim n n n
n→∞ 2 2 2
2
+ 3
+ 4
178
n n
ln 12 + 2
3
ln 13 + 2
4
ln 14 ln 12 1
= n→∞
lim n n = = ln
1+ 2
+ 2 1 2
3 4
So we get
s
n n n
1 1 1 1 1 n 1 n 1 n 1
= lim e n ln(( 2 ) +( 3 ) +( 4 ) ) = eln 2 =
n 1
lim + +
n→∞ 2 3 4 n→∞ 2
ś)
n n
1

1 1
ln(1− n )
lim 1− = lim e = lim en ln(1− n )
n→∞ n n→∞ n→∞
1

Now we will use l’Hôpital’s rule to find the limit of the exponent n ln 1 − .
n
1 (−1)(−1) 1 1

1
ln 1 − 1
n
1
1− n
· n2 1 · n2
1− n −1
lim n ln 1 − = lim 1 = lim −1 = lim −1 = lim = −1
n→∞ n n→∞
n
n→∞
n2
n→∞
n2
n→∞ 1 − n1
So we have
n
1 1

1
lim 1− = lim en ln(1− n ) = e−1 =
n→∞ n n→∞ e
t)
2n−1 2n−1
1

1 1
ln(1− )
lim 1− 2 = lim e n2 = lim e(2n−1) ln(1− n2 )
n→∞ n n→∞ n→∞
1

Now we will use l’Hôpital’s rule to find the limit of the exponent (2n − 1) ln 1 − .
n2
1 (−1)(−2)
1
ln 1 − 1
1− 12
· n3
2
n2 n3 −n
lim (2n − 1) ln 1 − 2 = n→∞
lim 1 = n→∞
lim n
−2 = lim −2
n→∞ n 2n−1 (2n−1)2
n→∞
(2n−1)2
−1
n3 −n −(2n − 1)2 − (4n2 − 4n + 1)
= lim 1 = lim = lim =0
n→∞
(2n−1)2
n→∞ n3 − n n→∞ n3 − n
So we get
2n−1
1

1
lim 1− 2 lim e(2n−1) ln(1− n2 ) = e0 = 1
= n→∞
n→∞ n
u)
2n+1 2n+1
1

ln(1+ 21n ) n+1 ln(1+ 21n )
lim 1+ n = lim 2 = lim 22
n→∞ 2 n→∞ n→∞
179
1

n+1
Now we will use l’Hôpital’s rule to find the limit of the exponent 2 ln 1 + n .
2

1 1

1
ln 1 + 2n
ln 1 + 2n ln (1 + 2−n )
lim 2n+1 ln 1 + n = lim 1 = lim 1 1 = lim 1
n→∞ 2 n→∞
2n+1
n→∞
2
· 2n
n→∞
2
· 2−n
1
1+2−n
ln 2(−1)2−n 2
= lim 1 = lim =2
n→∞ ln 2(−1)2−n n→∞ 1 + 2−n
2
So we get
2n+1
1

n+1 ln(1+ 21n )
lim 1+ = lim 22 = e2
n→∞ 2n n→∞
v)
2n+1
n−1
2n+1
n−1 n−1
lim lim eln n+3
= n→∞ lim e(2n+1) ln n+3
= n→∞
n→∞ n+3
n−1
Now we will use l’Hôpital’s rule to find the limit of the exponent (2n + 1) ln .
n+3
n−1 n+3 (1)(n+3)−(n−1)(1) n+3 4
n−1 ln n+3 n−1
· (n+3)2 n−1
·
(n+3)2
lim (2n + 1) ln = lim 1 = n→∞
lim −2 = n→∞
lim −2
n→∞ n + 3 n→∞ 2n+1 (2n+1)2 (2n+1)2
n + 3 −2(2n + 1)2 n + 3 −2 (4n2 + 4n + 1)

= n→∞
lim · = lim · = −8
n−1 (n + 3)2 n→∞ n − 1 n2 + 6n + 9
So we get
2n+1
n−1 1

n−1
lim = lim e(2n+1) ln n+3 = e−8 =
n→∞ n+3 n→∞ e8
w) We are interested in finding the limit of the sequence

!(−1)n n
(−1)n
an = 1 + .
n
We know that n
1

lim 1 + = e.
n→∞ n
We need to check whether the limit of the sequence
−n
−1

an = 1 +
n
is also e.
From part ś), we know that n
−1 1

lim 1 + = .
n→∞ n e
180
Thus −n n −1 n −1 −1
−1 −1 −1 1

lim 1+ = lim 1+ = lim 1+ = = e.
n→∞ n n→∞ n n→∞ n e
So we have !(−1)n n
(−1)n
lim 1+ = e.
n→∞ n
x)
!2n2 −n 2n2 −n
n2 − 1 n2 −1 2

ln 2 −n) ln n −1
lim = lim e n2 = lim e(2n n2
n→∞ n2 n→∞ n→∞
n2 − 1
Now we will use l’Hôpital’s rule to find the limit of the exponent (2n2 − n) ln .
n2
2 n2 2n(n2 )−2n(n2 −1) n2
2 n2 − 1 ln nn−1
2 n2 −1
· n4
· 2n
n2 −1 n4
lim (2n − n) ln lim = lim −4n+1 = lim
n→∞ n 2 n→∞ 1 n→∞ n→∞ −4n+1
2n2 −n (2n2 −n)2 (2n2 −n)2
1
·2
n2 −1 n
2
n3 −n 2(2n2 − n)2
= lim = lim = lim
n→∞ −4n+1 n→∞ −4n+1 n→∞ (n3 − n)(−4n + 1)
(2n2 −n)2 (2n2 −n)2
2 (4n4 − 4n3 + n2 ) 8
= lim = = −2
n→∞ (n − n)(−4n + 1)
3 −4
So we have
!2n2 −n
n2 − 1 2
2 −n) ln n −1 1
lim lim e(2n
= n→∞ n2 = e−2 =
n→∞ n2 e2
y)
√ !n √ n √
1+ n2 n n

ln 1+2 2 n ln 1+2 2
lim = lim e = lim e
n→∞ 2 n→∞ n→∞
√
1+ n2
Now we will use l’Hôpital’s rule to find the limit of the exponent n ln .
2
√ √
n 1 2
1+ n2 ln 1+2 2 ln 1+22 n 1
1+2 n
lim n ln = lim 1 = lim 1 = lim
n→∞ −1
n→∞ 2 n→∞
n
n→∞
n n2
2 ln 2 1 −1
1 · 2
· 2n · n2 2 ln 2 1
1+2 n
= lim −1 = lim 1 · · 2n
n→∞
n2
n→∞ 1+2 n 2
1
2n 1 1
= lim 1 · ln 2 = ln 2 = ln 2
n→∞ 1+2 n 1+1 2
So we have
√ !n
1+ n2 √
1+ n 2 1
1 1 √
lim lim en ln 2 = e 2 ln 2 = eln 2 2 = 2 2 = 2
= n→∞
n→∞ 2
181
Academic Year 2015/16, Colloquium II, 15 January, 2016
Question 1. Calculate the value of the following series:
∞ n
X π
cos
n=1 6
Answer:
The sum of the inside angles of a triangle is π, which is the same as 180◦ . In the picture, we have an
π
√
π π 3
equilateral triangle divided in two, α = 3 = and hence cos = .
2 6 6 2
Now we use the geometric series
1
= 1 + x + x2 + x3 + x4 + · · ·
1−x
√
π 3
which converges for |x| < 1. With x = cos = we get
6 2
∞ n 1 2 3 4
π π π π π
X
cos = cos + cos + cos + cos + ···
n=1 6 6 6 6 6
2 3 4
π π π π

= cos + cos + cos + cos + ···
6 6 6 6
1 1 2
= π −1 =
√ −1 = √ −1
1 − cos 6 1 − 23 2− 3
√ √
2 2+ 3 4+2 3 √
= √ √ −1= −1=3+2 3
2− 3 2+ 3 4−3
182
Question 2. Determine whether the following series is convergent or divergent
∞
X (n!)3 (28)n
n=1 (3n)!
Answer: We will use the ratio test.

[(n+1)!]3 (28)n+1
an+1 [3(n+1)]! [(n + 1)!]3 · (3n)! · 28 [(n + 1)!]3 · (3n)! · 28
lim = lim (n!)3 (28)n
= lim = lim
n→∞ a
n
n→∞ n→∞ (n!)3 · [3(n + 1)]! n→∞ (n!)3 · (3n + 3)!
(3n)!
[(n + 1)(n)(n − 1) · · · 2 · 1]3 · [3n(3n − 1)(3n − 2) · · · 2 · 1] · 28

= n→∞
lim
[n(n − 1)(n − 2) · · · 2 · 1]3 · [(3n + 3)(3n + 2)(3n + 1) · · · 2 · 1]
(n + 1)3 · 28 28 28
= lim = = >1
n→∞ (3n + 3)(3n + 2)(3n + 1) 3·3·3 27
So by the ratio test, the series diverges.
Question 3. For each α ∈ R, find the radius of convergence of the following series:
∞ nα
2 sin n1
xn
X
n=0 3
∞
cn xn , which says that the
X
Answer: There is a formula for the convergence of a power series
n=0
radius of convergence is
1
R= ,
s
where q
n
s = lim sup |cn |
1
n
and lim sup denotes the supremum limit. We will use the fact that lim = 1.
n→∞ sin n1
v v 1
u nα u nα nα1 n

q u 2 1 u
n 2 sin n1 2
n n sin n sin n
s = lim sup |cn | = lim sup = lim sup = lim sup 
u t 
t
3 3 3
1· nα nα nα

2 n sin 1
n 2 2
= lim sup = lim sup = n→∞
lim
3 3 3
2
lim nα = n→∞
For α = 0, we have n→∞ lim n0 = n→∞
lim 1 = 1, thus s = and the radius of convergence is
3
1 3
R= =
s 2
1
lim nα = ∞, thus s = 0 and the radius of convergence is R =
For α > 0, we have n→∞ =∞
0
1
lim nα = 0, thus s = 1 and the radius of convergence is R =
For α < 0, we have n→∞ =1
1
183
Question 4. Find a simple expression for h−1 (f (g(x))), where f (x) = xx , g(x) = x9 and h(x) = x3 .
1
(x9 )
Answer: We have h−1 (x) = x 3 and f (g(x)) = x9 . Thus
!1
(x9 ) 3 9
1 9
−1 9
h (f (g(x))) = x = x9x 3
= x3x
1
Question 5. Determine whether the function f : (2, 4) → R, f (x) = √
2− x
(a) has a maximum
(b) has a minimum
(c) is strictly increasing
Answer: We need to look at the derivative
!0
0 1 −1 √ 0 −1 −1 1
f (x) = √ = √ 2 2− x = √ 2· √ = √ √ 2
2− x (2 − x) (2 − x) 2 x 2 x (2 − x)
The derivative is always positive. So the function is strictly increasing. The domain of definition is
open on both sides, and since the function is strictly increasing, it follows that the function does not
have a minimum and does not have a maximum.
Question 6. Calculate the following limit if it exists, otherwise state that it does not exist
cos(2x)
lim
x→π
√
1 − 2 sin x
Answer:
cos(2x) 1
lim √ = √ =1
x→π
1 − 2 sin x 1− 2·0
√ √
lim
x→∞
ln x + 1 − ln x − 1
Answer:
√ √
√ √ ln x + 1 − ln x − 1
lim ln x + 1 − ln x − 1 = lim
x→∞ x→∞ 1
√ √ √ √
ln x + 1 − ln x − 1 ln x + 1 + ln x − 1
= lim √ √
x→∞ ln x + 1 + ln x − 1
√ 2 √ 2
ln x + 1 − ln x − 1 ln x + 1 − (ln x − 1)
= x→∞
lim √ √ lim √
= x→∞ √
ln x + 1 + ln x − 1 ln x + 1 + ln x − 1
2
lim √
= x→∞ √ =0
ln x + 1 + ln x − 1
184
1
lim e− x
x→0
Answer: When approaching zero from the right, we have the „right–side” limit
1
lim = +∞
x→0+ x
When approaching zero from the left, we have the „left–side” limit
1
lim− = −∞
x→0 x
So
1
lim+ e− x = e−∞ = 0
x→0
and
1
lim− e− x = e+∞ = +∞
x→0
Since the „right–side” and the „left–side” limits are different, the limit does not exist.
Question 9. Find all triples of real numbers (a, b, c) ∈ R3 , for which the function f : R → R given by

7
(ax + b)
 if x ¬ 0
f (x) = c
x cos

if x > 0
x
is continuous.
c
Answer: For any real number c, the value of cos will always be between 0 and 1, so when x
x
c
is approaching 0, the value of the expression x cos will go to zero. This means that (ax + b)7
x
evaluated at x = 0 needs to be zero. So b = 0 and a can be chosen arbitrarily. Thus for b = 0 and
any real numbers a and c, the function is continuous.
π

Question 10. Lef f : R → R, f (x) = ln arc tg x + .
2
(a) Calculate f 0 (x).
π

0
(b) Find f tg .
6
Answer: (a) We have
0 0
π 1 π 1

0
f (x) = ln arc tg x + = π · arc tg x + = π · (arc tg x)0
2 arc tg x + 2
2 arc tg x + 2
Now we need to know the derivative of arc tg x. For g(x) = tg x, we have g −1 (x) = arc tg x. We need
the derivative of g(x).
0
sin x (sin x)0 cos x − sin x(cos x)0

0 0
g (x) = (tg x) = =
cos x (cos x)2
185
cos x cos x − sin x(− sin x) cos2 x + sin2 x 1
= = =
(cos x)2 cos2 x cos2 x
So we get
h i0 1 1 h i
(arc tg x)0 = g −1 (x) = = 1 = cos2 g −1 (x) = cos2 (arc tg x)
g 0 [g −1 (x)] cos2 [g −1 (x)]
and
1 1 cos2 (arc tg x)
f 0 (x) = π · (arc tg x)0 = π · cos2 (arc tg x) =
arc tg x + 2
arc tg x + 2
arc tg x + π2
(b)
√ 2
0

π
cos2 arc tg tg π6 cos2 π 2
3
9
f tg = = π π6 = 2π =
6 arc tg tg π6 + π
6
+2 3
8π
2
Academic Year 2016/17, Colloquium II, 24 January, 2017

Question 1. Find the radius of convergence of the following series:
∞
X (3n)! · xn
n
n=1 9 · (n!)
3
Answer: We will use the ratio test

[3(n+1)]!·xn+1 (3n+3)!·xn+1 (3n+3)(3n+2)(3n+1)···2·1·xn+1
an+1 9n+1 ·[(n+1)!]3 9n+1 ·[(n+1)!]3 9n+1 ·[(n+1)n(n−1)···2·1]3
lim
n→∞ a
= n→∞
lim (3n)!·xn
= lim
n→∞ (3n)!·xn
= lim
n→∞ (3n)(3n−1)(3n−2)···2·1·xn
n
9n ·(n!)3 9n ·(n!)3 9n ·[n(n−1)(n−2)···2·1]3
(3n + 3)(3n + 2)(3n + 1) (3n + 3)(3n + 2)(3n + 1)

= lim · x = lim · lim |x|
n→∞ 9 · (n + 1) 3 n→∞ 9 · (n + 1)3 n→∞
33
= · |x| = 3 · |x| < 1
9
1
So for the series to converge absolutely, we need to have |x| < , which means that the radius of
3
1
convergence is .
3
Question 2. Is the following series convergent or divergent:

∞
1
X
cos(nπ) · ln 1 +
n=1 n
1

Answer: ln 1 + is always positive and monotone decreasing. It approaches zero when n tends
n
towards infinity, because ln(1) = 0. The multiplicative factor before it, cos(nπ), alternates between
−1 and 1, because cos(nπ) = 1 when n is even and cos(nπ) = −1 when n is odd. Thus by the
alternating series test, the series is convergent.
186
Question 3. Calculate the following limit if it exists, otherwise state that it does not exist:
sin(4x)
limπ √
x→ 4 (1 − 2 cos x)
Answer:
The sum of the inside angles of a triangle is π, which is the same as 180◦ .
π
π− π 2 π π 1
In the picture, α = . Thus cos = sin = √
=
2 4 4 4 2
0
The expression is of the form . So we will use l’Hôpital’s rule.
0
sin(4x) [sin(4x)]0 4 cos(4x) 4 cos π 4 · (−1)
limπ √ = limπ √ = lim √ = √ π
= √ = −4
x→ 4 (1 − 2 cos x) x→ 4 (1 − 2 cos x)0 x→ 4 ( 2 sin x)
π
( 2 sin 4 ) 2 · √12
Question 4. Determine whether the following function is continuous at x = 1

1

e− 1−x2 − 1 if |x| < 1
f (x) =
ln x2 if |x| 1
Answer: We have
1
lim− = +∞
x→1 1 − x2
Thus
1 1

1
−
lim− e 1−x2 − 1 = e−∞ − 1 = − 1 = − 1 = 0 − 1 = −1
x→1 e∞ ∞

We have ln 12 = ln 1 = 0. Thus the function f (x) is not continuous at x = 1.
187
Question 5. Let f : R → R be the function defined by
x2 + 2
f (x) = .
x2 + 1
Find the least upper bound (sup), and greatest lower bound (inf), for the values of f . Give all points
where f reaches a local or global extremum.
Answer: The first thing to notice is that this is an even function, meaning f (−x) = f (x). So we can
use symmetry with respect to the y-axis. At x = 0, the value of the function is 2. When moving away
from 0, both the numerator and the denominator increase. The numerator is always bigger than the
denominator, so it increases less in proportion when moving away from 0. As a result, the value of
the function decreases when moving away from 0. When x gets very large, the value of the function
approaches 1. Thus:
- The least upper bound is 2, and it is reached at x = 0.
- The greatest lower bound is 1, and it is never reached.
- There are no points where an extremum is reached other than the point (0, 2).
Alternatively, one may use calculus to check for extrema. The derivative of f (x) is
2x (x2 + 1) − 2x (x2 + 2) −2x

f 0 (x) = 2 =
(x2 + 1) (x2 + 1)2
So the derivative is zero only when x = 0.
Question 6. Let f : R → R be the function defined by
f (x) = x2 + 3x.
Give the equation of the tangent line of the function f at the point (1, 4).
Answer: The derivative of f (x) is
f 0 (x) = 2x + 3
The slope of the graph of the function f (x) at the point (1, 4) is f 0 (1) = 5. So we need the equation
of a line which goes through the point (1, 4) and which has slope 5. A Cartesian equation of a line
with slope 5 is of the form y = 5x + b. Since the line needs to go through (1, 4), we get b = −1. Thus
the equation of the tangent line of the function f at the point (1, 4) is
y = 5x − 1
188
Academic Year 2015/16, Examination, 25 January, 2016
Question 1. Calculate the following limit:
1 · 2 + 3 · 3 + 5 · 4 + · · · + (2n − 1)(n + 1)
lim
n→∞ n3
Answer: We will make use of the known facts that
n
X
k = 1 + 2 + 3 + ··· + k
k=1
n(n + 1)
=
2
and
n
k 2 = 12 + 22 + 32 + · · · + k 2
X
k=1
n(n + 1)(2n + 1)
=
6
Note also that n
X
1 = |1 + 1 + 1{z+ · · · + 1} = n
k=1 n times
So we get
n
1 · 2 + 3 · 3 + 5 · 4 + · · · + (2n − 1)(n + 1) X (2k − 1)(k + 1)
lim 3
= lim
n→∞ n n→∞
k=1 n3
n
X 2k 2 + k − 1
= n→∞
lim
k=1 n3
n n n
X 2k 2 X k X 1
= lim 3
+ lim 3
− lim 3
k=1 n k=1 n k=1 n
n→∞ n→∞ n→∞
Pn Pn Pn
2 k=1 k2 k=1 k k=1 1
= lim + lim − lim
n→∞ n3 n→∞ n3 n→∞ n3
n(n+1)(2n+1) n(n+1)
2· 6 2 n
= lim + lim − lim
n→∞ n3 n→∞ n3 n→∞ n3
4 2
= +0−0=
6 3
Question 2. Does the following series converge or diverge?

∞
X 2n · n!
n=1 nn
189
Answer: We will make use of the „arithmetic mean” „geometric mean” inequality, which states that
for any number n of nonnegative real numbers
x1 + x2 + · · · + xn √
n x1 x2 · · · xn
n
Justify your answer.
We will use the Cauchy condensation test in combination with the nth root test. By the Cauchy
condensation test, the series
∞
X 2n · n!
n=1 nn
converges if and only if the series
2(2 ) · 2k !
k

∞
k
X
2 ·
(2k )(2 )
k
k=1
converges. Now we will use the kth root test (same as the nth root test, just with k)
v v  1
2(2 ) 2k ! 
k k

2(2 ) (2k )! (2k ) (2k )!
k
u u
u
k
u
k 2 k
lim 2k · = lim t2k · = lim 2 ·
u u 
t
(2k )(2 ) (2k )(2 ) (2k )(2 )
k k k

k→∞ k→∞ k→∞

  k1 2k i1
( ) 2k ! 2k
h
 k 2 2 k
2k ! k
= lim 2 · = lim 2 ·

2(k·2 )
k

k→∞ k→∞ k·2k
k
2

2k h i1
k k k
2 2 !
= lim 2 ·
2(2 )
k
k→∞

2k h i1
2 k
2k 2k − 1 2k − 2 · · · 3 · 2 · 1 k
= lim 2 ·
2(2 )
k
k→∞

2k
k 2k +(2k −1)+(2k −2)+···+3+2+1
2 · 2k
¬ lim 2 ·
k→∞ (2k ) 2
k

2 h i
2 k
2k + 2k − 1 + 2k − 2 + · · · + 3 + 2 + 1
= lim 2 ·
2k · 2(2 )
k
k→∞

2k

2k
( )(
2k 2k +1 )

2 k
· 2
2 k
2k
2k + 1
= lim 2 · = lim
· 2(2 ) 2k · 2(2 )
k k
k→∞ k→∞
2k
k

2
2 k
2k + 1 2k + 1
= lim = lim (2k ) = 0
2(2 )
k
k→∞ k→∞
2
1
2k
2
The last equality follows from the fact that for k 2, we have 1 > 1. This implies that the
2k
190
numerator grows linearly with 2k and the denominator grows exponentially with 2k , which makes
the limit zero. Since v
2(2 ) (2k )!
k
u
u
k k
lim t 2 · = 0 < 1,
u
(2k )(2 )
k
k→∞
by kth root test, the series

(2k ) 2k !

∞ 2
2k ·
X
(2k )(2 )
k
k=1
converges. Hence by the Cauchy condensation test, the series

∞
X 2n · n!
n=1 nn
converges.
Question 3. Calculate the limit

1 1

lim (1 + x) x − e
x→0 x
Answer: Since the question implies that the limit exists, we only need to compute the „right–side”
1
limit. When x approaches zero from the right, then y = approaches infinity. So
x
!y
1 1
lim (1 + x) = lim x 1+ =e
x→0+ y→+∞ y
0
This means that the limit that we want to compute is of the form . We will use l’Hôpital’s rule
0
three times
1
1 1
1
(1 + x) x − e eln(1+x) x
−e 1
e x ·ln(1+x) − e
lim (1 + x) x − e = lim = lim = lim
x→0 x x→0 x x→0 x x→0 x
1
0 1
0 1
0
e x ·ln(1+x) − e e x ·ln(1+x) − (e)0 e x ·ln(1+x) − 0
= lim = lim = lim
x→0 (x)0 x→0 (x)0 x→0 1
0
1
0
1 1

= lim e x ·ln(1+x) = lim e x ·ln(1+x) · ln(1 + x)
x→0 x→0 x
0 0
1 1

1
= lim (1 + x) x · ln(1 + x) = lim e · · ln(1 + x)
x→0 x x→0 x
0 " 0 #
1 1 1

= e · lim · ln(1 + x) = e · lim · ln(1 + x) + · (ln(1 + x))0
x→0 x x→0 x x
−1 1 1 −1 1 x

= e · lim 2 · ln(1 + x) + · = e · lim 2 · ln(1 + x) + 2 ·
x→0 x x 1+x x→0 x x 1+x
x 1
− ln(1 + x) + 1+x
− ln(1 + x) + 1 − 1+x
x→0 x2 x→0 x2
0
1 1 −1
− ln(1 + x) + 1 − 1+x
− 1+x − (1+x)2
x→0 (x2 )0 x→0 2x
191
0
1 1 1 1
− 1+x + (1+x)2
− 1+x + (1+x)2
x→0 2x x→0 (2x)0
−1 −2 1 2
− (1+x) 2 + (1+x)3 (1+x)2
− (1+x)3 1−2 e
= e · lim = e · lim =e· =−
x→0 2 x→0 2 2 2
Question 4. Let T denote the tangent line to the graph of the function f (x) = 3x at the point x = 1.
Find the point where T intersects the line y = x.
Answer: We need the derivative of the function f (x)
h x i 0 0
f 0 (x) = (3x )0 = eln 3 = ex ln 3 = ex ln 3 (x ln 3)0 = 3x (x ln 3)0 = 3x ln 3
No we need the slope of f (x) at the point x = 1. To slope is
f 0 (1) = 31 ln 3 = 3 ln 3
The tangent line must be a line with slope 3 ln 3. A Cartesian equation for a line with slope 3 ln 3 is
y = (3 ln 3)x + b
Since the point (1, 3) is part of the line, we get b = 3 − 3 ln 3. The intersection with the line y = x is
obtained by putting
x = y = (3 ln 3)x + 3 − ln 3
!
3 − ln 3 3 − ln 3 3 − ln 3
which gives x = . Hence the point of intersection is , .
1 − 3 ln 3 1 − 3 ln 3 1 − 3 ln 3
Question 5. Prove that for all x, y ∈ R the following inequality holds:
| arc tg(x) − arc tg(y)| ¬ |x − y|
Answer: If we can show that the absolute value of the derivative of arc tg(x) is always smaller
than or equal to 1, the result follows by the mean value theorem. So we calculate the derivative of
arc tg(x). Let x = tg(y). Then y = arc tg(x) and we have
!0
= (tg(y))0 = = 2
=
dy cos y cos y cos2 y
cos2 y + sin2 y 1
= 2
=
cos y cos2 y

= = + = 1 + tg2 y
dy cos2 y cos2 y cos2 y
dy 1 1 1 1 1
(arc tg x)0 = = dx = 2
= 2
= 2
=
dx dy
192
1
Since ¬ 1 for all x, we have |(arc tg x)0 | ¬ 1 for all x, which is what we need.
1 + x2
Question 6. Find the point on the graph of the parabola y = x2 that is closest to the point (3, 0).
Answer: We will use a function whose value is the square of the distance and call this function S(x).
2
S(x) = (x − 3)2 + (y − 0)2 = (x − 3)2 + x2 − 0 = (x − 2)2 + x4
The distance is minimal when the square of the distance is minimal. We calculate the the derivative
of S(x)
S 0 (x) = 2(x − 3) + 4x3 = 4x3 + 2x − 6
The derivative needs to be zero for an extremum to occur. By inspection, we find that S 0 (1) = 0.
Could there be more critical points ? We can factor
h i
S 0 (x) = 4x3 + 2x − 6 = (x − 1)(4x2 + 4x + 6) = (x − 1) (2x + 1)2 + 5
and the second factor is never zero. So the only possibility for an extremum is at x = 1. This must
be a minimum. So the point on the graph of the parabola y = x2 that is closest to the point (3, 0) is
(1, 1).
Academic Year 2015/16, Examination, 2 February, 2016

6 · 2 + 13 · 3 + 20 · 4 + · · · + (7n − 1)(n + 1)
lim
n→∞ n3

n
X
k = 1 + 2 + 3 + ··· + k
k=1
n(n + 1)
=
2
and
n
k 2 = 12 + 22 + 32 + · · · + k 2
X
k=1
n(n + 1)(2n + 1)
=
6
Note also that n
X
1 = |1 + 1 + 1{z+ · · · + 1} = n
k=1 n times
So we get
n
6 · 2 + 13 · 3 + 20 · 4 + · · · + (7n − 1)(n + 1) X (7k − 1)(k + 1)
lim = lim
n→∞ n3 n→∞
k=1 n3
193
n
X 7k 2 + 6k − 1
= lim
n→∞
k=1 n3
n n n
7k 2
X X 6k X 1
= n→∞
lim 3
+ n→∞
lim 3
− n→∞
lim 3
k=1 n k=1 n k=1 n
Pn Pn Pn
7 k=1 k2 6 k=1 k k=1 1
= lim + lim − lim
n→∞ n3 n→∞ n3 n→∞ n3
n(n+1)(2n+1) n(n+1)
7· 6
6· 2 n
= lim + lim − lim
n→∞ n3 n→∞ n3 n→∞ n3
14 7
= +0−0=
6 3

∞
X 3n (n + 4)!
n=1 nn
Answer: We will make use of the „arithmetic mean” „geometric mean” inequality, which states that
for any number n of nonnegative real numbers
x1 + x2 + · · · + xn √
n x1 x2 · · · xn
n
We will use the Cauchy condensation test in combination with the nth root test. By the Cauchy
condensation test, the series
∞
X 3n (n + 4)!
n=1 nn
converges if and only if the series
(2k ) 2k + 4 !

∞ 3
2k ·
X
(2k )(2 )
k
k=1
converges. Now we will use the kth root test (same as the nth root test, just with k)
v v  1
3(2 ) 2k + 4 ! 
k k

3(2 ) (2k + 4)! 3(2 ) (2k + 4)!
k k
u u
u u
k k k
lim 2k · k
= lim t2 · = lim 2 ·
u u 
t
(2k )(2 ) (2k )(2 ) (2k )(2 )
k k k

k→∞ k→∞ k→∞

  k1 2k i1
( ) 2k + 4 !2k
h
 k 3 3 k
2k + 4 ! k
= lim 2 · = lim 2 ·

2(k·2 )
k

k→∞ k→∞ k·2k
k
2

2k h i1
3 k
2k + 4 ! k
= lim 2 ·
2(2 )
k
k→∞

2k h i1
3 k
2k + 4 2k + 3 2k + 2 · · · 3 · 2 · 1 k
= lim 2 ·
2(2 )
k
k→∞
194

2k
k (2k +4)+(2k +3)+(2k +2)+···+3+2+1
3 2k +4
¬ lim 2 ·
k→∞
2( ) 2k

2k h i
3 k
2k + 4 + 2k + 3 + 2k + 2 + · · · + 3 + 2 + 1
= lim 2 ·
(2k + 4) · 2(2 )
k
k→∞

2k

2k
(2k +4)(2k +5)

3 k
· 3 k
2k + 4 2k + 5
2
= lim 2 · = lim
k→∞
(2k + 4) · 2 (2k ) k→∞
(2k + 4) · 2(2 )
k

2k
3 k
2k + 5 2k + 5
= lim = lim (2k ) = 0
2(2 )
k
k→∞ k→∞
2
1
3k
2
The last equality follows from the fact that for k 2, we have 1 > 1. This implies that the
3k
numerator grows linearly with 2k and the denominator grows exponentially with 2k , which makes
the limit zero. Since v
3(2 ) (2k + 4)!
k
u
u
k k
lim t 2 · = 0 < 1,
u
(2k )(2 )
k
k→∞
by kth root test, the series

(2k ) 2k + 4 !

∞ 3
2k ·
X
(2k )(2 )
k
k=1
converges. Hence by the Cauchy condensation test, the series

∞
X 3n (n + 4)!
n=1 nn
converges.
Question 3. Calculate the limit:

1
lim (cos x) x2
x→0
Answer: 1
1 1
lim (cos x) x2 = lim eln cos x x2
= lim e x2 ·ln cos x
x→0 x→0 x→0
Now we need to find the limit of the exponent. We will use l’Hôpital’s rule twice.
1 ln cos x (ln cos x)0 1

cos x
(cos x)0 1
cos x
(− sin x)
lim · ln cos x = lim = lim = lim = lim
x→0 x2 x→0 x2 x→0 (x2 )0 x→0 2x x→0 2x
0 0 cos x−(cos x)0 sin x

sin x
− cos − cos x sin x
− (sin x) (cos x)2
x
= lim = lim = lim
x→0 2x x→0 (2x)0 x→0 2
(sin x)0 cos x − (cos x)0 sin x (cos x) cos x − (− sin x) sin x
= lim − 2
= lim −
x→0 2(cos x) x→0 2(cos x)2
195
(cos x)2 + (sin x)2 1 1
= lim − = lim − = −
x→0 2(cos x)2 x→0 2(cos x)2 2
So we get
1 1 1 1 1
lim (cos x) x2 = lim e x2 ·ln cos x = e− 2 = 1=√
x→0 x→0 e2 e
√
Question 4. Let T be the tangent line to the graph of the function f (x) = − 2x − x2 at the point
3 1
x = . Find the points where T intersects the parabola y = − x2 .
2 12
Answer: We need the derivative of f (x)
0
√ 0 1 0
2 2 1 1 −1 0
f (x) = − 2x − x = − 2x − x
2 = − 2x − x2 2 · 2x − x2
2
0
1 1
2 −2 2 0
(2x − x2 )
= − 2x − x · 2x − x = − 1
2 2 (2x − x2 ) 2
0
(2x − x2 ) 2 − 2x 2x − 2
=− √ =− √ = √
2 2x − x 2 2 2x − x 2 2 2x − x2
We have
3 3−2 1 1 1 1

f0 = q = q = q = √ = √
2 2 3 − 94 2 12 − 9
2 34 2 · 23 3
4 4
We have s s s √
3 9 12 9 3 3

f =− 3− =− − =− =−
2 4 4 4 4 2
1
So the tangent line is a line with slope √ . A Cartesian equation of such a line is given by
3
1
y = √ ·x+b
3
√ !
3 3 √
Since it needs to contain the point ,− , we get b = − 3. We need the points of intersection
2 2
of the line
1 √
y = √ ·x− 3
3
and the parabola
1 2
y=− x
12
1 √ 1
√ · x − 3 = − x2
3 12
1 2 1 √
x + √ ·x= 3
12 3
1 2 1 √
x + √ · x + 1 −1 = 3
12 3
| {z }
complete square
196
1 2 1 √
x + √ ·x+1= 3+1
12 3
| {z }
complete square
!2
1 √
√ ·x+1 = 3+1
2 3
q√
1
√ ·x+1=± 3+1
2 3
q√
1
√ · x = −1 ± 3+1
2 3
√ q√
x1,2 = 2 3 −1 ± 3+1
To get the y coordinate, we use the equation of the tangent line

1 √
y = √ ·x− 3
3
and we get the points
√
q√ q√ √
P1 = 2 3 −1 + 3 + 1 , 2 −1 + 3+1 − 3
and √ q√ q√ √
P2 = 2 3 −1 − 3 + 1 , 2 −1 − 3+1 − 3
Question 5. Prove that for all x 0 the following inequality holds:

x
e x+1 ¬ 1 + x ¬ ex
Answer: First we will establish the inequality
1 + x ¬ ex
The proof will be by contradiction. Assume there existed an xp > 0 such that exp < 1 + xp . Then by
the mean value theorem, there would be an xm in the open interval (0, xp ) such that the derivative
of ex evaluated at xm is smaller than the slope of the line going through (0, 1) and (xp , 1 + xp ). But
the slope of this line is 1, and the derivative of ex is ex which is greater than 1 when x is greater
than zero. So we get a contradiction and 1 + x ¬ ex .
Now we will establish the inequality x
e x+1 ¬ 1 + x
xp
The proof will be by contradiction. Assume there existed an xp > 0 such that 1 + xp < e xp +1 . Then
by thex
mean value theorem, there would be an xm in the open interval (0, xp ) such that the derivative
of e x+1 evaluated at xm is greater than the slope of the line going through (0, 1) and (xp , 1 + xp ). But
the slope of this line is 1, and
0 0 0 !
x x+1 1 1 −1
0
x x x x x

e x+1 =e x+1 =e x+1 − =e x+1 1− =e x+1 −
x+1 x+1 x+1 x+1 (x + 1)2
197
x 1 1
x 1 e x+1 e(1− x+1 ) e · e− x+1 e
=e x+1 · = = = = 1
(x + 1)2 (x + 1)2 (x + 1)2 (x + 1)2 e x+1 · (x + 1)2
x
Evaluated at x = 0, the derivative of e x+1 is 1. And the denominator is growing, since
1
0 1
0 1
0
e x+1 · (x + 1)2 = e x+1 · (x + 1)2 + e x+1 · (x + 1)2
0
1

1 1
= e x+1 · · (x + 1)2 + e x+1 · 2(x + 1)2
x+1
1 −1 1 1 −1 1
= e x+1 · 2
· (x + 1)2 + e x+1 · 2(x + 1)2 = e x+1 · + e x+1 · 2(x + 1)2
(x + 1) x+1
1

1
2
= |e {z } 2(x + 1) −
x+1 >0 for x 0
>0 | {z
x + 1 }
>0 for x0
x
So the derivative of e x+1 is smaller than 1 for x > 0, which leads to a contradiction. We conclude
that x
e x+1 ¬ 1 + x
Question 6. Show that the function

1
f (x) = x3 cos
x
1
extends to a C (continuously differentiable) function of R.
Answer: The function f (x) is not defined at x = 0. But it is clear that when approaching zero
from the right or from the left, the x3 progressively reduces the amplitude to zero. So we define the
extended function, which we call by the same name, as
1

x3 cos

if x 6= 0
f (x) =  x
0 if x = 0
We need to show that f (x) is differentiable at x0 = 0, it is clear that everywhere else it is differentiable.
We will use the definition of the derivative. Approaching zero from the right, keeping in mind that
the value of cos is always between 0 and 1, we get

1 1
f (0 + h) − f (0) (0 + h)3 cos 0+h
−0 h3 cos h 2 1

lim+ = lim+ = lim+ = lim+ h cos =0
h→0 h h→0 h h→0 h h→0 h
Approaching zero from the left, keeping in mind that the value of cos is always between 0 and 1, we
get

1 1
f (0 + h) − f (0) (0 + h)3 cos 0+h
−0 h3 cos h 1

lim− = lim− = lim− = lim− h2 cos =0
h→0 h h→0 h h→0 h h→0 h
f 0 (0) = 0.
198
Academic Year 2015/16, Examination, 2 March, 2016
4 · 3 + 9 · 4 + 14 · 5 + · · · + (5n − 1)(n + 2)
lim
n→∞ n3

n
X
k = 1 + 2 + 3 + ··· + k
k=1
n(n + 1)
=
2
and
n
k 2 = 12 + 22 + 32 + · · · + k 2
X
k=1
n(n + 1)(2n + 1)
=
6
Note also that n
X
1 = |1 + 1 + 1{z+ · · · + 1} = n
k=1 n times
So we get
n
4 · 3 + 9 · 4 + 14 · 5 + · · · + (5n − 1)(n + 2) X (5k − 1)(k + 2)
lim = lim
n→∞ n3 n→∞
k=1 n3
n
5k 2 + 9k − 2
X
= n→∞
lim
k=1 n3
n n n
X 5k 2 X 9k X 2
= lim 3
+ lim 3
− lim 3
k=1 n k=1 n k=1 n
n→∞ n→∞ n→∞
Pn Pn Pn
5 k=1 k2 9 k=1 k 2 k=1 1
= n→∞
lim + n→∞
lim − n→∞
lim
n3 n3 n3
n(n+1)(2n+1) n(n+1)
5· 6
9· 2 2n
= lim + lim − lim
n→∞ n3 n→∞ n3 n→∞ n3
10 5
= +0−0=
6 3

∞ n2
1 n+1
X
n n
n=1 2
199
n
1

Answer: We will use the fact that n→∞
lim 1+ = e. We use the nth root test.
n
n2 # n1 n2
v s
u n2 n2 "
1 n+1 1 n+1 1 n+1 1 n+1

u n n
n
lim t = n→∞
lim = n→∞
lim n = n→∞
lim n
n→∞ 2n n 2n n 2 n 2n n
n n n
1 n+1 1 n 1 1 1 e 2.72

= lim 1 = lim + = lim 1+ = ≈ >1
n→∞ 2 n n→∞ 2 n n n→∞ 2 n 2 2
So by the nth root test, the series
∞ n2
1 n+1
X
n n
n=1 2
diverges.

1 1

lim −
x→0 x sin x
Answer: It is possible to get the answer without calculation. When x is very small, sin x ≈ x, so the
0
limit we will calculate will be zero. We use a common denominator, this will be of the form and
0
we can apply l’Hôpital’s rule. We actually need to apply it twice.
1 1 sin x − x (sin x − x)0 (cos x − 1) cos x − 1

lim − = lim = lim 0
= lim 0 0
= lim
x→0 x sin x x→0 x sin x x→0 (x sin x) x→0 (x) sin x + x(sin x) x→0 sin x + x cos x
(cos x − 1)0 − sin x

= lim = lim
x→0 (sin x + x cos x)0 x→0 cos x + (x)0 cos x + x(cos x)0
− sin x 0 0
= lim = = =0
x→0 cos x + cos x + x(− sin x) 1+1+0 2
1
Question 4. Find the equations of all the tangent lines to the graph of f (x) = x2 + 3x + 3 which
2
pass through the point (0, 1)
Answer: The first thing we need is the derivative
f 0 (x) = x + 3
We need lines which have a point P = (xp , yp ) in common with the parabola f (x), and the derivative
f 0 (x) evaluated at xp needs to be equal to the slope of the line going through P and the point (0, 1).
We get the system of equations
yp − 1 yp − 1
f 0 (xp ) = xp + 3 = =
xp − 0 xp
1
yp = f (xp ) = x2p + 3xp + 3
2
Which can be combined to one equation and solved as follows
x2p + 3xp = yp − 1
200
1
x2p + 3xp = x2p + 3xp + 3 − 1
2
1
x2p = x2p + 2
2
2x2p = x2p + 4
x2p = 4
xp = ±2
So we get two solutions. For P1 = (2, 11), we need a line with slope f 0 (2) = 5 and going through the
point (0, 1). A Cartesian equation for this line is
T1 : y = 5x + 1
For P2 = (−2, −1), we need a line with slope f 0 (−2) = 1 and going through the point (0, 1). A
Cartesian equation for this line is
T1 : y = x + 1
π

Question 5. Prove that for all x, y ∈ 0, , the function f (t) = cos t2 satisfies the following
2
inequality:
|f (x) − f (y)| ¬ π|x − y|
Answer: If we can show that the absolute value of the derivative of f (t) is smaller than or equal to
π

π when t ∈ 0, , the result follows by the mean value theorem. So we calculate the derivative of
2
201

cos t2 .
0 h i 0 h i
cos t2 = − sin t2 2
t = − sin t2 · 2t = −2t sin t2
π

Since the value of sin is between −1 and 1, it follows that −2t sin t2 ¬ π when t ∈ 0, , which
2
is what we need.
√
Question 6. Let f (x) = x and a = 16. Using the Taylor polynomial of degree 2 of f at a, find an
√
approximate value of 18. If η is the approximate value, show that the true value is in the interval
1 1

η− ,η + .
2048 2048
√
Answer: We will use a Taylor polynomial of f (x) = x at a = 16. For a Taylor polynomial of degree
2 with error estimate, we need the first three derivatives.
√ 1 √
f (x) = x = (x) 2 f (16) = 16 = 4
1 1 1 1
f 0 (x) = · (x)− 2 f 0 (16) = √ =
2 2· 16 8
1 −1 3 −1 −1
f 00 (x) = · · (x)− 2 f 00 (16) = √ 3 =
2 2 4· 16 256
1 −1 −3 5
f 000 (x) = · · · (x)− 2
2 2 2
So we get the Taylor polynomial of degree 2 of f at a = 16
f (16) f 0 (16) f 00 (16)

f (x) ≈ · (x − 16)0 + · (x − 16)1 + · (x − 16)2
0! 1! 2!
0 f 00 (16)
= f (16) + f (16) · (x − 16) + · (x − 16)2
2!
1 1
= 4 + · (x − 16) − · (x − 16)2
8 256 · 2!
Hence
√ 1 1
18 = f (18) ≈ 4 + · (18 − 16) − · (18 − 16)2
8 256 · 2!
1 1
= 4 + · (2) − · (2)2
8 256 · 2!
2 22
= 22 + −
23 28 · 2
29 25 1 29 + 25 − 1 512 + 32 − 1 543
= 7
+ 7
− 7
= 7
= =
2 2 2 2 128 128
f 000 (c)
R2 (x) = (18 − 16)3 ,
3!
202
where c is between a = 16 and x = 18. To get the margin for the possible error, we choose c in the
interval [16, 18] which maximizes the absolute value f 000 (c). Since
1 −1 −3 5
f 000 (x) = · · · (x)− 2
2 2 2
the c in the interval [16, 18] which maximizes the absolute value of f 000 (c) is c = 16.
3 3
f 000 (16) = √ 5 =
8· 16 8192
Hence the error is less than

f 000 (16) 3 · 23 3·8 4 1
(18 − 16)3 = = = =
3! 3! · 8192 6 · 8192 8192 2048
Academic Year 2016/17, Examination, 1 February, 2017

n
!
1 X 1 1
lim √ √ −√
n→∞ n k=1 k k+1
Answer:
n
!
1 X 1 1
lim √ √ −√ =
n→∞ n k=1 k k+1
" ! ! !#
1 1 1 1 1 1 1
= lim √ √ −√ + √ −√ + ··· + √ − √
n→∞ n 1 1+1 2 2+1 n n+1
" ! ! !#
1 1 1 1 1 1
= lim √ 1 − √ + √ − √ + ··· + √ − √
n→∞ n 2 2 3 n n+1
" #
1 1
lim √ 1 − √
= n→∞ =0
n n+1
Question 2. Let a and b be real numbers such that b < a and b is not a negative integer. Is the
following series convergent or divergent?
∞ 2
X n+a n
n=1 n+b

1 1
Answer: We use the n-th root test. Note that for large n, we have ≈ .
n+b n
n2 # n1 n2
s
n2 " n
n+a n+a n+a n+a

n n
lim = lim = lim = lim
n→∞ n+b n→∞ n+b n→∞ n+b n→∞ n+b
203
n !n !n
n a n+b a−b a−b

= lim + = lim + = lim 1+
n→∞ n+b n+b n→∞ n+b n+b n→∞ n+b
!n
a−b
= lim 1+ = ea−b > 1 (since a − b > 0)
n→∞ n
Hence by the nth root test, the series diverges.

x 1

lim −
x→1 x − 1 ln x
Answer: This is of the form ∞ − ∞. In order to use l’Hôpital’s rule, we need to put it in the form
0
.
0
!
x 1 x ln x x−1 x ln x − x + 1

lim − = lim − = lim
x→1 x − 1 ln x x→1 (x − 1) ln x (x − 1) ln x x→1 (x − 1) ln x
(x ln x − x + 1)0 x0 ln x + x(ln x)0 − 1

= lim = lim
x→1 ((x − 1) ln x)0 x→1 (x − 1)0 ln x + (x − 1)(ln x)0
ln x + xx − 1 ln x (ln x)0
= lim x−1 = lim 1 = lim 0
x→1 ln x + x→1 ln x + 1 − x→1 1
x x ln x + 1 − x
1
x 1 1
= lim −1 = =
x→1 1 − 1 − (−1) 2
x x2
Question 4. Let f : R → R be the function given by

3t2
f (t) =
1 + t6
a) Does f have a maximum or a minimum? If so, then state the values of x at which they occur.
b) Sketch the graph of f .
Answer: a) To find the possible points for local extrema, we need the derivative.
!0 0 0
3t2 (3t2 ) (1 + t6 ) − (3t2 ) (1 + t6 ) 6t (1 + t6 ) − (3t2 ) 6t5
f (t) = = =
1 + t6 (1 + t6 )2 (1 + t6 )2
6t + 6t7 − 18t7 6t − 12t7 6t (1 − 2t6 )
= = =
(1 + t6 )2 (1 + t6 )2 (1 + t6 )2
1
The derivative needs to be zero for local extrema to occur. This happens when t = 0 and t = ± 1 .
26
The function f (t) is an even function, so it is symmetric with respect to the ordinate axis. We have
f (0) = 0, and when t approaches plus or minus infinity, f (t) approaches zero. So there is a minimum
at (0, 0) and there are maxima at
 3   3 
1 1 1 1 1 2 1
1 1

2
23 23
− 1 ,f − 1 = − 1 , 1
 = − 1 , 3
 = − 1 , 1 = − 1 ,2 3
26 26 26 1 + 2 26 2 26 23 26
204
and
1 1 1

2
1 ,f = 1 ,2 .
3
1
26 26 26
(b) from part a), we get
Question 5. Prove that for all x and y in R, the following inequality is true:
2

arc tg x3 − arc tg x3 ¬ 2 3 |x − y|
Answer: By the mean value theorem, it suffices to show that the maximum of the absolute value of
2

3
the derivative of F (t) = arc tg t is less than 2 3 . If one knows the derivative of arc tg, then finding
the derivative of F (t) is a simple application of the chain rule. If one does not, it is good to first find
the derivative of arc tg, which we will now do. Let x = tg(y). Then y = arc tg(x) and we have
!0
= (tg(y))0 = = 2
=
dy cos y cos y cos2 y
cos2 y + sin2 y 1
= 2
=
cos y cos2 y

= = + = 1 + tg2 y
dy cos2 y cos2 y cos2 y
dy 1 1 1 1 1
(arc tg x)0 = = dx = 2
= 2
= 2
=
dx dy
Using the chain rule, we get

0 1 0 3t2
f (t) = F 0 (t) = arc tg t3 = · t3
=
1 + (t3 )2 1 + t6
From the previous question (Question 5), we know that the maximum of the absolute value of F 0 (t)
2
is 2 3 .
Question 6. Let > 0 and suppose that, f : (−, ) → (−1, ∞) is a function which satisfies f (0) = 0
and
x + f (x) + ln(1 + f (x)) = 0. (2)
205
a) Find the values of f 0 (0) and f 00 (0). (Hint: differentiate equation (2)).
b) Give the second order Taylor polynomial of f around the point x0 = 0.
Answer: Differentiating equation (2), we get
[x + f (x) + ln(1 + f (x))]0 = (0)0

f 0 (x)
1 + f 0 (x) + =0
1 + f (x)
!
0 1
1 + f (x) 1 + =0
1 + f (x)
!
0 1
f (x) 1 + = −1
1 + f (x)
−1
f 0 (x) =
1 + 1+f1(x)
−1 −1 1
f 0 (0) = 1 = 1 = −
1 + 1+f (0) 1 + 1+0 2
Differentiating again, we get

" !#0
1
f 0 (x) 1 + = (−1)0
1 + f (x)
! !0
00 1 1
f (x) 1 + + f 0 (x) 1 + =0
1 + f (x) 1 + f (x)
−f 0 (x)
!
00 1
f (x) 1 + + f 0 (x) · =0
1 + f (x) (1 + f (x))2
[f 0 (x)]2
!
00 1
f (x) 1 + − =0
1 + f (x) (1 + f (x))2
[f 0 (x)]2
!
1
f 00 (x) 1 + =
1 + f (x) (1 + f (x))2
[f 0 (x)]2
(1+f (x))2
f 00 (x) =
1 + 1+f1(x)
2
[f 0 (0)]2 (− 21 ) 1
1
00 (1+f (0))2 (1+0)2
4
1 4 1
f (0) = = = = =
1 + 1+f1(0) 1
1 + 1+0 1+ 1
1
2 8
b) The second order Taylor polynomial of f around x0 = 0 is
f (0) 0 f 0 (0) 1 f 00 (0)

P2 (x) = (x − x0 ) + (x − x0 ) + (x − x0 )2
0! 1! 2!
f (0) f 0 (0) f 00 (0)
= (x − 0)0 + (x − 0)1 + (x − 0)2
0! 1! 2!
206
f (0) f 0 (0) f 00 (0) 2
= ·1+ ·x+ ·x
1 1 2
1
f 00 (0) 2 1 x x2
= f (0) + f 0 (0) · x + · x = 0 − · x + 8 · x2 = − +
2 2 2 2 16
Temat 15: Indefinite integrals.

Z
Question 1. Calculate sin x cos x dx using the following hints:
a) by parts with f (x) = sin x, g(x) = cos x

b) by parts with f (x) = cos x, g(x) = sin x
c) change of variables y = sin x
d) change of variables y = cos x
1
e) the double angle formula sin x cos x = sin(2x)
2
Z
Answer: a) We need to calculate sin x cos x dx by parts with f (x) = sin x, g(x) = cos x.
Z Z Z
sin x cos x dx = f (x)g(x) dx = F (x)g(x) − F (x)g 0 (x) dx
Z
= (− cos x) cos x − (− cos x)(− sin x) dx
Z
2
= − cos x − sin x cos x dx
Z
2 sin x cos x dx = − cos2 x
Z
cos2 x
sin x cos x dx = − +C
2
for any constant C.
Z
b) We need to calculate sin x cos x dx by parts with f (x) = cos x, g(x) = sin x.
Z Z Z
cos x sin x dx = f (x)g(x) dx = F (x)g(x) − F (x)g 0 (x) dx
Z
= sin x sin x − sin x cos x dx
Z
2
= sin x − cos x sin x dx
Z
2 cos x sin x dx = sin2 x
Z
sin2 x
cos x sin x dx = +C
2
207
for any constant C. Since sin2 x + cos2 x = 1, given that we have additive constants, this is consistent
with the result in a).
Z
c) We need to calculate sin x cos x dx by change of variables y = sin x. We have
dy dy
y = sin x = cos x dx =
dx cos x
Z Z
dy Z
y2 sin2 x
sin x cos x dx = y cos x · = y dy = +C = +C
cos x 2 2
Z
d) We need to calculate sin x cos x dx by change of variables y = cos x. We have
dy dy
y = cos x = − sin x dx = −
dx sin x
Z Z
−dy Z
y2 cos2 x
cos x sin x dx = y sin x · = − y dy = − + C = − +C
sin x 2 2
Z
1
e) We need to calculate sin x cos x dx using the double angle formula sin x cos x = sin(2x). We
2
have
Z Z
1 − cos 2x 2 sin2 x − 1 sin2 x
sin x cos x dx = sin(2x) dx = +C = +C = + C∗
2 4 4 2
2. Calculate:
Z
a) ln x dx
Z
b) arc sin x dx
Z
c) x arc tg x dx
Z
d) x arc cos x dx
Z
e) x2 e−x dx
Z
f) (ln x)2 dx
Z
g) xex sin x dx
Z
x2
h) dx
1 + x2
Z
dx
i)
2x2 + 3
Z
dx
j)
(1 + x2 )2
208
Z
k) (cos x)3 dx
Z √
x
l) e dx
√
Z 6
x dx
m) √
1+ 3x
Answer: a) We will us intgration by parts with f (x) = 1 and g(x) = ln x
Z Z Z Z
ln x dx = 1 · ln x dx = f (x)g(x) dx = F (x)g(x) − F (x)g 0 (x) dx
Z
1 Z
= x ln x − x· dx = x ln x − 1 dx = x ln x − x + C
x
for any constat C.
dx
b) We use a change of variables. Let x = sin y. Then = cos y and arc sin x = arc sin(sin y) = y.
dy
Hence Z Z
arc sin x dx = y cos y dy
Now we use integration by parts with f (y) = cos y and g(y) = y.

