Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Math 5C, Summer 2011

Take-Home Midterm
Due August 23, 2011 in Lecture
Name 1
Perm No. 1
Name 2
Perm No. 2
Style
Substance
Total
Directions:
1. Attach this page as a coversheet to your paper.
2. You may work in groups of 2 and use course materialsincluding the TA and myself, but no other resources
are allowed. Especially the internet!
3. There are 30 points on this exam.
4. Substance: Each part of each problem is worth 1 point for correctness20 points possible in total.
5. Style: There are 10 points for overall presentation; see below.
6. First and foremost your writing should be in fully developed sentences; a mathematical formula should be
treated as part of a sentence.
Bad:
calculus fact:
_
f

g = fg
_
g

f f = 0
Good: From the product rule (fg)

= f

g + g

f we can deduce that


_
f

g = fg
_
g

f,
which is known as integration by parts. Our remarks from before imply that f is identically 0.
7. It is encouraged that you craft your midterm into a coherent piece of prose, though you may simply answer
parts (a), (b), etc. if you wish.
8. Finally, Im unashamedly a sucker for mathematical aesthetic. Your nal product should be on crisp
beautiful paper with nicely drawn symbols and handwriting (or computer typesetting if youre super-cool).
Remember the old saying: If momma [the grader] aint happy, aint nobody happy.
Problem 1
In this problem were going to understand = =
2
x
+
2
y
a bit better. Let D be the unit disk in R
2
centered at the origin.
1. Use derivative rules to compute (ln(x
2
+ y
2
)).
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
(ln(x
2
+ y
2
)) = ((ln(x
2
+ y
2
)))
=
_

x

i +

y

j
_

_
2x
x
2
+ y
2

i +
2y
x
2
+ y
2

j
_
=

x
_
2x
x
2
+ y
2
_
+

y
_
2y
x
2
+ y
2
_
= 2(x
2
+ y
2
)
1
4x
2
(x
2
+ y
2
)
2
+ 2(x
2
+ y
2
)
1
4y
2
(x
2
+ y
2
)
2
= 4(x
2
+ y
2
)
1
4(x
2
+ y
2
)
1
= 0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2. Assuming that f = 0, show that
_
D
f
n
ds = 0.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
f = 0 (f) = 0

2
f
x
2
+

2
f
y
2
= 0
Using Greens Theorem:
_
R


FdA =
_
R

F nds
Let f =

F, and substitute into Greens Theorem.
_
D
(f)dA =
_
D
f nds
Using the fact: (f) = f and f n =
f
n
_
D
fdA =
_
D
f
n
ds
0 =
_
D
f
n
ds
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3. Ignoring the previous two parts, directly compute
_
D

n
ln(x
2
+ y
2
) ds.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
_
D

n
ln(x
2
+ y
2
) ds =
_
D
ln(x
2
+ y
2
) nds
For a unit disk, n = r = (x, y) and x
2
+ y
2
= 1 on the boundary.
_
D
ln(x
2
+ y
2
) nds =
_
D
_
2x
x
2
+ y
2
,
2y
x
2
+ y
2
_
(x, y)ds
=
_
D
(2x
2
+ 2y
2
)ds = 2
_
D
ds = 2(circumference) = 4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Something should look wrong; the last three questions seem to contradict each other. The problem is that
ln(x
2
+ y
2
) has a singularity at the origin, so our goal is to understand whats going on. For this were
going to need a smooth test function . Assume that is zero on and near D; in particular, and all
of its derivatives are zero on the boundary of D.
4. Evaluate
_
D
_


n
ln(x
2
+ y
2
) ln(x
2
+ y
2
)

n

_
ds.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Since we are on boundary of D, we know and all derivatives of = 0, thus both the rst and second term
go to zero.
_
D
_
0