Z Z Z
y cos y dy = g(y)f (y) dy = g(y)F (y) − g 0 (y)F (y) dy
Z
= y sin y − 1 · sin y dy = y sin y + cos y + C
q √
= (arc sin x)x + 1 − sin2 y + C = x arc sin x + 1 − x2 + C
for any constant C.

1
c) We know that (arc tg x)0 = . We use integration by parts with f (x) = x and g(x) = arc tg x.
1 + x2
Z Z Z
x arc tg x dx = f (x)g(x) dx = F (x)g(x) − F (x)g 0 (x) dx
x2 Z 2
x 1 x2 1Z 2 1
= · arc tg x − · 2
dx = · arc tg x − x · dx
2 2 1+x 2 2 1 + x2
x2 1Z 1 x2 1

= · arc tg x − 1− dx = · arc tg x − (x − arc tg x) + C
2 2 1 + x2 2 2
(x2 + 1) arc tg x − x
= +C
2
for any constant C.
dx
d) We use a change of variables. Let x = cos y. Then = − sin y and arc cos x = arc cos(cos y) = y.
dy
Hence Z Z Z
x arc cos x dx = (cos y)y(− sin y) dy = − y cos y sin y dy
From Question 1, we know that
Z
sin2 x
sin x cos x = +C
2
209
And we have
Z Z Z
2
sin x dx = sin x sin x dx = − cos x sin x − − cos x cos x dx
Z Z
2
= − cos x sin x + cos x dx = − cos x sin x + 1 − sin2 x dx
Z Z
2
2 sin x dx = − cos x sin x + 1 dx = − cos x sin x + x + C
Z
− cos x sin x + x + C x − cos x sin x + C x − cos x sin x
sin2 x dx = = = + C∗
2 2 2
Thus
Z Z
sin2 y Z sin2 y sin2 y Z sin2 y
x arc cos x dx = − y cos y sin y dy = −y · − − dy = −y · + dy
2 2 2 2
sin2 y y − cos y sin y
= −y · + +C
2 4
√
1 − x2 arc cos x − x 1 − x2
= − arc cos x · + +C
2 4
for any constant C.
e)
Z Z Z Z
2 −x 2 −x −x 2 −x −x 2 −x −x
xe dx = −x e − −2xe dx = −x e + 2xe dx = −x e − 2xe − −2e−x dx
Z
2 −x −x
= −x e − 2xe + 2e−x dx = −x2 e−x − 2xe−x − 2e−x + C

= −e−x x2 + 2x + 2 + C
for any constant C.

f)
Z Z
1 Z Z
(ln x)2 dx = 1 · (ln x)2 dx = x · (ln x)2 − dx = x · (ln x)2 − 2 ln x dx
x · 2 ln x ·
x
2
Z
2
Z
1
= x · (ln x) − 2 1 · ln x dx = x · (ln x) − 2 · x ln x + 2 x · dx
x
Z
2
= x · (ln x) − 2 · x ln x + 2 1 dx = x · (ln x)2 − 2 · x ln x + 2x + C

= x ln2 x − 2 ln x + 2 + C
for any constant C.

g) We will need the following result
Z Z
x x
e (cos x − sin x) dx = e (cos x − sin x) − ex (cos x − sin x)0 dx
Z
x
= e (cos x − sin x) − ex (− sin x − cos x) dx
210
Z
x
= e (cos x − sin x) + ex (sin x + cos x) dx
Z
x x
= e (cos x − sin x) + e (sin x + cos x) − ex (sin x + cos x)0 dx
Z
x x
= e (cos x − sin x) + e (sin x + cos x) − ex (cos x − sin x) dx
Z
2 ex (cos x − sin x) dx = ex (cos x − sin x) + ex (sin x + cos x) = 2ex cos x
Z
ex (cos x − sin x) dx = ex cos x
for any constant C. Using this result, we get

Z Z Z
xex sin x dx = ex (x sin x) dx = ex x sin x − ex (x sin x)0 dx
Z Z Z
= ex x sin x − ex (sin x + x cos x) dx = ex x sin x − ex sin x dx − ex (x cos x) dx
Z Z
= ex x sin x − ex sin x dx − ex x cos x + ex (x cos x)0 dx
Z Z Z
= ex x sin x − ex sin x dx − ex x cos x + ex cos x dx − ex x sin x dx
Z Z Z
2 xex sin x dx = ex x sin x − ex x cos x − ex sin x dx + ex cos x dx
Z
x ex x sin x − ex x cos x 1 Z x
xe sin x dx = + e (cos x − sin x) dx
2 2
ex x sin x − ex x cos x ex cos x ex
= + + C = (x sin x − x cos x + cos x) + C
2 2 2
for any constant C.
h)
x2 1
Z Z
dx = 1− dx = x − arc tg x + C
1 + x2 1 + x2
for any constant C.
i)
s s 
Z
dx 1Z dx 1Z dx 1 3 2 
= 2 = 2 = arc tg  ·x +C
2x2 + 3 3 ·x +1
2 3 3 2 3
q
2
3
3
·x +1
s s  s 
3 2  1 2 
= arc tg  · x + C = √ arc tg  ·x +C
9·2 3 6 3
for any constant C.

j)
!
Z
dx Z
1 x2 Z
1 Z
x2
= − dx = dx − dx
(1 + x2 )2 1 + x2 (1 + x2 )2 1 + x2 (1 + x2 )2
211
Z
x − 21 Z
− 12
= arc tg x − x· dx = arc tg x − x · + dx
(1 + x2 )2 1 + x2 1 + x2
x x
2 1Z 1 2 1
= arc tg x + − dx = arc tg x + − · arc tg x + C
1 + x2 2 1 + x2 1 + x2 2
1 x

= + arc tg x + C
2 1 + x2
for any constant C.
k)
Z Z Z Z
3 2 2
(cos x) dx = cos x cos x dx = 1 − sin x cos x dx = cos x − sin2 x cos x dx
1 1
Z Z
= cos x dx − sin x cos x dx = sin x − sin3 x + C = sin x 1 − sin2 x + C
2
3 3
for any constant C.
√ 1 dy 1 1 1 √
l) We do the change of variables y = x = x 2 . Then = x− 2 = √ , hence dx = 2 x dy and
dx 2 2 x
we get
Z √
x
Z √
x
√ Z √ √ Z Z
e dx = e · 2 x dy = 2 x · e x dy = 2 yey dy = 2yey − 2 ey dy = 2yey − 2ey + C
√ √
= 2ey (y − 1) + C = 2e x
x−1 +C
for any constant C.

√ 1 dy 1 5 5
m) We do the change of variables y = 6
x = x 6 . Then = x− 6 , hence dx = 6x 6 dy and we get
dx 6
√ 1 1 1 5 1 5 1 5
Z 6
x dx Z
x 6 dx Z
x 6 dx Z
x 6 · 6x 6 dy Z
x 6 · x 6 dy Z
x 6 + 6 dy
√ = 1 = 1 2 = 1 2 = 6 1 2 = 6 1 2
1+ 3x 1 + x3 1 + x6 1 + x6 1 + x6 1 + x6
1
6
6
x6 dy
!
Z
x dy 6
Z Z
y 6 dy Z
1
=6 2 =6 2 =6 =6 y4 − y2 + 1 − dy
1 + y2 1 + y2
1
1
1 + x6 1 + x6
!
y5 y3 6y 5
=6 − + y − arc tg y + C = − 2y 3 + 6y − 6 arc tg y + C
5 3 5
1
5
6 x6 1
3 1 1
= − 2 x6 + 6x 6 − 6 arc tg x 6 + C
5
5 5
6x 3 61 1 6x 6 √ √ √
= − 2x 6 + 6x 6 − 6 arc tg x 6 + C = − 2 x + 6 6 x − 6 arc tg 6 x + C
5 5
for any constant C.
Question 3. Use partial fractions to calculate the following:

Z
x3 Z
x Z
x3 − 1 Z
x3 + 1
(a) dx (b) dx (c) dx (d) dx
x+1 2x2 − 3x − 2 4x3 − x x3 − x2
212
Z
x4 Z
x3 − 1 Z
1 Z
1
(e) dx (f) dx (g) dx (h) dx
x2 + 1 x3 − 5x2 + 6x x4 − 1 x (1 + x2 )
Answer: a)
x3 1 x3 x2
Z Z
2
dx = x −x+1− dx = − + x − ln |x + 1| + C
x+1 x+1 3 2
for any constant C.

b)
x x A B Ax − 2A + 2Bx + B
= = + =
2x2 − 3x − 2 (2x + 1)(x − 2) 2x + 1 x − 2 (2x + 1)(x − 2)
A + 2B = 1
−2A + B = 0
1 2
A= B=
5 5
1 2
!
Z
x Z
5 5 1 2
dx = + dx = ln |2x + 1| + ln |x − 2| + C
2x2 − 3x − 2 2x + 1 x−2 10 5
for any constant C.
c)
1
x3 − 1 1 4
x−1 1 Ax + B C 1 Ax2 + Bx + 4Cx2 − C
= + 3 = + 2 + = +
4x3 − x 4 4x − x 4 4x − 1 x 4 4x3 − x
A + 4C = 0
1
B=
4
−C = −1
1
!
x3 − 1 1 x−1 Z
1 Ax + B C
Z Z
dx = + 43 dx = + + dx
4x3 − x 4 4x − x 4 4x2 − 1 x
1 −4x + 41
!
Z
1
= + + dx
4 4x − 1
2 x
−4x + 41
!
Z
1 1
= + + dx
4 (2x − 1)(2x + 1) x
!
Z
1 D E 1
= + + + dx
4 (2x − 1) (2x + 1) x
213
2D + 2E = −4
1
D−E =
4
− 78 − 89
!
Z
x3 − 1 Z
1 1
dx = + + + dx
4x3 − x 4 (2x − 1) (2x + 1) x
x 7 9
= − ln |2x − 1| − ln |2x + 1| + ln |x| + C
4 16 16
for any constant C.
d)
x3 + 1 x2 + 1 x2 + 1 Ax + B C Ax2 + Bx − Ax − B + Cx2
= 1 + = 1 + = 1 + + = 1 +
x3 − x2 x3 − x2 x2 (x − 1) x2 x−1 x2 (x − 1)
A+C =1
B−A=0
−B = 1
x3 + 1 Ax + B C −x − 1 2
Z Z Z
dx = 1+ + dx = 1+ + dx
x3 − x2 x 2 x−1 x 2 x−1
Z
x 1 2 1 1 2
Z
= 1− 2 − 2 + dx = 1− − 2 + dx
x x x−1 x x x−1
1
= x − ln |x| + + 2 ln |x − 1| + C
x
for any constant C.
e)
x4 2 x2 1
2
=x − 2 = x2 − 1 + 2
x +1 x +1 x +1
x4 1 x3
Z Z
2
dx = x −1+ 2 dx = − x + arc tg x + C
x2 + 1 x +1 3
for any constant C.
f)
x3 − 1 5x2 − 6x − 1 5x2 − 6x − 1
= 1 + = 1 +
x3 − 5x2 + 6x x3 − 5x2 + 6x (x − 2)(x2 − 3x)
A Bx + C Ax2 − 3Ax + Bx2 − 2Bx + Cx − 2C
=1+ + =1+
(x − 2) (x2 − 3x) (x − 2)(x2 − 3x)
A+B =5
−3A − 2B + C = −6
214
−2C = −1
− 72 17
x + 21
! !
Z
x3 − 1 Z
A Bx + C Z
dx = 1+ + 2 dx = 1+ + 22 dx
x3 − 5x2 + 6x (x − 2) (x − 3x) (x − 2) (x − 3x)
− 27 17
x + 21 − 72
! !
Z Z
D E
= 1+ + 2 dx = 1+ + + dx
(x − 2) x(x − 3) (x − 2) x x−3
− 27
!
Z
Dx − 3D + Ex
= 1+ + dx
(x − 2) x(x − 3)
17
D+E =
2
1
−3D =
2
− 72 −1 26
!
Z
x3 − 1 Z
dx = 1+ + 6 + 3 dx
x3 − 5x2 + 6x (x − 2) x x−3
7 1 26
=x− ln |x − 2| − ln |x| + ln |x − 3| + C
2 6 3
for any constant C.
g)
1 1
!
Z
1 Z
1 Z
2 2
dx = dx = − dx
x4 − 1 (x2 − 1) (x2 + 1) x2 − 1 x2 + 1
1 1 1 1 1
Z ! Z !
2 2 4 4 2
= − dx = − − dx
(x − 1) (x + 1) x2 + 1 x−1 x+1 x2 + 1
1 1 1
= ln |x − 1| − ln |x + 1| − arc tg x + C
4 4 2
for any constant C.
h)
1 1 x 1
Z Z
dx = − dx = ln |x| − ln |1 + x2 | + C
x (1 + x2 ) x 1 + x2 2
for any constant C.
Temat 16: Indefinite, definite and improper integrals

Question 1. Calculate
√
Z
dx Z
dx Z
1 + x2
a) b) c) dx
1 − x2 sin x x2
215
Answer: a)
1 1
!
Z
dx Z
dx Z
2 2
= = + dx
1 − x2 (1 − x)(1 + x) 1−x 1+x
1 1 1
= − ln |1 − x| + ln |1 + x| + C = (ln |1 + x| − ln |1 − x|) + C
2 2 2
for any constant C.
b)
dx 1 + cos x sin2 x + cos2 x + cos x sin x cos x

Z Z Z Z
= dx = dx = + dx
sin x (1 + cos x) sin x (1 + cos x) sin x 1 + cos x sin x
sin x
= − ln |1 + cos x| + ln | sin x| + C = ln +C
1 + cos x
for any constant C.
ex − e−x
c) We will need the following: sinh x = . We have
2
ex − (−e−x ) ex + e−x
(sinh x)0 = = = cosh x
2 2
and
!2 !2
2 2 ex + e−x ex − e−x e2x + 2ex−x + 2−2x − (e2x − 2ex−x + 2−2x )
cosh x − sinh x = − =
2 2 4
4ex−x 4e0
= = =1
4 4
Now we put x = sinh y. Then y = sinh−1 x and
0 dy 1 1 1 1 1
sinh−1 x = y0 = = dx = 0
= =q =√
dx dy
(sinh y) cosh y 2
1 + sinh y 1 + x2
We will use this result in the following calculation

Z √
1 + x2 Z
1 √ −1 √ Z
−1 1
2
· 2x
dx = · 1 + x 2 dx = · 1 + x 2− · √ dx
x2 x2 x x 1 + x2
√ √
1 + x2 Z 1 1 + x2
=− + √ dx = − + sinh−1 x + C
x 1+x 2 x
for any constant C.
Question 2. Calculate
π
Z 2
2x2 Z
4
Z 1
a) dx b) tg x dx c) arc tg x dx
1 x3 + 1 0 0
π Z 4 Z 2
Z
ln x
xe(x ) dx
2 2
d) x sin x dx e) dx f)
0 1 x 0
216
Z 3 2 Z 3 √ Z π
2
g) x2 2x3 + 4 dx h) x 9 − x2 dx i) sin x cos x dx
0 0 0
Answer: a) We will first calculate the indefinite integral

Z
2x2 2
3
dx = ln x3 + 1 + C
x +1 3
for any constant C.
2
Z 2
2x2 2 2h 3 2 2
i
3
dx = ln x3 + 1 = ln 2 + 1 − ln 13 + 1 = (ln 9 − ln 2) = (2 ln 3 − ln 2)
1 x +1 3 1 3 3 3
b) We will first calculate the indefinite integral

Z Z
sin x
tg x dx = dx = − ln | cos x| + C
cos x
for any constant c.
π
!
π 1
Z π

4
tg x dx = [− ln | cos x|]0 = − ln 4
cos − (− ln | cos 0|) = − ln √ + ln 1
0 4 2
√ √ √
! !
1 1 1
= − ln √ + 0 = − ln √ = − ln 1 − ln 2 = − 0 − ln 2 = ln 2 = ln 2
2 2 2
c) We will first calculate the indefinite integral. We put x = tg y. Then arc tg x = y and
!0
0 dx sin y (sin y)0 cos y − sin y(cos y)0 cos2 y + sin2 y 1
x = = (tg y)0 = = 2
= 2
=
dy cos y cos y cos y cos2 y
dy
Hence dx = and
cos2 y
Z Z
dy Z Z
sin y
arc tg x dx = y· 2
= y tg y − tg y dy = y tg y − dy = y tg y − [− ln(cos y)] + C
cos y cos y
" !#  !2 
1 1 1
= y tg y − ln + C = y tg y −  ln +C
cos y 2 cos y
sin2 y + cos2 y
" !# " !#
1 1 1
= y tg y − ln + C = y tg y − ln +C
2 cos2 y 2 cos2 y
sin2 y
" !#
1 1 2

= y tg y − ln +1 + C = y tg y − ln tg y + 1 + C
2 cos2 y 2
1 2 1 2

= (arc tg x)x − ln x + 1 + C = x arc tg x − ln x + 1 + C
2 2
for any constant C. So we get
Z 1 1
1 2

arc tg x dx = x arc tg x − ln x + 1
0 2 0
217
1 2 1 2

= 1 · arc tg 1 − ln 1 + 1 − 0 · arc tg 0 − ln 0 + 1
2 2
π 1 π 1
= − ln 2 − (0 − 0) = − ln 2
4 2 4 2
1
An alternative way of finding the indefinite integral, using the fact that (arc tg x)0 = , is the
x2 + 1
following
Z Z Z
x 1 2
arc tg x dx = 1 · arc tg x dx = x arc tg x − dx = x arc tg x − ln x + 1 +C
x2 + 1 2
for any constant C.
d) We will first calculate the indefinite integral
Z Z Z
x sin x dx = −x cos x − (− cos x) dx = −x cos x + cos x dx = −x cos x + sin x + C

π
π π π π
Z π

2
x sin x dx = [−x cos x + sin x]02 = − · cos + sin − [−0 · cos 0 + sin 0] = − · 0 + 1 − 0 = 1
0 2 2 2 2
e) We will first calculate the indefinite integral

Z
ln x 1 Z
1
dx = · ln x dx = ln x ln x − ln x · dx
x x x
Z
ln x
2 dx = ln x ln x
x
Z
ln x ln x ln x
dx = +C
x 2
" #4
Z 4
ln x ln x ln x ln 4 ln 4 ln 1 ln 1 ln 4 ln 4 2 ln 2 · 2 ln 2
dx = = − = −0= = 2(ln 2)2
1 x 2 1
2 2 2 2
f)
Z 2 2
1 2 1 2 1 2 1 4

xe(x ) = e(x ) = e(2 ) − e(0 ) =
2

e −1
0 2 0 2 2 2
g)
Z 3 3
1 1 3 1 1 1 1
2 3 3 3
2 3
x 2x + 4 dx = · 2x + 4 = · 2 · 33 + 4 − · 2 · 03 + 4
0 3 2·3 0 3 2·3 3 2·3
583 43
= −
18 18
h)
Z 3 √
2

1
3 3
1 3 3
x 9 − x2 dx = · − 9 − x2 2 = − · 9 − x2 2
0 3 2 0 3 0
218
1 3 1 3 1 3

2 2 2 2
= − · 9−3 − − · 9−0 = − − · 92
3 3 3
1 3 1 1 3 1 √ 3 1 3
= · 92 = · 92 = · 9 = · 3 = 32 = 9
3 3 3 3
i) We will first calculate the indefinite integral

Z Z
sin x cos x dx = (− cos x) cos x − (− cos x)(− sin x) dx
Z
2
= − cos x − sin x cos x dx
Z
2 sin x cos x dx = − cos2 x
Z
cos2 x
sin x cos x dx = − +C
2
#π π
π
cos2
" " # " # " # " #
Z
2 cos2 x 2
2 cos2 0 02 12 1
sin x cos x dx = − = − − − = − − − =
0 2 0
2 2 2 2 2
Question 3. Calculate the derivatives of the following functions:

Z x 5 Z x2 Z x2
1
a) F (x) = t 1 + t2 dt b) F (x) = t− 3 dt c) F (x) = t sin t dt
1 0 0
Z sin x Z 1+x2 √ Z 1+x2
d) F (x) = 1 − t2 dt e) F (x) = 1 + t dt f) F (x) = t2 dt
0 1 1
Z x2
g) F (x) = t−2 dt
x
Answer: a) Z x
d

2 5
5
0
f (x) = F (x) = t 1+t dt = x 1 + x2
dx 1
b)
!
d Z x2 − 1 − 1 d 1
x2 = x2·(− 3 ) · 2x = x− 3 · 2x = 2x− 3 +1 = 2x 3
2 2 1
0
f (x) = F (x) = t 3 dt = x2 3 ·
dx 0 dx
c) !
Z x2
d d 2
f (x) = F 0 (x) = t sin t dt = x2 (sin x2 ) · x = x2 (sin x2 ) · 2x = 2x3 sin x2
dx 0 dx
d)
Z sin x
d d

f (x) = F 0 (x) = 1 − t2 dt = 1 − (sin x)2 · sin x
dx 0 dx
219

= 1 − sin2 x · cos x = cos2 x cos x = cos3 x
e)
d Z 1+x2 √ √ √ √
!
0 d
f (x) = F (x) = 1 + t dt = 1 + 1 + x2 · 1 + x2 = 2 + x2 · 2x = 2x · 2 + x2
dx 1 dx
f)
!
0 d Z 1+x2 2 2 d 2 2
f (x) = F (x) = t dt = 1 + x2 · 1 + x2 = 1 + x2 · 2x = 2x · 1 + x2
dx 1 dx
g)
Z x2 !
d −2 d −2
f (x) = F 0 (x) = t−2 dt = x2 · x2 − x−2 = x2 · 2x − x−2 = x2·(−2) · 2x − x−2
dx x dx
2 1
= x−4 · 2x − x−2 = 2x−4 · x − x−2 = 2x−4+1 − x−2 = 2x−3 − x−2 = 3
− 2
x x
Question 4. Calculate the following “improper” integrals:

Z 0 Z 1 Z ∞
dx dx
a) √ b) √ c) e−x dx
−2
3
x −1 1 − x2 0
Z ∞ Z ∞ Z ∞
dx dx
d) e) xe−x dx f)
−∞ 4
x +4 0 2 x2 +x−2
π
Z
4 dx
g)
0 sin x cos x
Answer: a)
Z 0 Z 0 0
dx dx Z 0 − 1 3 2 3 3 3

2 2 1
√ = 1 = x 3 dx = x 3 =0− (−2) 3 = (−2) 3 = (−2)2· 3
−2
3
x −2 x 3 −2 2 −2 2 2 2
1
3 3 1 3 √

3
(−2)2
3
=− = − · 43 = − · 4
2 2 2
b) We will first calculate the indefinite integral, using the substitution x = sin u. Then u = arc sin x
dx
and = cos u, hence dx = (cos u) du.
du
Z
dx Z
(cos u) du Z
(cos u) du Z (cos u) du Z
√ = √ = √ = = 1 du = u + C = arc sin x + C
1 − x2 1 − sin2 u cos2 u cos u
for any constant C. We get
Z 1
dx π π

√ = [arc sin x]1−1 = arc sin 1 − arc sin(−1) = − − =π
−1 1−x 2 2 2
220
c) Z ∞
1
h i∞
−x
e dx = −e−x = lim −e −x
− −e 0
= lim − x − (−1) = 0 + 1 = 1
0 0 x→∞ x→∞ e
d) We will first calculate the indefinite integral
dx dx Ax + B Cx + D
Z Z Z
= = + 2 dx
x4 + 4 (x2 − 2x + 2) (x2 + 2x + 2) x − 2x + 2 x + 2x + 2
2
Z
Ax3 + 2Ax2 + 2Ax + Bx2 + 2Bx + 2B + Cx3 − 2Cx2 + 2Cx + Dx2 − 2Dx + 2D
= dx
(x2 − 2x + 2) (x2 + 2x + 2)
0=A+C
0 = 2A + B − 2C + D = 4A + B + D
0 = 2A + 2B + 2C − 2D = 2B − 2D
1 = 2B + 2D = 4B = 4D
− 81 x + 14 1
x + 14
!
dx 1Z 2−x x+2
Z Z
8
= + dx = + 2 dx
x4 + 4 x2 − 2x + 2 x2 + 2x + 2 8 x − 2x + 2 x + 2x + 2
2
1Z −x + 1 x+1 1 1

= + 2 + 2 + 2 dx
8 x − 2x + 2 x + 2x + 2 x − 2x + 2 x + 2x + 2
2
!
1Z −x + 1 x+1 1 1
= + 2 + + dx
8 x − 2x + 2 x + 2x + 2 (x − 1) + 1 (x + 1)2 + 1
2 2
1 1 1

= − ln x2 − 2x + 2 + ln x2 + 2x + 2 + arc tg(x − 1) + arc tg(x + 1) + C
8 2 2
1
for any constant C. The function is even, so
x4 +4
Z ∞ Z ∞
dx dx
4
= 2 4
−∞ x +4 0 x +4
∞
2 1 1

= − ln x2 − 2x + 2 + ln x2 + 2x + 2 + arc tg(x − 1) + arc tg(x + 1)
8 2 2 0
1 1 1

= lim − ln x2 − 2x + 2 + ln x2 + 2x + 2 + arc tg(x − 1) + arc tg(x + 1)
x→∞ 4 2 2
1 1 1

− − ln |2| + ln |2| + arc tg(−1) + arc tg(1)
4 2 2
1 1 1

= x→∞
lim − ln x2 − 2x + 2 + ln x2 + 2x + 2 + arc tg(x − 1) + arc tg(x + 1)
4 2 2
1 1 1

− − ln |2| + ln |2| − arc tg(1) + arc tg(1)
4 2 2
1 1 1

= lim − ln x2 − 2x + 2 + ln x2 + 2x + 2 + arc tg(x − 1) + arc tg(x + 1)
x→∞ 4 2 2
1 1 1

= x→∞
lim − ln (x − 1)2 + 1 + ln (x + 1)2 + 1 + arc tg(x − 1) + arc tg(x + 1)
4 2 2
221
1 1 1

= x→∞
lim − ln (x − 1)2 + ln (x + 1)2 + arc tg(x − 1) + arc tg(x + 1)
4 2 2
1 2 2

= x→∞
lim − ln |x − 1| + ln |x + 1| + arc tg(x − 1) + arc tg(x + 1)
4 2 2
1
= lim (− ln |x| + ln |x| + arc tg(x − 1) + arc tg(x + 1))
x→∞ 4
π π
1 2
+ 2 π
= x→∞
lim (arc tg x + arc tg x) = =
4 4 4
e) We will first calculate the indefinite integral

Z Z Z
xe−x dx = x −e−x − −e−x dx = −xe−x + e−x dx = −xe−x − e−x + C = e−x (−1 − x) + C

Z ∞ h i∞
xe−x dx = e−x (−1 − x) = lim e−x (−1 − x) − e0 (−1 − 0)
0 0 x→∞
−1 − x

= x→∞
lim − 1 · (−1) = 0 + 1 = 1
ex
f) We will first calculate the indefinite integral

dx dx A B Ax − A + Bx + 2B
Z Z Z Z
= = + dx = dx
x2 + x − 2 (x + 2)(x − 1) x+2 x−1 (x + 2)(x + 1)
A+B =0
−A + 2B = 1
− 13 1
!
Z
dx Z
3 1 1
= + dx = − ln |x + 2| + ln |x − 1| + C
x2 + x − 2 x+2 x−1 3 3

Z ∞ ∞
dx 1 1

= − ln |x + 2| + ln |x − 1|
2 x +x−2
2 3 3 2
1 1 1 1

lim − ln |x + 2| + ln |x − 1| − − ln |0 + 2| + ln |0 − 1|
= x→∞
3 3 3 3
1 1 1 1 1 1

lim − ln |x| + ln |x| − − ln 2 + ln 1 = 0 − − ln 2 + 0 = ln 2
= x→∞
3 3 3 3 3 3
g)
2
sin2 x
π π π
!
2
Z
4 dx Z
4 sin x + cos x
Z
4 cos2 x
= dx = + dx
0 sin x cos x 0 sin x cos x 0 sin x cos x sin x cos x
π
sin x cos x
Z π
4
= + dx = [− ln | cos x| + ln | sin x|]04
0 cos x sin x
222
π π

= − ln cos + ln sin − [− ln | cos 0| + ln | sin 0|]
4 4
" #
1 1
= − ln √ + ln √ − [− ln 1 + ln 0] = 0 − (0 + ln 0) = − ln 0 = −(−∞) = ∞
2 2
Temat 17: Norms and inner products

Question 1. Using triangle, inequality show that:
|kxk − kyk| ¬ kx − yk
Answer: The triangle inequality states that
ka + bk ¬ kak + kbk
We will use the triangle inequality with a = x − y and b = y. We get
ka + bk ¬ kak + kbk
kx − y + yk ¬ kx − yk + kyk
kxk ¬ kx − yk + kyk
kxk − kyk ¬ kx − yk
If kxk kyk then

|kxk − kyk| = kxk − kyk ¬ kx − yk
If kyk > kxk then
|kxk − kyk| = kyk − kxk ¬ ky − xk = k(−1)(x − y)k = | − 1|kx − yk = kx − yk
Question 2. Show that the Euclidean norm on the space R2 given by the formula
q
kxk = k(x1 , x2 )k = x21 + x22
is indeed a norm (in particular, the triangle inequality holds).

Answer: We have
q q √ q q
kkxk = k(kx1 , kx2 )k = k 2 x21 + k 2 x22 = k 2 (x21 + x22 ) = k 2 x21 + x22 = |k| x21 + x22 = |k|kxk
We have that q
kxk = k(x1 , x2 )k = x21 + x22 = 0
implies x1 = x2 = 0, hence x = (x1 , x2 ) = (0, 0) = 0
It remains to show that the triangle inequality holds. The triangle inequality states that
kx + yk ¬ kxk + kyk
223
So we need to show that
q q q
(x1 + y1 )2 + (x2 + y2 )2 ¬ x21 + x22 + y12 + y22 (3)
Squaring the left side of equation (3), we get
(x1 + y1 )2 + (x2 + y2 )2 = x21 + 2x1 y1 + y12 + x22 + 2x2 y2 + y22 = x21 + y12 + x22 + y22 + 2x1 y1 + 2x2 y2
| {z }
left
Squaring the right side of equation (3), we get

q q q
x21 + x22 + 2 x21 + x22 y12 + y22 + y12 + y22 = x21 + x22 + 2 (x21 + x22 ) (y12 + y22 ) +y12 + y22
| {z }
right
Now we need to show that the braced term on the left is is smaller than or equal to the braced term
on the right.
? q
2x1 y1 + 2x2 y2 ¬ 2 (x21 + x22 ) (y12 + y22 )
? q
x 1 y 1 + x2 y 2 ¬ (x21 + x22 ) (y12 + y22 )
?
q 2
(x1 y1 + x2 y2 )2 ¬ (x21 + x22 ) (y12 + y22 )
?
x21 y12 + 2x1 y1 x2 y2 + x22 y22 ¬ x21 + x22 y12 + y22
?
x21 y12 + 2x1 y1 x2 y2 +x22 y22 ¬ x21 y12 + x22 y12 + x21 y22 +x22 y22
left right
To finish, we will show that the bracketed term on the left is smaller than or equal to the bracketed
term on the right.
0 ¬ (x2 y1 − x1 y2 )2 = x22 y12 − 2x2 y1 x1 y2 + x21 y22

2x2 y1 x1 y2 ¬ x22 y12 + x21 y22
Question 3. Show that the Euclidean norm on the space Rn given by the formula
q
kxk = k(x1 , x2 , . . . , xn )k = x21 + x22 + · · · + x2n
is indeed a norm (in particular, the triangle inequality holds).

Hint: Use the Schwartz inequality.
Answer: We have
q q
kkxk = k(kx1 , kx2 , . . . , kxn )k = k 2 x21 + k 2 x22 + · · · + k 2 x2n = k 2 (x21 + x22 + · · · + x2n )
√ q q
= k 2 x21 + x22 + · · · + x2n = |k| x21 + x22 + · · · + x2n = |k|kxk
We have that q
kxk = k(x1 , x2 , . . . , xn )k = x21 + x22 + · · · + x2n = 0
224
implies x1 = x2 = · · · = xn = 0, hence x = (x1 , x2 , . . . , xn ) = (0, 0, . . . , 0) = 0
It remains to show that the triangle inequality holds. The triangle inequality states that
We have the inner product
hx, yi = h(x1 , x2 , . . . , xn ) , (y1 , y2 , . . . , yn )i = x1 y1 + x2 y2 + · · · + xn yn
and q q
kxk = k(x1 , x2 , . . . , xn )k = x21 + x22 + · · · + x2n = hx, xi
So the triangle inequality becomes
q q q
hx + y, x + yi ¬ hx, xi + hy, yi
We will use the fact that in Rn we have hx, yi = hy, xi. This is not true in Cn . Squaring both sides
of the triangle inequality, we get
q 2 q q 2
hx + y, x + yi ¬ hx, xi + hy, yi
q 2 q q q 2
hx + y, x + yi ¬ hx, xi + 2 hx, xi hy, yi + hy, yi
q q
hx, x + yi + hy, x + yi ¬ hx, xi + 2 hx, xi hy, yi + hy, yi
q q
hx + y, xi + hx + y, yi ¬ hx, xi + 2 hx, xi hy, yi + hy, yi
q q
hx, xi + hy, xi + hx, yi + hy, yi ¬ hx, xi + 2 hx, xi hy, yi + hy, yi
q q
hx, xi + 2hx, yi +hy, yi ¬ hx, xi + 2 hx, xi hy, yi +hy, yi
| {z } | {z }
left right
The Schwartz inequality states that
|hx, yi|2 ¬ hx, xi · hy, yi
so q
|hx, yi| ¬ hx, xi · hy, yi
which implies that the braced term on the left is smaller than or equal to the braced term on the
right, and hence the triangle inequality holds.
Question 4. Show that any inner product induces a norm, i.e. the formula
q
kvk = hv, vi
defines a norm.
Answer: We only consider norms (k·k) and inner products on n-dimensional Cartesian space Rn . We
have
q q q q q q
kkvk = hkv, kvi = hkv, kvi = khv, kvi = khkv, vi = k 2 hv, vi = |k| hv, vi = kkvk
225
We have that q
kvk = hv, vi = 0
implies that v = 0, the zero vector.
It remains to show that the triangle inequality holds. This is done exactly as in the previous exercise
(Question 3), using the Schwartz inequality.
Question 5. Prove that any norm induced by some inner product satisfies the following formula (the
parallelogram law):
kv + wk2 + kv − wk2 = 2 kvk2 + kwk2
Answer: It is useful to understand why this is called the parallelogram law: In a parallelogram, the
sum of the squares of the diagonals is equal to the the sum of the squares of the sides.
We only consider norms (k·k) and inner products on n-dimensional Cartesian space Rn .
kv + wk2 + kv − wk2 = hv + w, v + wi + hv − w, v − wi
= hv, v + wi + hw, v + wi + hv, v − wi − hw, v − wi
= hv + w, vi + hv + w, wi + hv − w, vi − hv − w, wi
= hv, vi + hw, vi + hv, wi + hw, wi + hv, vi + h−w, vi − hv, wi − h−w, wi
= hv, vi + hw, vi + hv, wi + hw, wi + hv, vi − hw, vi − hv, wi + hw, wi
= hv, vi + hw, wi + hv, vi + hw, wi

= 2 (hv, vi + hw, wi) = 2 kvk2 + kwk2
Question 6. Prove that if a norm k·k is induced by some inner product h·, ·i, then the following is
true
1
hv, wi = kv + wk2 − kv − wk2
4
Answer: We only consider norms (k·k) and inner products on n-dimensional Cartesian space Rn .
1
kv + wk2 − kv − wk2 =
4
226
1
= (hv + w, v + wi − hv − w, v − wi)
4
1
= (hv, v + wi + hw, v + wi − hv, v − wi − h−w, v − wi)
4
1
= (hv, v + wi + hw, v + wi − hv, v − wi + hw, v − wi)
4
1
= (hv + w, vi + hv + w, wi − hv − w, vi + hv − w, wi)
4
1
= (hv, vi + hw, vi + hv, wi + hw, wi − hv, vi − h−w, vi + hv, wi + h−w, wi)
4
1
= (hv, vi + hw, vi + hv, wi + hw, wi − hv, vi + hw, vi + hv, wi − hw, wi)
4
1 1
= (hw, vi + hv, wi + hw, vi + hv, wi) = (hv, wi + hv, wi + hv, wi + hv, wi) = hv, wi
4 4
Question 7. Assume without proof that if the norm k·k fulfills the parallelogram law, then the
formula in Question 6 defines an inner product h·, ·i (Jordan - von Neumann Theorem). Prove that
the norm induced by the inner product h·, ·i is equal to the norm k·k, i.e.
q
hv, vi = kvk
Answer:
s s s
q 1 1 1
hv, vi = (kv + vk2 − kv − vk2 ) = (k2 · vk2 − k0k2 ) = (k2 · vk2 − 0)
4 4 4
s s s
1 1 4 q
= · k2 · vk2 = (2 · kvk)2 = · kvk2 = kvk2 = kvk
4 4 4
Question 8. [Comparison of norms] Let k·k(1) and k·k(2) be two norms defined in Rn . We say that
k·k(2) is stronger (equiv. not weaker) than k·k(1) , if there exists C > 0 such that for every vector v
in Rn , the following inequality holds: kvk(1) ¬ Ckvk(2) . Show that the infinity norm
kxk∞ = max(|x1 |, |x2 |, . . . , |xn |) is stronger than any p-norm on Rn .
Answer: For p 1, the p-norm on Rn is defined as follows
q
p
k(x1 , x2 , . . . , xn ) kp = |x1 |p + |x2 |p + · · · + |xn |p
The Euclidean norm is the 2-norm

q
k(x1 , x2 , . . . , xn ) k2 = |x1 |2 + |x2 |2 + · · · + |xn |2
The infinity norm is obtained by taking the limit of p tending towards infinity
q
p
k(x1 , x2 , . . . , xn ) k∞ = lim k(x1 , x2 , . . . , xn ) kp = lim |x1 |p + |x2 |p + · · · + |xn |p
p→∞ p→∞
227
1 h p +|x p +···+|x p)
i1
= lim (|x1 |p + |x2 |p + · · · + |xn |p ) p = lim eln(|x1 | 2| n| p
p→∞ p→∞
1 p +|x p +···+|x p)
lim e p ·ln(|x1 |
= p→∞ 2| n|
1
Now we need to find the limit of the exponent · ln (|x1 |p + |x2 |p + · · · + |xn |p ).
p
Let |x|max = max(|x1 |, |x2 |, . . . , |xn |) and assume that there are m occurrences of
|x|max in {|x1 |, |x2 |, . . . , |xn |}. Using l’Hôpital’s rule, we get
1
lim · ln (|x1 |p + |x2 |p + · · · + |xn |p ) =
p→∞ p
ln (|x1 |p + |x2 |p + · · · + |xn |p )
= lim
p→∞ p
d
dp
ln (|x1 |p + |x2 |p + · · · + |xn |p )
= p→∞
lim d
dp
p
d
dp
(|x1 |p +|x2 |p +···+|xn |p )
|x1 |p +|x2 |p +···+|xn |p
= lim
p→∞ 1
d
dp
(|x1 |p + |x2 |p + · · · + |xn |p )
= p→∞
lim
|x1 |p + |x2 |p + · · · + |xn |p
h p p p i
d
dp
eln |x1 | + eln |x2 | + · · · + eln |xn |
= p→∞
lim
|x1 |p + |x2 |p + · · · + |xn |p

d
dp
ep·ln |x1 | + ep·ln |x2 | + · · · + ep·ln |xn |
= lim
p→∞ |x1 |p + |x2 |p + · · · + |xn |p
ep·ln |x1 | · ln |x1 | + ep·ln |x2 | · ln |x2 | + · · · + ep·ln |xn | · ln |xn |
= p→∞
lim
|x1 |p + |x2 |p + · · · + |xn |p
p p p
eln |x1 | · ln |x1 | + eln |x2 | · ln |x2 | + · · · + eln |xn | · ln |xn |
= lim
p→∞ |p
|x1 + |x2 + · · · + |xn |p |p
|x1 |p · ln |x1 | + |x2 |p · ln |x2 | + · · · + |xn |p · ln |xn |
= lim
p→∞ |x1 |p + |x2 |p + · · · + |xn |p
|x1 |p ·ln |x1 |+|x2 |p ·ln |x2 |+···+|xn |p ·ln |xn |
|x|pmax m · ln (|x|max )
= lim |x1 | +|x2 |p +···+|xn |p
p = p→∞
lim = ln (|x|max )
p→∞
|x|pmax
m
So we get
1 p +|x p +···+|x p)
k(x1 , x2 , . . . , xn ) k∞ = lim e p ·ln(|x1 | 2| n|
= eln(|x|max ) = |x|max = max(|x1 |, |x2 |, . . . , |xn |)
p→∞
We need to show that there exists C > 0, such that for any vector in Rn , the following inequality
holds q
p
|x1 |p + |x2 |p + · · · + |xn |p ¬ C max(|x1 |, ..., |xn |)
| {z }
√
| {z }
a
¬ p ap +ap +···+ap
228
If a = max(|x1 |, |x2 |, . . . , |xn |), then the biggest the left side can get is
√ √ √ √ √
p
ap + ap + · · · + ap = p nap = p n p ap = p na
√
So the inequality holds with C = p n.
Question 9. Show that any p-norm in Rn is stronger than the infinity norm (see the answer to
Question 8 for the definitions). Together with the result from Question 8, this implies that any two
p-norms are equivalent, in the sense that either one is stronger than the other.
We need to show that there exists C > 0, such that for any vector in Rn , the following inequality
holds q
max(|x1 |, |x2 |, . . . , |xn |) ¬ C p |x1 |p + |x2 |p + · · · + |xn |p
| {z } | {z }
¬a a
q √
If a = p |x1 |p + |x2 |p + · · · + |xn |p , then the biggest the left side can get is p ap = a. This happens
when all the components are zero, except one whose absolute value is a. So the inequality holds with
C = 1.
Question 9. (∗) Show that any two norms on Rn are equivalent.

Answer: It is essential to have understood questions 8 and 9 and how together the results obtained
there imply that any two p-norms on Rn are equivalent. Norms other than p-norms on Rn can be
obtained by taking combinations such as for example the following norm in R4
q
kxkex = 2|x1 | + 3|x2 |2 + [max(|x3 |, 2|x4 |)]2
We will show that the infinity norm k·k∞ is stronger than the norm k·kex . We need to show that
there exists C > 0, such that for any vector in Rn , the following inequality holds
q
2|x1 | + 3|x2 |2 + [max(|x3 |, 2|x4 |)]2 ¬ C max(|x1 |, |x2 |, |x3 |, |x4 |)
√ 2{z
| } | {z }
a
¬2a+ 2
3a +[max(a,2a)]
If a = max(|x1 |, |x2 |, |x3 |, |x4 |), then the biggest the left side can get is
q q √ √ √
2a + 3a2 + [(max a, 2a)]2 = 2a + 3a2 + (2a)2 = 2a + 3a2 + 4a2 = 2a + 7a2 = 2 + 7 a
√
So the inequality holds with C = 2 + 7. There is nothing special about the combination that was
taken to arrive at the norm k·kex . For any norm in Rn , one obtains by a similar calculation a suitable
C. So the infinity norm is stronger than any norm on Rn .
We will now show that the norm k·kex is stronger than the infinity norm k·k∞ . We need to show that
there exists C > 0, such that for any vector in Rn , the following inequality holds
 
q
max(|x1 |, |x2 |, |x3 |, |x4 |) ¬ C 
2|x1 | + 3|x2 |2 + max(|x3 |, 2|x4 |)2 

| {z } | {z }
¬a a
q
If a = 2|x1 | + 3|x2 |2 + [max(|x3 |, 2|x4 |)]2 , then the biggest the left side can get is a. This happens
when |x3 | = a and x1 = x2 = x4 = 0. So the inequality holds with C = 1. There is nothing special
229
about the combination that was taken to arrive at the norm k·kex . For any norm in Rn , one obtains
by a similar calculation a suitable C. So any norm on Rn is stronger than the infinity norm.
Question 10. Show that any two norms k·k(1) and k·k(2) in Rn are equivalent if and only if for every
sequence of vectors v1 , v2 , . . . the following holds:
lim kvn k(1) = 0 ⇔ lim kvn k(2) = 0

n→∞ n→∞
Answer: We will first assume that two norms k·k(1) and k·k(2) are equivalent, in the sense that either
one is stronger than the other, and show that the two sided implication

n→∞ n→∞
holds.
The statement
lim kvn k(1) = 0
n→∞
means that for every > 0, there exits m, such that for all n m, we have kvn k(1) < . Assume this
statement is true.
The statement
lim kvn k(2) = 0
n→∞
means that for every δ > 0, there exits m, such that for all n m, we have kvn k(2) < δ.
The two norms are equivalent, in particular k·k(1) is stronger than k·k(2) . So there exists C > 0, such
that for every vector v in Rn , the following inequality holds: kvk(2) ¬ Ckvk(1) . When kvn k(1) < ,
δ
then kvn k(2) < C. So given δ > 0, we put = and there exists m, such that for all n m, we
C
have kvn k(1) < and consequently kvn k(2) < C = δ. This gives us the implication
lim kvn k(1) = 0 ⇒ n→∞

n→∞
lim kvn k(2) = 0
Similarly, one obtains the implication
lim kvn k(1) = 0 ⇐ n→∞

n→∞
lim kvn k(2) = 0
as a consequence of the fact that k·k(2) is stronger than k·k(1) . This finishes the „only if” part of what
the question asks us to prove.
Now we assume that we have the two sided implication

n→∞ n→∞
and we will show that this implies that the two norms k·k(1) and k·k(2) in Rn are equivalent, in the
sense that either one is stronger than the other. This can be done by showing that we do not have

n→∞ n→∞
if the two norms are not equivalent. So assume that that one of the norms is not stronger than the
other, say k·k(1) is not stronger than k·k(2) . This means that for any C > 0, regardless how big, there
230
exists a vector v such that kvk(2) > Ckvk(1) . Let vn be a sequence of vectors such that
kvn k(2) > nkvn k(1) , since k·k(1) is not stronger than k·k(2) , such a sequence must exist. The inequ-
ality kvn k(2) > nkvn k(1) only makes sense if kvn k(1) is finite and vn = 6 0 for all n. The sequence
(1)
1 1 1
vn = kvn k(1) = converges to zero. But the sequence
nkvn k(1) nkvn k(1) n
(2)
1 1 n
vn = kvn k(2) > kvn k(1) = 1
nkvn k(1) nkvn k (1) nkvn k(1)
does not converge to zero. So we do not have the implication
n→∞
lim kvn k(1) = 0 ⇒ n→∞
lim kvn k(2) = 0
Similarly, we obtain that in the event that k·k(2) is not stronger than k·k(1) , we do not have the
implication
lim kvn k(1) = 0 ⇐ lim kvn k(2) = 0
n→∞ n→∞
This finishes the „if” part of what the question asks us to prove.
Question 11. a) Show that the unit ball in any norm is a convex set.
b) Let W be a convex set in Rn such that:
• for every v ∈ Rn there exists t > 0 such that v ∈ tW , where tW = {tw : w ∈ W }
• for every w ∈ W , all vectors of the form rw, where r ∈ [−1, +1], belong to W
• there exists R > 0 such that for every w ∈ W , w12 + · · · + wn2 < R.
Prove that the formula
kvk = inf{t > 0 : v ∈ tW }
defines a norm.
Answer: a) The unit ball, B, in a norm k·k is the set of vectors which have norm less than or equal
to 1. To show that the unit ball is a convex set, we need to show that the line segment joining two
points in the unit ball is contained in the unit ball. So we need to show that for any two vectors
v, w ∈ B and a real number s such that 0 ¬ s ¬ 1, we have [v + s(w − v)] ∈ B, which means that
kv + s(w − v)k ¬ 1. By the triangle inequality, we get
kv + s(w − v)k = k(1 − s)v + swk ¬ k(1 − s)vk + kswk =

= |1 − s|kvk + |s|kwk ¬ |1 − s| + |s| = 1 − s + s = 1
Note that we used the fact that |1 − s| = 1 − s and |s| = s. This would not be true if we had
s > 1 or s < 0.
b) The first property tells us that W contains the zero vector and, in the case of R, R2 and R3 ,
where the word direction makes sense, for every direction, a vector pointing in that direction. The
second property tells us that if w ∈ W , then −w ∈ W . The third property tells us that the set W
is bounded.
It is clear that for every vector v and every real number s, we have
inf{t > 0 : sv ∈ tW } = |s| inf{t > 0 : v ∈ tW }
231
It is clear that for every vector v, we have
inf{t > 0 : v ∈ tW } 0
It is clear that inf{t > 0 : v ∈ tW } = 0 if and only if v = 0.