n
ln(x
2
+ y
2
) ln(x
2
+ y
2
)0
_
ds
_
D
(0 0) ds = 0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5. Assuming that we can make sense of ln(x
2
+ y
2
), show that it must satisfy
_
D
ln(x
2
+ y
2
) dA =
_
D
ln(x
2
+ y
2
)dA.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Working with only the left side of the equation, we add a form of zero: (
_
D
ln(x
2
+y
2
)dA+
_
D
ln(x
2
+
y
2
)dA). So on the left, we have:
_
D
ln(x
2
+ y
2
) dA
_
D
ln(x
2
+ y
2
)dA +
_
D
ln(x
2
+ y
2
)dA
=
_
D
(ln(x
2
+ y
2
) ln(x
2
+ y
2
)) dA +
_
D
ln(x
2
+ y
2
)dA
From Homework 1, we know the identity: (fg gf) = fg gf. We can then say:
=
_
D
(ln(x
2
+ y
2
) ln(x
2
+ y
2
)) dA +
_
D
ln(x
2
+ y
2
)dA
Using Greens Theorem:
=
_
D
(ln(x
2
+ y
2
) ln(x
2
+ y
2
)) nds +
_
D
ln(x
2
+ y
2
)dA
On the boundary of D, we know and all derivatives of = 0. Thus:
=
_
D
(0ln(x
2
+y
2
) ln(x
2
+y
2
)0) nds +
_
D
ln(x
2
+y
2
)dA =
_
D
(0) nds +
_
D
ln(x
2
+y
2
)dA

_
D
ln(x
2
+ y
2
) dA =
_
D
ln(x
2
+ y
2
)dA
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
In the next few parts well try to evaluate the integral on the right.
6. Given a small positive number a let B(a) be the annulus {(x, y) : a
2
x
2
+y
2
1} and C(a) be the circle
{(x, y) : x
2
+ y
2
= a
2
}. Given a vector eld F, rewrite
_
B(a)
FdA
in terms of line integrals on C(a) and D. Be very careful when considering boundaries and orientation!
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
As seen in the picture, B(a) is the annulus with outer radius 1. We can think of the annulus as a unit disk
D with a smaller disc A(a), whos boundary is C(a), taken out of it. Thus B(a) = D - A(a). Thus, if we
think about the boundary of B(a), we can say: B(a) = D - A(a) = D - C(a).
The subtraction of C(a) comes from the orientation of the inner circle and the normal pointing in the
opposite direction of the outer boundary.
Using this information, we can write:
_
B(a)
FdA =
_
B(a)
F nds =
_
D
F nds
_
C(a)
F nds
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7. Show that
_
B(a)
ln(x
2
+ y
2
)dA =
_
C(a)
_


n
ln(x
2
+ y
2
) ln(x
2
+ y
2
)

n

_
ds.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
From Homework 1, we know the identity: (fggf) = fggf. If we set g = and f = ln(x
2
+y
2
),
and take the integral over B(a) on both sides we get:
_
B(a)
[ln(x
2
+ y
2
) ln(x
2
+ y
2
)]dA =
_
B(a)
[ln(x
2
+ y
2
) ln(x
2
+ y
2
)]dA
On the right side, we use the identity derived in problem 6. On the left side, we know ln(x
2
+ y
2
) = 0.
Thus we get:
_
B(a)
[ln(x
2
+y
2
)(0)]dA =
_
D
[ln(x
2
+y
2
)ln(x
2
+y
2
)] nds
_
C(a)
[ln(x
2
+y
2
)ln(x
2
+y
2
)] nds
Over the boundary of D, and all derivatives of = 0, thus the entire integral over D goes to zero. So,
we are left with:
_
B(a)
ln(x
2
+ y
2
)dA =
_
C(a)
[ln(x
2
+ y
2
) + ln(x
2
+ y
2
)] nds
The outward normal derivative of f,
f
n
= f n. We can use this to change the form of our last equation:
_
B(a)
ln(x
2
+ y
2
)dA =
_
C(a)
[

n
ln(x
2
+ y
2
) ln(x
2
+ y
2
)

n
]ds
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8. As a 0 the continuity of means that (0, 0) (a constant) all along C(a). On the other hand /n
changes values wildly along C(a), but for some constant M we have |/n| < M. Use these facts to
evaluate
lim
a0
_
C(a)
_