Let k·k denote the „norm” defined by kvk= inf{t > 0 : v ∈ tW }. To verify that this indeed defines
a norm, we need to show that the triangle inequality
holds. The closure of of W , denoted by W , is defined as the union of W and all its limit points. For
every v ∈ Rn , we have inf{t > 0 : v ∈ tW }= inf{t > 0 : v ∈ tW }. So without loss of generality
we may assume that W is closed. Let x be a nonzero vector such that kxk = l. Then there exists
a vector v on the boundary of W such that x = lv. Let y be a nonzero vector such that kyk = k.
Then there exists a vector w on the boundary of W such that y = kw. Since the set W is convex,
we know that the line segment joining two points in W is contained in W . We want to express x + y
as a multiple of a vector whose head is on the line segment joining the heads of the vectors v and
w. Every vector whose head is on this line segment can be expressed as v + b(w − v), where b is a
real number such that 0 ¬ b ¬ 1. So
x + y = lv + kw = a[v + b(w − v)] = a[(1 − b)v + bw] = a(1 − b)v + abw
This gives the system of equations
l = a(1 − b)
k = ab
k
which has as only solution a = l + k and b = . So
l+k
" #
k
x + y = (l + k) v + (w − v)
l+k
" #
k
and v + (w − v) ∈ W , which implies that kx + yk ¬ l + k = kxk + kyk
l+k
Temat 18: Subsets of Rn. Limits of functions of several va-

riables. Continuity.
Question 1. Determine if the following sets are bounded, open, closed, compact, convex:
a) {(x, y) ∈ R2 : |x| ¬ 1, |y| < 2}
b) {(x, y, z) ∈ R3 : 2x + y − 3z ¬ 7}
c) {(x, y, z) ∈ R3 : |x + y + z| < 1}
d) {(x, y) ∈ R2 : x2 + 9y 2 > 1}
e) {(x, y) ∈ R2 : 3 + 4x + x2 + y 2 ¬ 0}
f) {(x, y) ∈ R2 : xy 1}
232
g) {(x, y) ∈ R2 : x + y 2 < 1}
h) {(x, y, z) ∈ R3 : x > |y|}
i) {(x, y, z) ∈ R3 : 4 ¬ x2 + y 2 + z 2 < 9}
j) {(x, y, z) ∈ R3 : x2 + y 2 + z 1, x2 + y 2 + z 2 < 4}
Answer: a) {(x, y) ∈ R2 : |x| ¬ 1, |y| < 2}. We need to determine whether the set is bounded, open,
closed, compact, convex. The set corresponds to the interior of a rectangle. The lateral sides are part
of the set, but the top and bottom sides are not. The set is bounded. The line segment joining any two
points in the set is completely contained in the set, so the set is convex. The set is not open, because
for example around the point (1, 0), there is no ball with radius greater than zero which is completely
contained in the set. The set is not closed, because the complement, {(x, y) ∈ R2 : |x| > 1 or |y| 2},
is not open. In order for a set to be compact, it needs to be closed and bounded. Since the set is not
closed, it is not compact.
b) {(x, y, z) ∈ R3 : 2x+y −3z ¬ 7}. We need to determine whether the set is bounded, open, closed,
compact, convex. The set corresponds to the part of R3 which is on or below the plane 2x+y−3z = 7.
The set is not bounded. The line segment joining any two points in the set is completely contained
in the set, so the set is convex. The set is not open, because for example around the point (0, 7, 0),
there is no ball with radius greater than zero which is completely contained in the set. The set is
closed, because the complement, {(x, y, z) ∈ R3 : 2x + y − 3z > 7}, is open. In order for a set to be
compact, it needs to be closed and bounded. Since the set is not bounded, it is not compact.
c) {(x, y, z) ∈ R3 : |x + y + z| < 1}. We need to determine whether the set is bounded, open,
closed, compact, convex. The set corresponds to the part of R3 which is strictly between the plane
x + y + z = −1 and the plane x + y + z = 1. The set is not bounded. The line segment joining
any two points in the set is completely contained in the set, so the set is convex. The set is open,
because around any point in the set, there is a ball with radius greater than zero which is completely
contained in the set. The set is not closed, because the complement, {(x, y, z) ∈ R3 : |x + y + z| 1},
d) {(x, y) ∈ R2 : x2 + 9y 2 > 1}. We need to determine whether the set is bounded, open, closed,
compact, convex. The set corresponds to the part of R2 which is outside the ellipse x2 + (3y)2 = 1.
The set is not bounded. The points (2, 0) and (−2, 0) are part of the set, but the line segment joining
these two points is not completely contained in the set. So the set is not convex. The set is open,
because around any point in the set, there is a ball with radius greater than zero which is completely
contained in the set. The set is not closed, because the complement, {(x, y) ∈ R2 : x2 + 9y 2 ¬ 1}, is
not open. In order for a set to be compact, it needs to be closed and bounded. Since the set is not
e) {(x, y) ∈ R2 : 3 + 4x + x2 + y 2 ¬ 0}. We need to determine whether the set is bounded, open,
closed, compact, convex. We have
3 + 4x + x2 + y 2 = 3 + (x + 2)2 − 4 + y 2 = (x + 2)2 + y 2 − 1 ¬ 0
So (x + 2)2 + y 2 ¬ 1, and the set corresponds to the part of R2 which is inside or on the circle
(x+2)2 +y 2 = 1. The set is bounded. The line segment joining any two points in the set is completely
contained in the set, so the set is convex. The set is not open, because for example around the point
233
(−1, 0), there is no ball with radius greater than zero which is completely contained in the set. The
set is closed, because the complement, {(x, y) ∈ R2 : 3 + 4x + x2 + y 2 > 0}, is open. In order for a
set to be compact, it needs to be closed and bounded. So the set is compact.
f) {(x, y) ∈ R2 : xy 1}. We need to determine whether the set is bounded, open, closed, compact,
1 1
convex. If x > 0, we have y . If x < 0, we have y ¬ . If x = 0, there is no solution, so the y-axis
x x
is not part of the set.
1
The set corresponds to the part of R2 which is above (respectively below) or on the hyperbola y = ,
x
depending on whether x is positive or negative. The set is not bounded. The points (1, 1) and (−1, −1)
are part of the set, but the line segment joining these two points is not completely contained in the
set. So the set is not convex. The set is not open, because for example around the point (1, 1), there
is no ball with radius greater than zero which is completely contained in the set. The set is closed,
because the complement, {(x, y) ∈ R2 : xy < 1}, is open. In order for a set to be compact, it needs
to be closed and bounded. Since the set is not bounded, it is not compact.
g) {(x, y) ∈ R2 : x + y 2 < 1}. We need to determine whether the set is bounded, open, closed,
compact, convex. We have x < 1 − y 2 .
234
The set corresponds to the part of R2 which is strictly inside the parabola x = 1 − y 2 . The set is not
bounded. The line segment joining any two points in the set is completely contained in the set, so the
set is convex. The set is open, because around any point in the set, there is a ball with radius greater
than zero which is completely contained in the set. The set is not closed, because the complement,
{(x, y) ∈ R2 : x + y 2 1}, is not open. In order for a set to be compact, it needs to be closed and
bounded. Since the set is not closed, it is not compact.
h) {(x, y, z) ∈ R3 : x > |y|}. We need to determine whether the set is bounded, open, closed,
compact, convex. If y 0, we have x > y. If y < 0, we have x > −y.
The set corresponds to the part of R3 which is strictly to the right of x = |y| (green). The green
half-lines are in fact the intersection of the half-planes x = y (for y 0) and x = −y (for y < 0) with
the plane z = 0, called the xy-plane. The set is not bounded. The line segment joining any two points
in the set is completely contained in the set, so the set is convex. The set is open, because around
any point in the set, there is a ball with radius greater than zero which is completely contained in
the set. The set is not closed, because the complement, {(x, y, z) ∈ R3 : x ¬ |y|}, is not open. In
235
order for a set to be compact, it needs to be closed and bounded. Since the set is not closed, it is not
compact.
i) {(x, y, z) ∈ R3 : 4 ¬ x2 + y 2 + z 2 < 9}. We need to determine whether the set is bounded, open,
closed, compact, convex. The set corresponds to the part of R3 which is on or outside the sphere
x2 +y 2 +z 2 = 22 and strictly inside the sphere x2 +y 2 +z 2 = 32 . The set is bounded. The points (2, 0, 0)
and (−2, 0, 0) are part of the set, but the line segment joining these two points is not completely
contained in the set. So the set is not convex. The set is not open, because for example around the
point (2, 0, 0), there is no ball with radius greater than zero which is completely contained in the set.
The set is not closed, because the complement, {(x, y, z) ∈ R3 : x2 + y 2 + z 2 < 4 or x2 + y 2 + z 2 9},
j) {(x, y, z) ∈ R3 : x2 + y 2 + z 1, x2 + y 2 + z 2 < 4}. We need to determine whether the set is
bounded, open, closed, compact, convex. The equation x2 + y 2 + z 2 = 22 describes a sphere, and
the equation z = 1 − x2 − y 2 describes a paraboloid (it is called a paraboloid, because putting
r2 = x2 + y 2 one gets z = 1 − r2 , which is a parabola). The set corresponds to the part of R3 which
is strictly inside the sphere x2 + y 2 + z 2 = 22 and on or above the the paraboloid z = 1 − x2 − y 2
(with the z-axis as vertical axis). The points (1, 0, 0) and (−1, 0, 0) are part of the set, but the line
segment joining these two points is not completely contained in the set. So the set is not convex. The
set is not open, because for example around the point (1, 0, 0), there is no ball with radius greater
than zero which is completely contained in the set. The set is not closed, because the complement,
{(x, y, z) ∈ R3 : x2 + y 2 + z < 1 or x2 + y 2 + z 2 4}, is not open. In order for a set to be compact,
it needs to be closed and bounded. Since the set is not closed, it is not compact.
Question 2. Determine whether the following limits exist. If the limit exists, then give the value of
the limit. If the limit does not exist, explain why not:
y 3 − x3
a) lim
(x,y)→(1,1) x − y
√
1 + x2 + y 2 − 1
b) lim
(x,y)→(0,0) x2 + y 2
sin(xy)
c) lim
(x,y)→(1,0) x
sin(xy)
d) lim
(x,y)→(0,1) x
sin(xy)
e) lim
(x,y)→(0,0) x
x2 + y 2
f) lim
(x,y)→(0,0) sin(xy)
xy
g) lim
(x,y)→(0,0) x2 + y 2
x2 y
h) lim
(x,y)→(0,0) x2 + y 2
236
x2 y 2
i) lim
(x,y)→(0,0) x2 + y 2
x2 y
j) lim
(x,y)→(0,0) x2 + y 3

k) lim y ln x2 + y 2
(x,y)→(0,0)
y 3 − x3
Answer: a) We want to determine whether lim exists, and if it does exist, find its
(x,y)→(1,1) x − y
value. We have
y 3 − x3 (x − y) (−y 2 − x2 − xy)
lim = lim = lim −y 2 − x2 − xy = −3
(x,y)→(1,1) x − y (x,y)→(1,1) x−y (x,y)→(1,1)
√
1 + x2 + y 2 − 1
b) We want to determine if lim exists, and if it does exist, find its value. We
(x,y)→(0,0) x2 + y 2
have
√ √ √
1 + x2 + y 2 − 1 1 + x2 + y 2 − 1 1 + x2 + y 2 + 1
lim = lim √
(x,y)→(0,0) x2 + y 2 (x,y)→(0,0) (x2 + y 2 ) 1 + x2 + y 2 + 1
√ 2
1 + x2 + y 2 − 12
= lim √
(x,y)→(0,0) (x2 + y 2 ) 1 + x2 + y 2 + 1
1 + x2 + y 2 − 1
= lim √
(x,y)→(0,0) (x2 + y 2 ) 1 + x2 + y 2 + 1
x2 + y 2
= lim √
(x,y)→(0,0) (x2 + y 2 ) 1 + x2 + y 2 + 1
1 1
= lim √ =
(x,y)→(0,0) 1 + x2 + y 2 + 1 2
sin(xy)
c) We want to determine whether lim exists, and if it does exist, find its value. We
(x,y)→(1,0) x
have
sin(xy) sin(1 · 0) sin 0 0
lim = = = =0
(x,y)→(1,0) x 1 1 1
sin(xy)
d) We want to determine whether lim
exists, and if it does exist, find its value. When
(x,y)→(0,1) x
α is close to zero, then sin α ≈ α. So we get
sin(xy) xy
lim = lim = lim y = 1
(x,y)→(0,1) x (x,y)→(0,1) x (x,y)→(0,1)
sin(xy)
e) We want to determine whether lim exists, and if it does exist, find its value. When
(x,y)→(0,0) x
237
α is close to zero, then sin α ≈ α. So we get
sin(xy) xy
lim = lim = lim y = 0
(x,y)→(0,0) x (x,y)→(0,0) x (x,y)→(0,0)
x2 + y 2
f) We want to determine whether limexists, and if it does exist, find its value. When
(x,y)→(0,0) sin(xy)
α is close to zero, then sin α ≈ α. So if the limit existed, we would have
x2 + y 2 x2 + y 2
lim = lim
(x,y)→(0,0) sin(xy) (x,y)→(0,0) xy
x2 + y 2
We will show that lim does not exist. Approaching (0, 0) along the line y = x, we get
(x,y)→(0,0) xy
x2 + y 2 x2 + x2 2x2
lim = lim = lim 2 = lim 2 = 2
(x,y)→(0,0) xy x→0 x · x x→0 x x→0
Approaching (0, 0) along the line y = −x, we get
x2 + y 2 x2 + (−x)2 2x2
lim = lim = lim = lim (−2) = −2
(x,y)→(0,0) xy x→0 x · (−x) x→0 −x2 x→0
x2 + y 2 x2 + y 2
So lim does not exist, which implies that lim does not exist either.
(x,y)→(0,0) xy (x,y)→(0,0) sin(xy)
xy
g) We want to determine whether lim exists, and if it does exist, find its value. We
(x,y)→(0,0) x + y 2
2
xy
will show that lim does not exist. Approaching (0, 0) along the line y = x, we get
(x,y)→(0,0) x + y 2
2
xy x·x x2 1 1
lim 2 2
= lim 2 2
= lim 2
= lim =
(x,y)→(0,0) x +y x→0 x + x x→0 2x x→0 2 2
Approaching (0, 0) along the line y = −x, we get
xy x · (−x) −x2 −1 1
lim 2 2
= lim 2 2
= lim 2
= lim =−
(x,y)→(0,0) x +y x→0 x + (−x) x→0 2x x→0 2 2
xy
So lim does not exist.
(x,y)→(0,0) x2 + y2
x2 y
h) We want to determine whether lim exists, and if it does exist, find its value.
(x,y)→(0,0) x2 + y 2
Approaching (0, 0) along the line x = 0, we get
x2 y 0·y 0
lim = lim = lim =0
(x,y)→(0,0) x2 + y 2 y→0 0 + y 2 y→0 y 2
Approaching (0, 0) along the line y = mx, for any real number m, we get
x2 y x2 mx mx3 mx
lim 2 2
= lim 2 2
= lim 2 2
= lim =0
(x,y)→(0,0) x +y x→0 x + (mx) x→0 (1 + m )x x→0 1 + m2
238
Approaching (0, 0) along the parabola y = mx2 , for any real number m, we get
x2 y x2 mx2 mx4 mx2

lim = lim = lim = lim =0
(x,y)→(0,0) x2 + y 2 x→0 x2 + (mx2 )2 x→0 x2 + m2 x4 x→0 1 + m2 x2
So we start to suspect that the limit exits and is equal to zero. To prove it, let
q
k(x, y) − (0, 0)k = k(x − 0, y − 0)k = (x − 0)2 + (y − 0)2 < δ =
q q x2
Then we have |y| = y2 ¬ x2 + y 2 < and since ¬ 1, we get
x2 + y 2
x2 y x2 x2
− 0 = y · = |y| · ¬ |y| <
x2 + y 2 x2 + y 2 x2 + y 2
So for every > 0, we can find a δ > 0 (namely δ = ), such that when k(x, y) − (0, 0)k < δ, we have
x2 y x2 y
− 0 < . This proves that lim = 0.
x2 + y 2 (x,y)→(0,0) x2 + y 2
x2 y 2
i) We want to determine whether lim exists, and if it does exist, find its value. As a
(x,y)→(0,0) x2 + y 2
consequence of the result from part (h), we get that this limit must be zero: the only difference is an
additional factor of y in the numerator, and y tends towards zero.
x2 y
j) We want to determine whether lim exists, and if it does exist, find its value. We
(x,y)→(0,0) x2 + y 3
1
are going to prove that the limit does not exist. Approaching (0, 0) along the curve y = x3 − x2 3
,
we get
1 1 1
x2 y x2 (x3 − x2 ) 3 x2 (x3 − x2 ) 3 x2 (x3 − x2 ) 3
lim = lim = lim = lim
x2 + y 3 x→0 x2 + (x3 − x2 ) 31 3 x→0 x2 + (x3 − x2 ) 33 x→0 x2 + x3 − x2
h i
(x,y)→(0,0)
1 1 !1 1
x2 (x3 − x2 ) 3 (x3 − x2 ) 3 x3 − x2 3
1

3
= lim = lim = lim = lim 1−
x→0 x3 x→0 x x→0 x3 x→0 x
1
1

3
The limit lim 1− does not exist, because the „left-side” limit is
x→0 x
1
1

3
lim− 1− =∞
x→0 x
whereas the „right-side” limit is
1
1

3
lim+ 1− = −∞
x→0 x
x2 y
So the limit lim does not exist.
(x,y)→(0,0) x2 + y 3

k) We want to determine whether lim y ln x2 + y 2 exists, and if it does exist, find its value.
(x,y)→(0,0)
239
q q
Let r = x2 + y 2 . Then |y| = y 2 ¬ r, and using l’Hôpital’s rule, we get
ln r
lim y ln x2 + y 2 ¬ lim r ln r2 = lim 2r ln (r) = 2 lim r ln (r) = 2 lim 1
(x,y)→(0,0) r→0 r→0 r→0 r→0
r
1
(ln r)0 r
= 2 lim 0 = 2 lim = 2 lim(−r) = 0
r→0 1 r→0 −1 r→0
r r2
So we have
lim y ln x2 + y 2 = 0
(x,y)→(0,0)
Question 3. Let f be defined by the following formula:

x2 y

if (x, y) 6= (0, 0)



f (x, y) = x4 + y 2



0 if (x, y) = (0, 0)
Show that the limit of f as we approach (0, 0) along any straight line through (0, 0) exists, but despite
this fact, f is not continuous at (0, 0).
Answer: If we approach (0, 0) along the line x = 0, we get
x2 y 02 · y 0
lim 4 2
= lim 4 2
= lim 2 = 0
(x,y)→(0,0) x +y y→0 0 +y y→0 y
If we approach (0, 0) along any line through (0, 0) which is not vertical, the equation of the line is
y = mx, for some m. We get
x2 y x2 · mx mx3 mx
lim 4 2
= lim 4 2
= lim 4 2 2
= lim 2 =0
(x,y)→(0,0) x +y x→0 x + (mx) x→0 x +m x x→0 x + m2
There are actually two cases to consider, m = 0 and m 6= 0. If m = 0, then the numerator is 0 while
the denominator is tending towards zero, so the limit is zero. If m 6= 0, then the numerator tends
towards zero whereas the denominator tends towards a number different from zero, so the limit is
zero.
Now we approach (0, 0) along the parabola y = x2 . We get
x2 y x2 · x2 x4 x4 1 1
lim = lim 2 = lim = lim = lim =
(x,y)→(0,0) x4 + y 2 x→0 x4 + (x2 ) x→0 x4 + x4 x→0 2x4 x→0 2 2
x2 y
So the limit lim does not exist, and f is not continuous at (0, 0).
(x,y)→(0,0) x4 + y 2
Question 4. Let f : R2 \ {(0, 0)} → R be defined by

x2 y 2
f (x, y) =
x2 y 2 + (x − y)2
Show that
lim lim f (x, y) = 0 = lim lim f (x, y)
x→0 y→0 y→0 x→0
240
but despite this fact the limit lim f (x, y) does not exist.
(x,y)→(0,0)
Answer:
! !
x2 y 2 x2 · 02 0

lim lim f (x, y) = lim lim 2 2 = lim = lim =0
x→0 y→0 x→0 y→0 x y + (x − y)2 x→0 x2 · 02 + (x − 0)2 x→0 x2
! !
x2 y 2 02 · y 2 0

lim lim f (x, y) = lim lim 2 2 = lim 2 2 = lim =0
y→0 x→0 y→0 x→0 x y + (x − y)2 y→0 0 · y + (0 − y)2 y→0 y 2
Approaching (0, 0) along the line y = x, we get
x2 y 2 x2 x2 x4
lim f (x, y) = lim = lim = lim = lim 1 = 1
(x,y)→(0,0) (x,y)→(0,0) x2 y 2 + (x − y)2 x→0 x2 x2 + (x − x)2 x→0 x4 x→0
Approaching (0, 0) along the line y = 0, we get

!
x2 y 2 x2 · 02 0
lim f (x, y) = lim = lim = lim =0
(x,y)→(0,0) (x,y)→(0,0) x2 y 2 + (x − y)2 x→0 x2 · 02 + (x − 0)2 x→0 x2
So the limit lim f (x, y) does not exist.

(x,y)→(0,0)
Question 5. Show that any scalar product h·, ·i : Rk × Rk → R is continuous.

Answer: Let k·k denote the Euclidean norm. We need to show that for any > 0, there exists δ > 0,
such that for every vector v, kx − yk < δ implies |hv, xi − hv, yi| < . We have kx − yk < δ if and
only if y = x + λz for a vector z with kzk = 1 and a scalar λ ∈ R with 0 ¬ λ < δ. By the properties
of the inner product and the Schwartz inequality, we have
|hv, xi − hv, yi| = |hv, xi − hv, x + λzi| = |hv, xi − hx + λz, vi| = |hv, xi − (hx, vi + hλz, vi)|
= |hv, xi − hx, vi − hλz, vi| = |hv, xi − hv, xi − λhz, vi| = | − λhz, vi| = λ|hz, vi|
q
¬ λ hz, zihv, vi

So for a given > 0, we can choose δ = q and then kx − yk < δ implies
hz, zihv, vi
q q q
|hv, xi − hv, yi| ¬ λ hz, zihv, vi < δ hz, zihv, vi = q · hz, zihv, vi =
hz, zihv, vi
We need to make sure to avoid division by zero. Since kzk = 1, we know that z 6= 0 and by the
properties of an inner product, hz, zi 6= 0. If v = 0, by the properties of an inner product we have
hv, xi = hv, yi = 0, and therefore |hv, xi − hv, yi| = 0 < .
Question 6. Prove that a function f : Rk → R is continuous if and only if for every open set G ⊂ R,
the set
{(x1 , x2 , . . . , xk ) ∈ Rk : f (x1 , x2 , . . . , xk ) ∈ G}
is an open set in Rk .
241
Answer: Let k·k denote the Euclidean norm. Let
f −1 (G) = {(x1 , x2 , . . . , xk ) ∈ Rk : f (x1 , x2 , . . . , xk ) ∈ G}
We will first prove that if f is continuous and G is open, then f −1 (G) is open. Suppose that p ∈ Rk
and f (p) ∈ G. Then p ∈ f −1 (G). Since G is open, there exists > 0 such that f (x) ∈ G if
|f (x) − f (p)| < . Since f is continuous, there exists δ > 0 such that |f (x) − f (p)| < if kx − pk < δ.
Thus kx − pk < δ implies f (x) ∈ G, so x ∈ f −1 (G). Thus f −1 (G) is open.
Now we will prove that if for every open set G, f −1 (G) is open, then f is continuous. Given p ∈ Rk
and > 0, we must find δ > 0 such that kx − pk < δ implies |f (x) − f (p)| < . Since the set
G = {y ∈ R : |y − f (p)| < }
is open, by assumption, f −1 (G) is an open set. By the definition of open set and the fact that
p ∈ f −1 (G), there exists a δ > 0 such that {x ∈ Rk : kx − bk < δ} ⊂ f −1 (G). This is another way
of saying that
(x ∈ Rk and kx − pk < δ) ⇒ (|f (x) − f (p)| < ),
so f is continuous.
Temat 19: Functions of several variables

Question 1. Determine whether the following limits exist. If the limit exists, then give the value of
the limit. If the limit does not exist, explain why not:
x2 − y 2 x2 − y 2
a) lim √ 2 b) lim
(x,y)→(0,0) x + y2 (x,y)→(0,0) x2 + y 2
xyz xy + yz + zx
c) lim d) lim
(x,y,z)→(0,0,0) x2 + y2 + z2 (x,y,z)→(0,0,0) x2 + y 2 + z 2
Answer a)
x2 − y 2 x2 y2
lim √ 2 = lim √ − lim √
(x,y)→(0,0) x + y 2 (x,y)→(0,0) x2 + y 2 (x,y)→(0,0) x2 + y 2
x y
= lim x · √ 2 − lim y · √ =0−0=0
(x,y)→(0,0) x + y 2 (x,y)→(0,0) x2 + y 2
| {z } | {z }
|···|¬1 |···|¬1
In the last line, the absolute value of the fractions is less than or equal to one, because the absolute
value of the denominator is greater than or equal to the absolute value of the numerator. Since x
and y both tend towards zero, both limits are zero.
x2 − y 2
b) We want to determine whether lim exists, and if it does exist, find its value.
(x,y)→(0,0) x2 + y 2
Approaching (0, 0) along the line y = 0, we get
x2 − y 2 x2 − 02 x2
lim = lim = lim = lim 1 = 1
(x,y)→(0,0) x2 + y 2 x→0 x2 + 02 x→0 x2 x→0
242
Approaching (0, 0) along the line x = 0, we get
x2 − y 2 02 − y 2 −y 2
lim = lim = lim = lim (−1) = −1
(x,y)→(0,0) x2 + y 2 y→0 02 + y 2 y→0 y 2 x→0
x2 − y 2
(x,y)→(0,0) x2 + y 2
c)
xyz yz
lim = lim x· 2
(x,y,z)→(0,0,0) x2 2
+y +z 2 (x,y,z)→(0,0,0) x + y2 + z2
y z
= lim x· √ 2 · √ =0
(x,y,z)→(0,0,0) x + y2 + z2 x2 + y 2 + z 2
| {z } | {z }
|···|¬1 |···|¬1
In the last line, the absolute value of the fractions is less than or equal to one, because the absolute
value of the denominator is greater than or equal to the absolute value of the numerator. Since x
tends towards zero, the limit is zero.
xy + yz + zx
d) We want to determine whether lim exists, and if it does exist, find its
(x,y,z)→(0,0,0) x2 + y 2 + z 2
value. Approaching (0, 0, 0) along the line y = z = 0 (called the x-axis) we get
xy + yz + zx x·0+0·0+0·x 0
lim 2 2 2
= lim 2 2 2
= lim 2 = 0
(x,y,z)→(0,0,0) x + y + z x→0 x +0 +0 x→0 x
Approaching (0, 0, 0) along the line x = y = z, we get
xy + yz + zx xx + xx + xx 3x2
lim = lim = lim = lim 1 = 1
(x,y,z)→(0,0,0) x2 + y 2 + z 2 x→0 x2 + x2 + x2 x→0 3x2 x→0
xy + yz + zx
(x,y,z)→(0,0,0) x2 + y 2 + z 2
Question 2. Determine if the following functions are continuous:

x−y
 3
x + y3

3

 if y 6= x 
 if (x, y) 6= (0, 0)
x3 − y
 
a) f (x, y) =  b) f (x, y) =  x2 + y 2
1 if y = x3
 

0 if (x, y) = (0, 0)
 3
xy + xy 2 2xy 3 + x2 y 3

if (x, y) 6= (0, 0) if (x, y) 6= (0, 0)

 

 
c) f (x, y) =  x2 + y 2 d) f (x, y) =  x4 + 2y 4
 

0 if (x, y) = (0, 0) 
0 if (x, y) = (0, 0)
Answer: a) At the point (0, 0), the equation y = x3 is satisfied, so f (0, 0) = 1. Approaching (0, 0)
along the line y = 0, we get
x−y x−0 x 1
lim = lim 3 = lim 3 = lim 2 = ∞
(x,y)→(0,0) x −y
3 x→0 x −0 x→0 x x→0 x
243
So the function is not continuous.
b) We have
x3 + y 3 x3 y3
lim = lim + lim
(x,y)→(0,0) x2 + y 2 (x,y)→(0,0) x2 + y 2 (x,y)→(0,0) x2 + y 2
x2 y2
= lim x· + lim y · =0+0=0
(x,y)→(0,0) x2 + y 2 (x,y)→(0,0) x2 + y 2
| {z } | {z }
¬1 ¬1
In the last line, the value of the braced fractions is between zero and one, because the value of the
denominator is greater than or equal to the value of the numerator. Since x and y both tend towards
zero, both limits are zero. Since the limit exists and is equal to zero, the function is continuous.
c) We have
x3 y + xy 2 x3 y xy 2
lim = lim + lim
(x,y)→(0,0) x2 + y 2 (x,y)→(0,0) x2 + y 2 (x,y)→(0,0) x2 + y 2
x2 y2
= lim xy · + lim x · =0+0=0
(x,y)→(0,0) x2 + y 2 (x,y)→(0,0) x2 + y 2
| {z } | {z }
¬1 ¬1
In the last line, the value of the braced fractions is between zero and one, because the value of the
denominator is greater than or equal to the value of the numerator. Since x and y both tend towards
zero, both limits are zero. Since the limit exists and is equal to zero, the function is continuous.
d) Approaching (0, 0) along the line y = 0, we get
2xy 3 + x2 y 3 2x · 03 + x2 · 03 0
lim = lim = lim 4 = 0
4
(x,y)→(0,0) x + 2y 4 x→0 x +2·0
4 4 x→0 x
2xy 3 + x2 y 3 2xx3 + x2 x3 2x4 + x5 2+x 2

lim = lim = lim = lim =
(x,y)→(0,0) x4 + 2y 4 x→0 4
x + 2x 4 x→0 3x 4 x→0 3 3
2xy 3 + x2 y 3
The limit lim does not exist, so the function is not continuous.
(x,y)→(0,0) x4 + 2y 4
Question 3. Prove that 

xy

 if x2 + y 2 6= 0
x2 + y2

f (x, y) =

0 if x = y = 0

is not continuous at (0, 0).

Answer: The condition x2 + y 2 6= 0 is equivalent to the condition (x, y) 6= (0, 0). Approaching (0, 0)
along the line y = 0, we get
xy x·0 0
lim = lim 2 = lim 2 = 0
(x,y)→(0,0) x2 +y 2 x→0 x + 0 2 x→0 x
244
xy xx x2 1 1
lim = lim = lim = lim =
(x,y)→(0,0) x2 + y 2 x→0 x2 + x2 x→0 2x2 x→0 2 2
xy
The limit lim does not exist, so the function is not continuous at (0, 0).
(x,y)→(0,0) x2 + y2
Question 4. Find the limits
f (x + h, y) − f (x, y) f (x, y + k) − f (x, y)

lim and lim
h→0 h k→0 k
for each of the following functions:
a) f (x, y) = x + y b) f (x, y) = xy c) f (x, y) = x2 + y 2 d) f (x, y) = x2 y 3 − 2x
Answer: a) We are actually asked to calculate the first partial derivative of f = x + y with respect
to x, and the first partial derivative of f = x + y with respect to y
∂f f (x + h, y) − f (x, y) x + h + y − (x + y) h
= lim = lim = lim = lim 1 = 1
∂x h→0 h h→0 h h→0 h h→0
∂f f (x, y + k) − f (x, y) x + y + k − (x + y) k
∂y k→0 k k→0 k k→0 k k→0
b) We are actually asked to calculate the first partial derivative of f = xy with respect to x, and
the first partial derivative of f = xy with respect to y
∂f f (x + h, y) − f (x, y) (x + h)y − xy xy + hy − xy hy
= lim = lim = lim = lim = lim y = y
∂x h→0 h h→0 h h→0 h h→0 h h→0
∂f f (x, y + k) − f (x, y) x(y + k) − xy xy + xk − xy xk

= lim = lim = lim = lim = lim x = x
∂y k→0 k k→0 k k→0 k k→0 k k→0
c) We are actually asked to calculate the first partial derivative of f = x2 + y 2 with respect to x,
and the first partial derivative of f = x2 + y 2 with respect to y
∂f f (x + h, y) − f (x, y) (x + h)2 + y 2 − (x2 + y 2 )

= lim = lim
∂x h→0 h h→0 h
x2 + 2xh + h2 + y 2 − (x2 + y 2 ) 2xh + h2
= lim = lim = lim (2x + h) = 2x
h→0 h h→0 h h→0
∂f f (x, y + k) − f (x, y) x2 + (y + k)2 − (x2 + y 2 )

= lim = lim
∂y k→0 k k→0 k
x2 + y 2 + 2yk + k 2 − (x2 + y 2 ) 2yk + k 2
= lim = lim = lim (2y + k) = 2y
k→0 k k→0 k k→0
d) We are actually asked to calculate the first partial derivative of f = x2 y 3 − 2x with respect to
245
x, and the first partial derivative of f = x2 y 3 − 2x with respect to y
∂f f (x + h, y) − f (x, y) (x + h)2 y 3 − 2(x + h) − (x2 y 3 − 2x)

= lim = lim
∂x h→0 h h→0 h
(x2 + 2xh + h2 ) y 3 − 2x − 2h − x2 y 3 + 2x
= lim
h→0 h
x2 y 3 + 2xhy 3 + h2 y 3 − 2x − 2h − x2 y 3 + 2x 2xhy 3 + h2 y 3 − 2h
= lim = lim
h→0 h h→0 h

= lim 2xy 3 + hy 3 − 2 = 2xy 3 − 2
h→0
∂f f (x, y + k) − f (x, y) x2 (y + k)3 − 2x − (x2 y 3 − 2x)

= lim = lim
∂y k→0 k k→0 k
x2 (y 3 + 3y 2 k + 3yk 2 + k 3 ) − 2x − x2 y 3 + 2x
= lim
k→0 k
x2 y 3 + 3x2 y 2 k + 3x2 yk 2 + x2 k 3 − 2x − x2 y 3 + 2x 3x2 y 2 k + 3x2 yk 2 + x2 k 3
= lim = lim
k→0 k k→0 k

= lim 3x2 y 2 + 3x2 yk + x2 k 2 = 3x2 y 2
k→0
Temat 20: Partial derivatives and differentiation of func-

tions of several variables.
Question 1. Let 
xy

 if x2 + y 2 6= 0
x2 + y2

f (x, y) =

0 if x = y = 0

∂ ∂
Determine if the partial derivatives f (0, 0) and f (0, 0) exist or not.
∂x ∂y
Answer: The condition x2 + y 2 6= 0 is equivalent to the condition (x, y) =
6 (0, 0). We know from
Question 3 of Temat 19 that the function is not continuous at (0, 0). But here the question is not
about continuity, the question is about the existence of the limits
∂ f (0 + h, 0) − f (0, 0) ∂ f (0, 0 + k) − f (0, 0)

f (0, 0) = lim and f (0, 0) = lim
∂x h→0 h ∂y k→0 k
We have
(0+h)·0
∂ f (0 + h, 0) − f (0, 0) (0+h)2 +02
−0 0
−0
h2
f (0, 0) = lim = lim = lim =
∂x h→0 h h→0 h h→0 h
0−0 0
= lim = lim = lim 0 = 0
h→0 h h→0 h h→0
246
∂
So the partial derivative f (0, 0) exists and is equal to zero. We have
∂x
0·(0+k)
∂ f (0, 0 + k) − f (0, 0) 02 +(0+k)2
−0 0
−0
k2
f (0, 0) = lim = lim = lim =
∂y k→0 k k→0 k k→0 k
0−0 0
= lim = lim = lim 0 = 0
k→0 k k→0 k k→0
∂
So the partial derivative f (0, 0) exists and is equal to zero.
∂y
Question 2. Calculate the partial derivatives of the following functions:
a) f (x, y) = xy b) f (x, y, z) = x2 y 3 z 4 c) f (x, y, z) = x2 + y 3 + z 4

d) f (x, y, z) = exyz e) f (x, y, z) = x4 − 16yz f) f (x, y, z) = xey + yez + zex
!
x
2 y
g) f (x, y, z) = x e ln z h) f (x, y) = arc tg i) f (x, y, z) = 1 − x2 − y 2 − z 2 e−xyz
y
Answer: a)
∂f ∂ ∂ y
= f (x, y) = x = yxy−1
∂x ∂x ∂x
∂f ∂ ∂ y ∂ ln x y ∂ y ln x ∂
= f (x, y) = x = e = e = ey ln x y ln x
∂y ∂y ∂y ∂y ∂y ∂y
y
= ey ln x · ln x = eln x · ln x = xy · ln x
b)
∂f ∂ ∂ 2 3 4
= f (x, y, z) = x y z = 2x2−1 y 3 z 4 = 2xy 3 z 4
∂x ∂x ∂x
∂f ∂ ∂ 2 3 4
= f (x, y, z) = x y z = 3x2 y 3−1 z 4 = 3x2 y 2 z 4
∂y ∂y ∂y
∂f ∂ ∂ 2 3 4
= f (x, y, z) = x y z = 4x2 y 3 z 4−1 = 4x2 y 3 z 3
∂z ∂z ∂z
c)
∂f ∂ ∂ 2
= f (x, y, z) = x + y 3 + z 4 = 2x2−1 = 2x
∂x ∂x ∂x
∂f ∂ ∂ 2
= f (x, y, z) = x + y 3 + z 4 = 3y 3−1 = 3y 2
∂y ∂y ∂y
∂f ∂ ∂ 2
= f (x, y, z) = x + y 3 + z 4 = 4z 4−1 = 4z 3
∂z ∂z ∂z
247
d)
∂f ∂ ∂ xyz ∂
= f (x, y, z) = e = exyz xyz = exyz · x1−1 yz = exyz · x0 yz = exyz · yz
∂x ∂x ∂x ∂x
∂f ∂ ∂ xyz ∂
= f (x, y, z) = e = exyz xyz = exyz · xy 1−1 z = exyz · xy 0 z = exyz · xz
∂y ∂y ∂y ∂y
∂f ∂ ∂ xyz ∂
= f (x, y, z) = e = exyz xyz = exyz · xyz 1−1 = exyz · xyz 0 = exyz · xy
∂z ∂z ∂z ∂z
e)
∂f ∂ ∂ 4
= f (x, y, z) = x − 16yz = 4x4−1 = 4x3
∂x ∂x ∂x
∂f ∂ ∂ 4
= f (x, y, z) = x − 16yz = −16y 1−1 z = −16y 0 z = −16z
∂y ∂y ∂y
∂f ∂ ∂ 4
= f (x, y, z) = x − 16yz = −16yz 1−1 = −16yz 0 = −16y
∂z ∂z ∂z
f)
∂f ∂ ∂
= f (x, y, z) = (xey + yez + zex ) = x1−1 ey + zex = x0 ey + zex = ey + zex
∂x ∂x ∂x
∂f ∂ ∂
= f (x, y, z) = (xey + yez + zex ) = xey + y 1−1 ez = xey + y 0 ez = xey + ez
∂y ∂y ∂y
∂f ∂ ∂
= f (x, y, z) = (xey + yez + zex ) = yez + z 1−1 ex = yez + z 0 ex = yez + ex
∂z ∂z ∂z
g)
∂f ∂ ∂ 2 y
= f (x, y, z) = x e ln z = 2x2−1 ey ln z = 2x1 ey ln z = 2xey ln z
∂x ∂x ∂x
∂f ∂ ∂ 2 y
= f (x, y, z) = x e ln z = x2 ey ln z
∂y ∂y ∂y
∂f ∂ ∂ 2 y 1
= f (x, y, z) = x e ln z = x2 ey ·
∂z ∂z ∂z z
sin β
h) We need to know the derivative of the arc tg function. Let α = tg β = . Then
cos β
arc tg α = arc tg(tg β) = β
and we have
d d
dα d d d sin β cos β dβ sin β − sin β dβ cos β
= α= tg β = = =
dβ dβ dβ dβ cos β (cos β)2
248
cos β cos β − sin β(− sin β) (cos β)2 + (sin β)2 1
= = =
(cos β)2 (cos β)2 (cos β)2
dβ
We are interested in the derivative of arc tg α = β. So we are interested in (arc tg α)0 = . We
dα
sin β
know that cos β = . So we get
α
dβ 1 1 (sin β)2
(arc tg α)0 = = dα = 1 = (cos β)2 =
dα dβ (cos β)2
α2
hence
(1 + α2 )(arc tg α)0 = (arc tg α)0 + α2 (arc tg α)0 = (cos β)2 + (sin β)2 = 1
and
1
(arc tg α)0 =
1 + α2
x
Now we use this with α =
y
!
∂f ∂ ∂ x 1 ∂ x 1 x1−1 1 x0
= f (x, y) = arc tg = 2 · = 2 · = 2 ·
∂x ∂x ∂x y 1 + xy ∂x y 1 + xy y 1 + xy2 y
1
1 1 y y
= x2
· = x2
=
1 + y2 y 1+ y2
y2 + x2
!
∂f ∂ ∂ x 1 ∂ x 1 ∂ −1 1 −1−1
= f (x, y) = arc tg = 2 · = 2 · xy = 2 · (−1)xy
∂y ∂y ∂y y 1+ x ∂y y 1+ yx ∂y 1+ yx
y
1 −2 1 −2 −xy −2 − yx2 x
y2 x
= 2 · (−1)xy = x2
· (−1)xy = x2
= x2
=− x2
=− 2
1+ x 1 + y2 1 + y2 1 + y2 1 + y2 y + x2
y
i)
∂f ∂ ∂
= f (x, y, z) = 1 − x2 − y 2 − z 2 e−xyz
∂x ∂x ∂x
∂ ∂
= e−xyz 1 − x2 − y 2 − z 2 + 1 − x2 − y 2 − z 2 e−xyz
∂x ∂x
∂
= e−xyz −2x2−1 + 1 − x2 − y 2 − z 2 e−xyz (−xyz)
∂x

= e−xyz −2x2−1 + 1 − x2 − y 2 − z 2 e−xyz (−x1−1 yz)

= e−xyz −2x1 + 1 − x2 − y 2 − z 2 e−xyz (−x0 yz)

= −2xe−xyz − yz 1 − x2 − y 2 − z 2 e−xyz = −2x − yz + x2 yz + y 3 z + yz 3 e−xyz
∂f ∂ ∂
= f (x, y, z) = 1 − x2 − y 2 − z 2 e−xyz
∂y ∂y ∂y
249
∂ ∂
= e−xyz 1 − x2 − y 2 − z 2 + 1 − x2 − y 2 − z 2 e−xyz
∂y ∂y
∂
= e−xyz −2y 2−1 + 1 − x2 − y 2 − z 2 e−xyz (−xyz)
∂y

= e−xyz −2y 2−1 + 1 − x2 − y 2 − z 2 e−xyz (−xy 1−1 z)

= e−xyz −2y 1 + 1 − x2 − y 2 − z 2 e−xyz (−xy 0 z)

= −2ye−xyz − xz 1 − x2 − y 2 − z 2 e−xyz = −2y − xz + x3 z + xy 2 z + xz 3 e−xyz
∂f ∂ ∂
= f (x, y, z) = 1 − x2 − y 2 − z 2 e−xyz
∂z ∂z ∂z
∂ ∂
= e−xyz 1 − x2 − y 2 − z 2 + 1 − x2 − y 2 − z 2 e−xyz
∂z ∂z
∂
= e−xyz −2z 2−1 + 1 − x2 − y 2 − z 2 e−xyz (−xyz)
∂z

= e−xyz −2z 2−1 + 1 − x2 − y 2 − z 2 e−xyz (−xyz 1−1 )

= e−xyz −2z 1 + 1 − x2 − y 2 − z 2 e−xyz (−xyz 0 )

= −2ze−xyz − xy 1 − x2 − y 2 − z 2 e−xyz = −2z − xy + x3 y + xy 3 + xyz 2 e−xyz
Question 3. Find all functions f (x, y), such that
∂
f (x, y) = 2xy 3 + ex sin y
∂x
and
∂
f (x, y) = 3x2 y 2 + ex cos y + 1
∂y
Answer: We get Z
f (x, y) = 2xy 3 + ex sin y dx = x2 y 3 + ex sin y + ϕ(y)
where ϕ(y) is a function which depends only on y. Differentiating f (x, y) with respect to y, we get
∂ d
f (x, y) = 3x2 y 2 + ex cos y + ϕ(y)
∂y dy
∂ d
Comparing this with the original expression for f (x, y) we find that ϕ(y) = 1. Thus
∂y dy
Z
ϕ(y) = 1 dy = y + C
for an arbitrary constant C. So we have
f (x, y) = x2 y 3 + ex sin y + y + C
250
Question 4. Examine the differentiability of the following functions:
x2 − y 2

xy · if x2 + y 2 > 0

 q
a) f (x, y) =  x2 + y 2 b) f (x1 , . . . , xk ) = x21 + · · · + x2k
if x2 + y 2 = 0


0
1

2
 x

 + y 2 sin if x2 + y 2 > 0 q
c) f (x, y) =  x2 + y 2 d) f (x, y) = 3
x3 + y 3
if x2 + y 2 = 0

0
1



 sin(xy) if y 6= 0 √
e) f (x, y) = y f) f (x, y, z) = xyz

x

if y = 0
Answer: Before solving this problem, one needs to understand what we are setting out to decide.
The function being differentiable at a point means that it can be linearly approximated in the
neighbourhood of that point in the following way: using a tangent line for functions of one variable,
and using a tangent plane for functions of two variables. To illustrate the concept, we will give two
examples . The function f (x) = |x| does not allow a linear approximation in the neighbourhood
of the point (0, 0) using a tangent line, because it does not have a tangent line there. So it is not
differentiable at the point (0, 0).
The function f (x) = x2 has slope 2 at the point (1, 1). It can be linearly approximated in the
neighbourhood of the point (1, 1) using the tangent line g(x) = −1 + 2x.
251
We will show that the linear approximation using the line g(x) = −1 + 2x is a good approximation
in the neighbourhood of the point (1, 1). We use the fact that g(x) has slope 2, and when moving
away from x = 1 by a distance ∆x, the value of f (x) changes by 2∆x plus a correctional term ∆x.
f (1 + ∆x) = f (1) + 2∆x + ∆x
What is left to show is that tends towards zero when ∆x tends towards zero. We have
f (1 + ∆x) − f (1) − 2∆x
=
∆x
f (1 + ∆x) − f (1) − 2∆x (1 + ∆x)2 − 12 − 2∆x
lim = lim
∆x→0 ∆x ∆x→0 ∆x
12 + 2∆x + (∆x)2 − 12 − 2∆x (∆x)2
= lim = lim = lim ∆x = 0
∆x→0 ∆x ∆x→0 ∆x ∆x→0
a) It is essential to have understood the examples at the beginning of the answer to this question
starting on page 251. We need to determine whether the function
x2 − y 2

xy · if x2 + y 2 > 0


f (x, y) = x2 + y 2

if x2 + y 2 = 0


0
is differentiable. We will show that the function is differentiable. We will first calculate the partial
derivatives for (x, y) 6= (0, 0).
∂ ∂
∂ x2 − y 2 ∂ x3 y − xy 3 (x2 + y 2 ) ∂x (x3 y − xy 3 ) − (x3 y − xy 3 ) ∂x (x2 + y 2 )
xy 2 = =
∂x x + y 2 ∂x x2 + y 2 (x2 + y 2 )2
(x2 + y 2 ) · (3x2 y − y 3 ) − (x3 y − xy 3 ) · 2x
=
(x2 + y 2 )2
∂ ∂
∂ x2 − y 2 ∂ x3 y − xy 3 (x2 + y 2 ) ∂y (x3 y − xy 3 ) − (x3 y − xy 3 ) ∂y (x2 + y 2 )
xy = =
∂y x2 + y 2 ∂y x2 + y 2 (x2 + y 2 )2
(x2 + y 2 ) · (x3 − 3xy 2 ) − (x3 y − xy 3 ) · 2y
=
(x2 + y 2 )2
252
These partial derivatives are clearly defined and continuous, so the function is differentiable every-
where except possibly at the point (0, 0, 0).
Now we will check whether the partial derivatives exist at the point (0, 0, 0). We have
(0+h) −0 2 2
2
∂ f (0 + h, 0) − f (0, 0) (0 + h) · 0 · (0+h)2 +02 − 0 0 · hh2 − 0
f (0, 0) = lim = lim = lim
∂x h→0 h h→0 h h→0 h
0·1−0 0
= lim = lim = lim 0 = 0
h→0 h h→0 h h→0
∂
So the partial derivative fx (0, 0) = f (0, 0) exists and is equal to zero. We have
∂x
2 2
∂ f (0, 0 + k) − f (0, 0) 0 · (0 + k) · 002 −(0+k)
+(0+k)2
−0 0 · −k
2
−0
k2
f (0, 0) = lim = lim = lim
∂y k→0 k k→0 k k→0 k
0 · (−1) − 0 0
= lim = lim = lim 0 = 0
k→0 k k→0 k k→0
∂
So the partial derivative fy (0, 0) = f (0, 0) exists and is equal to zero.
∂y
We are going to show that the function f (x, y) can be linearly approximated by a tangent plane in
the neighbourhood of the point (0, 0, 0).
q
f (0 + ∆x, 0 + ∆y) = f (0, 0) + fx (0, 0) · ∆x + fy (0, 0) · ∆y + (∆x)2 + (∆y)2
q q
The term (∆x)2 + (∆y)2 is a correctional term, where (∆x)2 + (∆y)2 is the distance in the
q
xy-plane from the point (0, 0). We will show that tends towards zero as the distance (∆x)2 + (∆y)2
tends towards zero.
f (0 + ∆x, 0 + ∆y) − f (0, 0) − fx (0, 0) · ∆x − fy (0, 0) · ∆y
= q
(∆x)2 + (∆y)2
f (0 + ∆x, 0 + ∆y) − f (0, 0) − fx (0, 0) · ∆x − fy (0, 0) · ∆y

lim q
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2
(∆x)2 −(∆y)2 (∆x)2 −(∆y)2
(∆x)(∆y) · (∆x)2 +(∆y)2
− 0 − 0 · ∆x − 0 · ∆y (∆x)(∆y) · (∆x)2 +(∆y)2
= lim q = lim q
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2 (∆x,∆y)→(0,0) (∆x)2 + (∆y)2
(∆y) (∆x)2 − (∆y)2

= lim (∆x) q · 2 2
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2 (∆x) + (∆y)
 
(∆y)  (∆x)2 (∆y)2 

= lim (∆x) q · − =0
 
2 2 2 2
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2  (∆x) + (∆y) (∆x) + (∆y) 
| {z } ¬1 ¬1
|···|¬1
The absolute value of the braced term is less than or equal to one, because the value of the denomi-
nator is greater than or equal to the absolute value of the numerator. For the same reason, the value
253
of the bracketed terms is between zero and one. Since ∆x tends towards zero, the limit is zero. So
the of the correctional term tends towards zero when (∆x, ∆y) approaches (0, 0), and this means
that the function f (x, y) can be linearly approximated by the tangent plane in the neighbourhood
of the point (0, 0, 0). Hence the function f (x, y) is differentiable at the point (0, 0, 0).
b) It is essential to have understood the examples at the beginning of the answer to this question
q
f (x1 , . . . , xk ) = x21 + · · · + x2k
is differentiable.
For an integer i with 1 ¬ i ¬ k, we have the partial derivative
∂ ∂ q 2 ∂ 2 1
f (x1 , . . . , xi , . . . , xk ) = x1 + · · · + x2i + · · · + x2k = x1 + · · · + x2i + · · · + x2k 2
∂xi ∂xi ∂xi
1 2 1 −1 ∂ 2
= · x1 + · · · + x2i + · · · + x2k 2 · x1 + · · · + x2i + · · · + x2k
2 ∂xi
1 2 − 1 ∂ 2
= · x1 + · · · + x2i + · · · + x2k 2 · x1 + · · · + x2i + · · · + x2k
2 ∂xi
1 1 ∂ 2 2 2

= · · x + · · · + x i + · · · + x k
2 (x2 + · · · + x2 + · · · + x2 ) 12 ∂xi 1
1 i k
1 1
= · · 2xi2−1
2 (x2 + · · · + x2 + · · · + x2 ) 12
1 i k
1 1
= · · 2x1i
2 (x2 + · · · + x2 + · · · + x2 ) 12
1 i k
xi xi
= =q 1
(x21 + · · · + x2i + · · · + x2k ) x21 + · · · + x2i + · · · + x2k
2
This expression is defined everywhere, except at the point (0, 0, . . . , 0). Now we will check whether
the partial derivatives exist at the point (0, 0, . . . , 0). We have
∂ f (0, . . . , 0, 0 + h, 0, . . . , 0) − f (0, 0, . . . , 0)
f (0, 0, . . . , 0) = lim
∂xi h→0 h
q √
02 + · · · + 02 + (0 + h)2 + 02 + · · · + 02 − 02 + 02 + · · · + 02
= lim
h→0 h
q √
(0 + h)2 − 0
= lim
h→0 h
√ √
h2 − 0 h2 |h|
= lim = lim = lim
h→0 h h→0 h h→0 h
This limit depends on the sign of h. We get
|h| h
lim+ = lim+ = lim+ 1 = 1
h→0 h h→0 h h→0
254
and
|h| −h
lim− = lim− = lim− (−1) = −1
h→0 h h→0 h h→0
∂
So the partial derivative f (0, 0, . . . , 0) does not exist. Since the partial derivatives do not exist,
∂xi
the function f (x1 , . . . , xk ) is not differentiable.
c) It is essential to have understood the examples at the beginning of the answer to this question
1

 x2 + y 2 sin

if x2 + y 2 > 0
2 2

f (x, y) = x + y

if x2 + y 2 = 0

0
!
∂ 2 1 1 ∂ 2 ∂ 1
x + y 2 sin 2 2
= sin 2 x + y 2 + x2 + y 2 sin 2
∂x x +y x + y2 ∂x ∂x x + y2
! !
1
2 2
1 ∂ 1
= sin 2 · 2x + x + y cos 2
x + y2 x + y2 ∂x x + y 2
2
! !
1
2 2
1 −2x
= sin 2 · 2x + x + y cos 2
x + y2 x + y2 (x2 + y 2 )2
!
∂ 2 1 1 ∂ 2 ∂ 1
x + y 2 sin 2 = sin x + y 2 + x2 + y 2 sin 2
∂y x + y2 x2 + y 2 ∂y ∂y x + y2
! !
1
2 2
1 ∂ 1
= sin 2 · 2y + x + y cos 2
x + y2 x + y2 ∂y x2 + y 2
! !
1
2 2
1 −2y
= sin 2 · 2y + x + y cos 2
x + y2 x + y2 (x2 + y 2 )2
where except possibly at the point (0, 0, 0).
Now we will check whether the partial derivatives exist at the point (0, 0, 0). We have
∂ f (0 + h, 0) − f (0, 0) ((0 + h)2 + 02 ) sin (0+h)12 +02 − 0
f (0, 0) = lim = lim
∂x h→0 h h→0 h
h2 sin h12 1
= lim = lim h sin 2 = 0
h→0 h h→0
| {zh }
|···|¬1
The value of the sin function is always between −1 and 1. Since h tends towards zero, the limit is
∂
zero. So the partial derivative fx (0, 0) = f (0, 0) exists and is equal to zero. We have
∂x
1
∂ f (0, 0 + k) − f (0, 0) (02 + (0 + k)2 ) sin 02 +(0+k)2 − 0
f (0, 0) = lim = lim
∂y k→0 k k→0 k
255
k 2 sin k12 1
= lim = lim k sin 2 = 0
k→0 k k→0
| {zk }
|···|¬1
The value of the sin function is always between −1 and 1. Since k tends towards zero, the limit is
∂
zero. So the partial derivative fy (0, 0) = f (0, 0) exists and is equal to zero.
∂y
q
f (0 + ∆x, 0 + ∆y) = f (0, 0) + fx (0, 0) · ∆x + fy (0, 0) · ∆y + (∆x)2 + (∆y)2
q q
q
tends towards zero.
f (0 + ∆x, 0 + ∆y) − f (0, 0) − fx (0, 0) · ∆x − fy (0, 0) · ∆y
= q
(∆x)2 + (∆y)2
f (0 + ∆x, 0 + ∆y) − f (0, 0) − fx (0, 0) · ∆x − fy (0, 0) · ∆y

lim q
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2
1
((∆x)2 + (∆y)2 ) sin (∆x)2 +(∆y)2 − 0 − 0 · ∆x − 0 · ∆y
= lim q
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2
1
((∆x)2 + (∆y)2 ) sin (∆x)2 +(∆y)2 (∆x)2 + (∆y)2 1
= lim q = lim q · sin
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2 (∆x,∆y)→(0,0) (∆x)2 + (∆y)2 (∆x)2 + (∆y)2
 
2 2
(∆x) (∆y)  · sin
1
= lim q +q
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2 2
(∆x) + (∆y) 2 (∆x) + (∆y)2
2
 
 
∆x ∆y 1
 
 
= lim  ∆x · q +∆y · q  · sin =0
(∆x,∆y)→(0,0) 
 2
(∆x) + (∆y) 2 2 2
(∆x) + (∆y) 

(∆x)2 + (∆y)2
 
| {z } | {z } |···|¬1
|···|¬1 |···|¬1
The absolute value of the braced terms is less than or equal to one, because the value of the
denominator is greater than or equal to the absolute value of the numerator. The absolute value of
the bracketed term is less than or equal to one, because the value of the sin function is between
−1 and 1. Since ∆x and ∆y tend towards zero, the limit is zero. So the of the correctional term
tends towards zero when (∆x, ∆y) approaches (0, 0), and this means that the function f (x, y) can
be linearly approximated by the tangent plane in the neighbourhood of the point (0, 0, 0). Hence the
function f (x, y) is differentiable at the point (0, 0, 0).
d) It is essential to have understood the examples at the beginning of the answer to this question
q
3
f (x, y) = x3 + y 3
256
q
is differentiable. We will show that f (x, y) is not differentiable. Putting x = 0, we get z = 3 y 3 = y,
√
3
which is a line with slope 1 in the yz-plane. Putting y = 0, we get z = x3 = x, which is line with
slope 1 in the xz-plane. If the function f (x, y) was differentiable at the point (0, 0, 0), it would allow
for a linear approximation in the neighbourhood of the point (0, 0, 0) using a tangent plane. This
tangent plane would have to contain the line x = 0, z = y, and it would have to contain the line
y = 0, z = x. So the only possibility for a tangent plane is the plane z = x + y. But in the plane
x = y, we have q √
3
√3
√
3
√3
√3
z = 3 x3 + y 3 = x3 + x3 = 2x3 = 2 · x3 = 2 · x
We compare this with the line of intersection of the „tangent plane” z = x + y with the plane√x = y,
3
which is given by x = y, z = x + x = 2x. This line is different from the line x = y, z = 2 · x.
This means that z = x + y cannot be the tangent plane. Since the function f (x, y) does not have a
tangent plane at the point (0, 0, 0), a linear approximation is not possible in the neighbourhood of
that point. Hence the function f (x, y) it is not differentiable.
e) It is essential to have understood the examples at the beginning of the answer to this question
1



 sin(xy) if y 6= 0
e) f (x, y) =  y
x

if y = 0
∂ 1 1 ∂ 1
sin(xy) = cos(xy) · (xy) = cos(xy) · y = cos(xy)
∂x y y ∂x y
∂ 1 ∂ 1 1 ∂ −1 1 ∂
sin(xy) = sin(xy) · + · sin(xy) = sin(xy) · 2 + · cos(xy) · (xy)
∂y y ∂y y y ∂y y y ∂y
−1 1 x 1
= sin(xy) · + · cos(xy) · x = · cos(xy) − sin(xy)
y2 y y y2
where except possibly at y = 0.
Now we will check whether the partial derivatives exist at y = 0. We have
∂ f (x + h, 0) − f (x, 0) x+h−x h
f (x, 0) = lim = lim = lim = lim 1 = 1
∂x h→0 h h→0 h h→0 h h→0
∂
So the partial derivative fx (0, 0) = f (0, 0) exists and is equal to 1. For α close to zero, we have
∂x
sin α ≈ α. Using this, we get
1
∂ f (x, 0 + k) − f (x, 0) 0+k
sin(x · (0 + k)) − x
f (x, 0) = lim = lim
∂y k→0 k k→0 k
1 1 sin(xk)
k
sin(x · 0 + xk) − x k
sin(xk) − x k
−x
= lim = lim = lim
k→0 k k→0 k k→0 k
xk
k
−x x−x 0
k→0 k k→0 k k→0 k k→0
257
∂
So the partial derivative fy (0, 0) = f (0, 0) exists and is equal to zero.
∂y
q
f (0 + ∆x, 0 + ∆y) = f (0, 0) + fx (0, 0) · ∆x + fy (0, 0) · ∆y + (∆x)2 + (∆y)2
q q
q
tends towards zero. For α close to zero, we have sin α ≈ α.
f (0 + ∆x, 0 + ∆y) − f (0, 0) − fx (0, 0) · ∆x − fy (0, 0) · ∆y
= q
(∆x)2 + (∆y)2
f (0 + ∆x, 0 + ∆y) − f (0, 0) − fx (0, 0) · ∆x − fy (0, 0) · ∆y

lim q
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2
1 sin[(∆x)(∆y)]
∆y
sin[(∆x)(∆y)] − 0 − 1 · ∆x − 0 · ∆y ∆y
− ∆x
= lim q = lim q
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2 (∆x,∆y)→(0,0) (∆x)2 + (∆y)2
(∆x)(∆y)
∆y
− ∆x ∆x − ∆x
= lim q = lim q
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2 (∆x,∆y)→(0,0) (∆x)2 + (∆y)2
0
= lim q = lim 0=0
(∆x,∆y)→(0,0) (∆x)2 + (∆y)2 (∆x,∆y)→(0,0)
So the of the correctional term tends towards zero when (∆x, ∆y) approaches (0, 0), and this means
that the function f (x, y) can be linearly approximated by the tangent plane in the neighbourhood
of the point (0, 0, 0). Hence the function f (x, y) is differentiable at the point (0, 0, 0).
f) It is essential to have understood the examples at the beginning of the answer to this question
√
f (x, y, z) = xyz
is differentiable. We will show that the function f (x, y, z) is not differentiable. We have
∂ ∂ √ ∂ 1 1 1 ∂ 1 1 ∂ 1 ∂
f (x, y, z) = xyz = (xyz) 2 = (xyz) 2 −1 xyz = (xyz)− 2 xyz = 1 · xyz
∂x ∂x ∂x 2 ∂x 2 ∂x 2 (xyz) 2 ∂x
1 1 1 yz yz
= 1 · x1−1 yz = 1 · x0 yz = 1 · yz = 1 = √
2 (xyz) 2 2 (xyz) 2 2 (xyz) 2 2 (xyz) 2 2 · xyz
This expression is not defined when xyz is negative, but xyz < 0 is not part of the domain of the
function f (x, y, z). The expression is defined everywhere else, except when at least one of x, y, and
z is equal to zero.
We will check whether the partial derivative with respect to x exists when x = 0. We have
q √
∂ f (0 + h, y, z) − f (0, y, z) (0 + h)yz − 0 · yz
f (0, y, z) = lim = lim
∂x h→0 h h→0 h
258
√ √ √ √ √ √
0 · yz + hyz − 0 · xy 0 + hyz − 0 hyz − 0 hyz
= lim = lim = lim = lim
h→0 h h→0 h h→0 h h→0 h
When y = 0 or z = 0, this limit is zero. But since the power of h in the denominator is higher than
the power of h in the numerator, when y 6= 0 and z 6= 0, this limit can only be infinity or minus
infinity. If a partial derivative does not have a finite value at a point, a linear approximation is not
possible in the neighbourhood of that point. So the function f (x, y, z) is not differentiable.
Question 5. Calculate the derivative in matrix form of the function h = g ◦ f , where:

√
g : R2 → R, g (x1 , x2 ) = e−(x1 +x2 )
2 2

a) f : R+ → R2 , f (x) = x, x
1
b) f : R → R2 , f (x) = (cos(2x), sin(2x)) g : R2 \ {(0, 0)} → R, g (x1 , x2 ) =
x21 + x22
√
c) f : R2 → R3 , f (x, y) = (x − y, x + y, 2 xy) g : R3 \ {(0, 0, 0)} → R,

g (x1 , x2 , x3 ) = ln x21 + x22 + x23

d) f : R2 → R3 , f (x, y) = x2 − y 2 , x2 + y 2 , x2 y 2 g : R3 → R,
g (x1 , x2 , x3 ) = x1 x2 + x2 x3 + x3 x1
e) f : R2 → R2 , f (x, y) = (x − y, x + y) g : R2 → R2 ,
g (x1 , x2 ) = (ex1 cos x2 , ex1 sin x2 )
Answer: a) We need to calculate the derivative in matrix form of the function h = g ◦ f , where:
√
g : R2 → R, g (x1 , x2 ) = e−(x1 +x2 )
2 2

f : R+ → R2 , f (x) = x, x
The composition h = g ◦ f is a function of the variable x. We get

√ 2

− x2 + ( x)
= e−(x )
2 +x
h=g◦f =e
dh d −(x2 +x) d h 2
= e−(x +x) − x + x = e−(x +x) [− (2x + 1)] = (−2x − 1)e−(x +x)
2 2 2
i
= e
dx dx dx
Since h = g ◦ f has one component and is a function of one variable, we get the 1 × 1 matrix
" #
dh

= (−2x − 1)e−(x +x)
2
Dh(x) =
dx
b) We need to calculate the derivative in matrix form of the function h = g ◦ f , where:

1
f : R → R2 , f (x) = (cos(2x), sin(2x)) g : R2 \ {(0, 0)} → R, g (x1 , x2 ) =
x21 + x22
The composition h = g ◦ f is a function of the variable x. We get
1 1
h=g◦f = = =1
(cos(2x))2 + (sin(2x))2 1
259
dh d
= 1=0
dx dx
Since h = g ◦ f has one component and is a function of one variable, we get the 1 × 1 matrix
" #
dh
Dh(x) = = [0]
dx
c) We need to calculate the derivative in matrix form of the function h = g ◦ f , where:

√
f : R2 → R3 , f (x, y) = (x − y, x + y, 2 xy) g : R3 \ {(0, 0, 0)} → R,

g (x1 , x2 , x3 ) = ln x21 + x22 + x23
The composition h = g ◦ f is a function of the variables x and y. We get

√
h = g ◦ f = ln (x − y)2 + (x + y)2 + (2 xy)2 = ln x2 − 2xy + y 2 + x2 + 2xy + y 2 + 4xy
  

= ln 2x2 + 4xy + 2y 2 = ln   2 2 
2 x + 2xy + y  = ln 2(x + y)
2
| {z }
complete square

= ln 2 + ln (x + y)2 = ln 2 + 2 ln(x + y)
∂h ∂ 2 ∂ 2 2
= [ln 2 + 2 ln(x + y)] = 0 + · (x + y) = ·1=
∂x ∂x x + y ∂x x+y x+y
∂h ∂ 2 ∂ 2 2
= [ln 2 + 2 ln(x + y)] = 0 + · (x + y) = ·1=
∂y ∂y x + y ∂y x+y x+y
Since h = g ◦ f has one component and is a function of two variables, we get the 1 × 2 matrix
" #
∂h ∂h 2 2

Dh(x, y) = =
∂x ∂y x+y x+y
d) We need to calculate the derivative in matrix form of the function g ◦ f , where:

f : R2 → R3 , f (x, y) = x2 − y 2 , x2 + y 2 , x2 y 2 g : R3 → R,
g (x1 , x2 , x3 ) = x1 x2 + x2 x3 + x3 x1

h = g ◦ f = x2 − y 2 x2 + y 2 + x2 + y 2 x2 y 2 + x2 y 2 x2 − y 2
= x4 − y 4 + x4 y 2 + x2 y 4 + x4 y 2 − x2 y 4 = x4 − y 4 + 2x4 y 2
∂h ∂ 4
= x − y 4 + 2x4 y 2 = 4x3 − 0 + 8x3 y 2 = 4x3 + 8x3 y 2
∂x ∂x
∂h ∂ 4
= x − y 4 + 2x4 y 2 = 0 − 4y 3 + 4x4 y = −4y 3 + 4x4 y
∂y ∂y
Since h = g ◦ f has one component and is a function of two variables, we get the 1 × 2 matrix
" #
∂h ∂h h i
Dh(x, y) = = 4x3 + 8x3 y 2 −4y 3 + 4x4 y
∂x ∂y
260
e) We need to calculate the derivative in matrix form of the function g ◦ f , where:
f : R2 → R2 , f (x, y) = (x − y, x + y) g : R2 → R2 ,
g (x1 , x2 ) = (ex1 cos x2 , ex1 sin x2 )

h = g ◦ f = e(x−y) cos(x + y), e(x−y) sin(x + y)
∂h1 ∂ (x−y) ∂ ∂
= e cos(x + y) = cos(x + y) e(x−y) + e(x−y) cos(x + y)
∂x ∂x ∂x ∂x
∂ ∂
= cos(x + y)e(x−y) (x − y) − e(x−y) sin(x + y) (x + y)
∂x ∂x
= cos(x + y)e(x−y) · (1) − e(x−y) sin(x + y) · (1) = cos(x + y)e(x−y) − e(x−y) sin(x + y)
= e(x−y) cos(x + y) − e(x−y) sin(x + y) = e(x−y) [cos(x + y) − sin(x + y)]

∂h1 ∂ (x−y) ∂ ∂
= e cos(x + y) = cos(x + y) e(x−y) + e(x−y) cos(x + y)
∂y ∂y ∂y ∂y
∂ ∂
= cos(x + y)e(x−y) (x − y) − e(x−y) sin(x + y) (x + y)
∂y ∂y
= cos(x + y)e(x−y) · (−1) − e(x−y) sin(x + y) · (1) = − cos(x + y)e(x−y) − e(x−y) sin(x + y)
= −e(x−y) cos(x + y) − e(x−y) sin(x + y) = −e(x−y) [cos(x + y) + sin(x + y)]

∂h2 ∂ (x−y) ∂ ∂
= e sin(x + y) = sin(x + y) e(x−y) + e(x−y) sin(x + y)
∂x ∂x ∂x ∂x
∂ ∂
= sin(x + y)e(x−y) (x − y) + e(x−y) cos(x + y) (x + y)
∂x ∂x
= sin(x + y)e(x−y) · (1) + e(x−y) cos(x + y) · (1) = sin(x + y)e(x−y) + e(x−y) cos(x + y)
= e(x−y) sin(x + y) + e(x−y) cos(x + y) = e(x−y) [sin(x + y) + cos(x + y)]

∂h2 ∂ (x−y) ∂ ∂
= e sin(x + y) = sin(x + y) e(x−y) + e(x−y) sin(x + y)
∂y ∂y ∂y ∂y
∂ ∂
= sin(x + y)e(x−y) (x − y) + e(x−y) cos(x + y) (x + y)
∂y ∂y
= sin(x + y)e(x−y) · (−1) + e(x−y) cos(x + y) · (1) = − sin(x + y)e(x−y) + e(x−y) cos(x + y)
= −e(x−y) sin(x + y) + e(x−y) cos(x + y) = e(x−y) [cos(x + y) − sin(x + y)]
261
Since h = g ◦ f has two components and is a function of two variables, we get the 2 × 2 matrix
 
∂h1 ∂h1
 ∂x ∂y 
 
Dh(x, y) = 
 


 ∂h2 ∂h2 

∂x ∂y
 
e(x−y) [cos(x + y) − sin(x + y)] −e(x−y) [cos(x + y) + sin(x + y)]
=



(x−y) (x−y)
e [sin(x + y) + cos(x + y)] e [cos(x + y) − sin(x + y)]
Temat 21: Gradient and growth

Question 1. In which direction does the function f grow most rapidly at P where:
πxy
a) f (x, y) = sin , P = (3, −1)
4
b) f (x, y, z) = ex sin y + ey sin z + ez sin x, P = (0, 0, 0)
Answer: a) The growth is fastest in the direction of the gradient. The gradient is given by
! !
∂ ∂ ∂ πxy ∂ πxy
grad f (x, y) = ∇f (x, y) = f (x, y), f (x, y) = sin , sin
∂x ∂y ∂x 4 ∂y 4
!
πxy ∂ πxy πxy ∂ πxy πy πxy πx πxy

= cos · , cos · = cos , cos
4 ∂x 4 4 ∂y 4 4 4 4 4
The gradient evaluated at the point P = (3, −1) is

!
π · (−1) π · 3 · (−1) π · 3 π · 3 · (−1)
grad f (3, −1) = ∇f (3, −1) = cos , cos
4 4 4 4
!
π −3π 3π −3π π −1 3π −1

= − cos , cos = − ·√ , ·√
4 4 4 4 4 2 4 2
!
π −3π π
= √ , √ = √ (1, −3)
4· 2 4· 2 4 2
So at P , the function growths most rapidly in the direction of the vector (1, −3).
b) The growth is fastest in the direction of the gradient. To get the gradient, we need the partial
derivatives.
∂ ∂ x
f (x, y, z) = (e sin y + ey sin z + ez sin x) = ex sin y + ez cos x
∂x ∂x
∂ ∂ x
f (x, y, z) = (e sin y + ey sin z + ez sin x) = ex cos y + ey sin z
∂y ∂y
∂ ∂ x
f (x, y, z) = (e sin y + ey sin z + ez sin x) = ey cos z + ez sin x
∂z ∂z
262
The gradient is given by
!
∂ ∂ ∂
grad f (x, y, z) = ∇f (x, y, z) = f (x, y, z), f (x, y, z), f (x, y, z)
∂x ∂y ∂z
= (ex sin y + ez cos x, ex cos y + ey sin z, ey cos z + ez sin x)
The gradient evaluated at the point P = (0, 0, 0) is

grad f (0, 0, 0) = ∇f (0, 0, 0) = e0 sin 0 + e0 cos 0, e0 cos 0 + e0 sin 0, e0 cos 0 + e0 sin 0
= (1 · 0 + 1 · 1, 1 · 1 + 1 · 0, 1 · 1 + 1 · 0) = (1, 1, 1)
So at P , the function growths most rapidly in the direction of the vector (1, 1, 1).
Question 2. Suppose we are at the point (−100, −100, 430) on a mountain described by the function
z = 500 − 0.003x2 − 0.004y 2 .
a) In which direction is the slope of the mountain steepest ?
b) How steep is the mountain at this point ? i.e, calculate the angle that an appropriate tangent
makes with the flat ground.
Answer: a) We have
z = f (x, y) = 500 − 0.003x2 − 0.004y 2
The fastest increase occurs in the direction of the gradient. The gradient is given by
!
∂ ∂
grad f (x, y) = ∇f (x, y) = f (x, y), f (x, y)
∂x ∂y
!
∂ ∂
= 500 − 0.003x2 − 0.004y 2 , 500 − 0.003x2 − 0.004y 2
∂x ∂y
= (0 − 2 · 0.003x − 0, 0 − 0 − 2 · 0.004y) = (−0.006x, −0.008y)
The gradient evaluated at (x, y) = (−100, −100) is

1
grad f (−100, −100) = ∇f (−100, −100) = ((−0.006)(−100), (−0.008)(−100)) = (0.6, 0.8) = (3, 4)
5
So at the point (−100, −100, 430), the slope of the mountain is steepest in the direction of the vector
(3, 4).
b) We need the directional derivative in the direction of the gradient. For a vector v, the directional
derivative at (x, y) is given by
v
dv f (x, y) = · ∇f (x, y)
kvk
The dot here stands for the dot product. So we get that at the point (−100, −100, 430), the directional
derivative in the direction of fastest increase is
(3, 4) 1 1 (3, 4) · (3, 4) 1 9 + 16 25
d(3,4) f (−100, −100) = · (3, 4) = = √ = =1
k(3, 4)k 5 5 k(3, 4)k 5 9 + 16 5(5)
Since the slope is 1, the tangent in the direction of fastest increase at the point (−100, −100, 430)
makes an angle of 45◦ with the flat ground. Note that this is a particular case, in general when
263
calculation a directional derivative, the direction vector v can be any vector and does not need to
have the same direction as the gradient.
Question 3. Determine the equation of the tangent line to the curve at the given point P
a) 2x3 + 2y 3 − 9xy = 0 where P = (1, 2)
b) x4 + xy + y 2 = 19 where P = (2, −3)
Answer: To see how this is done, we will first find the tangent line to a curve which is easy to
visualize. We will find the tangent line to the ellipse x2 + 2y 2 = 3 at the point (1, 1). This is an
√
ellipse
s centered at the origin which intersects the x-axis at x = 3 and which intersects the y-axis at
3
y= . This ellipse is the level 3 curve of the function A(x, y) = x2 + 2y 2 . The gradient of A(x, y)
2
is
! !
∂ ∂ ∂ 2 ∂
grad A(x, y) = ∇A(x, y) = A(x, y), A(x, y) = x + 2y 2 , x2 + 2y 2 = (2x, 4y)
∂x ∂y ∂x ∂y
Evaluated at the point (x, y) = (1, 1), the gradient is (2, 4) = 2(1, 2). This is a vector in the direction
of fastest increase of the function A(x, y), so it is perpendicular to the ellipse. To see why the gradient
is perpendicular to the level curve, we sketch here the level A = 3 (red) and the level A = 4 (blue).
This is for illustration purposes only and does not imply that A(x, y) is an integer valued function.
„The next level” does not exist, but when choosing the shortest path to a higher level, taking the
limit case of the higher level approaching the present level, one gets a direction which is perpendicular
to the curve.
The tangent line needs to go through the point (1, 1) and its slope is given by the slope of a vector
which is perpendicular to the gradient vector at that point. Two vectors are perpendicular if they are
both different from the zero vector and their dot product is zero. The vector (−2, 1) is perpendicular
∆y 1 1
to the gradient vector (2, 4). So the slope of the tangent line is = = − . The equation of
∆x (−2) 2
1 1
a line with slope − is given by y = − x + b. Since the tangent line contains the point (1, 1), we
2 2
3
get b = and the equation of the tangent line is
2
1 3
y =− x+
2 2
264
a) It is essential to have understood the example with a curve which is easy to visualize, starting on
page 264.
We need to determine the equation of the tangent line to the curve 2x3 + 2y 3 − 9xy = 0 at the
point P = (1, 2). The curve is the level zero curve of the function A(x, y) = 2x3 + 2y 3 − 9xy. The
gradient of A(x, y) is
!
∂ ∂
grad A(x, y) = ∇A(x, y) = A(x, y), A(x, y)
∂x ∂y
!
∂ 3 ∂
= 2x + 2y 3 − 9xy , 2x3 + 2y 3 − 9xy = 6x2 − 9y, 6y 2 − 9x
∂x ∂y
Evaluated at the point (x, y) = (1, 2), the gradient is (6 − 18, 24 − 9) = (−12, 15). This is a vector
in the direction of fastest increase of the function A(x, y), so it is perpendicular to the curve. The
tangent line needs to go through the point (1, 2) and its slope is given by the slope of a vector which
is perpendicular to the gradient vector at that point. Two vectors are perpendicular if they are both
different from the zero vector and their dot product is zero. The vector (15, 12) is perpendicular to
∆y 12 4
the gradient vector (−12, 15). So the slope of the tangent line is = = . The equation of
∆x 15 5
4 4
a line with slope is given by y = x + b. Since the tangent line contains the point (1, 2), we get
5 5
6
b = and the equation of the tangent line is
5
4 6
y = x+
5 5
b) It is essential to have understood the example with a curve which is easy to visualize, starting
on page 264.
We need to determine the equation of the tangent line to the curve x4 + xy + y 2 = 19 at the point
P = (2, −3). The curve is the level 19 curve of the function A(x, y) = x4 + xy + y 2 . The gradient of
A(x, y) is
!
∂ ∂
grad A(x, y) = ∇A(x, y) = A(x, y), A(x, y)
∂x ∂y
!
∂ 4 ∂
= x + xy + y 2 , x4 + xy + y 2 = 4x3 + y, x + 2y
∂x ∂y
Evaluated at the point (x, y) = (2, −3), the gradient is (32 − 3, 2 − 6) = (29, −4). This is a vector
in the direction of fastest increase of the function A(x, y), so it is perpendicular to the curve. The
tangent line needs to go through the point (2, −3) and its slope is given by the slope of a vector
which is perpendicular to the gradient vector at that point. Two vectors are perpendicular if they are
both different from the zero vector and their dot product is zero. The vector (4, 29) is perpendicular
∆y 29
to the gradient vector (29, −4). So the slope of the tangent line is = . The equation of a line
∆x 4
29 29
with slope is given by y = x + b. Since the tangent line contains the point (2, −3), we get
4 4
35
b=− and the equation of the tangent line is
2
29 35
y = x−
4 2
265
Academic Year 2014/15, Colloquium I, 17 May, 2015
Z
Question 1. Calculate the indefinite integral x arc tg x dx.
1
Answer: We know that (arc tg x)0 = . We use integration by parts with f (x) = x and
1 + x2
g(x) = arc tg x.
Z Z Z
x arc tg x dx = f (x)g(x) dx = F (x)g(x) − F (x)g 0 (x) dx
x2 Z 2
x 1 x2 1Z 2 1
= · arc tg x − · 2
dx = · arc tg x − x · dx
2 2 1+x 2 2 1 + x2
x2 1Z 1 x2 1

= · arc tg x − 1− dx = · arc tg x − (x − arc tg x) + C
2 2 1 + x2 2 2
(x2 + 1) arc tg x − x
= +C
2
for any constant C.
sin β
We used the derivative of the arc tg function. Here is a way to obtain it. Let α = tg β = . Then
cos β
arc tg α = arc tg(tg β) = β
and we have
d d
dα d d d sin β cos β dβ sin β − sin β dβ cos β
= α= tg β = = =
dβ dβ dβ dβ cos β (cos β)2
cos β cos β − sin β(− sin β) (cos β)2 + (sin β)2 1
= = =
(cos β)2 (cos β)2 (cos β)2
dβ
We are interested in the derivative of arc tg α = β. So we are interested in (arc tg α)0 = . We
dα
sin β
know that cos β = . So we get
α
dβ 1 1 (sin β)2
(arc tg α)0 = = dα = 1 = (cos β)2 =
dα dβ (cos β)2
α2
hence
(1 + α2 )(arc tg α)0 = (arc tg α)0 + α2 (arc tg α)0 = (cos β)2 + (sin β)2 = 1
and
1
(arc tg α)0 =
1 + α2
Z 2
Question 2. Calculate the definite integral x ln x dx.
1
Z
Answer: We will first calculate the indefinite integral x ln x dx. We use integration by parts with
266
f (x) = x and g(x) = ln x.
Z Z Z
x ln x dx = f (x)g(x) dx = F (x)g(x) − F (x)g 0 (x) dx
x2 Z 2
x 1 x2 1Z
= ln x − · dx = ln x − x dx
2 2 x 2 2
x2 1 x2 x2 1

= ln x − · +C = ln x − +C
2 2 2 2 2
" # 2
Z 2
x2 1 22 1 12 1 4 1 1 1

x ln x dx = ln x − = ln 2 − − ln 1 − = ln 2 − − 0−
1 2 2 1
2 2 2 2 2 2 2 2
1 1 1 3

= 2 ln 2 − + = 2 ln 2 − 1 + = 2 ln 2 −
2 4 4 4
Z +∞
dx
Question 3. Calculate the improper integral
1 (x + 1)x
Z
dx
Answer: We will first calculate the indefinite integral
(x + 1)x
dx 1 1
Z Z
= − dx = ln |x| − ln |x + 1| + C
(x + 1)x x x+1
for any constant C. Now we evaluate the improper integral

Z +∞
dx

= [ln |x| − ln |x + 1|]+∞
1 = lim [ln |x| − ln |x + 1|] − [ln 1 − ln |1 + 1|]
1 (x + 1)x x→+∞
x x

= lim ln − [0 − ln 2] = ln lim + ln 2
x→+∞ x+1 x→+∞ x + 1
1 1
= ln lim 1 + ln 2 = ln + ln 2 = ln 1 + ln 2 = 0 + ln 2 = ln 2
x→+∞ 1 +
x
1+0
Z +∞ √
Question 4. Calculate the improper integral e− x
dx
0
Z √
Answer: We will first calculate the indefinite integral e− x
dx. We use the substitution
√ 1
u = − x = −x 2 . Then
du 1 0
1 1 1 1 1 1 1
= u0 = −x 2 = − x 2 −1 = − x− 2 = 1 = √ =
dx 2 2 −2x 2 −2 x 2u
So dx = 2u du, and using integration by parts with f (u) = eu and g(x) = u we get
Z √ Z Z Z Z
e− x
dx = eu · 2u du = 2 eu · u du = 2 f (u)g(u) du = 2F (u)g(u) − 2 F (u)g 0 (u) du
267
Z
u
= 2e · u − 2 eu · 1 du = 2eu · u − 2eu + C = 2eu (u − 1) + C =
√
for any constant C. Substituting back u = − x, we get
Z √ √ √ √ √
e− x
dx = 2e− x
− x − 1 + C = −2e− x x+1 +C
Now we evaluate the improper integral

Z +∞ √ h √ √ i+∞ h √ √ i h √ √ i
− x − x − x
e dx = −2e x+1 = lim −2e x+1 − −2e− 0
0+1
0 0 x→+∞
√ ! √ !
−2 ( x + 1) h
−0
i x+1
−0

= lim √ − −2e (0 + 1) = −2 lim √ − −2e
x→+∞ e x x→+∞ e x
√ !!
x 1
= −2 lim √ + √ + 2 · e0 = −2 (0 + 0) + 2 · 1 = 2
x→+∞ e x e x
We used the fact that the denominator grows faster than the numerator to get the limit zero.
n o
Question 5. Is the set (x, y, z) ∈ R3 : z > x2 + y 2 convex? closed? compact?
Answer: To get the picture of what this set looks like, we put r2 = x2 +y 2 . The set z = r2 = x2 +y 2 is
a paraboloid with vertex (0, 0, 0), one may think of it as a parabola rotating around the z-axis. With
the z-axis as vertical axis, the inequality z > x2 + y 2 corresponds to everything above the paraboloid.
One could also say that z > x2 + y 2 corresponds to everything inside the paraboloid. The paraboloid
itself is not part of the set. The set is convex, since given any two points of the set, the straight line
segment joining them is completely contained in the set. The complement, z ¬ x2 + y 2 , is not open.
To see this, we look at the point (0, 0, 0), for example, which is contained in the complement. The
point (0, 0, ), for any > 0 is inside the paraboloid, so it is not contained in the complement. So
there is no open ball around the point (0, 0, 0) which is completely contained in the complement. This
implies that the complement is not open. Since the complement is not open, the set is not closed. To
be compact, a set needs to be closed and bounded. Since the set is not closed, it is not compact.
Question 6. Find the following limit or state that it does not exist:
√ √ !
sin (n2 ) + n2 n+1− n
lim , √1
n→∞ n2 + n e n −1
Answer: Using the fact that the value of the sin function is always between 1 and −1, dividing the
numerator and the dominator by n2 , we get
sin(n2 )
sin (n2 ) + n2 n2
+1 0+1
lim = n→∞
lim 1 = =1
n→∞ n2 + n 1+ n
1+0
Using l’Hôpital’s rule, we get

√ √ √ √ √ √ √ 2
√ 2
n+1− n n+1− n n+1+ n n + 1 − ( n)
lim √1
= n→∞
lim
√ √ = n→∞
lim 1
√ √
n→∞ √1 √
e n −1 e n −1 n+1+ n e −1
n n+1+ n
268
n+1−n 1
= n→∞
lim 1
√ √ = n→∞
lim 1
√ √
√ √
e n −1 n+1+ n e n −1 n+1+ n
1 1 √1
n
= n→∞
lim 1
√ √ = n→∞
lim 1
√ = n→∞
lim 1
√ √ √
e n − 1 ( n + n) 2 e n −1 n 2 e n −1
0 0 0
√1 √1 √1
L0 HR n n n
= lim
n→∞
0 = n→∞
lim 1 0 = lim
n→∞ √1
0
√1
2 e n −1
√
2 e n √1n − 0 2e n √1n
1 1 1
= lim √1
= 0
=
n→∞
2e n 2e 2
So we have √ √ !
sin (n2 ) + n2 n+1− n 1

lim , √1
= 1,
n→∞ n2 + n e n −1 2
Question 7. Find the following limit or state that it does not exist:
√
1 + xy − 1
lim
(x,y)→(0,0) y sin x
Answer: We will use the fact that for α close to zero, we have sin α ≈ α.
√ √ √
√ 2
1 + xy − 1 1 + xy − 1 1 + xy + 1 1 + xy − 1
lim = lim √ = lim √
(x,y)→(0,0) y sin x (x,y)→(0,0) y sin x 1 + xy + 1 (x,y)→(0,0) y sin x 1 + xy + 1
1 + xy − 1 xy
= lim √ = lim √
(x,y)→(0,0) y sin x 1 + xy + 1 (x,y)→(0,0) y sin x 1 + xy + 1
xy 1
= lim √ = lim √
(x,y)→(0,0) yx 1 + xy + 1 (x,y)→(0,0) 1 + xy + 1
1 1
= √ =
1+0+1 2
Question 8. Let f : R2 → R,
 3
x + y3
√ if (x, y) 6= (0, 0)


f (x, y) =  x2 + y 2


1 if (x, y) = (0, 0)
a) Is the function f continuous at the point (0, 0)?

b) Is the function f continuous at the point (−1, 1)?
c) Is the function f bounded?
269
Answer: a) We have
x3 + y 3 x3 y3
lim √ 2 = lim √ + lim √
(x,y)→(0,0) x + y 2 (x,y)→(0,0) x2 + y 2 (x,y)→(0,0) x2 + y 2
x y
= lim x2 √ 2 2
+ lim y 2 √ 2 =0+0=0
(x,y)→(0,0) x +y (x,y)→(0,0) x + y2
| {z } | {z }
|···|¬1 |···|¬1
The absolute value of the braced terms is less than or equal to one, because the absolute value of
the numerator is less than or equal to the value of the denominator. Since x and y both tend towards
zero, the limit is zero. Since
x3 + y 3
lim √ 2 = 0 6= 1 = f (0, 0),
(x,y)→(0,0) x + y2
the function f is not continuous at the point (0, 0).

b) The function f is continuous at the point (−1, 1), because a function which is given by a po-
lynomial divided by a polynomial is continuous at all points where the denominator does not tend
towards zero.
c) The function f is not bounded. For example we can put y = 0 and we get
x3
lim √ = lim x2 = +∞
x→+∞ x2 x→+∞
Question 9. Calculate the partial derivatives of the function f : {(x, y, z) ∈ R3 : x 6= 0} → R where
arc tg(yz)
f (x, y, z) =
x
d 1
Answer: We know that (arc tg a) = (arc tg a)0 = . We get
da 1 + a2
∂ ∂ arc tg(yz) ∂ 1 ∂
f (x, y, z) = = arc tg(yz) = arc tg(yz) x−1 = arc tg(yz)(−1)x−1−1
∂x ∂x x ∂x x ∂x
− arc tg(yz)
= arc tg(yz)(−1)x−2 =
x2
∂ ∂ arc tg(yz) 1 ∂ 1 1 ∂ 1 z ∂
f (x, y, z) = = · arc tg(yz) = · · yz = · · y
∂y ∂y x x ∂y x 1 + (yz)2 ∂y x 1 + (yz)2 ∂y
1 z ∂ 1 1 z 1 z z
= · 2
· y = · 2
· y 1−1 = · 2
· y0 =
x 1 + (yz) ∂y x 1 + (yz) x 1 + (yz) x + xy 2 z 2
∂ ∂ arc tg(yz) 1 ∂ 1 1 ∂ 1 y ∂
f (x, y, z) = = · arc tg(yz) = · 2
· yz = · 2
· z
∂z ∂z x x ∂z x 1 + (yz) ∂z x 1 + (yz) ∂z
1 y ∂ 1 y 1 y y
= · 2
· z1 = · 2
· z 1−1 = · 2
· z0 =
x 1 + (yz) ∂z x 1 + (yz) x 1 + (yz) x + xy 2 z 2
270
Question 10. Let f : {(x, y, z) ∈ R3 : z > 0} → R be the function given by the formula
f (x, y, z) = xe−y + y 2 ln z.
Calculate ∇f (1, 2, 3) · (1, 2, 3).

Answer:
!
∂ ∂ ∂
∇f (x, y, z) = f (x, y, z), f (x, y, z), f (x, y, z)
∂x ∂y ∂z
!
∂ −y 2
∂
−y 2
∂
−y 2

= xe + y ln z , xe + y ln z , xe + y ln z
∂x ∂y ∂z
!
1−1 −y −y ∂ y2
= x e + 0, xe (−y) + 2y 2−1 ln z, 0 +
∂y z
! !
0 −y −y y2 1 −y −y y2
= x e , xe (−1) + 2y ln z, = e , −xe + 2y ln z,
z z
!
22 1 1 4

−2 −2
∇f (1, 2, 3) = e , (−1) · e + 2 · 2 ln 3, = 2 , − 2 + 4 ln 3,
3 e e 3
1 1 4 1 1 4

∇f (1, 2, 3) · (1, 2, 3) = , − + 4 ln 3, · (1, 2, 3) = (1) + − + 4 ln 3 (2) + (3)
e2 e2 3 e2 e2 3
1 2 1
= 2
− 2 + 8 ln 3 + 4 = − 2 + 8 ln 3 + 4
e e e
Question 11. (additional question) For the following sets, determine whether they are open, closed,
convex, bounded, compact, connected.
n o
a) (x, y) ∈ R2 : x2 > y 2
n o
b) (x, y) ∈ R2 : x2 y 2
n o
c) (x, y) ∈ R2 : x2 < y 2
n o
d) (x, y) ∈ R2 : x2 ¬ y 2
n o
e) (x, y, z) ∈ R3 : x2 + y 2 > z 2
n o
f) (x, y, z) ∈ R3 : x2 + y 2 z 2
n o
g) (x, y, z) ∈ R3 : x2 + y 2 < z 2
n o
h) (x, y, z) ∈ R3 : x2 + y 2 ¬ z 2
Answer:
271
a) The red lines correspond to x2 = y 2 . The inequality x2 > y 2 is equivalent to the inequality
|x| > |y|. The set is open, because around every point in this set exists an open ball which is
completely contained in the set. The set is not closed, because the complement, x2 ¬ y 2 , is not open.
A set is convex if for any two points in the set, the straight line segment joining them is completely
contained in the set. The points (−2, 1) and (2, 1) are part of the set, but the line segment joining
them is not completely contained in the set. So the set is not convex. The set is not bounded. A set
is compact if it is closed and bounded. Since the set is not bounded, it is not compact. In Euclidean
space (such as R, R2 , and R3 , with the Euclidean distance as metric), a set is connected if any two
points can be joined by a path which is completely contained in the set. The point (0, 0) is not part
of the set. So for example the points (−1, 0) and (1, 0), which are part of the set, cannot be joined
by a path which is completely contained in the set. So the set is not connected.
b) The red lines correspond to x2 = y 2 . The inequality x2 y 2 is equivalent to the inequality
|x| |y|. The set is not open, because for example around the point (0, 0), which is part of the set,
there does not exist an open ball which is completely contained in the set. The set is closed, because
the complement, x2 < y 2 , is open. A set is convex if for any two points in the set, the straight line
segment joining them is completely contained in the set. The points (−2, 1) and (2, 1) are part of
the set, but the line segment joining them is not completely contained in the set. So the set is not
convex. The set is not bounded. A set is compact if it is closed and bounded. Since the set is not
bounded, it is not compact. In Euclidean space (such as R, R2 , or R3 , with the Euclidean distance as
metric), a set is connected if any two points can be joined by a path which is completely contained
in the set. The point (0, 0) is part of the set, and therefore any two points in the set can be joined
by a path which is completely contained in the set. So the set is connected.
c) In comparison with part a), by symmetry, the set x2 < y 2 is open, is not closed, is not convex, is
not bounded, is not compact, is not connected.
d) In comparison with part b), by symmetry, the set x2 ¬ y 2 is not open, is closed, is not convex,
is not bounded, is not compact, is connected.
e) One may put r2 = x2 + y 2 . Then the set r2 = x2 + y 2 > z 2 , with the z-axis as vertical axis, is
272
like the solid of revolution obtained by rotating the region which corresponds to the set from part
a) around the vertical axis. By analogy with part a), the set x2 + y 2 > z 2 is open, is not closed, is
not convex, is not bounded, is not compact. It is connected, because in the three dimensional case,
any two points can be connected by a path which is completely contained in the set: one can „go”
around the z-axis and remain in the set.
f) One may put r2 = x2 + y 2 . Then the set r2 = x2 + y 2 z 2 , with the z-axis as vertical axis, is
like the solid of revolution obtained by rotating the region which corresponds to the set from part b)
around the vertical axis. By analogy with part b), the set x2 + y 2 z 2 is not open, is closed, is not
convex, is not bounded, is not compact, is connected.
g) One may put r2 = x2 + y 2 . Then the set r2 = x2 + y 2 < z 2 , with the z-axis as vertical axis, is
like the solid of revolution obtained by rotating the region which corresponds to the set from part c)
around the vertical axis. By analogy with part c), the set x2 + y 2 < z 2 is open, is not closed, is not
convex, is not bounded, is not compact. It is not connected, because since (0, 0, 0) is not part of the
set, for example the points (0, 0, 1) and (0, 0, −1), which are part of the set, cannot be joined by a
path which is completely contained in the set.
h) One may put r2 = x2 + y 2 . Then the set r2 = x2 + y 2 ¬ z 2 , with the z-axis as vertical axis, is
like the solid of revolution obtained by rotating the region which corresponds to the set from part d)
around the vertical axis. By analogy with part d), the set x2 + y 2 ¬ z 2 is not open, is closed, is not
convex, is not bounded, is not compact, is connected.
Academic Year 2016/17, Colloquium I, 11 April, 2017, Ver-

sion A
Z
1
Question 1. Calculate the indefinite integral dx
x2 (x + 1)
Answer:
1 1−x 1 1 1 1 1 1
Z Z Z Z
dx = + dx = − + dx = x−2 − + dx
x2 (x + 1) x 2 x+1 x 2 x x+1 x x+1
1 x+1 1 1
= −x−1 − ln |x| + ln |x + 1| + C = − + ln + C = − + ln 1 + + C
x x x x
for any constant C.
Z e
Question 2. Calculate the definite integral √
3
3x2 ln x3 dx
e
Z
Answer: We will first calculate the indefinite integral 3x2 ln x3 dx. We will use integration by

parts with f (x) = 3x2 and g(x) = ln x3 .
Z Z Z
2
3x ln x 3
dx = f (x)g(x) dx = F (x)g(x) − F (x)g 0 (x) dx
3

3
Z
3x2 Z
= x ln x − x · 3 dx = x ln x3 − 3x2 dx
3 3
x

= x3 ln x3 − x3 + C = x3 ln x3 − 1 + C = x3 (3 ln x − 1) + C
273
Z e h ie √ 3 √
√ 3x2 ln x3 dx = x3 (3 ln x − 1) √
3
= e3 (3 ln e − 1) − 3
e 3 ln 3
e −1
3 e e
1

1

3 3
= e (3 · 1 − 1) − e 3 ln e 3 − 1 = e (3 − 1) − e 3 · ln (e) − 1
3
= 2e3 − e (ln (e) − 1) = 2e3 − e (1 − 1) = 2e3 − e · 0 = 2e3 − 0 = 2e3
Z +∞ (−x 13 )
e
Question 3. Calculate the improper integral 2 dx
0 x3
1
Z
e(−x 3 )
Answer: We will first calculate the indefinite integral 2 dx. We notice that the derivative of
x3
1 1 1 2 1
the exponent is − x 3 −1 = − x− 3 = − 2 . So we get
3 3 3x 3
1
Z
e(−x 3 ) 1
2 dx = −3e(−x 3 ) + C
x3
for any constant C. Now we evaluate the improper integral.
Z +∞ (−x 13 ) +∞
e 1 1 1

2 dx = −3e(−x 3 ) = lim −3e(−x 3 ) − −3e(−0 3 )
0 x3 1 x→+∞
" #
−3
= lim 1 − −3e0 = 0 − (−3 · 1) = 3
x→+∞ (x 3 )
e
Question 4. Determine whether the set {(x, y) ∈ R2 : |x| + |y| ¬ 1} is open, closed, bounded,
compact, connected, convex.
Answer: The set corresponds to everything inside or on the boundary of the square with vertices
(1, 0), (0, 1), (−1, 0), (0, −1). The point (1, 0) is part of this set. But for any > 0, the point (1 + , 0)
is not part of the set. So around the point (1, 0), there is no open ball which is completely contained
in the set. Hence the set is not open. The set is closed, because the complement,
{(x, y) ∈ R2 : |x| + |y| > 1}, is open. The set is bounded. A set is compact if it is closed and bounded.
So the set is compact. In Euclidean space (such as R, R2 , R3 , with the Euclidean distance as metric),
a set is connected if any two points belonging to the set can be joined by a path which is completely
contained in the set. The set is connected. A set is convex if given any two points of the set, the
straight line segment joining them is completely contained in the set. The set is convex.
Question 5. Find the following limit or state that it does not exist
n
√
!
cos n 1

lim √ , 1+ , n
n
n→∞ n n
cos n 1 n ln(n)

lim √ = 0. We know that n→∞
Answer: cos n is bounded, so n→∞ lim 1 + = e. Since n→∞
lim = 0,
n n n
we have
√ 1
1 1 ln n
lim
n→∞
n
n = lim
n→∞
n n = lim
n→∞
e ln n n
lim e n ·ln n = n→∞
= n→∞ lim e n = e0 = 1
274
So we get
n
√
!
cos n 1

lim √ , 1+ , n
n = (0, e, 1)
n→∞ n n
x3 y 2
lim
(x,y)→(0,0) x5 + 2y 5
Answer: Approaching (0, 0) along the line x = 0, we get
x3 y 2 03 · y 2 0
lim 5 5
= lim 5 5
= lim 5 = lim 0 = 0
(x,y)→(0,0) x + 2y y→0 0 + 2y y→0 2y y→0
x3 y 2 x3 · x2 x5 1 1
lim 5 5
= lim 5 5
= lim 5
= lim =
(x,y)→(0,0) x + 2y x→0 x + 2x x→0 3x x→0 3 3
So the limit does not exist.
Question 7. Determine whether the function f : R2 → R,

1

2
(x

 + y 2 ) sin if (x, y) 6= (0, 0)
f (x, y) = x2 + y2

if (x, y) = (0, 0)

1
is continuous.
Answer: The function is continuous everywhere except possibly at (0, 0), where the denominator
1 1
of the expression 2 2
is zero. The value of sin 2 is always between −1 and 1, and since
x +y x + y2
x2 + y 2 tends towards zero, we get
1
lim (x2 + y 2 ) sin = 0 6= 1 = f (0, 0)
(x,y)→(0,0) x2 + y2
So the function is not continuous at (0, 0).
Question 8. Let f : {(x, y, z) ∈ R3 : z > −1} → R be the function given by the formula
f (x, y, z) = (z + 1)x + y 2 ln(z + 1)
Find a unit vector w which gives the direction of maximum rate of increase of f at the point (1, 1, 0).
Answer: The maximum rate of increase occurs in the direction of the gradient. So we need to
calculate the gradient of the function f (x, y, z).
!
∂ ∂ ∂
∂x ∂y ∂z
275
!
∂ ∂ ∂
= (z + 1)x + y 2 ln(z + 1) , (z + 1)x + y 2 ln(z + 1) , (z + 1)x + y 2 ln(z + 1)
∂x ∂y ∂z
!
∂ ln(z+1) x ∂ ∂
= e + y 2 ln(z + 1) , (z + 1)x + y 2 ln(z + 1) , (z + 1)x + y 2 ln(z + 1)
∂x ∂y ∂z
!
∂ x ln(z+1) 2
∂
x 2
∂
x 2

= e + y ln(z + 1) , (z + 1) + y ln(z + 1) , (z + 1) + y ln(z + 1)
∂x ∂y ∂z
!
x ln(z+1) ∂ ∂ 1 ∂
= e x ln(z + 1) + 0, 0 + 2y ln(z + 1), x(z + 1)x−1 (z + 1) + y 2 (z + 1)
∂x ∂z z + 1 ∂z
!
x ln(z+1) ∂ 1
= e x ln(z + 1), 2y ln(z + 1), x(z + 1)x−1 · 1 + y 2 ·1
∂x z+1
1

x ln(z+1) x−1 2
= e ln(z + 1), 2y ln(z + 1), x(z + 1) +y
z+1
1
x
= eln(z+1) ln(z + 1), 2y ln(z + 1), x(z + 1)x−1 + y 2
z+1
1

x x−1 2
= (z + 1) ln(z + 1), 2y ln(z + 1), x(z + 1) +y
z+1
Evaluated at (x, y, z) = (1, 1, 0), we have
grad f (1, 1, 0) = ∇f (1, 1, 0)

1

= (0 + 1)1 ln(0 + 1), 2 · 1 · ln(0 + 1), 1 · (0 + 1)1−1 + 12
0+1

= ln 1, 2 · ln 1, 10 + 1 = (0, 2 · 0, 1 + 1) = (0, 0, 2)
So the maximum rate of increase of f at the point (1, 1, 0) occurs in the direction of the vector
(0, 0, 2). Since we need a unit vector (a vector of length one), we need to divide (0, 0, 2) by its length,
and we get
w = (0, 0, 1)
Qustion 9. Let f be the function described in Question 8. Give the value of the directional derivative
dv f (1, 1, 0)
where v = (4, 0, 3).