n
ln(x
2
+ y
2
) ln(x
2
+ y
2
)

n

_
ds.
To make the computations easier, notice that for a circle centered at the origin /n = /r, the derivative
with respect to the polar coordinate r.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
lim
a0
_
C(a)
_


n
ln(x
2
+ y
2
) ln(x
2
+ y
2
)

n

_
ds
= lim
a0
_
_
C(a)
_


n
ln(x
2
+ y
2
)
_
ds
_
C(a)
_
ln(x
2
+ y
2
)

n

_
ds
_
We take (0, 0) which we will notate as
0
, which is constant.
= lim
a0
_

0
_
C(a)
_

n
ln(x
2
+ y
2
)
_
ds
_
C(a)
_
ln(x
2
+ y
2
)

n

_
ds
_
Converting to spherical coordinates:
= lim
a0
_

0
_
C(a)
_

r
ln(r
2
)
_
rd
_
C(a)
_
ln(r
2
)

r

_
rd
_
= lim
a0
_

0
_
C(a)
_
2
r
_
rd
_
C(a)
_
2r ln(r)

n

_
d
_
Since we dened r = a, lim
a0
2r ln(r) = 0; and since

n
is bounded by a constant M, we can say that
lim
a0
2r ln(r)

n
= 0.
=
_
2
0
_
C(a)
d
_
C(a)
(0) d
_
= 2
0
_
|
2
0
_
= 4
0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9. Evaluate the following integrals using the previous part. By denition,
_
D
ln(x
2
+ y
2
) dA =
_
D
ln(x
2
+ y
2
)dA = lim
a0
_
B(a)
ln(x
2
+ y
2
)dA.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
From problem 8, we know:
lim
a0
_
C(a)
_


n
ln(x
2
+ y
2
) ln(x
2
+ y
2
)

n

_
ds = 4
0
From problem 7, we have the equality that:
_
B(a)
ln(x
2
+ y
2
)dA =
_
C(a)
_


n
ln(x
2
+ y
2
) ln(x
2
+ y
2
)

n

_
ds
Thus substituting into the equation in problem 8, we get:
lim
a0
_
B(a)
ln(x
2
+ y
2
)dA = 4
0
Thus, by denition:
_
D
ln(x
2
+ y
2
) dA =
_
D
ln(x
2
+ y
2
)dA = lim
a0
_
B(a)
ln(x
2
+ y
2
)dA = 4
0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10. The Dirac delta function is not really a function; nonetheless we pretend like it is a function and dene
it by the property that for any function f,
_
D
f dA = f(0, 0).
Write ln(x
2
+ y
2
) in terms of .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
From problem 9, we see that:
_
D
ln(x
2
+ y
2
) dA = 4
0
This shows that ln(x
2
+ y
2
) acts similarly to the Dirac delta function: when taking the integral over D
of a function multiplied by ln(x
2
+y
2
), we get that function evaluated at (0,0) multiplied by 4. We can
then relate ln(x
2
+ y
2
) and as:
ln(x
2
+ y
2
) = 4
Checking that this makes sense, we reevaluate the previous integral using the properties of the Dirac delta
function: _
D
ln(x
2
+ y
2
) dA =
_
D
4 dA = 4
_
D
dA = 4
0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11. In some appropriate sense the test functions can approximate nearly any function you like. Therefore we
can evaluate _
D
ln(x
2
+ y
2
) dA
for arbitrary , regardless of what it does on D. In a single step evaluate
_
D
e
2x+3y
ln(x
2
+ y
2
) dA.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
We take e
2x+3y
as our function. Knowing how ln(x
2
+ y
2
) acts like , we can easily nd a solution in a
single step:
_
D
(e
2x+3y
)ln(x
2
+ y
2
) dA = 4
_
D
(e
2x+3y
) dA = 4e
2x+3y