Answer:
v 1
dv f (1, 1, 0) = · grad f (1, 1, 0) = √ 2 (4, 0, 3) · (0, 0, 2)
kvk 4 + 02 + 3 2
1 1
= √ (4, 0, 3) · (0, 0, 2) = (4, 0, 3) · (0, 0, 2)
25 5
1 1 6
= (4 · 0 + 0 · 0 + 3 · 2) = · 6 =
5 5 5
276
Question 10. Let (u, v) denote coordinates on R2 and let f : R2 → R be a function described in the
coordinates (u, v). If g(r, θ) = f (r cos θ, r sin θ), then find the values of the constants A and B such
that
∂g cos θ ∂g ∂f ∂f
sin θ (r, θ) + (r, θ) = A (u, v) + B (u, v).
∂r r ∂θ ∂u ∂v
∂g ∂g
Hint: Compute and using the chain rule.
∂r ∂θ
Answer: We have g(r, θ) = f (r cos θ, r sin θ) = f (u, v). So u = r cos θ and we have
∂u ∂u
= cos θ and = −r sin θ
∂r ∂θ
Also, v = r sin θ and we have
∂v ∂v
= sin θ and = r cos θ
∂r ∂θ
The chain rule gives
∂g ∂f ∂u ∂f ∂v
= +
∂r ∂u ∂r ∂v ∂r
∂g ∂f ∂u ∂f ∂v
= +
∂θ ∂u ∂θ ∂v ∂θ
Substituting the partial derivatives which we computed, we get
∂g ∂f ∂f
(r, θ) = cos θ + sin θ
∂r ∂u ∂v
∂g ∂f ∂f
(r, θ) = (−r sin θ) + r cos θ
∂θ ∂u ∂v
These expressions we now put in the equation which needs to be satisfied
sin θ (r, θ) + (r, θ) = A (u, v) + B (u, v)
∂r r ∂θ ∂u ∂v
! !
∂f ∂f cos θ ∂f ∂f ∂f ∂f
sin θ cos θ + sin θ + (−r sin θ) + r cos θ = A (u, v) + B (u, v)
∂u ∂v r ∂u ∂v ∂u ∂v
∂f ∂f ∂f ∂f ∂f ∂f
sin θ cos θ + (sin θ)2 − sin θ cos θ + (cos θ)2 = A (u, v) + B (u, v)
∂u ∂v ∂u ∂v ∂u ∂v
∂f ∂f ∂f
(sin θ)2 + (cos θ)2 = A (u, v) + B (u, v)
∂v ∂u ∂v
∂f ∂f ∂f
= A (u, v) + B (u, v)
∂v ∂u ∂v
From the last line, we get B = 1 and A = 0.
Academic Year 2016/17, Colloquium I, 11 April, 2017, Ver-

sion B
Z
1
Question 1. Calculate the indefinite integral dx
x(x2 + 1)
277
Answer:
1 1 x 1 1 2x
Z Z Z
dx = − 2 dx = − · 2 dx
x(x2 + 1) x x +1 x 2 x +1
1 1
= ln |x| − · ln |x2 + 1| + C = ln |x| − · ln(x2 + 1) + C
2 2
for any constant C.
Z e
Question 2. Calculate the definite integral x2 ln x9 dx
1
Z
Answer: We will first calculate the indefinite integral x2 ln x9 dx. We will use integration by

parts with f (x) = x2 and g(x) = ln x9 .
Z Z Z
2
x ln x 9
dx = f (x)g(x) dx = F (x)g(x) − F (x)g 0 (x) dx
x3 9 Z x3 9x8 x3 9 Z
= ln x − · dx = ln x − 3x2 dx
3 3 x9 3
!
x3 9 ln (x9 )
= ln x − x3 + C = x3 −1 +C
3 3

" !#e ! !
Z e
2

9

3 ln (x9 ) 3 ln (e9 ) ln (19 )
x ln x dx = x −1 =e − 1 − 13 −1
1 3 1
3 3
! !
9 ln e ln 1 9·1 0

3
=e −1 − − 1 = e3 −1 − −1
3 3 3 3
= e3 (3 − 1) − (0 − 1) = 2e3 + 1
Z +∞
3x2 + 1
Question 3. Calculate the improper integral dx
1 (x3 + x)2
3x2 + 1 Z
Answer: We will first calculate the indefinite integral dx. We notice that in the nume-
(x3 + x)2
rator appears the derivative of the term in brackets in the denominator.
Z
3x2 + 1 Z
2

3
−2 Z
3
0
3
−2
dx = 3x + 1 x + x dx = x + x x + x dx
(x3 + x)2
−1 −1
= − x3 + x +C = +C
x3+x
+∞
Z +∞
3x2 + 1 −1 −1 −1 −1 1

2 dx = 3
= lim 3 − 3 =0− =
1 3
(x + x) x +x 1 x→+∞ x + x 1 +1 1+1 2
278
Question 4. Determine whether the set {(x, y) ∈ R2 : y ¬ x2 + 1} is open, closed, bounded,
compact, connected, convex.
Answer: The set corresponds to everything on or below the parabola y = x2 + 1. The point (0, 1)
is part of this set. But for any > 0, the point (0, 1 + ) is not part of the set. So around the point
(0, 1), there is no open ball which is completely contained in the set. Hence the set is not open. The
set is closed, because the complement, {(x, y) ∈ R2 : y > x2 + 1}, is open. The set is not bounded.
A set is compact if it is closed and bounded. Since the set is not bounded, it is not compact. In
Euclidean space (such as R, R2 , R3 , with the Euclidean distance as metric), a set is connected if any
two points belonging to the set can be joined by a path which is completely contained in the set.
The set is connected. A set is convex if given any two points of the set, the straight line segment
joining them is completely contained in the set. The points (−2, 2) and (2, 2) are part of the set. But
the straight line segment joining these two points is not completely contained in the set. So the set
is not convex.
ln(n) 1 √
!
lim , en , n n
n→∞ n
ln(n)
Answer: ln n grows more and more slowly as n gets big, so lim = 0.
n→∞ n
1 1 ln(n)
We have lim = 0, so lim e n = e0 = 1. Since lim = 0, we have
n→∞ n n→∞ n→∞ n
√ 1
1 1 ln n
lim n
n = lim n n = lim eln n n
= lim e n ·ln n = lim e n = e0 = 1
n→∞ n→∞ n→∞ n→∞ n→∞
So we get
ln(n) 1 √
!
lim , e n , n n = (0, 1, 1)
n→∞ n
x4 y
lim
(x,y)→(0,0) x4 + 2y 2
Answer:
x4 y x4
lim = lim y =0
(x,y)→(0,0) x4 + 2y 2 (x,y)→(0,0) x4 + 2y 2
| {z }
¬1
The value of the braced term is between zero and one, because the value of the denominator is greater
than or equal to the value of the numerator. Since y tends towards zero, the limit is zero.
Question 7. Determine whether the function f : R2 → R,

 2
x − y2
if (x, y) 6= (0, 0)



f (x, y) = x+y



0 if (x, y) = (0, 0)
279
is continuous.
Answer: The function is continuous everywhere except possibly at (0, 0), where the denominator of
x2 − y 2
the expression is zero. We have
x+y
x2 − y 2 (x + y)(x − y)
lim = lim = lim (x − y) = 0 = f (0, 0)
(x,y)→(0,0) x + y (x,y)→(0,0) x+y (x,y)→(0,0)
So the function is continuous at (0, 0), which implies that it is continuous.
Question 8. Let f : {(x, y, z) ∈ R3 } → R be the function given by the formula
f (x, y, z) = sin(zx) + x cos 2y
π π

Find a vector w which gives the direction of maximum rate of increase of f at the point
, ,0 .
2 4
Answer: The maximum rate of increase occurs in the direction of the gradient. So we need to
calculate the gradient of the function f (x, y, z).
!
∂ ∂ ∂
∂x ∂y ∂z
!
∂ ∂ ∂
= (sin(zx) + x cos 2y), (sin(zx) + x cos 2y), (sin(zx) + x cos 2y)
∂x ∂y ∂z
!
∂ ∂ ∂
= cos(zx) zx + cos 2y, 0 − x sin(2y) 2y, cos(zx) zx + 0
∂x ∂y ∂z
= (z cos(zx) + cos 2y, −2x sin(2y), x cos(zx))
π π

Evaluated at (x, y, z) = , , 0 , we have
2 4
π π π π

grad f , , 0 = ∇f , ,0
2 4 2 4
π π π π π π

= 0 · cos 0 · + cos 2 · , −2 · sin 2 · , · cos 0 ·
2 4 2 4 2 2
π π π π π

= 0 + cos , −π sin , · cos 0 = 0 + 0, −π · 1, · 1 = 0, −π,
2 2 2 2 2
π π

So the maximum rate of increase of f at the point , , 0 occurs in the direction of the vector
2 4
π π

w = 0, −π, = (0, −2, 1)
2 2
Qustion 9. Let f be the function described in Question 8. Give the value of the directional derivative
π π

dv f , ,0
2 4
280
where v = (4, −2, 4).
Answer:
π π v π π 1 π

dv f , ,0 = · grad f , ,0 = q (4, −2, 4) · 0, −π,
2 4 kvk 2 4 42 + (−2)2 + 42 2
1 π 1 π

= √ (4, −2, 4) · 0, −π, = (4, −2, 4) · 0, −π,
36 2 6 2
π π 8π 2π
= (4, −2, 4) · (0, −2, 1) = (4 · 0 + (−2) · (−2) + 4 · 1) = =
2·6 2·6 2·6 3
Question 10. Let (u, v) denote coordinates on R2 and let f : R2 → R be a function described in the
coordinates (u, v). If g(r, θ) = f (r cos θ, r sin θ), then find the values of the constants A and B such
that
sin θ (r, θ) + (r, θ) = A (u, v) + B (u, v).
∂r r ∂θ ∂u ∂v
∂g ∂g
Hint: Compute and using the chain rule.
∂r ∂θ
Answer: We have g(r, θ) = f (r cos θ, r sin θ) = f (u, v). So u = r cos θ and we have
∂u ∂u
= cos θ and = −r sin θ
∂r ∂θ
Also, v = r sin θ and we have
∂v ∂v
= sin θ and = r cos θ
∂r ∂θ
The chain rule gives
∂g ∂f ∂u ∂f ∂v
= +
∂r ∂u ∂r ∂v ∂r
∂g ∂f ∂u ∂f ∂v
= +
∂θ ∂u ∂θ ∂v ∂θ
Substituting the partial derivatives which we computed, we get
∂g ∂f ∂f
(r, θ) = cos θ + sin θ
∂r ∂u ∂v
∂g ∂f ∂f
(r, θ) = (−r sin θ) + r cos θ
∂θ ∂u ∂v
These expressions we now put in the equation which needs to be satisfied
sin θ (r, θ) + (r, θ) = A (u, v) + B (u, v)
∂r r ∂θ ∂u ∂v
! !
∂f ∂f cos θ ∂f ∂f ∂f ∂f
sin θ cos θ + sin θ + (−r sin θ) + r cos θ = A (u, v) + B (u, v)
∂u ∂v r ∂u ∂v ∂u ∂v
∂f ∂f ∂f ∂f ∂f ∂f
sin θ cos θ + (sin θ)2 − sin θ cos θ + (cos θ)2 = A (u, v) + B (u, v)
∂u ∂v ∂u ∂v ∂u ∂v
281
∂f ∂f ∂f
(sin θ)2 + (cos θ)2 = A (u, v) + B (u, v)
∂v ∂u ∂v
∂f ∂f ∂f
= A (u, v) + B (u, v)
∂v ∂u ∂v
From the last line, we get B = 1 and A = 0.
Temat 22: Extrema of functions of several variables.

Question 1. Determine whether at the point (x, y) = (0, 0), the following functions have a local
extremum:
a) z = z(x, y) = x2 + y 2
b) z = z(x, y) = x2 − y 2
Answer: a) We have z = x2 + y 2 . One may put r2 = x2 + y 2 , and the surface z = r2 = x2 + y 2 is a
paraboloid with vertex (0, 0, 0). So (x, y, z) = (0, 0, 0) is a local and global minimum of the function
z = z(x, y) = x2 + y 2 .
b) We have z = x2 − y 2 . Putting x = 0, moving away from the point (x, y) = (0, 0) along the
y-axis, the value of the function z(x, y) is decreasing. So (0, 0, 0) cannot be a local minimum. Putting
y = 0, moving away from the point (x, y) = (0, 0) along the x-axis, the value of the function z(x, y)
is increasing. So (0, 0, 0) cannot be a local maximum. Since (x, y, z) = (0, 0, 0) cannot be a local
minimum and it cannot be a local maximum, it cannot be a local extremum.
Question 2. Does the function w = f (x, y, z) = xy + yz + zx have local extrema?

Answer: For a local extremum to occur at a point (x, y, z), a necessary condition is that the first
order partial derivatives are all zero at that point. This means that
∂ ∂
f (x, y, z) = (xy + yz + zx) = y + z = 0
∂x ∂x
∂ ∂
f (x, y, z) = (xy + yz + zx) = x + z = 0
∂y ∂y
∂ ∂
f (x, y, z) = (xy + yz + zx) = y + x = 0
∂z ∂z
We get y = −z = x = −y, so y = −y which implies that y = 0 and consequently x = 0 and z = 0. So
the only possibility for a point at which a local extremum can occur is the point (x, y, z) = (0, 0, 0).
Putting z = 0, moving away from the point (x, y, z) = (0, 0, 0) along the line y = x, the value of the
function f (x, y, z) is increasing. So (0, 0, 0, 0) cannot be a local maximum. Putting z = 0, moving
away from the point (x, y, z) = (0, 0, 0) along the line y = −x, the value of the function f (x, y, z)
is decreasing. So (0, 0, 0, 0) cannot be a minimum. Since (x, y, z, w) = (0, 0, 0, 0) cannot be a local
maximum and it cannot be a local minimum, it cannot be a local extremum.
Question 3. Determine sup f (x, y) and inf f (x, y) for the following functions:
x,y∈D x,y∈D
q
a) f (x, y) = x2 + y 2 , D = {(x, y) ∈ R2 : x2 + y 2 ¬ 1}
282
b) f (x, y) = xy 2 , D = {(x, y) ∈ R2 : x2 + y 2 ¬ 3}
c) f (x, y) = xy − x − y + 3, D is the region bounded by the triangle with vertices (0, 0), (2, 0),
(0, 4)
d) f (x, y) = x2 + y 2 − x − y, D is the region bounded by the triangle with vertices (0, 0), (2, 0),
(0, 2)
e) f (x, y) = x + 2y, D is the region bounded by the square with vertices (±1, ±1)
f) f (x, y) = x2 + y 2 − x, D is the region bounded by the square with vertices (±1, ±1)
q
Answer: a) We have f (x, y) = x2 + y 2 , and D = {(x, y) ∈ R2 : x2 + y 2 ¬ 1}. We can put
q
x2 + y 2 = r2 , and the function z = x2 + y 2 = r describes a cone. The cone has absolute minimum
z = 0 at (x, y) = (0, 0). So inf f (x, y) = 0. On D, the cone reaches the maximum at the boundary,
x,y∈D
so sup f (x, y) = 1.
x,y∈D
b) We have f (x, y) = xy 2 and D = {(x, y) ∈ R2 : x2 + y 2 ¬ 3}. The region D is a disk with

√
radius 3. The function z = f (x, y) = xy 2 reaches extrema at the boundary of this disk. This is so,
because given the value of the function at an interior point of the disk D, moving radially outward
can only further increase or decrease the value without changing the sign. At the boundary, we have

x2 + y 2 = 3, so y 2 = 3 − x2 . Substituting this into z = f (x, y), we get z = x 3 − x2 = 3x − x3 . Now
this is a function of one variable of which we need to find local extrema. We take the derivative and
put it equal to zero. So 3 − 3x2 = 0, which has solutions x = ±1. We get z = 2 for x = 1, this is
a maximum and hence sup f (x, y) = 2. We get z = −2 for x = −1, this is a minimum and hence
x,y∈D
inf f (x, y) = −2.
x,y∈D
c) We have f (x, y) = xy − x − y + 3, and D is the region bounded by the triangle with vertices
(0, 0), (2, 0), (0, 4). We are going to check if f (x, y) has local extrema, so we calculate the partial
derivatives and put them equal to zero.
∂ ∂
f (x, y) = (xy − x − y + 3) = y − 1 = 0
∂x ∂x
y=1
∂ ∂
f (x, y) = (xy − x − y + 3) = x − 1 = 0
∂y ∂y
x=1
So there is a critical point at (1, 1) where the function has value 2. It can be shown that this is a
saddle point as follows. If we put y = x, which is a line containing the point (1, 1), we have
f (x, x) = x2 − x − x + 3 = x − 2x + 1 + 2 = (x − 1)2 + 2
which is a upwards shaped parabola with minimum at x = 1. So (1, 1) cannot be a maximum. If we

put y = 2 − x, which is a line containing the point (1, 1), we have
f (x, 2 − x) = x(2 − x) − x − (2 − x) + 3 = 2x − x2 − x − 2 + x + 3 = −(x2 − 2x + 1) + 2 = −(x − 1)2 + 2
283
which is a downwards shaped parabola with maximum at x = 1. So (1, 1) cannot be a minimum.
Hence extrema must occur on the boundary.
The boundary of D consists of three line segments. We will check them one by one for maximum
and minimum values.
The line segment joining (0, 0) and (2, 0) is part of the line y = 0. Replacing y = 0 in f (x, y) we get
f (x, 0) = x · 0 − x − 0 + 3 = −x + 3
The maximum that f (x, y) reaches on this segment is 3, the minimum is 1.

The line segment joining (0, 0) and (0, 4) is part of the line x = 0. Replacing x = 0 in f (x, y) we get
f (0, y) = 0 · y − 0 − y + 3 = −y + 3
The maximum that f (x, y) reaches on this segment is 3, the minimum is −1.
The line segment joining (2, 0) and (0, 4) is part of the line y = 4 − 2x. Replacing y = 4 − 2x in
f (x, y), we get

f (x, 4 − 2x) = x(4 − 2x) − x − (4 − 2x) + 3 = 4x − 2x2 − x − 4 + 2x + 3 = − 2x2 − 5x − 1
!2
2

25

25 √ 5 25
= − 2x − 5x + + −1=− 2x − √ + −1
| {z
8 } 8 2 2 8
complete square
!2
√ 5 17
=− 2x − √ +
2 2 8
5 5
Since x = √ √ = is between x = 0 and x = 2, the maximum that f (x, y) reaches on this
2 2 2 4
17
segment is . The minimum must occur at one of the endpoints, which are also part of one of the
8
other two line segments.
So sup f (x, y) = 3 and inf f (x, y) = −1.
x,y∈D x,y∈D
d) We have f (x, y) = x2 + y 2 − x − y, and D is the region bounded by the triangle with vertices
(0, 0), (2, 0), (0, 2). The shape of the surface determined by z = f (x, y) is a circular paraboloid, this
becomes evident if we express z as follows:
2 2
1 1 1 1 1 1

z = f (x, y) = x + y − x − y = x − x + + y 2 − y + − = x −
2 2 2
+ y− −
| {z 4} | {z 4} 2 2 2 2
1 1 1 1

The minimum occurs at (x, y) = and has value − . So inf f (x, y) = − . The maximum
,
2 2 2 x,y∈D 2
1 1

occurs at points in D which are furthest away from the vertex , . The points (0, 2) and (2, 0)
2 2
1 1

are furthest away from , , so sup f (x, y) = f (0, 2) = f (2, 0) = 2
2 2 x,y∈D
284
e) We have f (x, y) = x + 2y, and D is the region bounded by the square with vertices (±1, ±1). It
is clear that the maximum is reached when x = y = 1, so sup f (x, y) = f (1, 1) = 3. It is clear that
x,y∈D
the minimum is reached when x = y = −1, so inf f (x, y) = f (−1, −1) = −3.
x,y∈D
f) We have f (x, y) = x2 + y 2 − x, and D is the region bounded by the square with vertices (±1, ±1).
The shape of the surface determined by z = f (x, y) is a circular paraboloid, this becomes evident if
we express z as follows:
2
1 1 1 1

z = f (x, y) = x2 + y 2 − x = x2 − x + + y2 − = x − + y2 −
| {z 4} 4 2 4
complete square
1 1 1

The minimum occurs at (x, y) = , 0 and has value − . So inf f (x, y) = − . The maximum
2 4 x,y∈D 4
1

occurs at points in D which are furthest away from the vertex , 0 . The points (−1, 1) and (−1, −1)
2
1

are furthest away from , 0 , so sup f (x, y) = f (−1, 1) = f (−1, −1) = 3
2 x,y∈D
Question 4. Determine the maximum value of the following functions:

4 3
a) z = z(x, y) = 1 + x + 4y 3 − x4 − y 4
3
2 −y 2
b) z = z(x, y) = 1 + x2 e−x
4 3
Answer: a) We have z = z(x, y) = 1 + x + 4y 3 − x4 − y 4 . The biggest power of both x and y is 4,
3
and these biggest powers have a coefficient of −1. So when moving out on the axes, the value of the
function goes to minus infinity. This implies that the maximum value occurs at a local extremum. A
necessary condition for having a local extremum is that the first order partial derivatives are zero.
So we calculate them and put them equal to zero.
∂z ∂ 4 4

= 1 + x3 + 4y 3 − x4 − y 4 = 0 + 3 · x2 + 0 − 4x3 − 0 = 4x2 (1 − x) = 0
∂x ∂x 3 3
x = 0 or 1
∂z ∂ 4

= 1 + x3 + 4y 3 − x4 − y 4 = 0 + 0 + 3 · 4y 2 − 0 − 4y 3 = 4y 2 (3 − y) = 0
∂y ∂y 3
y = 0 or 3
So local extrema can occur at (0, 0), (0, 3), (1, 0), (1, 3). It suffices to check which of the four pairs
gives the biggest z-value. We find that the maximum occurs at (x, y) = (1, 3) and has value
4 3 4 4 85
1+ 1 + 4 · 33 − 14 − 34 = + (4 − 3)33 = + 27 =
3 3 3 3
2 −y 2
b) We have z = z(x, y) = 1 + x2 e−x . When x gets big in absolute value, the exponential
2
function grows faster than x . Since there is a minus sign in the exponent, when moving out on the
axes, the value of the function goes to zero. And the function is positive valued. So the maximum
285
value occurs at a local extremum. A necessary condition for having a local extremum is that the first
order partial derivatives are zero. So we calculate them and put them equal to zero.
∂z ∂ 2 2 2 2
2 2 ∂

= 1 + x2 e−x −y = 2xe−x −y + 1 + x2 e−x −y −x2 − y 2
∂x ∂x ∂x
2 −y 2
2 −y 2 2 −y 2
2 −y 2 2 −y 2
= 2xe−x + 1 + x2 e−x (−2x) = 2xe−x − 2x + 2x3 e−x = −2x3 e−x =0
x=0
∂z ∂ 2 2
2 2 ∂
2 2
= 1 + x2 e−x −y = 1 + x2 e−x −y −x2 − y 2 = 1 + x2 e−x −y (−2y)
∂y ∂y ∂y
y=0
So there is a critical point at (0, 0), the maximum occurs there and has value
2 −02
1 + 02 e−0 = e0 = 1
Question 5. Find the dimensions x, y, z of a rectangular box with fixed volume V = 1000 and
minimal surface area. Does the question make sense if we seek to maximize the surface area?
1000
Answer: The volume is V = 1000 = xyz. So z = , and the area, which we want to minimize, is
xy
given by
2000 2000
A = 2xy + 2yz + 2xz = 2xy + +
x y
We have A as a function of x and y, to check for local extrema we calculate the first order partial
!
∂A ∂ 2000 2000 2000 2000
= 2xy + + = 2y − 2
+ 0 = 2y − 2
∂x ∂x x y x x
!
∂A ∂ 2000 2000 2000 2000
= 2xy + + = 2x + 0 − 2
= 2x − 2
∂y ∂y x y y y
1000 1000
From the first equation, we get y = 2
. From the second equation, we get x = 2 . Substituting
x y
the expression for y into the expression for x, we get
1000 1000 1000 x4

x= = 2 = 10002
=
y2 1000

1000
x2 x4
So x3 = 1000. This means x = 10, and through back substitution, we get y = z = 10. So there is one
critical point, this must be a minimum. Since there is only one critical point which is a minimum,
seeking to maximize the surface area makes no sense.
Question 6. Determine the dimensions x, y, z of a rectangular box with maximum volume and total
surface area of the sides equal to 600 cm2
286
300 − xy
Answer: The surface area is 600 = 2xy + 2yz + 2xz. So z = , and the volume, which we
x+y
want to maximize, is given by
300 − xy 300xy − x2 y 2
V = xyz = xy =
x+y x+y
We have V as a function of x and y, to check for local extrema we calculate the first order partial
∂V (300y − 2xy 2 ) (x + y) − (300xy − x2 y 2 ) 300xy − 2x2 y 2 + 300y 2 − 2xy 3 − 300xy + x2 y 2
= =
∂x (x + y)2 (x + y)2
−x2 y 2 + 300y 2 − 2xy 3
= =0
(x + y)2
∂V (300x − 2x2 y) (x + y) − (300xy − x2 y 2 ) 300xy − 2x2 y 2 + 300x2 − 2x3 y − 300xy + x2 y 2
= =
∂y (x + y)2 (x + y)2
−x2 y 2 + 300x2 − 2x3 y
= =0
(x + y)2
A fraction can be zero only if the numerator is zero, so from the first equation we get
−x2 y 2 + 300y 2 − 2xy 3 = 0. Of course y = 0 will not maximize the volume, so we divide by y to get
300 − x2
−x2 + 300 − 2xy = 0 and hence y = .
2x
We also have −x2 y 2 + 300x2 − 2x3 y = 0 from which we get −y 2 + 300 − 2xy = 0. Substituting the
expression for y, we get
!2
300 − x2 300 − x2 90000 − 600x2 + x4
− + 300 − 2x =− 2
+ x2 = 0
2x 2x 4x
Multiplying by 4x2 , we get
0 = −90000 + 600x2 − x4 + 4x4 = 3x4 + 600x2 − 90000

0 = x4 + 200x2 − 30000 = x
|
4
+ 200x{z2 + 10000} −10000 − 30000
complete square
2
4
=x
|
+ 200x{z2 + 10000} −40000 = x2 − 100 − 40000
complete square
2
40000 = x2 − 100
200 = x2 − 100
x2 = 100
So we have x2 = 100, and since side lengths are positive, we get x = 10. By back substitution, we
get y = z = 10. Since we only got one critical point, this must be a maximum.
Question 7. A rectangular box without a lid is to have a volume of 4 liters. What dimensions x, y,
z should we choose so as to minimize the area of the sides?
287
4000
Answer: 4 liters are the same as 4000 cm3 . The volume is V = 4000 = xyz. So z = , and the
xy
area, which we want to minimize, is given by
8000 8000
A = xy + 2yz + 2xz = xy + +
x y
We have A as a function of x and y, to check for local extrema we calculate the first order partial
!
∂A ∂ 8000 8000 8000 8000
= xy + + =y− 2
+0=y− 2
∂x ∂x x y x x
!
∂A ∂ 8000 8000 8000 8000
= xy + + =x+0− 2
=x− 2
∂y ∂y x y y y
8000 8000
From the first equation, we get y = . From the second equation, we get x = . Substituting
x2 y2
the expression for y into the expression for x, we get
8000 8000 8000 x4

x= = 2 = 80002
=
y2 8000

8000
x2 x4
So x3 = 8000. This means x = 20, and through back substitution, we get y = 20 and z = 10. So
there is one critical point, this must be a minimum.
Question 8. A rectangular box is to have a volume of 45 liters. The cost of material is as follows:
1.5 zl per m2 for the sides, 2 zl per m2 for the lid, and 3 zl per m2 for the base. Determine the
minimum cost of such a box.
0.045
Answer: 45 liters are the same as 0.045 m3 . The volume is V = 0.045 = xyz. So z = , and the
xy
cost, which we want to minimize, is given by
0.135 0.135
C = (2 + 3)xy + (1.5 + 1.5)yz + (1.5 + 1.5)xz = 5xy + 3yz + 3xz = 5xy + +
x y
We have C as a function of x and y, to check for local extrema we calculate the first order partial
!
∂C ∂ 0.135 0.135 0.135 0.135
= 5xy + + = 5y − 2
+ 0 = 5y −
∂x ∂x x y x x2
!
∂C ∂ 0.135 0.135 0.135 0.135
= 5xy + + = 5x + 0 − 2
= 5x − 2
∂y ∂y x y y y
0.135 0.027 0.027

From the first equation, we get y = 2
= 2
. From the second equation, we get x = .
5x x y2
288
Substituting the expression for y into the expression for x, we get
0.027 0.027 0.027 x4
x= = 2 = 0.0272
=
y2 0.027

0.027
x4
x2
So x3 = 0.027. This means x = 0.3, and through back substitution, we get y = 0.3 and z = 0.5. So
there is one critical point, this must be a minimum. The minimum cost is
C(0.3, 0.3, 0.5) = 5 · 0.3 · 0.3 + 3 · 0.3 · 0.5 + 3 · 0.3 · 0.5 = 1.35
We kept the same units throughout (m and zl), so the minimum cost of the material for one box is
1.35 zl.
Temat 23: Extrema of functions of several variables II.

Question 1. Let
2
 xy (x − y2)

if (x, y) 6= (0, 0)


f (x, y) = x2 + y 2



0 if (x, y) = (0, 0)
Show that:
a) The point (0, 0) is a critical point of f .
∂ 2f ∂ 2f ∂ 2f ∂ 2f
b) The second order partial derivatives , , and all exist at (0, 0), however
∂x∂x ∂x∂y ∂y∂x ∂y∂y
∂ 2f ∂ 2f
6= .
∂x∂y ∂y∂x
c) The point (0, 0) is not a local extremum of the function f .
∂f ∂f
Answer: a) We have to show that (0, 0) = (0, 0) = 0. We have
∂x ∂y
(0+h)·0·((0+h)2 −02 )
∂f f (0 + h, 0) − f (0, 0) (0+h)2 +02
−0
(0, 0) = lim = lim
∂x h→0 h h→0 h
h·0·h2 0
h2 h2 0
h→0 h h→0 h h→0 h3 h→0
0·(0+k)·(02 −(0+k)2 )
∂f f (0, 0 + k) − f (0, 0) 02 +(0+k)2
−0
(0, 0) = lim = lim
∂y k→0 k k→0 k
0·k·(−k2 ) 0
k2 k2 0
k→0 k k→0 k k→0 k 3 k→0
So the point (0, 0) is a critical point of f .

b) We will calculate all second order partial derivatives using the definition. In order to do that, we
need to know the first order partial derivatives for (x, y) 6= (0, 0).
∂ ∂
∂f ∂ xy (x2 − y 2 ) ∂ x3 y − xy 3 (x2 + y 2 ) ∂x (x3 y − xy 3 ) − (x3 y − xy 3 ) ∂x (x2 + y 2 )
= = =
∂x ∂x x2 + y 2 ∂x x2 + y 2 (x2 + y 2 )2
289
(x2 + y 2 ) (3x2 y − y 3 ) − (x3 y − xy 3 ) (2x + 0) 3x4 y − x2 y 3 + 3x2 y 3 − y 5 − 2x4 y + 2x2 y 3
= =
(x2 + y 2 )2 (x2 + y 2 )2
x4 y + 4x2 y 3 − y 5
=
(x2 + y 2 )2
∂ ∂
∂f ∂ xy (x2 − y 2 ) ∂ x3 y − xy 3 (x2 + y 2 ) ∂y (x3 y − xy 3 ) − (x3 y − xy 3 ) ∂y (x2 + y 2 )
= = =
∂y ∂y x2 + y 2 ∂y x2 + y 2 (x2 + y 2 )2
(x2 + y 2 ) (x3 − 3xy 2 ) − (x3 y − xy 3 ) (0 + 2y) x5 − 3x3 y 2 + x3 y 2 − 3xy 4 − 2x3 y 2 + 2xy 4
= =
(x2 + y 2 )2 (x2 + y 2 )2
x5 − 4x3 y 2 − xy 4
=
(x2 + y 2 )2
Using the definition of partial derivatives, we get
(0+h)4 ·0+4(0+h)2 ·03 −05
∂ 2f ∂ ∂f ∂f
(0 + h, 0) − ∂f
(0, 0) ((0+h)2 +02 )2
−0
∂x ∂x
(0, 0) = (0, 0) = lim = lim
∂x∂x ∂x ∂x h→0 h h→0 h
h4 ·0+4h2 ·0 0
h4 h4 0
h→0 h h→0 h h→0 h5 h→0
∂f ∂f (0+h)5 −4(0+h)3 ·02 −(0+h)·04

∂ 2f ∂ ∂f ∂y
(0 + h, 0) − ∂y
(0, 0) ((0+h)2 +02 )2
−0
(0, 0) = (0, 0) = lim = lim
∂x∂y ∂x ∂y h→0 h h→0 h
h5 −4h3 ·0−h·0 h5
h4 h4 h5
h→0 h h→0 h h→0 h5 h→0
04 ·(0+k)+4·02 ·(0+k)3 −(0+k)5

∂ 2f ∂ ∂f ∂f
(0, 0 + k) − ∂f
(0, 0) (02 +(0+k)2 )2
−0
∂x ∂x
(0, 0) = (0, 0) = lim = lim
∂y∂x ∂y ∂x k→0 k k→0 k
0·k+4·0·k3 −k5 −k5
k4 k4 −k 5
= lim = lim = lim = lim (−1) = −1
k→0 k k→0 k k→0 k 5 k→0
∂f ∂f 05 −4·03 ·(0+k)2 −0·(0+k)4

∂ 2f ∂ ∂f ∂y
(0, 0 + k) − ∂y
(0, 0) (02 +(0+k)2 )2
−0
(0, 0) = (0, 0) = lim = lim
∂y∂y ∂y ∂y k→0 k k→0 k
−4·0·k2 −0·k4 0
k4 k4 0
k→0 k k→0 k k→0 k 5 k→0
∂ 2f ∂ 2f
We have (0, 0) = 1 6= −1 = (0, 0).
∂x∂y ∂y∂x
c) When moving away from the point (0, 0) along the line x = 2y, we get
2y · y ((2y)2 − y 2 ) 2y 2 (4y 2 − y 2 ) 2y 2 (3y 2 ) 6y 4 6
f (2y, y) = 2 2
= 2 2
= 2
= 2
= y2
(2y) + y 4y + y 5y 5y 5
and the value of the function is increasing. So (0, 0) cannot be a local maximum. When moving away
from the point (0, 0) along the line x = −2y, we get
−2y · y ((−2y)2 − y 2 ) −2y 2 (4y 2 − y 2 ) −2y 2 (3y 2 ) −6y 4 6
f (−2y, y) = 2 2
= 2 2
= 2
= 2
= − y2
(−2y) + y 4y + y 5y 5y 5
290
and the value of the function is decreasing. So (0, 0) cannot be a local minimum. Since (0, 0) is not
a local maximum and it is not a local minimum, it cannot be a local extremum.

Question 2. Let f (x, y) = y − x2 y − 3x2 .
a) Show that grad f (0, 0) = (0, 0).
b) Show that for all (α, β) ∈ R2 \ (0, 0), the function h(t) = f (αt, βt) has a local minimum at t = 0.
c) Show that the point (0, 0) is not a local extremum of the function f .
Answer: a) We have
! ! !
∂f ∂f ∂ ∂ ∂ ∂
grad f = ∇f = , = f, f = y − x2 y − 3x2 , y − x2 y − 3x2
∂x ∂y ∂x ∂y ∂x ∂y
!
∂ 2 2 2 4
∂
2 2 2 4

= y − x y − 3x y + 3x , y − x y − 3x y + 3x
∂x ∂y
!
∂ 2 ∂
= y − 4x2 y + 3x4 , y 2 − 4x2 y + 3x4
∂x ∂y

= 0 − 8xy + 12x3 , 2y − 4x2 + 0 = −8xy + 12x3 , 2y − 4x2
Evaluated at (0, 0), we get grad f (0, 0) = (0, 0).

b) We have

h(t) = f (αt, βt) = βt − (αt)2 βt − 3(αt)2 = (βt)2 − 4(αt)2 βt + 3(αt)4 = β 2 t2 − 4α2 βt3 + 3α4 t4
Taking the derivative with respect to t, we get

d 22
h0 (t) = β t − 4α2 βt3 + 3α4 t4 = 2β 2 t − 12α2 βt2 + 12α4 t3
dt
So h0 (0) = 0, which means that the necessary condition for having a local minimum at t = 0 is
fulfilled. Now we calculate the second derivative.
d 0 d 2
h00 (t) = h (t) = 2β t − 12α2 βt2 + 12α4 t3 = 2β 2 − 24α2 βt + 36α4 t2
dt dt
If β 6= 0, then h00 (0) = 2β 2 . This is a positive number, which means that at t = 0, the slope of
h(t) increases and changes from negative to positive. So there must be a local minimum at t = 0.
If however β = 0, we have α 6= 0. In that case h(t) = 3α4 t4 , which has clearly a local minimum at
t = 0.
c) If we put y = 2x2 , we get

f (x, 2x2 ) = 2x2 − x2 2x2 − 3x2 = x2 −x2 = −x4
So moving away from (0, 0) along the curve y = 2x2 , the value of f decreases and (0, 0) cannot be a
local minimum.
If we put y = −x2 , we get

f (x, −x2 ) = −x2 − x2 −x2 − 3x2 = −2x2 −4x2 = 8x4
291
So moving away from (0, 0) along the curve y = −x2 , the value of f increases and (0, 0) cannot be a
local maximum.
Since (0, 0) is not a local minimum and it is not a local maximum, it cannot be a local extremum.
Question 3. Let A = {(x, y, z) ∈ R3 : 2x − 3y + z = 1}. Find the point (x, y, z) ∈ A closest to the
point (3, −2, 1).
Answer: We have z = 3y − 2x + 1. The distance is minimal if the the square of the distance is
minimal. So we need to find the minimum of the function
S = (x − 3)2 + (y − (−2))2 + (z − 1)2 = (x − 3)2 + (y − (−2))2 + (3y − 2x + 1 − 1)2
= (x − 3)2 + (y + 2)2 + (3y − 2x)2 = x2 − 6x + 9 + y 2 + 4y + 4 + 9y 2 − 12xy + 4x2
= 5x2 − 6x − 12xy + 10y 2 + 4y + 13
We need to calculate the first order partial derivatives and put them equal to zero.
∂S ∂ ∂ 2
= S= 5x − 6x − 12xy + 10y 2 + 4y + 13 = 10x − 6 − 12y = 0
∂x ∂x ∂x
∂S ∂ ∂ 2
= S= 5x − 6x − 12xy + 10y 2 + 4y + 13 = −12x + 20y + 4 = 0
∂y ∂y ∂y
So 60x − 36 − 72y = 0 and −60x + 100y + 20 = 0. Adding these two equations, we get 28y − 16 = 0,
400
4 + 20 9 12 18 1
so y = , and through back substitution, we get x = 7 = and z = − +1 = .
7 60 7 7 7 7
There is only one critical point, this must be a minimum. So the point (x, y, z) ∈ A closest to the
9 4 1

point (3, −2, 1) is , , .
7 7 7
Question 4. Find the maximum possible volume of a circular cylinder whose height plus circumference
of the base does not exceed 108 cm.
Answer: Let the height be h and the radius be r. The circumference is 2rπ, so we have 108 = h+2rπ.
The volume is given by
V = πr2 · h = πr2 (108 − 2rπ) = 108πr2 − 2π 2 r3
We need to put the derivative equal to zero, note that this is a function of only one variable.
dV d d
= V = 108πr2 − 2π 2 r3 = 216πr − 6π 2 r2 = 6πr(36 − πr) = 0
dr dr dr
36 36
We get the critical points r = 0 and r = . The maximum must occur at r = . We have
π π
2 3
36 36 36 108 · 362 − 2 · 363 3 · 363 − 2 · 363 363

V = 108π − 2π 2 = = =
π π π π π π
363
So the maximum possible volume is cm3 .
π
Question 5. Find the maximum volume of a parallelepiped for which the sum of all three sides
(length, width, height) does not exceed 108 cm.
292
Answer: It is clear that this parallelepiped is right-angled. We have 108 = x + y + z, and the volume,
which we want to maximize, is given by
V = xyz = xy(108 − x − y) = 108xy − x2 y − xy 2
We have V as a function of x and y, to check for local extrema we calculate the first order partial
∂V ∂ ∂
= V = 108xy − x2 y − xy 2 = 108y − 2xy − y 2 = y(108 − 2x − y) = 0
∂x ∂x ∂x
∂V ∂ ∂
= V = 108xy − x2 y − xy 2 = 108x − x2 − 2xy = x(108 − x − 2y) = 0
∂y ∂y ∂y
14 PARTIAL DERIVATIVES
From the first equation, we get 108 − 2x − y = 0 or y = 0, of course we cannot have y = 0 if
we EXAMPLE Find the extreme
want to2 maximize the values
volume.of f �x, y� � ythe
From 2 2
.
� xsecond equation, we get 108 − x − 2y = 0 or x = 0,
of course we cannot
SOLUTION Since x and fyx�=2y0, the
fx � �2have if only
we want
criticalto
pointmaximize the volume.
is �0, 0�. Notice that So we have 108 − 2x − y = 0
andfor108 −onx the
points − x2y-axis
=we0 ,have
from 0, so f �x,
y �which wey� get
� �x1082
� 0−(if2(108
x � 0).−However,
2y) − fory = 0. So y = 36 and through
points
back on the y-axis we
substitution get xy�=0,zso=f �x,
we have � y 2 �we
36.y� Since 0 (ifonly ). Thus
y � 0got oneevery disk this must be a maximum. The
critical,
with center �0, 0� contains points
3 where f takes positive values as well as points where
maximum volume
f takes negative is 36
values. cm3 =
Therefore f �0,46656
0� � 0cm
3
can’t. be an extreme value for f , so f has
no extreme value. M
Question 6. 2Find
Example andthe
illustrates classify
fact thatthe local extrema
a function of the
need not have following
a maximum functions:
or minimum
value at a critical point. Figure
3 3 3shows how this is possible. The graph of f is the
2 hyper-
a) f (x,z y)
bolic paraboloid � y=2 x 2+ y + 3xy + 3 f (x,(zy)
� x , which has a horizontal tangentb)plane � 0=) at
3xthe + 6xy
origin. + 2y 3 + 12x − 24y
You can see that f �0, 0� � 0 4is a4maximum in the direction of the x-axis but a minimum
y c) f (x,ofy)
in the direction e−x Near
the=y-axis.
−y
the fshape
the origin the graph has d) (x, y)
of = x3 −and
a saddle 3xyso2
�0, 0� is called a saddle point of f .
We need to be able to determine whether or not a function has an extreme value at a
critical point. The following test, which is proved at the end of this section, is analogous
Answer:
to the Second Derivative Test for functions of one variable.
3 SECOND DERIVATIVES TEST Suppose the second partial derivatives of f are

continuous on a disk with center �a, b�, and suppose that fx �a, b� � 0 and
fy �a, b� � 0 [that is, �a, b� is a critical point of f ]. Let
D � D�a, b� � fxx �a, b� fyy �a, b� � � fx y �a, b�� 2
(a) If D � 0 and fxx �a, b� � 0, then f �a, b� is a local minimum.

(b) If D � 0 and fxx �a, b� � 0, then f �a, b� is a local maximum.
(c) If D � 0, then f �a, b� is not a local maximum or minimum.
NOTE 1 In case (c) the point �a, b� is called a saddle point of f and the graph of f
crosses its tangent plane at �a, b�.
NOTE 2 If D � 0, the test gives no information: f could have a local maximum or local
minimum at �a, b�, or �a, b� could be a saddle point of f .
NOTE 3 To remember the formula for D, it’s helpful to write it as a determinant:
D� � fxx
fyx
fx y
fyy �
� fxx fyy � � fx y �2
V EXAMPLE 3 Find the local maximum and minimum values and saddle points of
3 3
a) f �x,
We x 4 � fy(x,
have
y� � 4
� y) �x
4xy= 1. + y + 3xy + 3. We need to find the critical points, so we calculate the
first order partial derivatives and put them equal to zero.
SOLUTION We ﬁrst locate the critical points:
∂f ∂ fx � 4x
∂ 3 � 4y 3
� 4y+ �3 4x= 3x2 + 0 + 3y + 0 = 3 x2 + y = 0
x3 + y 3 +fy 3xy

= f=
∂x ∂x ∂x
Setting these partial derivatives equal to 0, we obtain the equations
x3 � y � 0 and 293
y3 � x � 0
To solve these equations we substitute y � x 3 from the ﬁrst equation into the second
one. This gives
0 � x 9 � x � x�x 8 � 1� � x�x 4 � 1��x 4 � 1� � x�x 2 � 1��x 2 � 1��x 4 � 1�
∂f ∂ ∂ 3
= f= x + y 3 + 3xy + 3 = 0 + 3y 2 + 3x + 0 = 3 y 2 + x = 0
∂y ∂y ∂y
From the first equation, we get y = −x2 , and putting this in the second equation, we get
2
−x2 + x = x4 + x = x(x3 + 1) = 0. This gives x = 0 or x = −1, and through back substitution
we calculate the corresponding y-values. The critical points so obtained are (0, 0) and (−1, −1).
When moving away from the point (0, 0) along the line y = 0, we have
f (x, 0) = x3 + 03 + 3x · 0 + 3 = x3 + 3
When x increases, the value of f (x, 0) increases, so (0, 0) cannot be a local maximum. When x
decreases, the value of f (x, 0) decreases, so (0, 0) cannot be a local minimum. Since (0, 0) is not a
local maximum and it is not a local minimum, it cannot be a local extremum.
Now we look at the critical point (−1, −1). We calculate the second order partial derivatives, to
understand why, see the explanation at the beginning of the answer to Question 6 on page 293.
∂ 2f ∂ ∂f ∂ 2 ∂ 2
fxx = = = 3 x +y = 3x + 3y = 6x + 0 = 6x
∂x∂x ∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂ 2 ∂ 2
fyx = = = 3 y +x = 3y + 3x = 0 + 3 = 3
∂x∂y ∂x ∂y ∂x ∂x
∂ 2f ∂ ∂f ∂ 2 ∂ 2
fxy = = = 3 x +y = 3x + 3y = 0 + 3 = 3
∂y∂x ∂y ∂x ∂y ∂y
∂ 2f ∂ ∂f ∂ 2 ∂ 2
fyy = = = 3 y +x = 3y + 3x = 6y + 0 = 6y
∂y∂y ∂y ∂y ∂y ∂y
We have fxx (−1, −1) = −6, fyx (−1, −1) = fxy (−1, −1) = 3, and fyy (−1, −1) = −6. So
D(−1, −1) = fxx (−1, −1)fyy (−1, −1) − [fxy (−1, −1)]2 = (−6) · (−6) − 9 = 27
Since D(−1, −1) = 27 > 0 and fxx (−1, −1) = −6 < 0, there is a local maximum at (−1, −1) [see
the explanation at the beginning of the answer to Question 6 on page 293]. The value of this local
maximum is 4.
We already know that (0, 0) is not a local extremum. But we can also use the second order partial
derivatives to verify this: we have fxx (0, 0) = 0, fyx (0, 0) = fxy (0, 0) = 3, and fyy (0, 0) = 0. So
D(0, 0) = fxx (0, 0)fyy (0, 0) − [fxy (0, 0)]2 = 0 − 9 = −9
Since D(0, 0) = −9 < 0, there is a saddle point at (0, 0) [see the explanation at the beginning of the
answer to Question 6 on page 293]. The coordinates of this saddle point are (0, 0, 3).
b) We have f (x, y) = 3x2 + 6xy + 2y 3 + 12x − 24y. We need to find the critical points, so we calculate
the first order partial derivatives and put them equal to zero.
∂f ∂ ∂ 2
= f= 3x + 6xy + 2y 3 + 12x − 24y = 6x + 6y + 0 + 12 + 0 = 6x + 6y + 12 = 0
∂x ∂x ∂x
∂f ∂ ∂ 2
= f= 3x + 6xy + 2y 3 + 12x − 24y = 0 + 6x + 6y 2 + 0 − 24 = 6x + 6y 2 − 24 = 0
∂y ∂y ∂y
294
From the first equation, we get 6x = −12 − 6y, and putting this in the second equation, we get
" 2 #
1 25 1 25

2 2 2 2
−12−6y+6y −24 = 6y −6y−36 = 6 y − y − 6 = 6 y − y + − =6 y− − =0
4 4 2 4
This gives y = 3 or y = −2, and through back substitution we calculate the corresponding x-values.
The critical points so obtained are (−5, 3) and (0, −2).
We calculate the second order partial derivatives, to understand why, see the explanation at the
beginning of the answer to Question 6 on page 293.
∂ 2f ∂ ∂f ∂
fxx = = = (6x + 6y + 12) = 6 + 0 + 0 = 6
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂
fyx = = = 6x + 6y 2 − 24 = 6 + 0 − 0 = 6
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ ∂f ∂
fxy = = = (6x + 6y + 12) = 0 + 6 + 0 = 6
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂
fyy = = = 6x + 6y 2 − 24 = 0 + 12y − 0 = 12y
∂y∂y ∂y ∂y ∂y
We have fxx (−5, 3) = 6, fyx (−5, 3) = fxy (−5, 3) = 6, and fyy (−5, 3) = 36. So
D(−5, 3) = fxx (−5, 3)fyy (−5, 3) − [fxy (−5, 3)]2 = 6 · 36 − 36 = 180
Since D(−5, 3) = 180 > 0 and fxx (−5, 3) = 6 > 0, there is a local minimum at (−5, 3) [see the
explanation at the beginning of the answer to Question 6 on page 293]. The value of this local
minimum is −93.
We have fxx (0, −2) = 6, fyx (0, −2) = fxy (0, −2) = 6, and fyy (0, −2) = −24. So
D(−5, 3) = fxx (0, −2)fyy (0, −2) − [fxy (0, −2)]2 = 6 · (−24) − 36 = −180
Since D(0, −2) = −180 < 0, there is a saddle point at (0, −2) [see the explanation at the beginning
of the answer to Question 6 on page 293]. The coordinates of this saddle point are (0, −2, 32).
4 4
c) We have f (x, y) = e−x −y . The value of the function is 1 at (0, 0). When moving radially
outwards, the exponent decreases and thus the value of the function decreases. So there is only one
local extremum, a maximum at (0, 0).
d) We have f (x, y) = x3 − 3xy 2 . We need to find the critical points, so we calculate the first order
partial derivatives and put them equal to zero.
∂f ∂ ∂ 3
= f= x − 3xy 2 = 3x2 − 3y 2 = 0
∂x ∂x ∂x
∂f ∂ ∂ 3
= f= x − 3xy 2 = 0 − 6xy = −6xy = 0
∂y ∂y ∂y
From the second equation, we get that one of x and y must be zero, and from the first equation, we
get that the other one must be zero as well. So there is one critical point, (0, 0). Moving away from
(0, 0) along the line y = 0, we have
f (x, 0) = x3 − 3x · 02 = x3
295
When x increases, the value of f (x, 0) increases, so (0, 0) cannot be a local maximum. When x
decreases, the value of f (x, 0) decreases, so (0, 0) cannot be a local minimum. Since (0, 0) is not a
local maximum and it is not a local minimum, it cannot be a local extremum.
Question 7. Find and classify the local extrema of the following functions:
a) f (x, y) = x3 y − 3xy 2
b) f (x, y) = (x − y)(xy − 1)
Answer: a) We have f (x, y) = x3 y − 3xy 2 . We need to find the critical points, so we calculate the
∂f ∂ ∂ 3
= f= x y − 3xy 2 = 3x2 y − 3y 2 = 3y x2 − y = 0
∂x ∂x ∂x
∂f ∂ ∂ 3
= f= x y − 3xy 2 = x3 − 6xy = x x2 − 6y = 0
∂y ∂y ∂y
There is an obvious solution y = 0 and x = 0. From the first equation, we also get the possibility
x2 = y and putting this in x2 − 6y = 0, we get that −5x2 = 0, wich means that x = y = 0. So the
only critical point is (0, 0). Moving away from (0, 0) along the curve y = x2 , we have
2
f x, x2 = x3 x2 − 3x x2 = x5 − 3x5 = −2x5

When x increases, the value of f x, x2 decreases, so (0, 0) cannot be a local minimum. When x

decreases, the value of f x, x2 increases, so (0, 0) cannot be a local maximum. Since (0, 0) is not a
local minimum and it is not a local maximum, it cannot be a local extremum.
b) We have f (x, y) = (x − y)(xy − 1) = x2 y − xy 2 − x + y. We need to find the critical points, so
we calculate the first order partial derivatives and put them equal to zero.
∂f ∂ ∂ 2
= f= x y − xy 2 − x + y = 2xy − y 2 − 1 + 0 = 2xy − y 2 − 1 = 0
∂x ∂x ∂x
∂f ∂ ∂ 2
= f= x y − xy 2 − x + y = x2 − 2xy − 0 + 1 = x2 − 2xy + 1 = 0
∂y ∂y ∂y
Adding the first and the second equation, we get x2 − y 2 = 0. So x = y or x = −1. If x = y, then
from the first equation we get 2x2 − x2 − 1 = 0, so x2 = 1 which gives the critical points (−1, −1)
and (1, 1). If x = −y, then from the first equation we get −2x2 − x2 − 1 = 0, which has no solution.
We have two critical points, (−1, −1) and (1, 1).
Moving away from (−1, −1) along the line x = −1, we have
f (−1, y) = y + y 2 + 1 + y = y 2 + 2y + 1 = (y + 1)2
This is a parabola with vertex y = −1, shaped upwards, so if there was an extremum at (−1, −1), it
would have to be a minimum.
Moving away from (−1, −1) along the line y = −1, we have
f (x, −1) = −x2 − x − x − 1 = −x2 − 2x − 1 = −(x + 1)2
296
This is a parabola with vertex x = −1, shaped downwards, so if there was an extremum at (−1, −1),
it would have to be a maximum.
We conclude that at (−1, −1) the function has a saddle point. The coordinates of this saddle point
are (−1, −1, 0).
Similarly, one can show that at (1, 1) the function has a saddle point. The coordinates of this saddle
point are (1, 1, 0).
Question 8. Show that the function f (x, y) = x2 (1 + y)3 + y 2 has exactly one critical point p. Show
that p is a local minimum, sup f (x, y) = +∞, and inf 2 f (x, y) = −∞
(x,y)∈R2 (x,y)∈R
Answer: We have f (x, y) = x2 (1 + y)3 + y 2 . We need to find the critical points, so we calculate the
∂f ∂ ∂ 2
= f= x (1 + y)3 + y 2 = 2x(1 + y)3 + 0 = 2x(1 + y)3 = 0
∂x ∂x ∂x
∂f ∂ ∂ 2
= f= x (1 + y)3 + y 2 = 3x2 (1 + y)2 + 2y = 0
∂y ∂y ∂y
From the first equation, we get x = 0 or y = −1. If x = 0, from the second equation we get that
y = 0. If y = −1, from the second equation we get that y = 0, a contradiction. So there is one critical
point, (0, 0).
beginning of the answer to Question 6 on page 293.
∂ 2f ∂ ∂f ∂
fxx = = = 2x(1 + y)3 = 2(1 + y)3
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂ 2
fyx = = = 3x (1 + y)2 + 2y = 6x(1 + y)2 + 0 = 6x(1 + y)2
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ ∂f ∂
fxy = = = 2x(1 + y)3 = 6x(1 + y)2
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂ 2
fyy = = = 3x (1 + y)2 + 2y = 6x2 (1 + y) + 2
∂y∂y ∂y ∂y ∂y
We have fxx (0, 0) = 2, fyx (0, 0) = fxy (0, 0) = 0, and fyy (0, 0) = 2. So
D(0, 0) = fxx (0, 0)fyy (0, 0) − [fxy (0, 0)]2 = 2 · 2 − 0 = 4
Since D(0, 0) = 4 > 0 and fxx (0, 0) = 2 > 0, there is a local minimum at (0, 0) [see the explanation
at the beginning of the answer to Question 6 on page 293]. The value of this local minimum is 0.
To see that sup f (x, y) = +∞, it suffices to let x and y tend to +∞.
(x,y)∈R2
To see that inf f (x, y) = −∞, it suffices to fix x = 1 and let y tend to −∞.
(x,y)∈R2
297
Temat 24: Additional exercises on the differential calculus
of functions of several variables.
Question 1. Let P , V and T denote pressure, volume, and temperature of a given gas. They satisfy
the equation P V = RT , where R is a constant. Prove that
∂P ∂V ∂T
· · = −1
∂V ∂T ∂P
Answer:
T T 1
P =R V =R T = PV
V P R
∂P RT ∂V R ∂T V
=− 2 = =
∂V V ∂T P ∂P R
∂P ∂V ∂T RT R V RT
· · =− 2 · · =− = −1
∂V ∂T ∂P V P R PV
Question 2. Find the maximum and minimum of the following function:
f (x, y) = sin x + sin y + sin(x + y)

π π

2
where D = (x, y) ∈ R : 0 ¬ x ¬ , 0 ¬ y ¬ .
2 2
Answer: A necessary condition for having a local extremum is that the first order partial derivatives
are both equal to zero.
∂f
= cos x + 0 + cos(x + y) = cos x + cos(x + y) = 0
∂x
∂f
= 0 + cos y + cos(x + y) = cos y + cos(x + y) = 0
∂y
So we have cos x = cos y , which on D means x = y. Then cos x = − cos 2x, which on D means
2π π π
x= = . So x = y = is a critical point, and
6 3 3
√ √ √ √
π π π π 2π 3 3 3 3 3

f , = sin + sin + sin = + + =
3 3 3 3 3 2 2 2 2
This must be a maximum, since on D the value of the function is always non-negative and f (0, 0) = 0,
which is a minimum.
Question 3. Among triangles inscribed inside a circle of radius R find the one with maximal area.
Answer:
298
qp = h2
aq
Assume we have found a which maximizes the area. Then b = c will maximize the area A = .
2
2
a
q(2R − q) =
2
q
aq 2q q(2R − q) q
A= = = 2q 3 R − q 4
2 2
dA 6q 2 R − 4q 3 q 2 (6R − 4q)
= √ 3 = √ =0
dq 2 2q R − q 4 2 2q 3 R − q 4
A fraction can be zero only if the numerator is zero. Clearly, q = 0 will not maximize the area. So
3R
6R − 4q = 0 and hence q = . This gives
2
s s √
3R 3R 3R 3R 3R R 3 3 R2

A= 2R − = · =
2 2 2 2 2 2 4
We have s s
3R 3R 3R R √
q
a = 2 q(2R − q) = 2 2R − =2 · = 3R
2 2 2 2
and
s v
√ !2 2 s s
12R2 √ 2 √
2 u
a u 3R 3R 3R 2 9R 2
b=c= + q2 = t
+ = + = = 3R = 3R
2 2 2 4 4 4
√
Hence the triangle which maximizes the area is an equilateral triangle of side length 3R.
Question 4. Among rectangular boxes inscribed inside a sphere of radius R find the one with
maximal volume.
Answer:
299
Let the hight of the box be z, the breadth be x, and the depth be y. Assume we found z which
maximizes the volume. Then V = xyz and to maximize the volume we need to have x = y, because
√ √
(2r)2 = x2 +y 2 and maximizing the area A = xy = x 4r2 − x2 = 4x2 r2 − x4 leads to the equation
dA 8xr2 − 4x3 2x (2r2 − x2 )

= √ 2 2 = √ =0
dx 2 4x r − x4 x2 r 2 − x4
√ √ √
which implies x = 2r. Consequently y = 4r2 − x2 = 2r = x. Now we want to maximize
V = xyz = x2 z , knowing that (2R)2 = (2r)2 + z 2 = x2 + y 2 + z 2 = 2x2 + z 2 . So
4R2 − z 2 z3
V = xyz = x2 z = z = 2zR2 −
2 2
dV 3z 2
= 2R2 − =0
dz 2
s
2 4R2 − z 2 2
So z = √ R and this gives x = y = = √ R and
3 2 3
2 2 2 8
V = xyz = √ R · √ R · √ R = √ R3
3 3 3 3 3
2
Since x = y = z = √ R, the rectangular box which maximizes the volume is a cube of side length
3
2
√ R.
3
Question 5. Give an example of a continuous function of two variables, which has two local maxima,
but no other local extrema.
2 2 2
Answer: f (x, y) = e−(x−1) (x+1) e−y . The expression −(x − 1)2 (x + 1)2 in the exponent is always less
than or equal to zero, it is zero only if x = 1 or x = −1. By symmetry, we know that −(x−1)2 (x+1)2
has a local minimum at x = 0. The exponent −y 2 has a local and global maximum at y = 0. So
(x, y) = (1, 0) and (x, y) = (−1, 0) are local maxima of f (x, y), and f (x, y) has no other local extrema.
3 2
Question 6. Show that the function f (x, y) = 2 1 − e2y + x2 − 3 1 − e2y + x2 − 24x2 e2y has
exactly one critical point at which the function has a strict local maximum, but the function is
neither bounded above or below.
300
Answer: We need to calculate the first order partial derivatives and put them equal to zero.
∂f ∂ 3 2
= 2 1 − e2y + x2 − 3 1 − e2y + x2 − 24x2 e2y
∂x ∂x
2
= 6 1 − e2y + x2 · 2x − 6 1 − e2y + x2 · 2x − 48xe2y

= 6 1 − e2y + x2 1 − e2y + x2 · 2x − 6 1 − e2y + x2 · 2x − 48xe2y

= 12x 1 − e2y + x2 1 − e2y + x2 − 12x 1 − e2y + x2 − 48xe2y
h i
= 12x 1 − e2y + x2 1 − e2y + x2 − 1 − 48xe2y = 12x 1 − e2y + x2 x2 − e2y − 48xe2y

= 12x 1 − e2y + x2 x2 − e2y − 48xe2y = 12x x2 − x2 e2y + x4 − e2y + e4y − x2 e2y − 48xe2y

= 12x x2 − 2x2 e2y + x4 − 5e2y + e4y = 0
∂f ∂ 3 2
= 2 1 − e2y + x2 − 3 1 − e2y + x2 − 24x2 e2y
∂y ∂y
2
= 6 1 − e2y + x2 −2e2y − 6 1 − e2y + x2 −2e2y − 48x2 e2y

= 6 1 − e2y + x2 1 − e2y + x2 −2e2y − 6 1 − e2y + x2 −2e2y − 48x2 e2y

= −12e2y 1 − e2y + x2 1 − e2y + x2 + 12e2y 1 − e2y + x2 − 48x2 e2y
h i
= 12e2y 1 − e2y + x2 1 − 1 − e2y + x2 − 48x2 e2y

= 12e2y 1 − e2y + x2 e2y − x2 − 48x2 e2y

= 12e2y e2y − e4y + x2 e2y − x2 + x2 e2y − x4 − 48x2 e2y

= 12e2y e2y − e4y + 2x2 e2y − 5x2 − x4 = 0
From the first equation, we see that x = 0 is one possibility. Putting x = 0 in the second equation,
we get e2y − e4y = 0 or e2y = 0 which implies y = 0 since the exponential function has only positive
values. So (0, 0) is a critical point. For another critical point, we would need to have the following
two equations satisfied
x2 − 2x2 e2y + x4 − 5e2y + e4y = 0

e2y − e4y + 2x2 e2y − 5x2 − x4 = 0
which is equivalent to
x2 − 5e2y = 2x2 e2y − x4 − e4y

−e4y + 2x2 e2y − x4 = −e2y + 5x2
From this we get x2 − 5e2y = −e2y + 5x2 . This means −4x2 = 4e2y which is a contradiction because
the exponential function has only positive values and a square is always non-negative.
301
beginning of the answer to Question 6 of Temat 23 on page 293.
∂ 2f ∂ ∂f ∂
fxx = = = 12x x2 − 2x2 e2y + x4 − 5e2y + e4y
∂x∂x ∂x ∂x ∂x
∂ 3
= 12 x − 2x3 e2y + x5 − 5xe2y + xe4y = 12 3x2 − 6x2 e2y + 5x4 − 5e2y + e4y
∂x
∂ 2f ∂ ∂f ∂ 2y 2y
fyx = = = 12e e − e4y + 2x2 e2y − 5x2 − x4
∂x∂y ∂x ∂y ∂x
∂ 4y
= 12 e − e6y + 2x2 e4y − 5x2 e2y − x4 e2y = 12 4xe4y − 10xe2y − 4x3 e2y
∂x
∂ 2f ∂ ∂f ∂
fxy = = = 12x x2 − 2x2 e2y + x4 − 5e2y + e4y
∂y∂x ∂y ∂x ∂y
∂ 3
= 12 x − 2x3 e2y + x5 − 5xe2y + xe4y = 12 −4x3 e2y − 10xe2y + 4xe4y
∂y
∂ 2f ∂ ∂f ∂ 2y 2y
fyy = = = 12e e − e4y + 2x2 e2y − 5x2 − x4
∂y∂y ∂y ∂y ∂y
∂ 4y
= 12 e − e6y + 2x2 e4y − 5x2 e2y − x4 e2y = 12 4e4y − 6e6y + 8x2 e4y − 10x2 e2y − 2x4 e2y
∂y
We have fxx (0, 0) = −48, fyx (0, 0) = fxy (0, 0) = 0, and fyy (0, 0) = −24. So
D(0, 0) = fxx (0, 0)fyy (0, 0) − [fxy (0, 0)]2 = (−48) · (−24) − 0 = 1152
Since D(0, 0) = 1152 > 0 and fxx (0, 0) = −48 < 0, there is a local maximum at (0, 0) [see the
explanation at the beginning of the answer to Question 6 of Temat 23 on page 293]. The value of
this local maximum is 0.
3 2
To see that the function f (x, y) = 2 1 − e2y + x2 − 3 1 − e2y + x2 − 24x2 e2y is not bounded
above, it suffices to put y = 0 and let x tend towards plus infinity. To see that the function is not
bounded below, it suffices to put x = 0 and let y tend towards plus infinity.
∂f
Question 7. Show that a function f (x, y) of class C 2 , such that (x, y) = 6xy 2 and
∂x
∂f
(x, y) = 8x2 y does not exist.
∂y
Answer: We need to calculate the following second order partial derivatives.
∂ 2f ∂ ∂f ∂
fyx = = = 8x2 y = 16xy
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ ∂f ∂
fxy = = = 6xy 2 = 12xy
∂y∂x ∂y ∂x ∂y
Since fyx = 16xy 6= 12xy = fxy , such a function is not of class C 2 .
Question 8. Determine if the following functions satisfy Laplace’s equation:

∂ 2f ∂ 2f
+ =0
∂x2 ∂y 2
302
q
a) f (x, y) = x2 + y 2
q
b) f (x, y) = ln x2 + y 2
y
c) f (x, y) = arc tg
x
d) f (x, y) = e−x sin y
q
Answer: a) We have f (x, y) = x2 + y 2
∂f ∂ q 2 2x x
= x + y2 = √ 2 2
=√ 2
∂x ∂x 2 x +y x + y2
√ 2
2 2 x + y 2 − x √ x2 2
∂ f ∂ f ∂ ∂f ∂ x x +y x2 + y 2 − x2 y2
= = = √ = = 3 = 3
∂x2 ∂x∂x ∂x ∂x ∂x x2 + y 2 x2 + y 2 (x2 + y 2 ) 2 (x2 + y 2 ) 2
∂f ∂ q 2 2y y
= x + y2 = √ 2 2
=√ 2
∂y ∂y 2 x +y x + y2
√ 2
x + y 2 − y √ 2y 2
∂ 2f ∂ 2f ∂ ∂f ∂ y x +y x2 + y 2 − y 2 x2
= = = √ = = 3 = 3
∂y 2 ∂y∂y ∂y ∂y ∂y x2 + y 2 x2 + y 2 (x2 + y 2 ) 2 (x2 + y 2 ) 2
We have
∂ 2f ∂ 2f y2 x2 x2 + y 2 1
+ = 3 + 3 = 3 = √ 6= 0
∂x2 ∂y 2 (x2 + y 2 ) 2 (x2 + y 2 ) 2 (x2 + y 2 ) 2 x2 + y 2
So f (x, y) doest not satisfy Laplace’s equation.
q
b) We have f (x, y) = ln x2 + y 2
∂f ∂ 1 ∂ q 2 1 x x
q
= ln x2 + y 2 = √ 2 x + y 2 = √ √ =
∂x ∂x x + y 2 ∂x x2 + y 2 x2 + y 2 x2 + y 2
∂ 2f ∂ 2f ∂ ∂f ∂ x x2 + y 2 − x · 2x x2 + y 2 − 2x2 y 2 − x2
= = = = = =
∂x2 ∂x∂x ∂x ∂x ∂x x2 + y 2 (x2 + y 2 )2 (x2 + y 2 )2 (x2 + y 2 )2
∂f ∂ 1 ∂ q 2 1 y y
q
= ln 2 2
x +y = √ x + y2 = √ 2 √ 2 = 2
∂y ∂y 2 2
x + y ∂y x +y 2 x +y 2 x + y2
∂ 2f ∂ 2f ∂ ∂f ∂ y x2 + y 2 − y · 2y x2 + y 2 − 2y 2 x2 − y 2
= = = = = =
∂y 2 ∂y∂y ∂y ∂y ∂y x2 + y 2 (x2 + y 2 )2 (x2 + y 2 )2 (x2 + y 2 )2
We have
∂ 2f ∂ 2f y 2 − x2 x2 − y 2 y 2 − x2 + (x2 − y 2 ) 0
+ = 2 + 2 = 2 = 2 =0
∂x 2 ∂y 2 2
(x + y )2 2 2
(x + y ) 2
(x + y )2 (x + y 2 )2
So f (x, y) satisfies Laplace’s equation.
y
c) We have f (x, y) = arc tg
x
∂f ∂ y 1 ∂ y − xy2
= arc tg = 2 = 2
∂x ∂x x 1 + xy ∂x x 1 + xy
303
2
2y y y 2y 2
∂ 2f ∂ 2f ∂ ∂f ∂ − xy2 x3
1+ x
− x2 x3
2y
+ 2y 3
− 2y 3 2y
x3 x5 x5 x3
= = = = = =
∂x2 ∂x∂x ∂x ∂x ∂x 1 + y 2 2 2 2 2 2 2

y y y
x 1+ x
1+ x
1+ x
1
∂f ∂ y 1 ∂ y x
= arc tg = 2 = 2
∂y ∂y x 1 + xy ∂y x 1 + xy
2 2
y 1 2y y 2y
∂ 2f ∂ 2f ∂ ∂f ∂ 1 − x12 1+ x
− x x2
− x12 1+ x
− x3
x
= = = = =
∂y 2 ∂y∂y ∂y ∂y ∂y 1 + y 2 2 2 2 2

y y
x 1+ x
1+ x
We have
2 2
y 2y y
∂ 2f ∂ 2f 2y − x12 1+ x
− x3
− x12 1+ x − x12 −1
x3
+ = 2 2 + 2 2 = 2 2 = 2 = 6= 0
∂x2 ∂y 2 x2 + y 2

y
1 + xy 1+ y
1+ y 1+ x
x x
So f (x, y) doest not satisfy Laplace’s equation.

d) We have f (x, y) = e−x sin y
∂f ∂ −x
= e sin y = −e−x sin y
∂x ∂x
∂ 2f ∂ 2f ∂ ∂f ∂ −x
= = = −e sin y = e−x sin y
∂x2 ∂x∂x ∂x ∂x ∂x
∂f ∂ −x
= e sin y = e−x cos y
∂y ∂y
∂ 2f ∂ 2f ∂ ∂f ∂ −x
2
= = = e cos y = −e−x sin y
∂y ∂y∂y ∂y ∂y ∂y
We have
∂ 2f ∂ 2f
+ = e−x sin y − e−x sin y = 0
∂x2 ∂y 2
So f (x, y) satisfies Laplace’s equation.
∂ 2f ∂ 2f
Question 9. Prove that if the function f (x, y) satisfies the condition + = 0, then the
∂x2 ∂y 2
∂ 2F ∂ 2F
function F (u, v) = f u2 − v 2 , 2uv satisfies the condition + = 0.
∂u2 ∂v 2
Answer: Putting x = u2 − v 2 and y = 2uv, we get
∂x ∂x
= 2u = −2v
∂u ∂v
∂y ∂y
= 2v = 2u
∂u ∂v
∂ 2F ∂ 2F
Now we calculate and using the chain rule.
∂u2 ∂v 2
∂F ∂f ∂x ∂f ∂y ∂f ∂f
= + = 2u + 2v
∂u ∂x ∂u ∂y ∂u ∂x ∂y
304
!
∂ 2F ∂ ∂f ∂f
= 2u + 2v
∂u2 ∂u ∂x ∂y
! !
∂ 2 f ∂x ∂ 2 f ∂y ∂f ∂ 2 f ∂x ∂ 2 f ∂y ∂f
= + 2u + 2+ + 2v + ·0
∂x∂x ∂u ∂y∂x ∂u ∂x ∂x∂y ∂u ∂y∂y ∂u ∂y
! !
∂ 2f ∂ 2f ∂f ∂ 2f ∂ 2f ∂f
= 2u + 2v 2u + 2 + 2u + 2v 2v + ·0
∂x2 ∂y∂x ∂x ∂x∂y ∂y 2 ∂y
∂ 2f ∂ 2f ∂f ∂ 2f 2
2 ∂ f
= 4u2 + 4uv + 2 + 4uv + 4v
∂x2 ∂y∂x ∂x ∂x∂y ∂y 2
∂F ∂f ∂x ∂f ∂y ∂f ∂f
= + = (−2v) + 2u
∂v ∂x ∂v ∂y ∂v ∂x ∂y
!
∂ 2F ∂ ∂f ∂f
2
= (−2v) + 2u
∂v ∂v ∂x ∂y
! !
∂ 2 f ∂x ∂ 2 f ∂y ∂f ∂ 2 f ∂x ∂ 2 f ∂y ∂f
= + (−2v) + · (−2) + + 2u + ·0
∂x∂x ∂v ∂y∂x ∂v ∂x ∂x∂y ∂v ∂y∂y ∂v ∂y
! !
∂ 2f ∂ 2f ∂f ∂ 2f ∂ 2f ∂f
= 2
(−2v) + 2u (−2v) + · (−2) + (−2v) + 2
2u 2u + ·0
∂x ∂y∂x ∂x ∂x∂y ∂y ∂y
∂ 2f ∂ 2f ∂f ∂ 2f 2
2 ∂ f
= 4v 2 − 4uv − 2 − 4uv + 4u
∂x2 ∂y∂x ∂x ∂x∂y ∂y 2
∂ 2f ∂ 2f
Since we have + = 0, we get
∂x2 ∂y 2
∂ 2F ∂ 2F 2
2 ∂ f ∂ 2f ∂f ∂ 2f 2
2 ∂ f
+ = 4u + 4uv + 2 + 4uv + 4v
∂u2 ∂v 2 ∂x2 ∂y∂x ∂x ∂x∂y ∂y 2
∂ 2f2 ∂ 2f ∂f ∂ 2f 2
2 ∂ f
+ 4v − 4uv −2 − 4uv + 4u
∂x2 ∂y∂x ∂x ∂x∂y ∂y 2
!

2 2
∂ 2f ∂ 2f
= 4u + 4v + =0
∂x2 ∂y 2
Question 10. Find and classify all the critical points of the following functions:
a) f (x, y) = e(xy) − 2xy
b) f (x, y) = x3 y 2 (1 − x − y)
2 −y 2
c) f (x, y) = 2x2 + y 2 e−x
d) f (x, y, z) = x2 + y 2 + z 2 − xy + x + 2z
Answer: a) We have f (x, y) = e(xy) − 2xy. We need to calculate the first order partial derivatives
and put them equal to zero.
∂f ∂ (xy)
= e − 2xy = ye(xy) − 2y = 0
∂x ∂x
305
∂f ∂ (xy)
= e − 2xy = xe(xy) − 2x = 0
∂y ∂y
We get the critical point (0, 0), and also infinitely many critical points for which e(xy) = 2. The
!
ln 2
latter can be expressed as x, .
x
∂ 2f ∂ ∂f ∂ (xy)
fxx = = = ye − 2y = y 2 e(xy)
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂ (xy)
fyx = = = xe − 2x = e(xy) + xye(xy) − 2
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ ∂f ∂ (xy)
fxy = = = ye − 2y = e(xy) + yxe(xy) − 2
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂ (xy)
fyy = = = xe − 2x = x2 e(xy)
∂y∂y ∂y ∂y ∂y
We have fxx (0, 0) = 0, fyx (0, 0) = fxy (0, 0) = −1, and fyy (0, 0) = 0. So
D(0, 0) = fxx (0, 0)fyy (0, 0) − [fxy (0, 0)]2 = 0 · 0 − (−1)2 = −1
Since D(0, 0) = −1 < 0, there is a saddle point at (0, 0) [see the explanation at the beginning of the
answer to Question 6 of Temat 23 on page 293]. The coordinates of this saddle point are (0, 0, 1).
!
ln 2
For the critical points x, , we have e(xy) = 2. So we have
x
! ! ! !
ln 2 2 ln 2 ln 2 ln 2
fxx x, = 2y , fyx x, = fxy x, = 2xy, and fyy x, = 2x2 . So
x x x x
! ! ! " !#2
ln 2 ln 2 ln 2 ln 2
D x, = fxx x, fyy x, − fxy x,
x x x x
= 2y 2 · 2x2 − (2xy)2 = 4x2 y 2 − 4x2 y 2 = 0
!
ln 2
Since D x, = 0, the test is inconclusive [see the explanation at the beginning of the answer
x
to Question 6 of Temat 23 on page 293]. The value of the function at these points is
!
ln 2
f x, = 2 − 2 ln 2
x
We consider the function g(t) = et − 2t. We have g 0 (t) = et − 2 and g 00 (t) = et . Since g 0 (ln 2) = 0
and g 00 (ln 2) = 2 > 0, we know that g has a local minimum at t = ln 2. Now we compare this with
ln 2
f (x, y) = e(xy) − 2xy. When moving away from a point at which y = , by comparison with the
x
function g(t), we know that the value of f (x, y) must either remain 2 − 2 ln 2 or increase, so the
!
ln 2
points x, are local minima.
x
306
b) We have f (x, y) = x3 y 2 (1 − x − y) = x3 y 2 − x4 y 2 − x3 y 3 . We need to calculate the first order
partial derivatives and put them equal to zero.
∂f ∂ 3 2
= x y − x4 y 2 − x3 y 3 = 3x2 y 2 − 4x3 y 2 − 3x2 y 3 = x2 y 2 (3 − 4x − 3y) = 0
∂x ∂x
∂f ∂ 3 2
= x y − x4 y 2 − x3 y 3 = 2x3 y − 2x4 y − 3x3 y 2 = x3 y(2 − 2x − 3y) = 0
∂y ∂y
Clearly x = 0 or y = 0 will make the two equations hold, and solving the system of two equations

3 − 4x − 3y
 =0
2 − 2x − 3y

=0
1 1

gives (x, y) = ,
2 3
∂ 2f ∂ ∂f ∂ 2 2
fxx = = = 3x y − 4x3 y 2 − 3x2 y 3 = 6xy 2 − 12x2 y 2 − 6xy 3
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂ 3
fyx = = = 2x y − 2x4 y − 3x3 y 2 = 6x2 y − 8x3 y − 9x2 y 2
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ ∂f ∂ 2 2
fxy = = = 3x y − 4x3 y 2 − 3x2 y 3 = 6x2 y − 8x3 y − 9x2 y 2
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂ 3
fyy = = = 2x y − 2x4 y − 3x3 y 2 = 2x3 − 2x4 − 6x3 y
∂y∂y ∂y ∂y ∂y
1 1 1 1 1 1 1 1 1 1 1

We have fxx , = − , fyx , = fxy , = − , and fyy , = − . So
2 3 9 2 3 2 3 12 2 3 8
2 2
1 1 1 1 1 1 1 1 1 1 1 1

D , = fxx , fyy , − fxy , = − · − − − =
2 3 2 3 2 3 2 3 9 8 12 144
307
1 1 1 1 1 1 1 1

Since D , = > 0 and fxx , = − < 0, there is a local maximum at , [see
2 3 144 2 3 9 2 3
the explanation at the beginning of the answer to Question 6 of Temat 23 on page 293]. The value
1
of this local maximum is .
432
At the critical points with x = 0 and the critical points with y = 0, we have fxx = fxy = 0 and
hence
D = fxx fyy − [fxy ]2 = 0
In these cases the test is inconclusive [see the explanation at the beginning of the answer to Question
6 of Temat 23 on page 293].
When moving away from (0, 0) along the line x = y, we have
f (x, x) = x3 x2 − x4 x2 − x3 x3 = x5 − 2x6
which, near the point (0, 0), increases when x increases, and, near the point (0, 0), decreases when x
decreases. So (0, 0) is not a local extremum.
For y = m 6= 0, when moving away from (0, m) along the line y = m, we have

f (x, m) = x3 m2 − x4 m2 − x3 m3 = m2 − m3 x3 − m2 x4
which, near the point (0, m), increases in one direction and decreases in the other, unless m = 1. If
m = 1, and hence m2 − m3 = 0 , the function f (x, m) decreases near the point (0, m) when moving
away from (0, m) along the line y = m.
For x = m 6= 0, when moving away from (m, 0) along the line x = m, we have

f (m, y) = m3 y 2 − m4 y 2 − m3 y 3 = m3 − m4 y 2 − m3 y 3
which, near the point (m, 0), decreases or increases in both directions, depending on the sign of
m3 − m4 , unless m = 1. If m = 1, and hence m3 − m4 = 0 , the function f (m, y) decreases near
the point (m, 0) when moving away from (m, 0) along the line x = m in positive y-direction, and
it increases near the point (m, 0) when moving away from (m, 0) along the line x = m in negative
y-direction.
2 −y 2
c) We have f (x, y) = 2x2 + y 2 e−x . We need to calculate the first order partial derivatives and
put them equal to zero.
∂f ∂ 2 2 2 2 2
2 2
= 2x + y 2 e−x −y = 4xe−x −y − 2x 2x2 + y 2 e−x −y
∂x ∂x
308
2 −y 2
= 2x 2 − 2x2 − y 2 e−x =0
∂f ∂ 2 2 2 2 2
2 2
= 2x + y 2 e−x −y = 2ye−x −y − 2y 2x2 + y 2 e−x −y
∂y ∂y
2 −y 2
= 2y 1 − 2x2 − y 2 e−x =0
We get the critical point (0, 0). It is obvious that this must be a local minimum, since f (0, 0) = 0 and
at all other points the function is positive valued. For x = 0, we have 1 − 2x2 − y 2 = 1 − y 2 , putting
this equal to zero gives the critical points (0, 1) and (0, −1). For y = 0, we have 2−2x2 −y 2 = 2−2x2
putting this equal to zero gives the critical points (1, 0) and (−1, 0).
In f (x, y), both variables only appear as a square. So the classification of the critical point (0, −1)
will be the same as the one of (0, 1), and the classification of the critical point (−1, 0) will be the
same as the one of (1, 0).
∂ 2f ∂ ∂f ∂ 2 2 ∂ 2 2
fxx = = = 2x 2 − 2x2 − y 2 e−x −y = 4x − 4x3 − 2xy 2 e−x −y
∂x∂x ∂x ∂x ∂x ∂x
2 −y 2
2 −y 2
= 4 − 12x2 − 2y 2 e−x − 2x 4x − 4x3 − 2xy 2 e−x
2 −y 2
2 −y 2
= 4 − 12x2 − 2y 2 e−x + −8x2 + 8x4 + 4x2 y 2 e−x
2 −y 2
= 4 − 20x2 − 2y 2 + 8x4 + 4x2 y 2 e−x
∂ 2f ∂ ∂f ∂ 2 2 ∂ 2 2
fyx = = = 2y 1 − 2x2 − y 2 e−x −y = 2y − 4x2 y − 2y 3 e−x −y
∂x∂y ∂x ∂y ∂x ∂x
2 −y 2
2 −y 2
= (−8xy) e−x − 2x 2y − 4x2 y − 2y 3 e−x
2 −y 2
2 −y 2
= (−8xy) e−x + −4xy + 8x3 y + 4xy 3 e−x
2 −y 2
2 −y 2
= 8x3 y + 4xy 3 − 12xy e−x = 4xy 2x2 + y 2 − 3 e−x
∂ 2f ∂ ∂f ∂ 2 2 ∂ 2 2
fxy = = = 2x 2 − 2x2 − y 2 e−x −y = 4x − 4x3 − 2xy 2 e−x −y
∂y∂x ∂y ∂x ∂y ∂y
2 −y 2
2 −y 2 2 −y 2
2 −y 2
= −4xye−x − 2y 4x − 4x3 − 2xy 2 e−x = −4xye−x + −8xy + 8x3 y + 4xy 3 e−x
2 −y 2
2 −y 2
= 8x3 y + 4xy 3 − 12xy e−x = 4xy 2x2 + y 2 − 3 e−x
∂ 2f ∂ ∂f ∂ 2 2 ∂ 2 2
fyy = = = 2y 1 − 2x2 − y 2 e−x −y = 2y − 4x2 y − 2y 3 e−x −y
∂y∂y ∂y ∂y ∂y ∂y
2 −y 2
2 −y 2
= 2 − 4x2 − 6y 2 e−x − 2y 2y − 4x2 y − 2y 3 e−x
2 −y 2
2 −y 2
= 2 − 4x2 − 6y 2 e−x + −4y 2 + 8x2 y 2 + 4y 4 e−x
2 −y 2
= 2 − 4x2 − 10y 2 + 8x2 y 2 + 4y 4 e−x
309
2 4
We have fxx (0, 1) = , fyx (0, 1) = fxy (0, 1) = 0, and fyy (0, 1) = − . So
e e
2 4 8

2
D(0, 1) = fxx (0, 1)fyy (0, 1) − [fxy (0, 1)] = · − − 02 = − 2
e e e
8
Since D(0, 1) = − , there is a saddle point at (0, 1) [see the explanation at the beginning of the
e2
1

answer to Question 6 of Temat 23 on page 293]. The coordinates of this saddle point are 0, 1, .
e
8 2
We have fxx (1, 0) = − , fyx (1, 0) = fxy (1, 0) = 0, and fyy (1, 0) = − . So
e e
8 2 16

2
D(1, 0) = fxx (1, 0)fyy (1, 0) − [fxy (1, 0)] = − · − − 02 = 2
e e e
16 8
Since D(1, 0) = 2
> 0 and fxx (1, 0) = − < 0, there is a local maximum at (1, 0) [see the
e e
2
this local maximum is .
e
d) We have f (x, y, z) = x2 + y 2 + z 2 − xy + x + 2z. We need to calculate the first order partial

∂f ∂ 2
= x + y 2 + z 2 − xy + x + 2z = 2x − y + 1 = 0
∂x ∂x
∂f ∂ 2
= x + y 2 + z 2 − xy + x + 2z = 2y − x = 0
∂y ∂y
∂f ∂ 2
= x + y 2 + z 2 − xy + x + 2z = 2z + 2 = 0
∂z ∂z
2 1

We get the critical point − , − , −1 . We will show that the function f (x, y, z) is bounded below
3 3
2 1

which will imply that the only critical point − , − , −1 must be a global minimum. The highest
3 3
power of z is even. So it suffices to show that the function g(x, y) = x2 + y 2 − xy − (−x) is bounded
below. The cases x = 0, y = 0, y = x, and y = −x are clear, this gives upward shaped parabolas.
310
If we let (x, y) approach (±∞, ±∞) along a give line other than y = x or y = −x, at least one
of x2 and y 2 is bigger than xy, and any one of x2 and y 2 , since it is a square, is bigger than −x.
So lim g(x, y) = +∞, which implies that g(x, y) is bounded below and hence the function
(x,y)→(±∞,±∞)
2 1 4

f (x, y, z) is bounded below. The value of the global minimum of f (x, y, z) at − , − , −1 is − .
3 3 3
Temat 25: Tangent planes and diffeomorphisms.

Question 1. Find the equation of the plane tangent to the surface described by the following
equations at the indicated point P :
√ √ √
a) 3 x + 3 y + 4 z = 1, P = (1, −1, 1)
b) xyz + x2 − 3y 2 + z 3 = 14, P = (5, −2, 3)
Answer: a) We will use the gradient as a vector which is perpendicular to the surface. It is essential
to have understood the√example in
√ two dimensions with a curve which is easy to visualize,
√ starting√on
√ √
3
page 264. The surface x+ y+ z = 1 is the level A = 1 of the function A(x, y, z) = x+ 3 y+ 3 z.
3 3 3
We will calculate the gradient of the function A(x, y, z) and evaluate it at P = (1, −1, 1).
!
∂ ∂ ∂
grad A(x, y, z) = ∇A(x, y, z) = A(x, y, z), A(x, y, z), A(x, y, z)
∂x ∂y ∂z
∂ √ √ ∂ √ √ ∂ √ √
!
3
√ √ √
= x+ 3y+ 3z , 3
x+ 3y+ 3z , 3
x+ 3y+ 3z
∂x ∂y ∂z
!
∂ 1 1 1
∂ 1 1 1
∂ 1 1 1

= x3 + y 3 + z 3 , x3 + y 3 + z 3 , x3 + y 3 + z 3
∂x ∂y ∂z
1 1 −1 1 1 −1 1 1 −1 1 −2 1 −2 1 −2

= x3 , y 3 , z 3 = x 3, y 3, z 3
3 3 3 3 3 3
 
1 −2 1 1 2 1 1 1 1 1 1 

2
grad A(1, −1, 1) = 1 3 , (−1)− 3 , 1− 3  √ 2 ,
= 2 ,
3 √
√ 2 
3 3 3 3 31 3 3 −1

3
1
!
1 1 1 1 1 1 1 1 1

= · 2, · 2, · = , ,
3 1 3 (−1) 3 12 3 3 3
Since the gradient is perpendicular to the surface, for a point (x, y, z) in the plane tangent to the
surface at (1, −1, 1), we have
1 1 1 1 1 1

0= , , · (x − 1, y − (−1), z − 1) = (x − 1) + (y + 1) + (z − 1)
3 3 3 3 3 3
We can multiply by 3 and the tangent plane is given by
x+y+z =1
b) We will use the gradient as a vector which is perpendicular to the surface. It is essential to
have understood the example in two dimensions with a curve which is easy to visualize, starting on
311
page 264. The surface xyz + x2 − 3y 2 + z 3 = 14 is the level A = 14 of the function A(x, y, z) =
xyz + x2 − 3y 2 + z 3 . We will calculate the gradient of the function A(x, y, z).
!
∂ ∂ ∂
grad A(x, y, z) = ∇A(x, y, z) = A(x, y, z), A(x, y, z), A(x, y, z) =
∂x ∂y ∂z
!
∂ ∂ ∂
= xyz + x2 − 3y 2 + z 3 , xyz + x2 − 3y 2 + z 3 , xyz + x2 − 3y 2 + z 3
∂x ∂y ∂z

= yz + 2x, xz − 6y, xy + 3z 2
We evaluate the gradient at P = (5, −2, 3).

grad A(5, −2, 3) = (−2) · 3 + 2 · 5, 5 · 3 − 6 · (−2), 5 · (−2) + 3 · 32 = (4, 27, 17)
Since the gradient is perpendicular to the surface, for a point (x, y, z) in the plane tangent to the
surface at (5, −2, 3), we have
0 = (4, 27, 17) · (x − 5, y − (−2), z − 3) = 4(x − 5) + 27(y + 2) + 17(z − 3)
and the tangent plane is given by
4x − 27y + 17z = 4 · 5 − 27 · 2 + 17 · 3 = 17
Question 2. In each of the following cases, find all points on the surface described by z = f (x, y)
at which the tangent plane is horizontal:
a) f (x, y) = −x2 − y 2 + 8x − 6y + 10
b) f (x, y) = x4 + y 3 − 3y
Answer: a) We have z = f (x, y) = −x2 − y 2 + 8x − 6y + 10 and hence
z + x2 + y 2 − 8x + 6y = 10. It is essential to have understood the previous question (Question 1 of
Temat 25, starting on page 311.). We will use the gradient as a vector which is perpendicular to the
surface. The surface z + x2 + y 2 − 8x + 6y = 10 is the level A = 10 of the function
A(x, y, z) = z + x2 + y 2 − 8x + 6y. We will calculate the gradient of the function A(x, y, z).
!
∂ ∂ ∂
∂x ∂y ∂z
!
∂ ∂ ∂
= z + x2 + y 2 − 8x + 6y , z + x2 + y 2 − 8x + 6y , z + x2 + y 2 − 8x + 6y
∂x ∂y ∂z
= (2x − 8, 2y + 6, 1)
If the gradient vector is in the direction (or opposite to the direction) of (0, 0, 1), the tangent plane
is horizontal. So we need 2x − 8 = 0 and 2y + 6 = 0 which implies (x, y) = (4, −3). So at the point
(4, −3, 35), the tangent plane is horizontal.
b) We have z = f (x, y) = x4 + y 3 − 3y and hence z − x4 − y 3 + 3y = 0. It is essential to have
understood the previous question (Question 1 of Temat 25, starting on page 311.). We will use the
gradient as a vector which is perpendicular to the surface. The surface z − x4 − y 3 + 3y = 0 is the
312
level A = 0 of the function A(x, y, z) = z − x4 − y 3 + 3y. We will calculate the gradient of the
function A(x, y, z).
!
∂ ∂ ∂
grad A(x, y, z) = ∇A(x, y, z) = A(x, y, z), A(x, y, z), A(x, y, z)
∂x ∂y ∂z
!
∂ 4 3
∂
4 3
∂
4 3

= z − x − y + 3y , z − x − y + 3y , z − x − y + 3y
∂x ∂y ∂z

= −4x3 , −3y 2 + 3, 1
If the gradient vector is in the direction (or opposite to the direction) of (0, 0, 1), the tangent plane is
horizontal. So we need −4x3 = 0 and −3y 2 +3 = 0 which implies (x, y) = (0, 1) or (x, y) = (0, −1).
So at the points (0, 1, −2 ) and (0, −1, 2), the tangent plane is horizontal.
Question 3. Determine all the points on the surface described by the equation
3 2 1 3 1 4
z=y + y − y − x2
4 24 32
where the tangent plane to the surface is horizontal. Which of these points are local extrema of the
function z = f (x, y)?
3 2 1 3 1 4 2 3 1 1
Answer: We have z = f (x, y) = y + y − y −x and hence z− y 2 − y 3 + y 4 +x2 = 0. It
4 24 32 4 24 32
is essential to have understood the previous question (Question 1 of Temat 25, starting on page 311.).
We will use the gradient as a vector which is perpendicular to the surface.
3 1 3 1 4
The surface z − y 2 − y + y + x2 = 0 is the level A = 0 of the function
4 24 32
3 1 3 1 4
A(x, y, z) = z − y 2 − y + y + x2 . We will calculate the gradient of the function A(x, y, z).
4 24 32
!
∂ ∂ ∂
∂x ∂y ∂z
∂ 3 1 3 1 4 ∂ 3 1 3 1 4

= z − y2 − y + y + x2 , z − y2 − y + y + x2 ,
∂x 4 24 32 ∂y 4 24 32
!
∂ 3 1 3 1 4

z − y2 − y + y + x2
∂z 4 24 32
3 1 1

= 2x, − y − y 2 + y 3 , 1
2 8 8
If the gradient vector is in the direction (or opposite to the direction) of (0, 0, 1), the tangent plane
3 1 1
is horizontal. So we need 2x = 0 and − y − y 2 + y 3 = 0. So x = 0 , and from the second
2 8 8
equation we get y(−12 − y + y 2 ) = y(y − 4)(y + 3) = 0. This implies (x, y) = (0, 0), (x, y) = (0, 4),
20 297

or (x, y) = (0, −3). So at the points (0, 0, 0 ), 0, 4, , and 0, −3, , the tangent plane is
3 96
horizontal.
313
∂f ∂f 3 1 1
We have = −2x and = y + y2 − y3.
∂x ∂y 2 8 8
∂ 2f ∂ ∂f ∂
fxx = = = (−2x) = −2
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂ 3 1 1

fyx = = = y + y2 − y3 = 0
∂x∂y ∂x ∂y ∂x 2 8 8
∂ 2f ∂ ∂f ∂
fxy = = = (−2x) = 0
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂ 3 1 1 3 1 3

fyy = = = y + y2 − y3 = + y − y2
∂y∂y ∂y ∂y ∂y 2 8 8 2 4 8
3
We have fxx (0, 0) = −2, fyx (0, 0) = fxy (0, 0) = 0, and fyy (0, 0) = . So
2
3
D(0, 0) = fxx (0, 0)fyy (0, 0) − [fxy (0, 0)]2 = (−2) · − 02 = −9
2
Since D(0, 0) = −9 < 0, there is a saddle point at (0, 0) [see the explanation at the beginning of
the answer to Question 6 of Temat 23 on page 293]. The coordinates of this saddle point are (0, 0, 0).
7
We have fxx (0, 4) = −2, fyx (0, 4) = fxy (0, 4) = 0, and fyy (0, 4) = − . So
2
7

2
D(0, 4) = fxx (0, 4)fyy (0, 4) − [fxy (0, 4)] = (−2) · − − 02 = 7
2
20
3
63
We have fxx (0, −3) = −2, fyx (0, −3) = fxy (0, −3) = 0, and fyy (0, −3) = − . So
24
63 63

2
D(0, −3) = fxx (0, −3)fyy (0, −3) − [fxy (0, −3)] = (−2) · − − 02 =
24 12
63
Since D(0, −3) = > 0 and fxx (0, −3) = −2 < 0, there is a local maximum at (0, −3) [see the
12
297
96
Question 4. Find a diffeomorphism between the following pairs of domains:

a) the interior of the triangle T1 with vertices (0, 0), (0, 1), (1, 0) and the interior of the triangle T2
with vertices (0, 0), (0, 1), (2, 0)
b) the interior of the triangle T with vertices (0, 0), (0, 1), (1, 0) and the interior of the square with
vertices (0, 0), (0, 1), (1, 1), (1, 0)
314
c) the interior of the triangle T with vertices (0, 0), (0, 1), (1, 0) and the interior of the unit circle
with centre (0, 0)
d) the interior of the triangle T with vertices (0, 0), (0, 1), (1, 0) and the domain
n o
P = (x, y) ∈ R2 : 0 < x < 1 and 0 < y < x2
e) the interior of the triangle T with vertices (0, 0), (0, 1), (1, 0) and the plane R2
Answer a) We need to find a diffeomorphism between the interior of the triangle T1 with vertices
(0, 0), (0, 1), (1, 0) and the interior of the triangle T2 with vertices (0, 0), (0, 1), (2, 0).
Let f (x, y) = (2x, y). This is a diffeomorphism from T1 to T2 . The segment of the line x = 1 + y,
although not part of the domain, would get mapped to the segment of the line x = 2 + 2y. The
x

inverse is given by f −1 (x, y) = ,y .
2
b) We need to find a diffeomorphism between the interior of the triangle T with vertices (0, 0),
(0, 1), (1, 0) and the interior of the square with vertices (0, 0), (0, 1), (1, 1), (1, 0)
y

Let f (x, y) = x, . This is a diffeomorphism from the interior of the triangle T with vertices
1−x
(0, 0), (0, 1), (1, 0) to the interior of the square with vertices (0, 0), (0, 1), (1, 1), (1, 0). The segment
of the line y = 1 − x, although not part of the domain, would get mapped to the segment of the line
y = 1. The inverse is given by f −1 (x, y) = (x, (1 − x)y).
c) We need to find a diffeomorphism between the interior of the triangle T with vertices (0, 0),
(0, 1), (1, 0) and the interior of the unit circle with center (0, 0). From part b), we already have a
315
y

diffeomorphism, f (x, y) = x, , from the interior of the triangle to the interior of a square.
1−x
The function g(x, y) = (2x − 1, 2y − 1) maps the interior of the square from part b) to the interior of
the square with vertices (1, 1), (−1, 1), (−1, −1), (1, −1). The unit circle centered at (0, 0) satisfies
√ √
the equation x2 + y 2 = 1, which gives y = ± 1 − x2 . Hence the function h(x, y) = x, y 1 − x2
maps the interior of the square with vertices (1, 1), (−1, 1), (−1, −1), (1, −1) to the interior of the
unit circle centered at (0, 0).
The diffeomorphism d which we are looking for is the composition of f , g and h. We have
2y

g ◦ f (x, y) = 2x − 1, −1
1−x
and
2y 2y
q q
d = h ◦ g ◦ f (x, y) = 2x − 1, −1 1 − (2x − 1) = 2x − 1,
2 − 1 · 2 · x(1 − x)
1−x 1−x
Now we calculate the inverse of d.
!
−1 y
h (x, y) = x, √
1 − x2
x 1 y 1 x+1 y+1

−1
g (x, y) = + , + = ,
2 2 2 2 2 2
f −1 (x, y) = (x, (1 − x)y)
√ y +1
! !
−1 −1 x+1 1−x2 x+1 y 1
g ◦ h (x, y) = , = , √ +
2 2 2 2 1 − x2 2
!!
x+1 x+1 y 1

−1 −1 −1 −1
d =f ◦g ◦h = , 1− √ +
2 2 2 1 − x2 2
316
d) We need toY find a diffeomorphism between the interior of the
Y triangle T with vertices (0, 0),
(0, 1), (1, 0) and the domain
n o
P = (x, y) ∈ R2 : 0 < x < 1 and 0 < y < x2
X X
+1.0 +1.0
From part b), we already know a diffeomorphism from the interior of the triangle with vertices (0, 0),
(0, 1), (1, 0) to the interior of the square with vertices (0, 0), (0, 1), (1, 1), (1, 0). This diffeomormism
y

is f (x, y) = x, . The function g(x, y) = (x, xy) is a diffeomorphism from the interior of the
1−x
square with vertices (0, 0), (0, 1), (1, 1), (1, 0) to the interior of the triangle with vertices (0, 0), (1, 1),
(1, 0). The function h(x, y) = (x, y 2 ) is a diffeomorphism from the interior of the triangle with
vertices (0, 0), (1, 1), (1, 0) to
n o
P = (x, y) ∈ R2 : 0 < x < 1 and 0 < y < x2
The diffeomorphism d which we are looking for is the composition of f , g and h. We have
xy

g ◦ f (x, y) = x,
1−x
and !
x2 y 2
d = h ◦ g ◦ f (x, y) = x,
(1 − x)2
√
h−1 (x, y) = (x, y)
y

g −1 (x, y) = x,
x
f −1 (x, y) = (x, (1 − x)y)
√ !
−1 −1 y
g ◦ h (x, y) = x,
x
√ !
−1 −1 −1 −1 (1 − x) y
d =f ◦g ◦h = x,
x
e) We need to find a diffeomorphism between the interior of the triangle T with vertices (0, 0), (0, 1),
(1, 0) and the plane R2 . Let f and g be the mappings from part c). The tg function is continuous
317
π π π π

on the open interval − , and has vertical asymptotes at − and , approaching −∞ and
2 2 2 2
πx πy

+∞ respectively. The function h(x, y) = tg , tg is a diffeomorphism from the square with
2 2
vertices (1, 1), (−1, 1), (−1, −1), (1, −1) to R2 . The diffeomorphism d which we are looking for is the
composition of f , g and h. We have
2y

g ◦ f (x, y) = 2x − 1, −1
1−x
and

2y

π −1
!
π(2x − 1) 1−x π(2x − 1) π(2y + x − 1)
d = h ◦ g ◦ f (x, y) = tg , tg  = tg , tg
2 2 2 2 − 2x

2 arc tg x 2 arc tg y

h−1 (x, y) = ,
π π
x 1 y 1

−1
g (x, y) = + , +
2 2 2 2
f −1 (x, y) = (x, (1 − x)y)
arc tg x 1 arc tg y 1

−1 −1
g ◦ h (x, y) = + , +
π 2 π 2
arc tg x 1 arc tg x 1 arc tg y 1

d−1 = f −1 ◦ g −1 ◦ h−1 = + , 1− − +
π 2 π 2 π 2
n o
Question 5. Find a diffeomorphism f : R1 → R2 where R1 = (x, y) ∈ R2 : 1 < x2 + y 2 < 4 and
n o
R2 = (x, y) ∈ R2 : 1 < x2 + y 2 < 9 .
Answer: R1 is the annulus with inner radius 1 and outer radius 2, and with center (0, 0). R2 is the
annulus with inner radius 1 and outer radius 3, and with center (0, 0).
318
Let
√ √ !
1 1 1q 2 1 x2 + y 2 + 1 x2 + y 2 + 1
q
f (x, y) = x x2 + y 2 + ,y x + y2 + = x ,y
2 2 2 2 2 2
This is a diffeomorphism from R1 to R2 . The circle of radius 1 centered at (0, 0), although not part
of the domain, would get mapped to itself. The circle of radius 2 centered at (0, 0), although not
part of the domain, would get mapped to the circle of radius 3 centered at (0, 0). The multplicative
1q 2 1
factor x + y 2 + was determined by solving the system of equations
2 2
1

a · 1 + b =


1


a · 2 + b =

 3
2
1 1
which gives a = and b = .
2 2
f (x, y) is bijective and hence invertible.
Temat 26: Local invertibility of functions. Implicit function

theorem. Extrema with constraints and Lagrange multi-
pliers.
Question 1. a) Show that there exists a unique function f (x) such that ∀x ∈ R, y = f (x) satisfies
the equation 3x + ex = y + ey .
b) Show that f is a diffeomorphism of class C 1 .
c) Compute the derivative of the inverse diffeomorphism f −1 at the point y = 0.
319
Answer: a) It is useful to illustrate what we are going to do by an example with a function for which
we can find an analytic expression of the inverse. Let’s look at x6 and y 3 . Clearly, y = f (x) = x2
satisfies the equation x6 = y 3 . If we put z = g(x) = x6 and z = h(y) = y 3 , then
1 6
y = f (x) = h−1 (g(x)) = h−1 x6 = x6 3
= x 3 = x2
Now we look at 3x + ex and y + ey . We put z = g(x) = 3x + ex and z = h(y) = y + ey . The function

h(y) is strictly increasing, which implies that it is injective. So h(y) is invertible, even though we
cannot find an analytic expression for h−1 . We have
y = f (x) = h−1 (g(x)) = h−1 (3x + ex )
b) The function f (x) from part a) is continuous. Since g(x) = 3x + ex is injective, it is invertible,
even though we cannot find an analytic expression for g −1 . The inverse of f is
x = f −1 (y) = g −1 (h(x)) = g −1 (y + ey )
Using z = 3x + ex and z = y + ey , we get

dz d dz d
= (3x + ex ) = 3 + ex = (y + ey ) = 1 + ey
dx dx dy dy
and hence
dz dz
0 df dy dx 3 + ex
−1 0
df −1 dx dy 1 + ey
f (x) = = = dz = f (y) = = = dz =
dx dx dy
1 + ey dy dy dx
3 + ex
Since f and f −1 have continuous first order partial derivatives, the function f is a diffeomorphism
of class C 1 .
c) Since we have 3x + ex = y + ey , y = 0 implies 3x + ex = 0 + e0 = 1. This equation only holds
for x = 0. So
0 1 + e0 2 1
f −1 (0) = 0
= =
3+e 4 2
Question 2. Show that the equations
F1 (x, y, t) = 3x2 y + t2 x − ty 2 − 3 = 0 and F2 (x, y, t) = tx2 + xy 2 − 2t2 y = 0
define x and y as functions of t such that x(1) = y(1) = 1 provided t is sufficiently close to 1.
Calculate a tangent vector v to the curve t → (x(t), y(t), t) at t = 1 using the equation
v = grad F1 (1, 1, 1) × grad F2 (1, 1, 1)
and explain why this cross product gives a vector which is tangent to the curve t → (x(t), y(t), t)
!
d dx dy
at t = 1 . The derivative of the curve t → (x(t), y(t), t) is (x(t), y(t), t) = (t), (t), 1 . Find
dt dt dt
the derivative at t = 1, i.e. calculate
!
dx dy
w= (1), (1), 1 .
dt dt
320
Check that w = λv (the vectors v and w are collinear).
Answer: We need to show that there exists a function g from a subset of R to a subset of R2 such
that g(1) = (1, 1) and g(t) = (x(t), y(t)) provided t is sufficiently close to 1. We need the Jacobian
matrix

∂F1 ∂F1 
6xy + t2 3x2 − 2ty
 
 ∂x ∂y 
 
J(x, y, t) =   = 
∂F2 ∂F2  2tx + y 2 2xy − 2t2
 

∂x ∂y
3−2
   
6+1 7 1
J(1, 1, 1) =   = 
2+1 2−2 3 0
 
7 1 7 1
det(J(1, 1, 1)) = det   = = 7 · 0 − 1 · 3 = −3
3 0 3 0
Since the determinant of the Jacobian matrix evaluated at (1, 1, 1) is −3 6= 0 , this matrix is invertible
and hence there exists a function g from a subset of R to a subset of R2 such that g(1) = (1, 1) and
g(t) = (x(t), y(t)) provided t is sufficiently close to 1.
We are going to use the gradient of F1 and the gradient of F2 , evaluated at the point (1, 1, 1), to
calculate the tangent vector to the curve t → (x(t), y(t), t) at the point (1, 1, 1).
grad F1 (x, y, t) =
! !
∂ ∂ ∂ ∂ ∂ ∂
= ∇F1 (x, y, t) = , , F1 (x, y, t) = F1 (x, y, t), F1 (x, y, t), F1 (x, y, t)
∂x ∂y ∂t ∂x ∂y ∂t
!
∂ 2 ∂ ∂
= 3x y + t2 x − ty 2 − 3 , 3x2 y + t2 x − ty 2 − 3 , 3x2 y + t2 x − ty 2 − 3
∂x ∂y ∂t

= 6xy + t2 , 3x2 − 2ty, 2tx − y 2
grad F1 (1, 1, 1) = (6 + 1, 3 − 2, 2 − 1) = (7, 1, 1)

grad F2 (x, y, t) =
! !
∂ ∂ ∂ ∂ ∂ ∂
= ∇F2 (x, y, t) = , , F2 (x, y, t) = F2 (x, y, t), F2 (x, y, t), F2 (x, y, t)
∂x ∂y ∂t ∂x ∂y ∂t
!
∂ 2 ∂ ∂
= tx + xy 2 − 2t2 y , tx2 + xy 2 − 2t2 y , tx2 + xy 2 − 2t2 y
∂x ∂y ∂t

= 2tx + y 2 , 2xy − 2t2 , x2 − 4ty
grad F2 (1, 1, 1) = (2 + 1, 2 − 2, 1 − 4) = (3, 0, −3)
i j k
1 1 7 1 7 1
v = (7, 1, 1) × (3, 0, −3) = 7 1 1 =i −j +k
0 −3 3 −3 3 0
3 0 −3
321
= i[1(−3) − 0 · 1] − j[7(−3) − 3 · 1] + k[7 · 0 − 3 · 1]
= −3i + 24j − 3k = (−3, 24, −3) = (−3)(1, −8, 1)
This cross product gives a vector which is tangent to the curve t → (x(t), y(t), t) at t = 1 because
it gives a vector which is perpendicular to the vector grad F1 (1, 1, 1) (which is in turn perpendicular
to the surface determined by F1 ) and also perpendicular to F2 (1, 1, 1) (which is in turn perpendicular
to the surface determined by F2 ). So v is tangent to both surfaces at the point (1, 1, 1), hence tangent
to the curve determined by the intersection of the two surfaces at the point (1, 1, 1).
We know the Jacobian matrix evaluated at the point (1, 1, 1). Its determinant is −3 and we are now
goint to calulate its inverse.
−1
1  0 −1
  
7 1
[J(1, 1, 1)]−1 =   = 
3 0 −3 −3 7
We also need
∂F1 ∂ 2
= 3x y + t2 x − ty 2 − 3 = 2tx − y 2
∂t ∂t
∂F1
which gives (1, 1, 1) = 1, and
∂t
∂F2 ∂ 2
= tx + xy 2 − 2t2 y = x2 − 4ty
∂t ∂t
∂F2 dx dy
which gives (1, 1, 1) = −3. Now we calculate (1) and (1) using the inverse of the Jacobian
∂t dt dt
matrix.
     
dx ∂F1 ∂F1
(1)  (1, 1, 1)  
0

−1  (1, 1, 1)
 dt −1  ∂t 1  ∂t
  
   
 = −[J(1, 1, 1)] 
  =   
 dy

  ∂F1

 3 −3 7  ∂F1
 

(1) (1, 1, 1) (1, 1, 1)
dt ∂t ∂t
    
1 0 −1 1 0 · 1 + (−1)(−3)
1
=   =  
3 −3 7 −3 3 (−3) · 1 + 7 · (−3)
   
1 3 1
=  = 
3 −24 −8
So we have
!
dx dy 1
w= (1), (1), 1 = (1, −8, 1) = (−3)(1, −8, 1) = λ(−3)(1, −8, 1) = λv
dt dt −3
Question 3. Show that the equation

1
x2 + y 2 + 3z 2 = xy + 6z 3
has a solution in the form z = f (x, y) such that f (0, 0) = 0.
322
Answer: We know that x2 + y 2 = k for a positive constant k is the equation of a circle centered at
(0, 0). By „stretching” the circle, we obtain an ellipse:
a(x + by)2 + (y + cx)2 = k
for real numbers b and c and a positive real number a. We are going to show that x2 + y 2 − xy = k
is the equation of an ellipse.

k = a(x + by)2 + (y + cx)2 = a x2 + 2bxy + b2 y 2 + y 2 + 2cxy + c2 x2

= a + c2 x2 + ab2 + 1 y 2 + (2ab + 2c)xy = x2 + y 2 − xy
gives the system of equations 



 a + c2 =1



ab2 + 1 = 1




2ab + 2c = −1
1 3
Since a > 0 , from the second equation we get b = 0 and thus c = − and a = . So for a positive
2 4
number k, we have the ellipse
2
3 x

k = x2 + y − = x2 + y 2 − xy
4 2
From the original equation
1
x2 + y 2 + 3z 2 = xy + 6z 3
1
we get that x2 + y 2 − xy = 6z 3 − 3z 2 . So we can define f (x, y) such that f (0, 0) = 0, and around
1
(x, y) = (0, 0), we let f (x, y) = z such that 6z 3 − 3z 2 = k for
2
3 x

k = x2 + y − = x2 + y 2 − xy
4 2
1 1
Since 6 · 0 3 − 3 · 02 = 0 and for z positive and close to zero, 6z 3 − 3z 2 > 0, the function f (x, y) is
well defined and continuous.
Alternatively, one may use the Jacobian matrix as in Question 2 of the present temat (the previous
question). We have one equation to be satisfied,
1
x2 + y 2 + 3z 2 − xy = 6z 3
We can take the third power on both sides which gives

3
x2 + y 2 + 3z 2 − xy = 216z
Thus 3
F = x2 + y 2 + 3z 2 − xy − 216z = 0
Differentiating this expression with respect to z gives
∂F 2
= 3 x2 + y 2 + 3z 2 − xy · 6z − 216
∂z
323
So we have the Jacobian matrix
" #
∂F
2
2 2 2
J(x, y, z) = = 3 x + y + 3z − xy · 6z − 216
∂z
Evaluated at the point (0, 0, 0), we get
J(0, 0, 0) = [−216]
1
which has determinant −216 6= 0 and is thus invertible. So the equation x2 + y 2 + 3z 2 − xy = 6z 3
defines z as a function f (x, y) in a neighbourhood of (0, 0) such that f (0, 0) = 0.
Question 4. Let g : R → R be of class C 1 such that for some 0 < k < 1 and every x ∈ R the
condition |g 0 (x)| ¬ k holds. Show that the mapping h(x) = x + g(x) is a diffeomorphism of class
C 1.
Answer: We have
d
h0 (x) = [x + g(x)] = 1 + g 0 (x)
dx
and we know that for every x ∈ R, |g 0 (x)| ¬ k for a fixed number k satisfying 0 < k < 1 . This
implies that h0 (x) 1 − k > 0 . Thus at any x, the function h(x) has slope which is greater than or
equal to a given positive number. A continuous function which has positive slope everywhere must be
injective, otherwise there would be a contradiction with the mean value theorem (stated at the end
of this answer). If in addition the slope is always greater than or equal to a fixed positive number,
the function must also be surjective and thus bijective. So h(x) = x + g(x) is invertible and thus a
diffeomorphism of class C 1 from R to R.
The mean value theorem states that for any function that is continuous on [a, b] and differentiable
on (a, b) there exists some c in the interval (a, b) such that the secant joining the endpoints of the
interval [a, b] is parallel to the tangent a c.
Question 5. Let E = {(x, y) ∈ R2 : x2 − 2xy + 2y 2 = 1}. Use the Lagrange multipliers method to
find points of E which are closest to and farthest from the origin of the coordinate system.
Answer: Before solving the problem, we are going to determine the shape of E. We know that
x2 + y 2 = 1 is the equation of a circle centered at (0, 0). By „stretching” the circle, we obtain an
ellipse:
a(x + by)2 + (y + cx)2 = 1
for real numbers b and c and a positive real number a. We are going to show that x2 − 2xy + 2y 2 = 1
is the equation of an ellipse.

1 = a(x + by)2 + (y + cx)2 = a x2 + 2bxy + b2 y 2 + y 2 + 2cxy + c2 x2
324

= a + c2 x2 + (2ab + 2c)xy + ab2 + 1 y 2 = x2 − 2xy + 2y 2
gives the system of equations 



 a + c2 =1



2ab + 2c = −2


ab2 + 1 = 2


From the second equation, we get c = −1 − ab. Putting this in the first equation gives
1
a + 1 + 2ab + a2 b2 = 1. From the third equation, we get a = 2 . Putting this in the equation
b
1 2 1
a + 1 + 2ab + a2 b2 = 1 gives 2 + 1 + + 2 = 1 . Hence b2 = −b, which implies b = −1, and through
b b b
back substitution, we get a = 1 and c = 0 . So we have
x2 − 2xy + 2y 2 = (x − y)2 + y 2 = 1
which is the equation of an ellipse.
325
Now we will solve the problem and find the points of this ellipse which are closest to and farthest
from (0, 0). The distance from the origin is biggest when its square is biggest. So we need to find x
and y such that f (x, y) = x2 + y 2 reaches an extremum, subject to the constraint
g(x, y) = x2 − 2xy + 2y 2 = 1. Using Lagrange multipliers, we solve the equations ∇f = λ∇g and
g(x, y) = 1, which can be written as
fx = λgx fy = λgy g(x, y) = 1
or as 
2x

 = λ(2x − 2y)



2y = λ(4y − 2x)


 2
− 2xy + 2y 2 = 1

x
q
From the third equation, we have x2 − 2xy + 2y 2 = (x − y)2 + y 2 = 1, hence x = y ± 1 − y 2 . We
are not interested in the values of λ, but solving the first two equations for λ allows us to equate the
expressions for λ which gives
x y
=λ=
x−y 2y − x
q
2 2 2 2
Thus 2xy − x = xy − y which is equivalent to 0 = xy − x + y . We know that x = y ± 1 − y2.
q
For x = y + 1 − y 2 , we get
q q 2
0= y+ 1− y2 y− y+ 1− y2 + y2
q q
2 2 2
=y +y 1− y2 − y + 2y 1 − y2 +1−y + y2
q
2
= 2y − y 1 − y 2 − 1
q
y 1 − y 2 = 2y 2 − 1
y 2 (1 − y 2 ) = 4y 4 − 4y 2 + 1
0 = 5y 4 − 5y 2 + 1
5 5 5 1
= 5y 4 − 5y 2 + − + 1 = 5y 4 − 5y 2 + −
| {z 4} 4 | {z 4} 4
√ !2
√ 52 1
= 5y − −
2 4
√ !2
1 √ 2 5
= 5y −
4 2
√
1 √ 5
± = 5y 2 −
2 2
1 1 1 1
y2 = + √ or − √
2 2 5 2 2 5
326
√ √
5 + 1 5−1
y2 = √ or √
2 5 2 5
v√ v√ v√ v√
u u u u
u 5+1 u 5+1 u 5−1 u 5−1
t √
y= or − t √ or t √
or − t √
2 5 2 5 2 5 2 5
We already have four values for y , since an ellipse has four points at which the distance from the
origin reaches
q an extremum, there is no need to do a calculation
q similar to theqpreceding one for
x = y − 1 − y . If we were going to use both x = y + 1 − y and x = y − 1 − y 2 to find the
2 2
corresponding x-values, this would give eight distinct points and we know that only two points can
have maximal distance from the origin and two q points can have minimalq distance from the origin. So
for each of the four points, only one of x = y + 1 − y 2 and x = y − 1 − y 2 gives the x coordinate
of a point which has maximal or minimal distance from the origin. After obtaining an x coordinate,
we will need to check whether 0 = xy − x2 + y 2 , and if this equation holds, the pair (x, y) gives the
coordinates of a extremum.
We are going to use the two positive y-values, calculate the corresponding x-value using
q q
x=y+ 1− y2 or x = y − 1 − y 2 , and check that 0 = xy − x2 + y 2 .
v√
u
u 5+1 q
t √
For y = , using x = y + 1 − y 2 , we get
2 5
v v
u√ u√
v√ 
u√
u v 2 v v√ v√ 
u u u u
5+1 u 5+1 u 5 + 1 u 5−1
t 5 + 1
u u u u 5+1
t √ t1 − 
+u t
√  ,
t √ 
=  √ + t √
, t √ 


2 5 2 5 2 5 
2 5 2 5 2 5
≈ (1.37638, 0.850651)
We check
u√
v√
u√ u√
v√
u√
v v v 2 v 2
u u
u 5−1 u 5+1 u 5−1
t 5 + 1 t 5 + 1
u u u 5+1
xy − x2 + y 2 = √ + t √ t √
− √ +t √  + t √ 
  
 
2 5 2 5 2 5 2 5 2 5 2 5
u√
v√ v√
u√
v 2 v 2
u u
u 5+1 u 5 + 1u 5 − 1 u 5−1
=
t
 √  − t √ t √
− t √ 
 
2 5 2 5 2 5 2 5
√v √
5+1 u 5 − 1 5−1 2 2
u
= √ −ut √ 2 − √ = √ − √ =0
2 5 2 5 2 5 2 5 2 5
v√
u
u 5−1 q
t √
For y = , using x = y − 1 − y 2 , we get
2 5
v v
u√ u√
v√ 
u√
u v 2 v v√ v√ 
u u u u
u
t √
5−1 u u 5−1 u 5 − 1
t 5 − 1
u u 5+1 u 5−1
t1 − 
−u
t
√  ,
t √ 
=  √ − t √
, t √ 


2 5 2 5 2 5
2 5 2 5 2 5
327
≈ (0.525731, −0.32492)
We check
u√
v√
u√ u√
v√
u√
v v v 2 v 2
u u
t 5 − 1
u 5+1 u 5−1
t 5 − 1
u u u 5+1 u 5−1
xy − x2 + y 2 =  √ − t √ t √
 − √ −t √   +

t √ 

2 5 2 5 2 5 2 5 2 5 2 5
u√
v√ v√
u√
v 2 v 2
u u
u 5−1 u 5 + 1u 5 − 1 u 5+1
= t
 √  + t √ t √
−

t √ 

2 5 2 5 2 5 2 5
√ v √
5−1 u 5 − 1 5+1 2 2
u
= √ +ut √ 2 − √ =− √ + √ =0
2 5 2 5 2 5 2 5 2 5
So we get the point on the ellipse E with maximal distance from the origin
u√
v v√ v√ 
u u
u 5−1
t 5 + 1
u u 5+1
 √ + t √
, t √ 
 ≈ (1.37638, 0.850651)
2 5 2 5 2 5
and reflecting at the origin, we get

u√
 v v√ v√ 
u u
u 5−1
 t 5+1
u u 5+1
− √ − t √
, −t √ 

≈ (−1.37638, −0.850651)
2 5 2 5 2 5
And we get the point on the ellipse E with minimal distance from the origin
u√
v v√ v√ 
u u
t 5 − 1
u u 5+1 u 5−1
 √ − t √
, t √ 
 ≈ (0.525731, −0.32492)
2 5 2 5 2 5
and reflecting at the origin, we get

u√
 v v√ v√ 
u u
 t 5−1
u u 5+1 u 5−1
− √ + t √
, −t √ 

≈ (−0.525731, 0.32492)
2 5 2 5 2 5
Question 6. Use the Lagrange multipliers method to find all those points on the ellipse x2 +2y 2 = 1,
which are nearest to and farthest from the line x + y = 2.
Answer: The first thing we need is a function f (x, y) which gives the distance of an arbitrary
point (x, y) from the line x + y = 2. This can be done using the dot product from linear algebra
as follows. The point (2, 0) is on the line x + y = 2. For any point (x, y) , we take the vector
(x, y) − (2, 0) = (x − 2, y) and project it on a unit vector which is perpendicular to the line x + y = 2.
1
The vector √ (1, 1) is a unit vector perpendicular to the line x + y = 2. The dot product gives
2
1 1 1
√ (x − 2, y) · (1, 1) = √ (x − 2 + y), so we get the function f (x, y) = √ (x − 2 + y). This function
2 2 2
328
gives a negative value for a point which lies below the line x + 2 = y, but we can later take the
absolute value.
1
We need to find x and y such that f (x, y) = √ (x − 2 + y) reaches an extremum, subject to the
2
2 2
constraint g(x, y) = x + 2y = 1. Using Lagrange multipliers, we solve the equations ∇f = λ∇g
and g(x, y) = 1, which can be written as
or as 
1
√ = 2xλ


2






1
√ = 4yλ
 2





x2 + 2y 2 =

1

From the first two equations, we get 2y = x, putting this in the third equation, we get
!
1 1 2 1
4y 2 + 2y 2 = 6y 2 = 1. Thus y = √ or y = − √ . So we get the point √ , √ whose distance
6 6 6 6
√2 − 2 + √1
! !
2 1 6 6 2 1
from the line x + y = 2 is f √ , √ = √ . And we get the point − √ , − √
6 6 2 6 6
2 1
− √6 − 2 − √6
!
2 1
whose distance from the line x + y = 2 is f − √ , − √ = √ . So the point
6 6 2
! !
2 1 2 1
√ ,√ is nearest to the line x + y = 2 , and the point − √ , − √ is farthest from the line
6 6 6 6
x + y = 2.
329
Question 7. Find supremum and infimum of f (x, y, z) = x2 − yz on the sphere x2 + y 2 + z 2 = 1.
Answer: We need to find x , y and z such that f (x, y, z) = x2 − yz reaches an extremum, subject
to the constraint g(x, y, z) = x2 + y 2 + z 2 = 1. Using Lagrange multipliers, we solve the equations
∇f = λ∇g and g(x, y, z) = 1, which can be written as
fx = λgx fy = λgy fz = λgz g(x, y, z) = 1
or as 


 2x = 2xλ



−z = 2yλ


−y = 2zλ







 2
x + y2 + z2 = 1

From the first equation, we get λ = 1 or x = 0. If λ = 1, from the second and the third equation,
we get −z = −4z , hence z = y = 0 and then from the fourth equation we get x = 1 or x = −1. If
330
x = 0, will will use the second and third equations to get
z y
− =λ=−
2y 2z
and thus z 2 = y 2 . Then from the fourth equation we get
! ! ! !
1 1 1 1 1 1 1 1
(y, z) = √ , √ or (y, z) = − √ , √ or (y, z) = √ , −√ or (y, z) = − √ , − √ .
2 2 2 2 2 2 2 2
We have
!
1 1 1
f (1, 0, 0) = 1 f (−1, 0, 0) = 1 f 0, √ , √ =−
2 2 2
! ! !
1 1 1 1 1 1 1 1 1
f 0, − √ , √ = f 0, √ , − √ = f 0, − √ , − √ =−
2 2 2 2 2 2 2 2 2
Thus the supremum and infimum of f (x, y, z) = x2 − yz on the sphere x2 + y 2 + z 2 = 1 are

1
respectively 1 and − .
2
Question 8. a) Find the maximal value of the function f (x, y, z) = x + y + z on the sphere
x 2 + y 2 + z 2 = a2 .
b) Prove the following inequality between the arithmetic and square mean, i.e.,
s
x+y+z x2 + y 2 + z 2
¬ for all x, y, z 0
3 3
Answer: a) We need to find x , y and z such that f (x, y, z) = x + y + z reaches an extremum,

subject to the constraint g(x, y, z) = x2 + y 2 + z 2 = a2 . Using Lagrange multipliers, we solve the
equations ∇f = λ∇g and g(x, y, z) = a2 , which can be written as
fx = λgx fy = λgy fz = λgz g(x, y, z) = a2
or as 
1

 = 2xλ



1 = 2yλ


1= 2zλ







 2
x + y 2 + z 2 = a2

1
From the first three equations we get x = y = z = . Then from the fourth equation we get
2λ
a
x = y = z = ± √ . Thus the maximal value of the function f (x, y, z) = x + y + z on the sphere
3
x2 + y 2 + z 2 = a2 is
√
!
a a a 3a
f √ ,√ ,√ = √ = 3a
3 3 3 3
given that in defining the sphere one uses a non-negative a.
q
b) Let a = x2 + y 2 + z 2 . From part a), we know that if x2 + y 2 + z 2 = a2 for a non-negative real
331
√
number a, then x + y + z ¬ 3 a. Thus
√ √ 2 s
x+y+z 3a a x + y2 + z2 x2 + y 2 + z 2
¬ =√ = √ =
3 3 3 3 3
Temat 27: More examples of conditional extrema and the

Kuhn-Tucker theorem.
Question 1. a) Show that among rectangles with vertices in the circle of radius 1, the square
maximizes the area.
b) If we inscribe rectangles in the remaining pieces of the circle so that one of the sides is contained
in an edge of the square, what dimensions should we choose so that the rectangles maximise area?
Answer: a)
We need a function which gives the area of the rectangle. Let (x, y) be the coordinates of the upper
right vertex of the rectangle. Then f (x, y) = (2x)(2y) = 4xy gives the area of the triangle. We
want to maximize f (x, y) = 4xy subject to the constraint g(x, y) = x2 + y 2 = 1 . Using Lagrange
multipliers, we solve the equations ∇f = λ∇g and g(x, y) = 1, which can be written as
or as 


 4y = 2xλ



4x = 2yλ


 2
+ y2 = 1

x
We are not interested in the values of λ, but we can solve the first two equations for λ to get
2y 2x
=λ=
x y
Hence x2 = y 2 , and since (x, y) are the coordinates of the upper right vertex of the rectangle, we
get x = y. Thus the rectangle which maximizes the area is a square.
b)
332
From x2 + y 2 = 1 , we get that the coordinates of the upper right vertex of the square found
!
1 1
in part a) are √ , √ . We need a function wich gives the area of the small rectangles. Let
2 2
(x, y) be the upper right vertex of the red rectangle. Then the area of this rectangle is given by
√
!
1
f (x, y) = x − √ · 2y . We want to maximize f (x, y) = 2x − 2 y subject to the constraint
2
2 2
g(x, y) = x +y = 1 . Using Lagrange multipliers, we solve the equations ∇f = λ∇g and g(x, y) = 1,
which can be written as
or as 

 2y = 2xλ
√




2x − 2 = 2yλ




 2
+ y2 = 1

x
y x − √12
=λ=
x y
x
Thus y 2 = x2 − √ , and putting this in g(x, y) = x2 + y 2 = 1, we get
2
x
x2 + x2 − √ = 1
2
x 1 1
2x2 − √ + − =1
2 16 16
| {z }
complete square
x 1 17
2x2 − √ + =
2 16 16
| {z }
complete square
√ 2
1 17
2x − =
4 16
333
√
√ 1 17
2x − = ±
4 4
√ √
1 17 1 17
4
+ 4 4
− 4
x= √ or √
2 2
Since (x, y) are the coordinates of the upper right vertex of the red rectangle, we use
1
√
17
√
+ 1 + 17
x= 4 √ 4 = √ , and from y 2 = 1 − x2 , we get
2 4 2
√ !2 √ √
2 1 + 17 32 1 + 2 17 + 17 7 − 17
y =1− √ = − =
4 2 32 32 16
s √ q √
7 − 17 7 − 17
So y = = , and consequently the dimensions of four small rectangles are
16 4
q √ q √ √ √
7 − 17 7 − 17 1 + 17 1 17 − 3
2 = for the long edge, and √ −√ = √ for the short edge.
4 2 4 2 2 4 2
Question 2. Among the points which belong to the intersection of the plane x + y + z = 12 and
the paraboloid z = x2 + y 2 , find those with maximal and minimal z coordinate.
Answer: We want to find the maximum and minimum of the function z = f (x, y) = x2 + y 2 subject
to the constraint g(x, y) = x + y + z = x + y + x2 + y 2 = 12. Using Lagrange multipliers, we solve
the equations ∇f = λ∇g and g(x, y) = 12, which can be written as
or as 
2x

 = λ(1 + 2x)



2y = λ(1 + 2y)


x + y + x2 + y 2 = 12


2x 2y
=λ=
1 + 2x 1 + 2y
So x + 2xy = y + 2xy which implies x = y. Using x = y, from the third equation we get x + x2 = 6.
Thus x = y = 2 or x = y = −3. Using x = y = 2, we get the minimal z value 8. Using x = y = −3,
we get the maximal z value 18.
Question 3. Among the points which belong to the intersection of the cone z 2 = x2 + y 2 and the
plane x + 2y + 3z = 3, find those closest to and farthest from the origin.
Answer: The distance to the origin reaches an extremum when the square of the distance reaches
an extremum. So we need to find the extrema of the function f (x, y) = x2 + y 2 + z 2 = 2x2 + 2y 2
subject to the constraint
q
g(x, y) = x + 2y + 3z = x + 2y + 3 x2 + y 2 = 3
334
or q
h(x, y) = x + 2y + 3z = x + 2y − 3 x2 + y 2 = 3
Using Lagrange multipliers, we solve the equations ∇f = λ∇g and g(x, y) = 3, which can be written
as
or as !
3x

4x =λ 1+ √ 2



x + y2







 !
3y
4y = λ 2 + √ 2
x + y2









 q
x + 2y + 3 x2 + y 2 = 3

4x 4y
=λ=
1 + √ 3x 2 + √ 3y
2x +y 2 x +y 2
2
Thus
3xy 3xy
2x + √ 2 2
=y+√ 2
x +y x + y2
q √
which implies y = 2x. Using y = 2x in x + 2y + 3 x2 + y 2 = 3, we get 5x + 3 5x2 = 3.
3 6
Thus x = √ and y = √ . Repeating this procedure with the constraint equation
5+3 5 5+3 5
q 3 6
h(x, y) = x+2y−3 x2 + y 2 = 3, one also obtains y = 2x, and then x = √ and y = √ .
5−3 5 5−3 5
We use x + 2y + 3z = 3 to get the z coordinate. The point which is closest to the origin is
 
3√
3− − 2 5+36√5
!
3 6 5+3 5 3 6 5
 √ , √ ,  = √ , √ , 1− √
5+3 5 5+3 5 3 5+3 5 5+3 5 5+3 5
√ !
3 6 3 5
= √ , √ , √
5+3 5 5+3 5 5+3 5
and the point which is farthest from the origin is
 
3√
3− − 2 5−36√5
!
3 6 5−3 5 3 6 5
 √ , √ ,  = √ , √ , 1− √
5−3 5 5−3 5 3 5−3 5 5−3 5 5−3 5
√ !
3 6 −3 5
= √ , √ , √
5−3 5 5−3 5 5−3 5
Question 4. Determine the maximum and minimum values of the function f (x, y, z) = x2 + y 2 − z
on the set S = {(x, y, z) ∈ R3 : x2 + y 2 − z 2 = 0, x2 + y 2 + (z − 2)2 = 2}.
Answer: We have the double cone determined by x2 + y 2 − z 2 = 0 and the sphere determined by
335
√
x2 + y 2 + (z − 2)2 = 2, which is a sphere of radius 2 centered at (0, 0, 2). The sphere and the cone
touch each other, they do not intersect. Since the locus where the two surfaces touch each other is a
circle given by z = 1 and x2 + y 2 = 1, the function f (x, y, z) = x2 + y 2 − z has constant value zero
on the set S.
Question 5. Determine the maximum and minimum values of the function f (x, y) on the set
S ⊂ R2 , where:
a) f (x, y) = x2 − y 2 S = {(x, y) ∈ R2 : x2 + y 2 = 4}
b) f (x, y) = x2 + y 2 S = {(x, y) ∈ R2 : 2x + 3y = 6}
c) f (x, y) = xy S = {(x, y) ∈ R2 : 4x2 + 9y 2 = 36}
d) f (x, y) = 4x2 + 9y 2 S = {(x, y) ∈ R2 : x2 + y 2 = 1}
e) f (x, y) = x2 y 2 S = {(x, y) ∈ R2 : x2 + 4y 2 = 4}
Answer: a) We need to determine the maximum and minimum values of the function
f (x, y) = x2 − y 2 subject to the constraint g(x, y) = x2 + y 2 = 4. Using Lagrange multipliers, we
solve the equations ∇f = λ∇g and g(x, y) = 4, which can be written as
or as 


 2x = 2xλ



−2y = 2yλ


 2
+ y2 = 4

x
From the first equation, we get x = 0 or λ = 1. If x = 0, then from the third equation we
get y = 2 or y = −2. If λ = 1, then from the second equation we get y = 0, and then either
x = 2 or x = −2. So there are four solutions, namely (x, y, λ) = (0, 2, −1), (x, y, λ) = (0, −2, −1),
(x, y, λ) = (2, 0, 1) and (x, y, λ) = (−2, 0, 1). So the maximum is f (2, 0) = f (−2, 0) = 4, and the
minimum is f (0, 2) = f (0, −2) = −4.
b) We need to determine the maximum and minimum values of the function
f (x, y) = x2 + y 2 subject to the constraint g(x, y) = 2x + 3y = 6. Using Lagrange multipliers, we
or as 


2x = 2λ



2y = 3λ




2x + 3y = 6
336
2y 2y 4y 13y
x=λ= . Thus x = , and putting this in the third equation, we get + 3y = = 6. So
3 3 3 3
18 12
we have y = and x = . Thus we get the minimum
13 13
12 18 122 + 182 468 36

f , = 2
= 2 =
13 13 13 13 13
The function g(x, y) = 2x + 3y = 6 describes a line in R2 , the function f (x, y) = x2 + y 2 does not
reach a maximum on that line.
c) We need to determine the maximum and minimum values of the function
f (x, y) = xy subject to the constraint g(x, y) = 4x2 + 9y 2 = 36. Using Lagrange multipliers, we
or as 


 y = 8xλ



x = 18yλ


4x2 + 9y 2 = 36


y x
=λ= . So 9y 2 = 4x2 , and from the third equation we get 8x2 = 36 and 18y 2 = 36. This
8x 18y
3 √ 3 √ √ √
! ! ! !
3 3
leads to the solutions √ , 2 , − √ , 2 , √ , − 2 , − √ , − 2 . So we get the maximum
2 2 2 2
√ √ √ √
! ! ! !
3 3 3 3
f √ , 2 = f − √ , − 2 = 3 and the minimum f − √ , 2 = f √ , − 2 = −3.
2 2 2 2
d) We need to determine the maximum and minimum values of the function
f (x, y) = 4x2 + 9y 2 subject to the constraint g(x, y) = x2 + y 2 = 1. Using Lagrange multipliers, we
or as 


 8x = 2xλ



18y = 2yλ


 2
+ y2 = 1

x
From the first equation, we get λ = 4 or x = 0. If λ = 4, the second equation implies y = 0,
and then from the third equation, we get x = 1 or x = −1. If x = 0, from the third equation we
get y = 1 or y = −1. This gives the solutions (1, 0), (−1, 0), (0, 1), and (0, −1). So we get the
maximum f (0, 1) = f (0, −1) = 9 and the minimum f (1, 0) = f (−1, 0) = 4.
e) We need to determine the maximum and minimum values of the function
f (x, y) = x2 y 2 subject to the constraint g(x, y) = x2 + 4y 2 = 4. Using Lagrange multipliers, we solve
the equations ∇f = λ∇g and g(x, y) = 4, which can be written as
337
or as 
2
2xy

 = 2xλ



2x2 y = 8yλ


 2
+ 4y 2 = 4

x
x2 x2
From the first two equations, we get y 2 = λ = , unless λ = 0. If y 2 = , from the third equation
4 4
√ √ √ 1 √ 1 √
! ! !
1
we get x = 2 or x = − 2. This gives the solutions 2, √ , − 2, √ , 2, − √ , and
2 2 2
√
!
1
− 2, − √ . If λ = 0, we get that one of x and y must be zero, which leads to the solutions (2, 0),
2
(−2, 0), (0, 1), and (0, −1).
√ 1 √ 1 √ √
! ! ! !
1 1
So we get the maximum f 2, √ = f − 2, √ =f 2, − √ = f − 2, − √ = 1,
2 2 2 2
and the minimum f (2, 0) = f (−2, 0) = f (0, 1) = f (0, −1) = 0.
Question 6. Find the points belonging to the set S = {(x, y) ∈ R2 : x2 + 12xy + 6y 2 = 130} which
are nearest to the origin.
Answer: The distance to the origin is smallest if its square is smallest. So we need to find the minimum
of the function f (x, y) = x2 + y 2 subject to the constraint g(x, y) = x2 + 12xy + 6y 2 = 130. Using
Lagrange multipliers, we solve the equations ∇f = λ∇g and g(x, y) = 130, which can be written as
or as 


2x = (2x + 12y)λ



2y = (12x + 12y)λ


 2
+ 12xy + 6y 2 = 130

x
From the first two equations we get
x y
=λ=
x + 6y 6x + 6y
Thus 6x2 + 6xy = xy + 6y 2 which is equivalent to 6x2 + 5xy − 6y 2 = 0. We get

25 2 25 2 25 169 2
0 = 6x2 + 5xy − 6y 2 = 6x2 + 5xy + y − y − 6y 2 = 6x2 + 5xy + y 2 − y
| {z 24 } 24 | {z 24 } 24
!2
√ 5 169 2
= 6x + √ y − y
2 6 24
!2
169 2 √ 5
y = 6x + √ y
24 2 6
13 √ 5
± √ y = 6x + √ y
2 6 2 6
338
±13y = 12x + 5y
3 2
y= x or y = − x
2 3
3 27 2 65 2
With y = x, from the equation x2 + 12xy + 6y 2 = 130, we get x2 + 18x2 + x = x = 130,
2 2 2
2
which implies x2 = 4 and consequently x = 2 or x = −2. With y = − x, from the equation
3
2 2 2 2 8 2 13 2
x + 12xy + 6y = 130, we get x − 8x + x = − x = 130, which does not have any solutions
3 3
for real numbers x. So the points which are nearest to the origin are (2, 3) and (−2, −3).
Question 7. Find the points belonging to the set S = {(x, y) ∈ R2 : 4x2 + 9y 2 = 36} nearest and
farthest from the point (1, 0).
The distance to the point (1, 0) reaches an extremum if its square reaches an extremum. So we need
to find the extrema of the function f (x, y) = (x − 1)2 + y 2 subject to the constraint
g(x, y) = 4x2 + 9y 2 = 36. Using Lagrange multipliers, we solve the equations ∇f = λ∇g and
g(x, y) = 36, which can be written as
or as 


 2x − 2 = 8xλ



2y = 18yλ


4x2 + 9y 2 = 36


We get the solution y = 0, x = ±3. And if y 6= 0, from the first two equations we get
2y 1 2x − 2 1 1
= =λ= = −
18y 9 8x 4 4x
9 36 · 25 − 4 · 81 64
which implies x = and then from equation 4x2 + 9y 2 = 36 we get y 2 = = , so
5 9 · 25 25
8 9 8 9 8 16 + 64 80

y = ± . We have f (0, 3) = 4, f (0, −3) = 16, f , =f ,− = = < 4. So the
5 5 5 5 5 25 25
9 8 9 8

points which are nearest from the point (1, 0) are , and , − , and the point which is
5 5 5 5
farthest from the point (1, 0) is (0, −3).
Question 8. Determine the maximum and minimum values of the function f (x, y, z) on the set
S ⊂ R3 where:
a) f (x, y, z) = 3x + 2y + z S = {(x, y, z) ∈ R2 : x2 + y 2 + z 2 = 1}
b) f (x, y, z) = x2 + y 2 + z 2 S = {(x, y, z) ∈ R2 : 3x + 2y + z = 6}
c) f (x, y, z) = xyz S = {(x, y, z) ∈ R2 : x2 + y 2 + z 2 = 1}
339
d) f (x, y, z) = xyz S = {(x, y, z) ∈ R2 : x2 + y 2 + z 2 = 1, x + y + z = 1}
e) f (x, y, z) = xyz S = {(x, y, z) ∈ R2 : x2 + y 2 + z 2 ¬ 1, x + y + z = 1}
f ) f (x, y, z) = x2 + y 2 + z 2 S = {(x, y, z) ∈ R2 : x + y + z = 1, x + 2y + 3z = 6}
Answer: a) We need to determine the maximum and minimum values of the function
f (x, y, z) = 3x + 2y + z subject to the constraint g(x, y, z) = x2 + y 2 + z 2 = 1. Using Lagrange
multipliers, we solve the equations ∇f = λ∇g and g(x, y, z) = 1, which can be written as
or as 


 3 = 2xλ



2 = 2yλ


1= 2zλ







 2
x + y2 + z2 = 1

From the first three equations, we get

3 1 1
λ= = =
2x y 2z
Hence y = 2z and x = 3z. Putting this in the fourth equation, we get 14z 2 = 1. This leads to the
! !
3 2 1 3 2 1
solutions √ , √ , √ and − √ , − √ , − √ .
14 14 14 14 14 14
√
!
3 2 1 14
We get the maximum f √ , √ , √ = √ = 14
14 14 14 14
√
!
3 2 1 14
and the minimum f − √ , − √ , − √ = − √ = − 14.
14 14 14 14
b) We need to determine the maximum and minimum values of the function f (x, y, z) = x2 +y 2 +z 2
subject to the constraint g(x, y, z) = 3x + 2y + z = 6. Using Lagrange multipliers, we solve the
equations ∇f = λ∇g and g(x, y, z) = 6, which can be written as
or as 


2x = 3λ



2y = 2λ


2z =λ









3x + 2y + z = 6
2x
λ= = y = 2z
3
340
Hence y = 2z and x = 3z. Putting this in the fourth equation, we get 14z = 6. This leads to the
9 6 3

solution , , .
7 7 7
9 6 3 126

We get the minimum f , , = .
7 7 7 49
There is no maximum, since in the plane g(x, y, z) = 3x + 2y + z = 6, there is no limit as to how far
away points can be from the origin.
c) We need to determine the maximum and minimum values of the function f (x, y, z) = xyz subject
to the constraint g(x, y, z) = x2 + y 2 + z 2 = 1. Using Lagrange multipliers, we solve the equations
or as 


 yz = 2xλ



xz = 2yλ


xy = 2zλ







 2
x + y2 + z2 = 1


yz xz xy
2λ = = =
x y z
unless one of x, y and z is zero. So unless one of x, y and z is zero, we get x2 = y 2 = z 2 , and then
! ! !
1 1 1 1 1 1 1 1 1
from the fourth equation we get the solutions √ , √ , √ , −√ , √ , √ , √ , −√ , √ ,
3 3 3 3 3 3 3 3 3
! ! ! !
1 1 1 1 1 1 1 1 1 1 1 1
√ , √ , −√ , −√ , −√ , √ , −√ , √ , −√ , √ , −√ , −√ ,
3 3 3 3 3 3 3 3 3 3 3 3
! !3 !3
1 1 1 1 1 1 1
− √ , − √ , − √ . We get the maximum √ = √ and the minimum − √ =− √ .
3 3 3 3 3 3 3 3 3
d) We need to determine the maximum and minimum values of the function f (x, y, z) = xyz
subject to the constraints x2 + y 2 + z 2 = 1 and x + y + z = 1. We can use z = 1 − x − y to get
f (x, y) = xy(1 − x − y) = xy − x2 y − xy 2 and the constraint equation
g(x, y) = x2 + y 2 + (1 − x − y)2 = 1 + 2x2 + 2y 2 − 2x − 2y + 2xy = 1. Using Lagrange multipliers,
we solve the equations ∇f = λ∇g and g(x, y) = 1, which can be written as
or as 


 y − 2xy − y 2 = λ (4x − 2 + 2y)



x − x2 − 2xy = λ (4y − 2 + 2x)


1 + 2x2 + 2y 2 − 2x − 2y + 2xy = 1


From the first two equations, we get

y − 2xy − y 2 x − x2 − 2xy
λ= =
4x − 2 + 2y 4y − 2 + 2x
341
unless 4x − 2 + 2y = 0 or 4y − 2 + 2x = 0. So if 4x − 2 + 2y 6= 0 and 4y − 2 + 2x 6= 0, we have
y − 2xy − y 2 1 − 2x − y 1 − 2x − y y
λ= =y =y =−
4x − 2 + 2y 4x − 2 + 2y −2(−2x + 1 − y) 2
x − x2 − 2xy 1 − x − 2y 1 − x − 2y x
= =x =x =−
4y − 2 + 2x 4y − 2 + 2x −2(−2y + 1 − x) 2
Thus x = y and from the the third equation, we get
1 + 2x2 + 2x2 − 2x − 2x + 2xx = 1
6x2 − 4x = 0
x(6x − 4) = 0
2
So x = 0 or x = , and using x = y and z = 1 − x − y, we get the solutions (0, 0, 1) and
3
2 2 1

, ,− .
3 3 3
If 4x − 2 + 2y = 0, the equation
y − 2xy − y 2 = λ (4x − 2 + 2y)
holds for any λ, since both sides contain the factor 2x − 1 + y = 0. From 2x − 1 + y = 0, we get
y = 1 − 2x and we will put this in the equation 1 + 2x2 + 2y 2 − 2x − 2y + 2xy = 1 .
1 + 2x2 + 2(1 − 2x)2 − 2x − 2(1 − 2x) + 2x(1 − 2x) = 1

1 + 2x2 + 2 1 − 4x + 4x2 − 2x − 2 + 4x + 2x − 4x2 = 1
1 + 2x2 + 2 − 8x + 8x2 − 2x − 2 + 4x + 2x − 4x2 = 1

6x2 − 4x = 0
x(3x − 2) = 0
2
So x = 0 or x = , and using y = 1 − 2x and z = 1 − x − y, we get the solutions (0, 1, 0) and
3
2 1 2

,− , .
3 3 3
1 2 2

Similarly, if 4y − 2 + 2x = 0, we get the solutions, (1, 0, 0) and − , , .
3 3 3
−1 2 2 4
So the maximum is 1 · 0 · 0 = 0 and the minimum is · · =− .
3 3 3 9
e) We need to determine the maximum and minimum values of the function f (x, y, z) = xyz subject
to the constraints x2 + y 2 + z 2 ¬ 1 and x + y + z = 1. In part d), we found the extrema at the
intersection of the surfaces determined by the equations x2 + y 2 + z 2 = 1 and x + y + z = 1.
Now we are going to find the extrema of the function f (x, y, z) = xyz subject to the constraint
g(x, y, z) = x + y + z = 1 and check whether we get values for x, y and z which satisfy the inequality
x2 + y 2 + z 2 ¬ 1. Using Lagrange multipliers, we solve the equations ∇f = λ∇g and g(x, y, z) = 1,
342
or as 
yz

 =λ



xz =λ


xy =λ









x+y+z =1
From the first two equations, we get yz = xz which implies y = x or z = 0. If z = 0, λ = 0
and this implies that one of x an y is zero as well. From the second and the third equations, we get
xz = xy which implies z = y or x = 0. If x = 0, λ = 0 and this implies that one of z an y is
1 1 1

zero as well. Then using the third equation, we get the solutions , , , (0, 1, 0), (1, 0, 0), and
3 3 3
4
(0, 0, 1). These are points where local extrema occur, and we already have the minimum − from
9
1 1 1 1
part d). This is still the minimum, and we get the maximum · · = .
3 3 3 9
f) We need to determine the maximum and minimum values of the function f (x, y, z) = x2 + y 2 + z 2
subject to the constraints x + y + z = 1 and x + 2y + 3z = 6. We can use z = 1 − x − y to get
f (x, y) = xy(1 − x − y) = x2 + y 2 + (1 − x − y)2 and the constraint equation
g(x, y) = x + 2y + 3(1 − x − y) = −2x − y + 3 = 6. Using Lagrange multipliers, we solve the equations
∇f = λ∇g and g(x, y) = 6, which can be written as
or as 
2x − 2(1 − x − y)

 = −2λ



2y − 2(1 − x − y) = −λ




−2x − y + 3 = 6
From the first two equations, we get
x − (1 − x − y) = −λ = 2y − 2(1 − x − y)
1
Thus x = 2y −(1−x−y) = 2y −1+x+y = 3y −1+x, hence y = . Then from the third equation, we
3
2 2
5 5 1 −5 1 −5 1 2 25 + 1 + 49 75

get x = − . So the minimum is f − , = + + 1− − = = .
3 3 3 3 3 3 3 9 9
Question 9. Using the Kuhn-Tucker theorem, find the maximum value of the function
f (x, y) = x + ay on the set M = {(x, y) ∈ R2 : x2 + y 2 ¬ 1, x + y 0} (the coefficient a is a fixed
parameter).
Answer: We need to find the maximum value of the function f (x, y) = x + ay, subject to the
additional constraints g(x, y) = 1 − x2 − y 2 0 and h(x, y) = x + y 0. Using Lagrange multipliers,
we solve ∇f = λ∇g + µ∇h, λg(x, y) = 0, µh(x, y) = 0, g(x, y) 0, h(x, y) 0, λ ¬ 0, and µ ¬ 0
fx = λgx + µhx fy = λgy + µhy λg(x, y) = 0 µh(x, y) = 0
343
g(x, y) 0 h(x, y) 0 λ¬0 µ¬0
or as 




1 = −2xλ + µ


a = −2yλ + µ








2 2
λ 1 − x − y =0








µ(x + y) = 0


1 − x2 − y 2 0







x+y 0








λ ¬ 0







µ ¬ 0

Subtracting the second equation from the first, we get 1 − a = (−2x + 2y)λ, thus
1−a
λ=
−2x + 2y
unless x = y. Subtracting x times the second equation from y times the first, we get y−xa = yµ−xµ,
thus
y − xa
µ=
y−x
1−a
unless x = y. We assume first that x 6= y. Then from λ = and the equation
−2x + 2y

λ 1 − x2 − y 2 = 0, we get

(1 − a) 1 − x2 − y 2 = 0
y − xa
and from µ = and the equation µ(x + y) = 0, we get
y−x
(y − xa)(x + y) = 0
If 1 − a= 0, then λ = 0 and from the first equation we get µ = 1 > 0, a contradiction. So from
(1 − a) 1 − x2 − y 2 = 0 we get 1 − x2 − y 2 = 0. From (y − xa)(x + y) = 0, we get y = xa or
x = −y. We will deal with the cases x = y and x = −y later. Putting y = xa in 1 − x2 − y 2 = 0,
we get
1 − x 2 − y 2 = 1 − x 2 − x 2 a2 = 1 − 1 + a2 x 2 = 0
1 a
which gives the possible solution x = √ 2
, y=√ , and also the possible solution
1+a 1 + a2
−1 −a
x= √ 2
, y= √ . The second solution will not give a positive value of f (x, y) = x + ay,
1+a 1 + a2
so we only keep the first solution. If x = y, from the first two equations, we get a = 1. And since
λ cannot be zero, from the third equation we get 1 = 2x2 , which together with x + y 0 implies
1 1 a
x=y= √ = √ = √ . If x = −y, from the first two equations, we get 1 + a = 2µ,
2 1 + a2 1 + a2
and since µ ¬ 0, this implies a ¬ −1. From the first two equations, we also get a − 1 = −4yλ,
344
and since λ ¬ 0 and a ¬ −1, we get y < 0. Since λ cannot be zero, from the third equation we get
1 1
1 = 2x2 , which together with y < 0 implies x = √ and y = − √ . In conclusion, if a > −1, the
2 2
1+a 2 √
maximum of f (x, y) = x + ay is √ = 1 + a2 . If a ¬ −1, the maximum of f (x, y) = x + ay
1 + a2
√ 1−a √
is the biggest of 1 + a2 and √ . If a = −1, the two expressions are equal, otherwise 1 + a2
2
√
is bigger. So the maximum is 1 + a2 for all a.
Academic Year 2015/16, Colloquium II, 3 June, 2016

Question 1. Let 
xy

 if x2 + y 2 6= 0
x2 + y2

f (x, y) = 
0 if x = y = 0

Calculate grad f (0, 0) or state that it does not exist.

Answer: The gradient of f at the point (0, 0) is defined as the vector whose dot product with any
unit vector u is the directional derivative of f at (0, 0) in the direction of u. That is,
(grad f (0, 0)) · u = du f (0, 0)
By definition, the directional derivative of f at a point (x0 , y0 ) in the direction of a unit vector
u = (a, b) is
f (x0 + ah, y0 + bh) − f (x0 , y0 )
du f (x0 , y0 ) = lim
h→0 h
if this limit exists.
For u = (1, 0), we have
(0+h)0
∂f f (0 + h, 0) − f (0, 0) (0+h)2 +02
−0 0
0
h2
du f (0, 0) = (0, 0) = lim = lim = lim = lim = lim 0 = 0
∂x h→0 h h→0 h h→0 h h→0 h3 h→0
So if grad f (0, 0) existed, its first component would have to be zero.
For u = (0, 1), we have
0(0+h)
∂f f (0, 0 + h) − f (0, 0) 02 +(0+h)2
−0 0
0
h2
du f (0, 0) = (0, 0) = lim = lim = lim = lim = lim 0 = 0
∂y h→0 h h→0 h h→0 h h→0 h3 h→0
So if grad f (0, 0) existed, its second component would have to be zero.

!
1 1
For u = √ , √ , we have
2 2

0+ √1 h 0+ √1 h
2 2
2 2 −0
f 0+ √1 h, 0 + √1 h − f (0, 0) 0+ √1 h + 0+ √1 h
2 2 2 2
du f (0, 0) = lim = lim
h→0 h h→0 h
345

√1 h √1 h
2 2
2 2 h2
2
√1 h + √1 h h2 2 1 1
2 2 2
+ h2 1+1 2 1
= lim = lim = lim = lim = lim = +∞
h→0 h h→0 h h→0 h h→0 h h→0 2h
!
1 1
The directional derivative of f at (0, 0) in the direction of u = √ , √ is not zero, and since
2 2
the gradient of f at (0, 0), if it existed, would have first component zero and second component
zero, this implies that the gradient of f at (0, 0) does not exist.
Question 2. Determine Df (x, y) where f = h ◦ g,

g(x, y) = (x + y, x − y), h(u, v) = u2 + v 2 , uv and x, y, u, v ∈ R

f (x, y) = (x + y)2 + (x − y)2 , (x + y)(x − y)

= x2 + 2xy + y 2 + x2 − 2xy + y 2 , x2 − y 2 = 2x2 + 2y 2 , x2 − y 2
∂f1 ∂ 2
= 2x + 2y 2 = 4x
∂x ∂x
∂f1 ∂ 2
= 2x + 2y 2 = 4y
∂y ∂y
∂f2 ∂ 2
= x − y 2 = 2x
∂x ∂x
∂f2 ∂ 2
= x − y 2 = −2y
∂y ∂y
Since f = h ◦ g has two components and is a function of two variables, we get the 2 × 2 matrix
 
∂f1 ∂f1  
 ∂x

∂y 
 4x 4y
Df (x, y) =  =
   
 

 ∂f2 ∂f2 
 2x −2y
∂x ∂y
Question 3. Let f = x2 + y. Find the maximum of f on the set

( )
x2 y 2 2
A = (x, y) ∈ R : + ¬1
4 9
Answer: The set A corresponds to a region whose boundary is an ellipse centered at the origin. The
maximum must occur at the boundary, as otherwise one could further increase the absolute value of
one of x and y while keeping the other one fixed and thus increase the value of the function. So we
x2 y 2
need to find the maximum of the function f = x2 +y subject to the constraint g(x, y) = + = 1.
4 9
Using Lagrange multipliers, we solve the equations ∇f = λ∇g and g(x, y) = 1, which can be written
346
as
or as  x
 2x =λ
2







 2y
1= λ



9

2
x y2



+
=1

4 9
If x = 0, from the third equation we get that y = 3 or y = −3, so in that case the value of the
function cannot be bigger than 3. If x 6= 0, from the first equation we get λ = 4 and then from
s
9 9 4(64 − 9)
the second equation, we get y = . Using y = , from the last equation we get x = ± .
8 8 64
Thus the maximum is
s   s 
4(64 − 9) 9  4(64 − 9) 9  4(64 − 9) 9 2 · 55 9 110 + 36 146 73
f , = f − , = + = + = = =
64 8 64 8 64 8 32 8 32 32 16
n o
Question 4. Let K = (x, y) ∈ R2 : |x| + |y| ¬ 1 and let f (x, y) = xy. Find the maximum and
the minimum of f on the set K.
Answer: The maximum must occur at the boundary, as otherwise one could further increase the
absolute value of one of x and y while keeping the other one fixed and thus increase the value of
the function. By symmetry with respect to the origin, we will get two points where the maximum
occurs, one of them lies on the line y = 1 − x. We will find the x coordinate of this point by taking
the derivative of f (x, 1 − x) = x − x2 and putting it equal to zero. Thus 1 − 2x = 0 and we get the
1 1 1 1 1

maximum f , = f − ,− = . By symmetry, we get the minimum
2 2 2 2 4
1 1 1 1 1

f − , =f ,− =− .
2 2 2 2 4
 
−1 0 2
 
 
Question 5. Decide whether the quadratic form given by the matrix 
 0 −2 0 
 is:
 
2 0 −3
positive definite, positive semidefinite, negative definite, negative semidefinite, or indefinite.
Answer: A quadratic form given by a matrix is positive definite if all eigenvalues of the matrix are
positive, positive semidefinite if all eigenvalues of the matrix are non-negative, negative definite if
all eigenvalues of the matrix are negative, negative semidefinite if all eigenvalues of the matrix are
non-positive, indefinite if the matrix has eigenvalues of both signs.
We need to find the eigenvalues of the matrix. The characteristic polynomial is
−1 − λ 0 2
0 −2 − λ 0 =
2 0 −3 − λ
347
−2 − λ 0 0 0 0 −2 − λ
= (−1 − λ) −0 +2
0 −3 − λ 2 −3 − λ 2 0
= (−1 − λ)[(−2 − λ)(−3 − λ) − 0 · 0] − 0 + 2[0 · 0 − 2(−2 − λ)]

= (−1 − λ) 6 + 5λ + λ2 + 4λ + 8 = −6 − 5λ − λ2 − 6λ − 5λ2 − λ3 + 4λ + 8

= −λ3 − 6λ2 − 7λ + 2 = (λ + 2) −λ2 − 4λ + 1
The characteristic polynomial is zero if λ = −2, so −2 is an eigenvalue. This eigenvalue was found
by guessing, this often works for third degree polynomials, trying with small integers. We need to
factor −λ2 − 4λ + 1 to find the remaining eigenvalues.
 
h i
0 = −λ2 − 4λ + 1 = − λ2 + 4λ − 1 = − λ 2
4λ + 4} −5 = − (λ + 2)2 − 5
+ {z
 
|
complete square
5 = (λ + 2)2
√ √
λ = 5 − 2 or − 5 − 2
√ √
We know that 2 < 5, and thus 5 − 2 > 0. Since the matrix has positive as well as negative
eigenvalues, the quadratic form given by the matrix is indefinite.
Question 6. Find all the points at which the function f : R2 → R, given by
f (x, y) = 2x4 + y 4 − x2 − y 2
has a local maximum or a local minimum.

Answer: We need to find the critical points, so we calculate the first order partial derivatives and
∂f ∂ ∂ 4
= f= 2x + y 4 − x2 − y 2 = 8x3 − 2x = 2x 4x2 − 1 = 0
∂x ∂x ∂x
∂f ∂ ∂ 4
= f= 2x + y 4 − x2 − y 2 = 4y 3 − 2y = 2y 2y 2 − 1 = 0
∂y ∂y ∂y
1 1
From the first equation, we get x = 0, x = , or x = − , and from the second equation we
2 2 ! !
1 1 1 1 1

get y = 0 , y = √ , or y = √ . So the critical points are (0, 0), 0, √ , 0, − √ , ,0 ,
2 2 2 2 2
! ! ! !
1 1 1 1 1 1 1 1 1

− ,0 , ,√ , − ,√ , , − √ , and − , − √ . We calculate the second order
2 2 2 2 2 2 2 2 2
partial derivatives, to understand why, see the explanation at the beginning of the answer to Question
6 of Temat 23 on page 293.
∂ 2f ∂ ∂f ∂ 3
fxx = = = 8x − 2x = 24x2 − 2
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂ 3
fyx = = = 4y − 2y = 0
∂x∂y ∂x ∂y ∂x
348
∂ 2f ∂ ∂f ∂ 3
fxy = = = 8x − 2x = 0
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂ 3
fyy = = = 4y − 2y = 12y 2 − 2
∂y∂y ∂y ∂y ∂y
D(0, 0) = fxx (0, 0)fyy (0, 0) − [fxy (0, 0)]2 = (−2) · (−2) − 0 = 4
this local maximum is f (0, 0) = 0.
! ! ! !
1 1 1 1
We have fxx 0, √ = −2, fyx 0, √ = fxy 0, √ = 0, and fyy 0, √ = 4. So
2 2 2 2
! ! ! " !#2
1 1 1 1
D 0, √ = fxx 0, √ fyy 0, √ − fxy 0, √ = (−2) · 4 − 0 = −8
2 2 2 2
! !
1 1
Since D 0, √ = −8 < 0, there is a saddle point at 0, √ [see the explanation at the beginning
2 2
of the answer to Question 6 of Temat 23 on page 293]. The coordinates of this saddle point are
! !
1 1 1
0, √ , − . By symmetry, there is also a saddle point at 0, − √ . The coordinates of this
2 4 2
!
1 1
saddle point are 0, − √ , − .
2 4
1 1 1 1

We have fxx , 0 = 4, fyx , 0 = fxy , 0 = 0, and fyy , 0 = −2. So
2 2 2 2
2
1 1 1 1

D , 0 = fxx , 0 fyy , 0 − fxy ,0 = 4 · (−2) − 0 = −8
2 2 2 2
1 1

Since D , 0 = −8 < 0 there is a saddle point at , 0 [see the explanation at the beginning
2 2
1 1 1

, 0, − . By symmetry, there is also a saddle point at − , 0 . The coordinates of this saddle
2 8 2
1 1

point are − , 0, − .
2 8
! ! ! !
1 1 1 1 1 1 1 1
We have fxx ,√ = 4, fyx ,√ = fxy ,√ = 0, and fyy ,√ = 4. So
2 2 2 2 2 2 2 2
! ! ! " !#2
1 1 1 1 1 1 1 1
D ,√ = fxx , √ fyy , √ − fxy ,√ = 4 · 4 − 0 = 16
2 2 2 2 2 2 2 2
! ! !
1 1 1 1 1 1
Since D ,√ = 16 > 0 and fxx ,√ = 4 > 0, there is a local minimum at ,√ [see
2 2 2 2 2 2
the explanation at the beginning of the answer to Question 6 of Temat 23 on page 293]. The value of
349
!
1 1 3
this local minimum is f ,√ = − . By symmetry, there are local minima with the same value
2 2 8
! ! !
1 1 1 1 1 1
at − ,√ , , − √ , and − , − √ .
2 2 2 2 2 2
Question 7. Let f : (0, 1) × (0, 1) → Ω ⊂ R2 be the function defined by

u 3u
r r
3
f (u, v) = ,v .
v v
The boundary of the image Ω is given by three curves described by equations of the form y = h(x)
if we use (x, y) to denote coordinates on the image. What are the three equations? Give a sketch of
Ω. Is f a diffeomorphism?
Answer: The first thing to notice is that
 s 
u 3u u uv 3  u √
r r r r
3 3
(x, y) = f (u, v) = 3
,v =3 , = 3
, uv 2 .
v v v v v
The boundary of (0, 1) × (0, 1) consists of four line segments.
The segment at which u has constant value 1 (blue), although not part of the domain, would get
mapped to s 
3 1
√
3
(x, y) =  , v2 .
v
1
Thus y = h(x) = . The segment at which v has constant value 1 (green), although not part of the
x2
domain, would get mapped to √ √
(x, y) = 3 u, 3 u .
Thus y = h(x) = x. The segment at which u has constant value zero (yellow), although not part of
the domain, would get mapped to
(x, y) = (0, 0) ,
this segment would get collapsed to a point. The segment at which v has constant value zero (red),
although not part of the domain, would get mapped to the line y = h(x) = 0. This calls for an
√
3
r
u
explanation. For any finite number u, we have lim 2
ûv = 0 . But the limit lim 3
does
(û,v)→(u,0) (u,v)→(0,0) v
s
u0
not exist, for any given positive number x we can find (u0 , v0 ) such that x = 3 and (u0 , v0 ) is
v0
350
as close as we wish to (0, 0).
We have  s 
u 3u u uv 3  u √
r r r r
3 3
(x, y) = f (u, v) = 3
,v =3 , = 3
, uv 2 ,
v v v v v
s
y3
Thus x3 v = u and y 3 = uv 2 . From these two equations we get x3 = u and thus
u
3
2 y
u = x3 y 2 3
= x2 y. And we get y 3 = x3 v 3 and thus v = . Since f is differentiable and
x
y

−1 2
f (x, y) = x y,
x
is differentiable, f is a diffeomorphism.
Question 8. Let y(x) be determined by the equation yx3 + xy 3 − y 3 − 1 = 0. Calculate y(1) and
y 0 (1).
Answer: Putting x = 1 in the equation yx3 + xy 3 − y 3 − 1 = 0, we get y + y 3 − y 3 − 1 = y − 1 = 0.
Thus y(1) = 1. Let z(x, y) = yx3 + xy 3 − y 3 − 1. We have z(x, y) = 0 and using the chain rule, we
get
dz ∂z dx ∂z dy ∂z ∂z dy ∂z ∂z dy
= + = ·1+ = + =0
dx ∂x dx ∂y dx ∂x ∂y dx ∂x ∂y dx
∂z dy ∂z
=−
∂y dx ∂x
∂z
dy ∂x 3yx2 + y 3 − 0 − 0 3yx2 + y 3
= − ∂z = − 3 = −
dx ∂y
x + 3xy 2 − 3y 2 − 0 x3 + 3xy 2 − 3y 2
Since we know that y(1) = 1, we get
dy 3 · 1 · 12 + 13 3+1
y 0 (1) = (1) = − 3 =− = −4
dx 1 +3·1·1 −3·12 2 1+3−3
Question 9. Find an equation of the plane normal to the manifold

n o
M = (x, y, z) ∈ R3 : x2 + y 2 = 4, x + y + z = 0
351
√ √ √
at the point 2, 2, −2 2 .
Answer: We will use the gradient to get a vector which is perpendicular to a level surface of the
function F1 = x2 + y 2
!
∂F1 ∂F1 ∂F1
grad F1 = ∇F1 = , , = (2x, 2y, 0)
∂x ∂y ∂z
√ √ √ √ √
We have grad F1 2, 2, −2 2 = 2 2, 2 2, 0 . We will use the gradient to get a vector which
is perpendicular to a level surface of the function F2 = x + y + z
!
∂F2 ∂F2 ∂F2
grad F2 = ∇F2 = , , = (1, 1, 1)
∂x ∂y ∂z
√ √ √
We have grad F2 2, 2, −2 2 = (1, 1, 1). The plane which is perpendicular to the manifold at
√ √ √ √ √ √
the point 2, 2, −2 2 must contain the point the point 2, 2, −2 2 and is spanned by the
√ √ √
vectors 2 2, 2 2, 0 = 2 2(1, 1, 0) and (1, 1, 1). So a parametric equation of this plane is given by
√ √ √
(x, y, z) = 2 + s + t, 2 + s + t, −2 2 + t
A Cartesian equation describing this plane is x − y = 0.

Question 10. Give an example of a non-zero tangent vector to the manifold
n o
M = (x, y, z) ∈ R3 : x2 + y 2 = 4, x + y + z = 0
√ √ √
at the point 2, 2, −2 2 .
Answer: In question 9, we found a Cartesian equation of the plane perpendicular to the manifold at
√ √ √
the point 2, 2, −2 2 . The plane is given by the equation x − y = 0, so the vector (1, −1, 0)
√ √ √
is perpendicular to this plane and thus tangent to the manifold at the point 2, 2, −2 2 . A
tangent vector can also be obtained using the cross product of two spanning vectors of the plane
perpendicular to the manifold
i j k
1 0 1 0 1 1
v = (1, 1, 0) × (1, 1, 1) = 1 1 0 =i −j +k
1 1 1 1 1 1
1 1 1
= i[1 · 1 − 1 · 0] − j[1 · 1 − 1 · 0] + k[1 · 1 − 1 · 1]

= i − j = (1, −1, 0)
Academic Year 2016/17, Colloquium II, 1 June, 2017

Question 1. Is there a C 2 function f : R2 → R such that
352
∂f ∂f
= x sin y and = y cos x
∂x ∂y
Answer: If
∂f ∂f
∂x ∂y
then we would have
∂ 2f ∂ ∂f ∂
= = (x sin y) = x cos y
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂
= = (y cos x) = −y sin x
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ 2f
So we would have 6= , and thus there is no C 2 function f : R2 → R such that
∂y∂x ∂x∂y
∂f ∂f
∂x ∂y
Question 2. Let f (x, y) = x + y 2 . Find the maximum of f on the set

n o
A = (x, y) ∈ R2 : 3x2 + 2y 2 + 8x ¬ 1 .
Answer: We have !2
2 2
√ 4 16
3x + 2y + 8x = 3x + √ − + 2y 2
3 3
So the region A is bounded by an ellipse. The maximum of f (x, y) = x + y 2 must occur on the
boundary, as otherwise one could increase x or increase the absolute value of y and thus increase
the value of f (x, y) = x + y 2 . We need to find x and y such that f (x, y) = x + y 2 reaches an
extremum, subject to the constraint g(x, y) = 3x2 + 2y 2 + 8x = 1. Using Lagrange multipliers, we
or as 


 1 = (6x + 8)λ



2y = 4yλ


3x2 + 2y 2 + 8x = 1


1
From the second equation, we get y = 0 or λ = . If y = 0 , the biggest that f (x, y) can get on A
2
is q q q
16 √4 16 √4
1 + 16 − √4 1 + 3
− 1 + 3
+ 1
√3 3
= √ q
3
3
= √ q <1
3 3 1 + 16 + √4 3 3
3 1 + 16 + √4 3 3
1
If λ = , from 1 = (6x + 8)λ, we get x = −1. Then from 3x2 + 2y 2 + 8x = 1, we get y 2 = 3. Thus
2
353
the maximum is √ √
f −1, 3 = f −1, − 3 = −1 + 3 = 2
Question 3. Let K = {(x, y) ∈ R2 : |x| ¬ 1 and |y| ¬ 1} and let f (x, y) = x2 + y 2 + xy. Find the
maximum and the minimum of f on the set K.
Answer: K is bounded by a square, and the maximum of f (x, y) = x2 + y 2 + xy occurs in the upper
right corner and the lower left corner. The maximum is
f (1, 1) = f (−1, −1) = 12 + 12 + 1 · 1 = 3

2
y 3 2

2 2
We have x + y + xy = x + + y 0. Thus the minimum is
2 4
f (0, 0) = 02 + 02 + 0 · 0 = 0
Question 4. Does the function f : R2 → R, given by
f (x, y) = 2x4 − y 4 − x2 − y 2
have saddle points ? If so, then answer yes and write down the saddle points, otherwise answer no.
∂f ∂ ∂ 4
= f= 2x − y 4 − x2 − y 2 = 8x3 − 2x = 2x 4x2 − 1 = 0
∂x ∂x ∂x
∂f ∂ ∂ 4
= f= 2x − y 4 − x2 − y 2 = −4y 3 − 2y = −2y 2y 2 + 1 = 0
∂y ∂y ∂y
1 1
From the first equation, we get x = 0, x = , or x = − , and from the second equation we get
2 2
1 1

y = 0. So the critical points are (0, 0), , 0 and − , 0 . We calculate the second order partial
2 2
derivatives, to understand why, see the explanation at the beginning of the answer to Question 6 of
Temat 23 on page 293.
∂ 2f ∂ ∂f ∂ 3
fxx = = = 8x − 2x = 24x2 − 2
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂
fyx = = = −4y 3 − 2y = 0
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ ∂f ∂ 3
fxy = = = 8x − 2x = 0
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂
fyy = = = −4y 3 − 2y = −12y 2 − 2
∂y∂y ∂y ∂y ∂y
D(0, 0) = fxx (0, 0)fyy (0, 0) − [fxy (0, 0)]2 = (−2) · (−2) − 0 = 4
354
this local maximum is f (0, 0) = 0.
1 1 1 1

We have fxx , 0 = 4, fyx , 0 = fxy , 0 = 0, and fyy , 0 = −2. So
2 2 2 2
2
1 1 1 1

D , 0 = fxx , 0 fyy , 0 − fxy ,0 = 4 · (−2) − 0 = −8
2 2 2 2
1 1

Since D , 0 = −8 < 0 there is a saddle point at , 0 [see the explanation at the beginning
2 2
1 1 1

, 0, − . By symmetry, there is also a saddle point at − , 0 . The coordinates of this saddle
2 8 2
1 1

point are − , 0, − .
2 8

Question 5. Let f (r, θ) = r2 cos(2θ), r2 sin(2θ) where (r, θ) ∈ (0, 1) × (0, 2π). Sketch the image
of f .
Answer: In the sketch, the coordinates are (x, y) = f (r, θ).
355
Question 6. Let f be the function defined in question 6. Is f a local diffeomorphism of the set
(0, 1) × (0, 2π) ? Is f a diffeomorphism of the set (0, 1) × (0, 2π) ?

Answer: The function f (r, θ) = r2 cos(2θ), r2 sin(2θ) is not invertible if the domain is
(0, 1) × (0, 2π), because with this domain, it is not injective. In fact the blue region in the answer to
question 5 gets colored twice: The radius, r2 , goes from zero to one (both extremities not included),
and the angle, 2θ, goes from zero to 4π (both extremities not included), thus twice around. Since f
is not invertible, it is not a diffeomorphism. But around every point in (0, 1) × (0, 2π), we can find
a neighbourhood such that if we restrict the domain to this neighbourhood, f is invertible. So f is
locally invertible if the domain is (0, 1) × (0, 2π). The function f is continuous and differentiable.
The local inverse at any particular point in the set (0, 1) × (0, 2π) is continuous and differentiable
on a neighbourhood around that point. Thus f is a local diffeomorphism of the set (0, 1) × (0, 2π).
Question 7. Let z(x, y) be determined by the equation xexz = yz + 2y and z(2, 1) = 0. Calculate
∂z ∂z
(2, 1) and (2, 1) .
∂x ∂y
Answer: Let g(x, y, z) = xexz − yz − 2y. The function z(x, y) is determined by g(x, y, z) = 0 and
z(2, 1) = 0.
∂g ∂g
= exz + xzexz (2, 1, 0) = 1
∂x ∂x
∂g ∂g
= −z − 2 (2, 1, 0) = −2
∂y ∂y
∂g ∂g
= x2 exz − y (2, 1, 0) = 3
∂z ∂z
When computing the partial derivative of z with respect to x, we treat y as a constant. Using the
356
chain rule, we get
dg ∂g dx ∂g dz ∂g ∂g dz ∂g ∂g dz
= + = ·1+ = + =0
dx ∂x dx ∂z dx ∂x ∂z dx ∂x ∂z dx
∂g
dz
= − ∂x
∂g
dx ∂z
∂g
∂z ∂x
(x, y) = − ∂g
∂x ∂z
∂g
∂z (2, 1, 0) 1
(2, 1) = − ∂x
∂g =−
∂x ∂z
(2, 1, 0) 3
When computing the partial derivative of z with respect to y, we treat x as a constant. Using the
chain rule, we get
dg ∂g dy ∂g dz ∂g ∂g dz ∂g ∂g dz
= + = ·1+ = + =0
dy ∂y dy ∂z dy ∂y ∂z dy ∂y ∂z dy
∂g
dz
= − ∂y
∂g
dy ∂z
∂g
∂z
(x, y) = − ∂y
∂g
∂y ∂z
∂g
∂z ∂y
(2, 1, 0) −2 2
(2, 1) = − ∂g =− =
∂y ∂z
(2, 1, 0) 3 3
Question 8. Find the equation of the plane tangent to the surface
S = {(x, y, z) ∈ R3 : xexz = yz + 2y}.
at the point (2, 1, 0).

Answer: We will use the gradient to get a vector which is perpendicular to a level surface of the
function g(x, y, z) = xexz − yz − 2y
!
∂g ∂g ∂g
grad g(x, y, z) = ∇g(x, y, z) = , , = exz + xzexz , −z − 2, x2 exz − y
∂x ∂y ∂z
We have grad g (2, 1, 0) = (1, −2, 3). So for any point (x, y, z) in the tangent plane to the surface
n o
M = (x, y, z) ∈ R3 : g(x, y, z) = 0 at the point (2, 1, 0), we have
0 = (1, −2, 3) · (x − 2, y − 1, z − 0) = 1(x − 2) − 2(y − 1) + 3 (z − 0) = x − 2 − 2y + 2 + 3z − 0
an thus the equation of the tangent plane is x − 2y + 3z = 0.
357
Question 9. Is the following subset of R3 a manifold
M = {(x, y, z) ∈ R3 : (x2 + y 2 + z 2 )2 = x2 − y 2 , x2 + y 2 + z 2 6= 0}.
Answer: There is one equation to be satisfied, g(x, y, z) = (x2 + y 2 + z 2 )2 − x2 + y 2 = 0 and from

x2 + y 2 + z 2 6= 0, we have the condition that x, y, and z are not all equal to zero.
We have
!
∂g ∂g ∂g
grad g(x, y, z) = ∇g(x, y, z) = , ,
∂x ∂y ∂z

= 2(x2 + y 2 + z 2 ) · 2x − 2x, 2(x2 + y 2 + z 2 ) · 2y + 2y, 2(x2 + y 2 + z 2 ) · 2z

= 4x(x2 + y 2 + z 2 ) − 2x, 4y(x2 + y 2 + z 2 ) + 2y, 4z(x2 + y 2 + z 2 )
1
If y = 0, z = 0, and x2 = , then grad g(x, y, z) = 0 . Thus r grad g(x, y, z) = 0 does not imply
2
r = 0, and therefore M is not a manifold.
 
−1 0 2
 
 
Question 10. Decide whether the quadratic form given by the matrix 
 0 1 0 
 is:
 
2 0 1
positive definite, positive semidefinite, negative definite, negative semidefinite, or indefinite.
Answer: A quadratic form given by a matrix is positive definite if all eigenvalues of the matrix are
positive, positive semidefinite if all eigenvalues of the matrix are non-negative, negative definite if
all eigenvalues of the matrix are negative, negative semidefinite if all eigenvalues of the matrix are
non-positive, indefinite if the matrix has eigenvalues of both signs.
We need to find the eigenvalues of the matrix. The characteristic polynomial is
−1 − λ 0 2
0 1−λ 0 =
2 0 1−λ
1−λ 0 0 0 0 1−λ
= (−1 − λ) −0 +2
0 1−λ 2 1−λ 2 0
= (−1 − λ)[(1 − λ)(1 − λ) − 0 · 0] − 0 + 2[0 · 0 − 2(1 − λ)]

= (−1 − λ) 1 − 2λ + λ2 + 4λ − 4 = −1 + 2λ − λ2 − λ + 2λ2 − λ3 + 4λ − 4

= −λ3 + λ2 + 5λ − 5 = (λ − 1) −λ2 + 5
The characteristic polynomial is zero if λ = 1, so 1 is an eigenvalue. This eigenvalue was found by

guessing, this often works for third degree polynomials, trying with small integers. After factoring
358
√ √
out λ − 1, the other factor is −λ2 + 5. Thus 5 and − 5 are also eigenvalues. Since the matrix
has positive as well as negative eigenvalues, the quadratic form given by the matrix is indefinite.
Academic Year 2014/15, Examination, 8 September, 2015

Question 1. a) Calculate the indefinite integral:
Z
cos(ln x) dx
b) Calculate the improper integral

Z ∞
1
dx
1 x (1 + (ln x)2 )
Answer: a) Using integration by parts, we get

Z Z Z
1
cos(ln x) dx = 1 · cos(ln x) dx = x cos(ln x) − x(− sin(ln x)) dx
x
Z Z
1
= x cos(ln x) + 1 · sin(ln x) dx = x cos(ln x) + x sin(ln x) − x cos(ln x) dx
x
Z
= x cos(ln x) + x sin(ln x) − cos(ln x) dx
Z
2 cos(ln x) dx = x cos(ln x) + x sin(ln x)
Z
x
cos(ln x) dx = [cos(ln x) + sin(ln x)] + C
2
for an arbitrary constant C.
Z
1
b) We will first calculate the indefinite integral dx. We know that
x (1 + (ln x)2 )
d 1 d 1
arc tg u = 2
and ln x = . Thus
du 1+u dx x
Z
1
dx = arc tg(ln x) + C
x (1 + (ln x)2 )
for an arbitrary constant C. Now we evaluate the improper integral.

Z ∞
1

dx = [arc tg(ln x)]∞
1 = lim arc tg(ln x) − arc tg(ln 1)
1 x (1 + (ln x)2 ) x→∞
π π

= lim arc tg s − arc tg 0 = −0=
s→∞ 2 2
Question 2. Find and classify the critical points of the function f (x, y) = x3 + 3x2 − 9x + y 3 − 12y.
359
∂f ∂ ∂ 3
= f= x + 3x2 − 9x + y 3 − 12y = 3x2 + 6x − 9
∂x ∂x ∂x

= 3 x2 + 2x − 3 = 3(x − 1)(x + 3) = 0
∂f ∂ ∂ 3
= f= x + 3x2 − 9x + y 3 − 12y = 3y 2 − 12 = 3 y 2 − 4 = 0
∂y ∂y ∂y
From the first equation, we get x = 1 or x = −3, and from the second equation we get y = 2 or
y = −2. So the critical points are (1, 2), (−3, 2), (1, −2) and (−3, −2). We calculate the second
order partial derivatives, to understand why, see the explanation at the beginning of the answer to
Question 6 of Temat 23 on page 293.
∂ 2f ∂ ∂f ∂ 2
fxx = = = 3x + 6x − 9 = 6x + 6
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂ 2
fyx = = = 3y − 12 = 0
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ ∂f ∂ 2
fxy = = = 3x + 6x − 9 = 0
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂ 2
fyy = = = 3 3y − 12 = 6y
∂y∂y ∂y ∂y ∂y
We have fxx (1, 2) = 12, fyx (1, 2) = fxy (1, 2) = 0, and fyy (1, 2) = 12. So
D(1, 2) = fxx (1, 2)fyy (1, 2) − [fxy (1, 2)]2 = (12) · (12) − 0 = 144
Since D(1, 2) = 144 > 0 and fxx (1, 2) = 12 > 0, there is a local minimum at (1, 2) [see the
this local minimum is f (1, 2) = −21
We have fxx (−3, 2) = −12, fyx (−3, 2) = fxy (−3, 2) = 0, and fyy (−3, 2) = 12. So
D(−3, 2) = fxx (−3, 2)fyy (−3, 2) − [fxy (−3, 2)]2 = (−12) · (12) − 0 = −144
Since D(−3, 2) = −144 < 0, there is a saddle point at (−3, 2) [see the explanation at the beginning
(−3, 2, 11).
We have fxx (1, −2) = 12, fyx (1, −2) = fxy (1, −2) = 0, and fyy (1, −2) = −12. So
D(1, −2) = fxx (1, −2)fyy (1, −2) − [fxy (1, −2)]2 = (12) · (−12) − 0 = −144
Since D(1, −2) = −144 < 0, there is a saddle point at (1, −2) [see the explanation at the beginning
(1, −2, 11).
We have fxx (−3, −2) = −12, fyx (−3, −2) = fxy (−3, −2) = 0, and fyy (−3, −2) = −12. So
D(−3, −2) = fxx (−3, −2)fyy (1, −2) − [fxy (−3, −2)]2 = (−12) · (−12) − 0 = 144
360
Since D(−3, −2) = 144 > 0, and fxx (−3, −2) = −12 < 0, there is a local maximum at (−3, −2)
[see the explanation at the beginning of the answer to Question 6 of Temat 23 on page 293]. The
value of this local maximum is f (−3, −2) = 43.
Question 3. Let g : R3 → R be given by the formula g(x, y, z) = 2y ln(z − 2) − xyz + 3x.

a) Show that there is a neighborhood of the point (x, y) = (1, 1) on which one can define a function
∂z
z = z(x, y) of class C 1 such that z(1, 1) = 3 and g(x, y, z(x, y)) = 0 . Calculate (1, 1) and
∂x
∂z
(1, 1).
∂y
b) Find the equation of the tangent plane to the surface
n o
M = (x, y, z) ∈ R3 : g(x, y, z) = 0
at the point (1, 1, 3).

Answer a) We have g(1, 1, 3) = 2 ln(1) − 3 + 3 = 0 − 3 + 3 = 0. And we have the one by one Jacobian
" #
∂g 2y

matrix J(x, y, z) = = − xy . Evaluated at the point (1, 1, 3), we have J(1, 1, 3) = [1].
∂z z−2
The Jacobian matrix is invertible, and thus one can define a function z = z(x, y) of class C 1 such
1

−1 −1
that z(1, 1) = 3 and g(x, y, z(x, y)) = 0 . We have (J(1, 1, 3)) = [1] = = [1], and we get
1
∂g
= −yz + 3
∂x
∂g
(1, 1, 3) = 0
∂x
" # " #
∂z ∂g
(1, 1) = − (J(1, 1, 3))−1 (1, 1, 3) = − (J(1, 1, 3))−1 [0] = −[1][0] = [0]
∂x ∂x
∂z
(1, 1) = 0
∂x
as well as
∂g
= 2 ln(z − 2) − xz
∂y
∂g
(1, 1, 3) = −3
∂y
" # " #
∂z ∂g
(1, 1, 3) = − (J(1, 1, 3))−1 (1, 1) = − (J(1, 1, 3))−1 [−3] = −[1][−3] = [3]
∂y ∂y
∂z
(1, 1) = 3
∂y
Using the chain rule, we can verify this result. When computing the partial derivative of z with
respect to x, we treat y as a constant.
= + = ·1+ = + =0
361
∂g
dz ∂x
= − ∂g
dx ∂z
∂g
∂z
(x, y) = − ∂x
∂g
∂x ∂z
∂g
∂z (1, 1, 3) 0
(1, 1) = − ∂x
∂g =− =0
∂x ∂z
(1, 1, 3) 1
When computing the partial derivative of z with respect to y, we treat x as a constant.

= + = ·1+ = + =0
∂g
dz
= − ∂y
∂g
dy ∂z
∂g
∂z
(x, y) = − ∂y
∂g
∂y ∂z
∂g
∂z ∂y
(1, 1, 3) −3
(1, 1) = − ∂g =− =3
∂y ∂z
(1, 1, 3) 1
b) We will use the gradient to get a vector which is perpendicular to a level surface of the function
g(x, y, z) = 2y ln(z − 2) − xyz + 3x
!
∂g ∂g ∂g 2y

grad g(x, y, z) = ∇g(x, y, z) = , , = −yz + 3, 2 ln(z − 2) − xz, − xy
∂x ∂y ∂z z−2
We have grad g(1, 1, 3) = (0, −3, 1). So for any point (x, y, z) in the tangent plane to the surface
n o
M = (x, y, z) ∈ R3 : g(x, y, z) = 0 at the point (1, 1, 3), we have
0 = (0, −3, 1) · (x − 1, y − 1, z − 3) = 0(x − 1) − 3(y − 1) + 1(z − 3) = −3y + 3 + z − 3 = −3y + z
an thus the equation of the tangent plane is −3y + z = 0.
Question 4. Let
n o
M = (x, y, z) ∈ R3 | (x − 1)2 + y 2 + z 2 = 1, x2 − y 2 = 1 .
Show that M is a one dimensional manifold and that the maximum and minimum values of f |M
where f (x, y, z) = x + y occur when z = 0 .
Answer: Each point of an n-dimensional manifold has a neighbourhood that is homeomorphic to
the Euclidean space of dimension n. One dimensional manifolds include lines and circles, but not
figure eights (because they have crossing points that are not locally homeomorphic to Euclidean
1-space). Two dimensional manifolds are also called surfaces. The plane, the sphere, and the torus
are examples of two dimensional manifolds.
362
M consists of the curve given by the intersection of a sphere and a hyperbolic sheet. This curve is a
loop, a deformed circle. So M is a one dimensional manifold.
We can verify that M is a manifold as follows.
There are two equations to be satisfied, g(x, y, z) = (x − 1)2 + y 2 + z 2 − 1 = 0 and
h(x, y, z) = x2 − y 2 − 1 = 0. We have
!
∂g ∂g ∂g
grad g(x, y, z) = ∇g(x, y, z) = , , = (2(x − 1), 2y, 2z) = (2x − 2, 2y, 2z)
∂x ∂y ∂z
and !
∂h ∂h ∂h
grad h(x, y, z) = ∇h(x, y, z) = , , = (2x, −2y, 0)
∂x ∂y ∂z
We have to show that for (x, y, z) ∈ M , the vectors (2x − 2, 2y, 2z) and (2x, −2y, 0) are linearly
independent, which amounts to showing that r(2x − 2, 2y, 2z) + s(2x, −2y, 0) = (0, 0, 0) implies
r = s = 0. From 2ry − 2sy = 0, we get r = s or y = 0. From 2rz = 0, we get r = 0 or z = 0. So
we have to consider four cases.
case 1: r = s and r = 0. Thus r = s = 0.
case 2: r = s and z = 0. Then from 2rx − 2r + 2sx = 0, we get 2rx − 2r + 2rx = 4rx − 2r =
1 1
2r(2x − 1) = 0. So r = s = 0 or x = . If x = , x2 − y 2 6= 1. Thus r = s = 0.
2 2
case 3: y = 0 and r = 0. Then from 2rx − 2r + 2sx = 0, we get 2sx = 0. Thus either x = 0 or
s = 0. If x = 0, then x2 − y 2 = 0 6= 1, so s = 0. Thus r = s = 0.
case 4: y = 0 and z = 0. From x2 − y 2 = 1, we get x = 1 or x = −1. Then (x − 1)2 + y 2 + z 2 6= 1.
So this case is not possible.
So r(2x − 2, 2y, 2z) + s(2x, −2y, 0) = (0, 0, 0) implies r = s = 0, which shows that M is a manifold.
We can verify that M has dimension one as follows.
There are two equations to be satisfied, g(x, y, z) = (x − 1)2 + y 2 + z 2 − 1 = 0 and
h(x, y, z) = x2 − y 2 − 1 = 0, and M is a subset of R3 . Thus the dimension of M is 3 − 2 = 1.
To find out at which z values the maximum and minimum values of f (x, y, z) = x + y restricted to
M can occur, we are looking for possible extrema of f (x, y, z) subject to the constraints
g(x, y, z) = (x − 1)2 + y 2 + z 2 − 1 = 0 and h(x, y, z) = x2 − y 2 − 1 = 0. Using Lagrange multipliers,
we solve ∇f = λ∇g + µ∇h, g(x, y, z) = 0, and h(x, y, z) = 0 which can be written as
fx = λgx + µhx fy = λgy + µhy fz = λgz + µhz g(x, y, z) = 1 h(x, y, z) = 1
or as 
1

 = 2(x − 1)λ + 2xµ



1 = 2yλ − 2yµ







0 = 2zλ


(x − 1)2 + y2 + z2 − 1 = 0








 2 2
x −y −1=0

363
From the third equation, we get z = 0 or λ = 0. If λ = 0, from the first two equations we get
2xµ = −2yµ, an then either µ = 0 as well which is clearly not possible, or x = −y in which case
the equation x2 − y 2 = 1 cannot hold. So z = 0.
Question 5. Find the volume of the solid that lies below the surface given by z = x2 y and lies
above the region in the xy-plane bounded by y = x + 2 and y = x2 .
Answer: The first thing we need are the x coordinates of the intersection points of the line and the
1 2 9 1 3

parabola. From x + 2 = x2 , we get x − = , thus x = ± . So x = −1 or x = 2. Between
2 4 2 2
the intersection points, the parabola lies below the line.
We get
Z x=2 Z y=x+2 Z x=2 " 2 2 #y=x+2

2 xy 1 Z x=2 h 2 2 iy=x+2
V = x y dy dx = dx = xy dx
x=−1 y=x2 x=−1 2 y=x2
2 x=−1 y=x2
1 Z x=2 2 1 Z x=2 4
2
2 2 2
= x (x + 2) − x x dx = x + 4x3 + 4x2 − x6 dx
2 x=−1 2 x=−1
" #x=2
1 x5 4x4 4x3 x7
= + + −
2 5 4 3 7 x=−1
! !
1 25 4 · 24 4 · 23 27 1 (−1)5 4 · (−1)4 4 · (−1)3 (−1)7
= + + − − + + −
2 5 4 3 7 2 5 4 3 7
!
1 25 26 25 27 1 1 4 4 1

= + + − + − + −
2 5 4 3 7 2 5 4 3 7
84 (25 + 1) + 105 (26 − 4) + 140 (25 + 4) − 60 (27 + 1)
=
840
364
21 (25 + 1) + 105 (24 − 1) + 35 (25 + 4) − 15 (27 + 1)
=
210
21 · 33 + 105 · 15 + 35 · 36 − 15 · 129 7 · 33 + 105 · 5 + 35 · 12 − 5 · 129
= =
210 70
231 + 525 + 420 − 645 531
= =
70 70
Academic Year 2015/16, Examination, 6 September, 2016

Question 1. a) Calculate the indefinite integral:
Z
(cos x)(sin x)ecos x dx.
b) Calculate the improper integral:

Z ∞
earc tg x
dx
−∞ 1 + x2
d cos x
Answer: a) We know that by the chain rule, we have e = (− sin x)ecos x . We will use integration
dx
by parts.
Z Z Z
(cos x)(sin x)ecos x dx = (− cos x)(− sin x)ecos x dx = (− cos x)ecos x − (sin x)ecos x dx
Z
= −(cos x)ecos x + (− sin x)ecos x dx = −(cos x)ecos x + ecos x
= ecos x (1 − cos x) + C
for an arbitrary constant C.

Z
earc tg x d 1
b) We will first calculate the indefinite integral 2
dx. We know that arc tg x = .
1+x dx 1 + x2
d arc tg x earc tg x
By the chain rule, e = . Thus
dx 1 + x2
Z
earc tg x
dx = earc tg x + C
1 + x2
for an arbitrary constant C. So we get
Z ∞
earc tg x 1

arc tg x ∞ π π π
h i
dx = e = lim e arc tg x
− lim e arc tg x
= e 2 − e− 2 = e 2 − π
−∞ 1 + x2 −∞ x→∞ x→∞ e2
Question 2. Let f : R2 → R be given by the formula
f (x, y) = x2 y − 2xy 2 + 4y.
365
Find all points (x, y) ∈ R2 such that ∇f (x, y) = (0, 0) and determine which of these points gives a
local minimum.
∂f ∂ ∂ 2
= f= x y − 2xy 2 + 4y = 2xy − 2y 2 = 2y(x − y) = 0
∂x ∂x ∂x
∂f ∂ ∂ 2
= f= x y − 2xy 2 + 4y = x2 − 4xy + 4 = 0
∂y ∂y ∂y
From the first equation, we get y = 0 or x = −y. If y = 0, from the second equation we get
x2 + 4 = 0 which has no solution in R. If x = y, from the second equation we get −3x2 + 4 = 0,
! !
2 2 ∂f ∂f 2 2
thus x = y = √ or x = y = − √ . So ∇f (x, y) = , = (0, 0) if (x, y) = √ , √ or
3 3 ∂x ∂y 3 3
!
2 2
(x, y) = − √ , − √ . We calculate the second order partial derivatives, to understand why, see
3 3
the explanation at the beginning of the answer to Question 6 of Temat 23 on page 293.
∂ 2f ∂ ∂f ∂
fxx = = = 2xy − 2y 2 = 2y
∂x∂x ∂x ∂x ∂x
∂ 2f ∂ ∂f ∂ 2
fyx = = = x − 4xy + 4 = 2x − 4y
∂x∂y ∂x ∂y ∂x
∂ 2f ∂ ∂f ∂
fxy = = = 2xy − 2y 2 = 2x − 4y
∂y∂x ∂y ∂x ∂y
∂ 2f ∂ ∂f ∂ 2
fyy = = = 3 x − 4xy + 4 = −4x
∂y∂y ∂y ∂y ∂y
! ! ! !
2 2 4 2 2 2 2 −4 2 2 −8
We have fxx √ ,√ = √ , fyx √ , √ = fxy √ ,√ = √ , and fyy √ , √ =√ .
3 3 3 3 3 3 3 3 3 3 3
So
! ! ! " !#2
2 2 2 2 2 2 2 2
D √ ,√ = fxx √ , √ fyy √ , √ − fxy √ , √
3 3 3 3 3 3 3 3
! ! !2
4 −8 −4 48
= √ √ − √ =−
3 3 3 3
! !
2 2 48 2 2
Since D √ , √ = − < 0, there is a saddle point at √ , √ [see the explanation at the
3 3 3 3 3
beginning of the answer to Question 6 of Temat 23 on page 293]. The coordinates of this saddle point
!
2 2 16
are √ , √ , √ .
3 3 3 3
! ! ! !
−2 −2 −4 −2 −2 −2 −2 4 −2 −2 8
We have fxx √ ,√ = √ , fyx √ , √ = fxy √ ,√ = √ , and fyy √ , √ =√ .
3 3 3 3 3 3 3 3 3 3 3
366
So
! ! ! " !#2
−2 −2 −2 −2 −2 −2 −2 −2
D √ ,√ = fxx √ , √ fyy √ , √ − fxy √ , √
3 3 3 3 3 3 3 3
! ! !2
−4 8 4 48
= √ √ − √ =−
3 3 3 3
! !
−2 −2 48 −2 −2
Since D √ , √ = − < 0, there is a saddle point at √ , √ [see the explanation at the
3 3 3 3 3
beginning of the answer to Question 6 of Temat 23 on page 293]. The coordinates of this saddle point
!
−2 −2 16
are √ , √ , − √ .
3 3 3 3
! !
2 2 2 2
So none of the points √ ,√ and −√ , −√ gives a local minimum.
3 3 3 3
Question 3. Let g : R3 → R be given by the formula

g(x, y, z) = 3z + 4xy ln z 2 + 1 + π.
a) Show that there is a neighborhood of the point (x, y) = (0, 0) on which one can define a function
π ∂z
z = z(x, y) of class C 1 such that z(0, 0) = − and g(x, y, z(x, y)) = 0. Calculate (0, 0) and
3 ∂x
∂z
(0, 0).
∂x
b) Find the equation of the tangent plane to the surface
n o
M = (x, y, z) ∈ R3 : F (x, y, z) = 0 .
π

at the point 0, 0, − .
3
π

Answer a) We have g 0, 0, − = 0. And we have the one by one Jacobian matrix
" #
3
∂g 8xyz π π

J(x, y, z) = = 3+ 2 . Evaluated at the point 0, 0, − , we have J 0, 0, − = [3].
∂z z +1 3 3
The Jacobian matrix is invertible, and thus one can define a function z = z(x, y) of class C 1 such
π π −1 1

−1
that z(0, 0) = − and g(x, y, z(x, y)) = 0 . We have 0, 0, − = [3] = , and we get
3 3 3
∂g
= 4y ln z 2 + 1
∂x
∂g π

0, 0, − =0
∂x 3
" # −1 " # −1
∂z π ∂g π π 1

(0, 0) = − J 0, 0, − 0, 0, − = − J 0, 0, − [0] = − [0] = [0]
∂x 3 ∂x 3 3 3
∂z
(0, 0) = 0
∂x
367
as well as
∂g
= 4x ln z 2 + 1
∂y
∂g π

0, 0, − =0
∂y 3
" # −1 " # −1
∂z π π ∂g π 1

0, 0, − = − J 0, 0, − (0, 0) = − J 0, 0, − [0] = − [0] = [0]
∂y 3 3 ∂y 3 3
∂z
(0, 0) = 0
∂y
Using the chain rule, we can verify this result. When computing the partial derivative of z with
respect to x, we treat y as a constant.
= + = ·1+ = + =0
∂g
dz
= − ∂x
∂g
dx ∂z
∂g
∂z
(x, y) = − ∂x
∂g
∂x ∂z

∂g
∂z ∂x
0, 0, − π3 0
(0, 0) = − ∂g =− =0
∂x 0, 0, − π 3
∂z 3
When computing the partial derivative of z with respect to y, we treat x as a constant.

= + = ·1+ = + =0
∂g
dz
= − ∂y
∂g
dy ∂z
∂g
∂z
(x, y) = − ∂y
∂g
∂y ∂z

∂g
∂z ∂y
0, 0, − π3 0
(0, 0) = − ∂g =− =0
∂y 0, 0, − π 3
∂z 3
b) We will use the gradient to get a vector which is perpendicular to a level surface of the function

g(x, y, z) = 3z + 4xy ln z 2 + 1 + π
!
∂g ∂g ∂g 8xyz

grad g(x, y, z) = ∇g(x, y, z) = , , = 4y ln z 2 + 1 , 4x ln z 2 + 1 , 3 +
∂x ∂y ∂z z2 + 1
368
π

We have grad g 0, 0, − = (0, 0, 3). So for any point (x, y, z) in the tangent plane to the surface
3
π
n o
3
M = (x, y, z) ∈ R : g(x, y, z) = 0 at the point 0, 0, − , we have
3
π π

0 = (0, 0, 3) · x − 0, y − 0, z − − = 0(x − 0) + 0(y − 0) + 3 z + = 3z + π
3 3
an thus the equation of the tangent plane is 3z + π = 0.
Question 4. Find maximum and minimum value of the function of f (x, y, z) = x − yz on the
sphere x2 + y 2 + z 2 = 1.
Answer: We need to find the maximum and the minimum of the function f = x − yz subject to
the constraint g(x, y, z) = x2 + y 2 + z 2 = 1. Using Lagrange multipliers, we solve the equations
or as 
1 = 2xλ





−z = 2yλ


−y = 2xλ







 2
x + y2 + z2 = 1

1 1
From −z = 2yλ and −y = 2xλ, we get zy = 4zyλ2 . This implies λ = or λ = − , unless z = 0
2 2
or y = 0. If z = 0 or y = 0, then x = y = 0, unless λ = 0 which would be in contradiction with
1 1
the equation 1 = 2xλ. So λ = or λ = − , and from 1 = 2xλ, we get x = 1 or x = −1. From
2 2
2 2 2
the equation x + y + z = 1, we then get y = z = 0. So the maximum is f (1, 0, 0) = 1 and the
minimum is f (−1, 0, 0) = −1.
Question 5. Calculate the double integrals:

ZZ Z 1Z 1
e(x ) dx dy
2

2 2
a) 2 cos x + y dx dy and b)
x2 +y 2 ¬1 0 y
Answer: a) We will use polar coordinates. We do the change of variables x = r cos θ and y = r sin θ.
We get the Jacobian matrix
 
∂x ∂x
−r sin θ
 
 ∂r

∂θ 
 cos θ
J =   = 
∂y ∂y 

 sin θ r cos θ
∂r ∂θ
cos θ −r sin θ
det(J) = = r(cos θ)2 − (−r)(sin θ)2 = r (cos θ)2 + (sin θ)2 = r
sin θ r cos θ
| det(J)| = r
369
The region of integration is a disc of radius one, centered at the origin. So the integral becomes
ZZ Z 2π Z 1 h i
2
2 cos x + y 2
dx dy = 2 cos (r cos θ)2 + (r sin θ)2 | det(J)| dr dθ
x2 +y 2 ¬1 0 0
Z 2π Z 1 h i
= 2 cos (r cos θ)2 + (r sin θ)2 r dr dθ
0 0
Z 2π Z 1
= 2r cos (r cos θ)2 + (r sin θ)2 dr dθ
0 0
Z 2π Z 1 h i
= 2r cos r2 (cos θ)2 + (sin θ)2 dr dθ
0 0
Z 2π Z 1 Z 2π h i1
2
= 2r cos r dr dθ = sin r2 dθ
0 0 0 0
Z 2π Z 2π
= (sin(1) − sin(0)) dθ = (sin(1) − 0) dθ
0 0
Z 2π
= sin(1) dθ = [θ sin(1)]2π
0 = 2π sin 1 − 0 · sin 1 = 2π sin 1
0
b) The antiderivative of e(u ) cannot be expressed by elementary functions, we will call this anti-
2
d
F (x) = e(u ) . And by integration by parts, we get the indefinite
2
derivative F (u). Then F 0 (u) =
du
integral
e(u )
Z Z Z 2
F (u) du = uF (u) − 0
uF (u) du = uF (u) − ue (u2 )
du = uF (u) −
2
Z 1Z 1
e(x ) dx dy.
2
Now we use the function F to evaluate
0 y
Z y=1 Z x=1 Z y=1 Z y=1

e(x ) dx dy =
2
[F (x)]x=1
x=y dy = (F (1) − F (y)) dy
y=0 x=y y=0 y=0
(y2 ) y=1 (y2 ) y=1

    
e e
= yF (1) − yF (y) −  = yF (1) − yF (y) + 
2 2
y=0 y=0
   
e(1 )   e(0 ) 
2 2
= 1 · F (1) − 1 · F (1) + − 0 · F (1) − 0 · F (0) +

2 2
e1 e0 e1 − e0 e−1
= F (1) − F (1) + − = =
2 2 2 2
Temat 28: Multiple integrals. Polar substitution.

Gaussian/Euler-Poisson integral.
Question 1. Derive the formula for volume and surface area:
370
a) cone with height ` and base radius r.
b) the torus obtained by rotation of the disk D around the OZ axis, where the disk D is in the
xz-plane with center (2, 0, 0) and radius r = 1.
a)
As a preparation for part b), we will solve this by rotation around the OZ axis.
`
In the xz-plane, we put z(x) = ` − x. Then a volume infinitesimal is given by dV = 2πxz(x) dx.
r
So we get
Z r Z r Z r ! Z r !
` `
V = 2πxz(x) dx = 2π xz(x) dx = 2π x ` − x dx = 2π `x − x2 dx
0 0 0 r 0 r
" #r ! !
x 2 ` x3 r2 ` r3 02 ` 03 1 1 `

= 2π ` − = 2π ` − − 2π ` − = 2π`r2 − = πr2
2 r 3 0
2 r 3 2 r 3 2 3 3
` dz `
With z(x) = ` − x, we have = − , and the area infinitesimal is given by
r dx r
v v
u !2 s
u (dx)2 (dz)2 dz `2
q u q u
dA = 2πx (dx)2 + (dz)2 = 2πxt + (dx)2 = 2πx 1 +
t
dx = 2πx 1 + dx
(dx)2 (dx)2 dx r2
So we get
s s  s r
Z r
`2 Z r
`2 x2 `2
A= 2πx 1 + 2 dx = 2π x 1 + 2 dx = 2π  1+ 2 
0 r 0 r 2 r
0
 s   s  s
r2 `2 02 `2 `2 √
= 2π  1 + 2  − 2π  1 + 2  = πr2 1 + 2 = π r4 + r2 `2
2 r 2 r r
To this we have to add the area of the base, so
√ √
Atotal = π r4 + r2 `2 + πr2 = π r2 + r4 + r2 `2
b)
371
We will do this for the upper half of the torus and multiply by two.
q
In the xz-plane, we have (x − 2)2 + z 2 = 1, so we put z(x) = 1 − (x − 2)2 . Then a volume
infinitesimal is given by dV = 2πxz(x) dx. So we get
Z 3 Z 3 Z 3 q
V =2 2πxz(x) dx = 4π xz(x) dx = 4π x 1 − (x − 2)2 dx
1 1 1
dx
We do the change of variables x = 2 + sin θ . Then = cos θ , so dx = (cos θ) dθ and we get
dθ
Z 3 q Z π q
2
V = 4π x 1 − (x − 2)2 dx = 4π π
(2 + sin θ) 1 − (sin θ)2 (cos θ) dθ
1 −2
Z π Z π
2 2
= 4π π
(2 + sin θ)(cos θ)(cos θ) dθ = 4π 2(cos θ)2 + (sin θ)(cos θ)2 dθ
−2 − π2
Z π Z π
2 2 2
= 8π (cos θ) dθ + 4π (sin θ)(cos θ)2 dθ
− π2 − π2
Z π Z π
2 2 2
Since (cos θ) dθ = (sin θ)2 dθ, we get
− π2 − π2
Z π Z π Z π Z π Z π Z π
2 2 2 2 2 2
π
(cos θ)2 dθ = 1 − (sin θ)2 dθ = dθ − (sin θ)2 dθ = dθ − (cos θ)2 dθ
−2 − π2 − π2 −2 π
− π2 − π2
π π h iπ π π
Z Z
2 2 2 2
2 (cos θ) dθ = dθ = θ = − − =π
− π2 − π2 − π2 2 2
π
Z
2 π
(cos θ)2 dθ =
−2 π 2
Since the function sin is odd, we have
Z π
2
(sin θ)(cos θ)2 dθ = 0
− π2
Thus π π
Z
2 2
Z
2 π
V = 8π (cos θ) dθ + 4π (sin θ)(cos θ)2 dθ = 8π · + 4π · 0 = 4π 2
− π2 − π2 2
q dz x−2
With z(x) = 1 − (x − 2)2 , we have =q , and the area infinitesimal is given by
dx 1 − (x − 2)2
v v
u !2
u (dx)2 (dz)2 q dz
q u u
dA = 2πx (dx)2 + (dz)2 = 2πx t + (dx)2 = 2πx 1 +
t
dx
(dx)2 (dx)2 dx
v  2 v
u
x−2 (x − 2)2
u u
u
= 2πx t1 +  q dx = 2πx t1 + dx
u 
1 − (x − 2)2 1 − (x − 2)2
v s
u 1 − (x − 2)2 + (x − 2)2 x
u
1
= 2πx t dx = 2πx dx = 2π q dx
1 − (x − 2)2 1 − (x − 2) 2
1 − (x − 2)2
372
dx
We do the change of variables x = 2 + sin θ . Then = cos θ , so dx = (cos θ) dθ and we get
dθ
Z 3 Z 3 π
x x Z
2 2 + sin θ
A=2 2π q dx = 4π q dx = 4π q (cos θ) dθ
1 1 − (x − 2)2 1 1 − (x − 2)2 − π2 1 − (sin θ)2
π π π
Z
2 2 + sin θ Z
2 2 + sin θ Z
2
= 4π q (cos θ) dθ = 4π (cos θ) dθ = 4π π (2 + sin θ) dθ
− π2 (cos θ)2 − π2 cos θ −2
iπ π π
h
= 4π 2θ − cos θ 2
= 4π π − cos − 4π −π − cos − = 4π(π − 0) − 4π(−π − 0) = 8π 2
− π2 2 2
373

SOLVED PROBLEMS MathematicalAnalysis

Uploaded by

Copyright:

Available Formats

You might also like

SOLVED PROBLEMS MathematicalAnalysis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SOLVED PROBLEMS MathematicalAnalysis

Uploaded by

Copyright:

Available Formats

Tutorial Mathematical Analysis

The exercises can be found at

Temat 2: School mathematics. Extension. 8

Temat 3: Induction, binomial theorem, inequalities. 25

Temat 4: Weighted mean, bounds of number sets, inequalities between means

Temat 6: Sequences II. 74

Academic Year 2015/16, Colloquium I, 20 November, 2015 93

Academic Year 2016/17, Colloquium I, 1 December, 2016 101

Temat 7: Series. 105

Temat 8: The limits of functions. 118

Temat 9: Functions. Continuity of functions. 122

Temat 10: Continuity continued. 127

Temat 11: The derivative of a function 140

Temat 12: Lagrange theorem, derivatives and properties of functions. 146

Temat 13: l’Hôpital’s rule, elasticity, and Taylor’s formula. 152

Questions revisited, using power series and l’Hôpital’s rule 166

Academic Year 2015/16, Colloquium II, 15 January, 2016 182

Academic Year 2016/17, Colloquium II, 24 January, 2017 186

Academic Year 2015/16, Examination, 25 January, 2016 189

Academic Year 2015/16, Examination, 2 March, 2016 199

Academic Year 2016/17, Examination, 1 February, 2017 203

Temat 15: Indefinite integrals. 207

Temat 16: Indefinite, definite and improper integrals 215

Temat 17: Norms and inner products 223

Temat 18: Subsets of Rn . Limits of functions of several variables. Continuity. 232

Temat 19: Functions of several variables 242

Temat 20: Partial derivatives and differentiation of functions of several variables.246

Temat 21: Gradient and growth 262

Academic Year 2014/15, Colloquium I, 17 May, 2015 266

Academic Year 2016/17, Colloquium I, 11 April, 2017, Version A 273

Academic Year 2016/17, Colloquium I, 11 April, 2017, Version B 277

Temat 22: Extrema of functions of several variables. 282

Temat 23: Extrema of functions of several variables II. 289

Temat 24: Additional exercises on the differential calculus of functions of several

Temat 25: Tangent planes and diffeomorphisms. 311

Temat 26: Local invertibility of functions. Implicit function theorem. Extrema

Academic Year 2015/16, Colloquium II, 3 June, 2016 345

Academic Year 2016/17, Colloquium II, 1 June, 2017 352

Academic Year 2014/15, Examination, 8 September, 2015 359

Academic Year 2015/16, Examination, 6 September, 2016 365

Temat 28: Multiple integrals. Polar substitution.

Question 2. Let a, b, c be natural numbers. Is the following sentence true?

Question 3. Is the following sentence a tautology:

Question 4. Consider the formula Φ(m, x) with two real-valued variables m, x :

Φ(m, x) ⇔ mx2 + mx + 4 > 0.

Which of the sentences are true:

Draw the following sets on the real plane:

(1 + a)n+1 = (1 + a)(1 + a)n = (1 + a)n + a(1 + a)n ­ 1 + na + a(1 + a)n .

Temat 2: School mathematics. Extension.

d) Solve the system of equations: 

Answer: a) We need to consider three cases. If x < 2, the equation |x − 2| + |x − 8| = 1 becomes

c) Case 1: x + 1 ­ 0 and y − 1 ­ 0. The system of equations becomes

So −x − 1 = 6 − y = 4y − 4, hence y = 2 and x = −5. We check that this solutions satisfies x + 1 < 0

d) It is easiest to first solve the equalities x2 − x − 6 = 0 and 2x = 15 − x2 .

Question 9. In Cartesian coordinate system, draw the plot of the function f :

Answer: a) Since |1 − x| = | − (1 − x)| = |x − 1|, it is convenient to write the function as

0 = 4x2 − 15x + 4m2

2 = W (17) − W (13) = W (13 + 4) − W (13)

(1 + a)n+1 = (1 + a)(1 + a)n = (1 + a)n + a(1 + a)n 1 + na + a(1 + a)n .

c) Case 1: x + 1 0 and y − 1 0. The system of equations becomes

We let 11 10 2