(0,0)
= 4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12. In R
3
it turns out that (1/
_
x
2
+ y
2
+ z
2
) = C for some constant C. Assuming this, integrate against
a very simple function to nd C.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using the property of the Dirac delta function described in problem 10, we know:
_
B
(1/
_
x
2
+ y
2
+ z
2
)dV =
_
B
CdV = C(0, 0) = C
Since C is a constant function, nding C at any point will tell us what C is at all points. Thus we solve the
integral to nd C. We use divergence theorem to change the integral from B to the boundary.
_
B
(1/
_
x
2
+ y
2
+ z
2
)dV =
_
B
((1/
_
x
2
+ y
2
+ z
2
))dV =
_
B
(1/
_
x
2
+ y
2
+ z
2
) d

A
=
_
B
_
x
(x
2
+ y
2
+ z
2
)
3/2
,
y
(x
2
+ y
2
+ z
2
)
3/2
,
z
(x
2
+ y
2
+ z
2
)
3/2
_
(x, y, z)d
=
_
B

x
2
+ y
2
+ z
2
(x
2
+ y
2
+ z
2
)
3/2
d
Converting to polar coordinates, and knowing on the boundary of B, r = 1, we solve the integral.
=
_
B

1
r
d =
_
B
d
= 4 = C
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
What weve done in this problem is shown that (1/2) ln
_
x
2
+ y
2
is the electrostatic potential of a point charge
in R
2
. In general if the density of charge in space is , then the electrostatic potential V solves V = . The
Dirac delta is simply the charge density of a point charge. In R
3
the potential of a point charge is proportional
to 1/
_
x
2
+ y
2
+ z
2
; if we write that instead as 1/r we get the (probably familiar) formula
V =
kQ
r
,
where k and Q are just a constant and unit of charge respectively. This is the beginning of electrostatics as well
as the mathematical subject of potential theory.
Problem 2
In this problem were going to use series to solve a problem in statistical physics and thermodynamics. Any
object whose temperature is above absolute zero will emit thermal radiationthat is, electromagnetic radiation
generated by the internal thermal energy of the object. This phenomenon is probably familiar; any object which
is raised to a high enough temperature will glow red-hot (in fact, other colors are possible too). Around the
turn of the 20th century, understanding the underlying physics of thermal radiation from known principles was
a big problem. When Lord Rayleigh attempted to derive the amount of electromagnetic radiation a blackbody
would emit at a certain frequency, a so-called ultraviolet catastrophe occurred: the rate at which the body emits
energy would be
P
_

0
d

4
=
(a nice instance of a divergent integral!). By looking at experimental data, Max Planck was able to deduce a
more accurate law. Though he amazingly guessed a complicated formula from the data, he later derived it by
assuming energy came in discrete incrementsthis was the rst time energy quantization and Plancks constant
h appeared in physics. Accepting something as crazy as energy quantization requires some convincing. Wed like
to compute the same rate of energy ow that caused Rayleighs law to fall apart. In doing so, well derive the
so-called Stefan-Boltzmann law of blackbody radiation.
1. Lets cut to the chase. After a derivation starting from Plancks law, the rate of energy ow (or power)
via thermal radiation for a blackbody is given by the formula
P =
2hA
c
2
_

0

3
exp(h/kT) 1
d,
where A is the surface area of the object, c is the speed of light, k is Boltzmanns constant, and T is
the temperature of the object (recall the notation exp t = e
t
). Lets remove all physical content from the
integral by substituting x = h/kT. From here deduce the Stefan-Boltzmann law: how does P depend on
T? Your expression for P will still contain the dimensionless integral
_

0
x
3
e
x
1
dx.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
We multiply by a form of the number 1:
k
3
T
3
h
2
h
2
k
3
T
3
P =
k
3
T
3
h
2
h
2
k
3
T
3
2hA
c
2
_

0

3
exp(h/kT) 1
d
=
k
3
T
3
h
2
2A
c
2
_

0
x
3
e
x
1
d
Since x =
h
kT
, dx =
h
kT
d d =
kT
h
dx.
P =
k
4
T
4
h
3
2A
c
2
_

0
x
3
e
x
1
dx
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
At this point we could say that were done. However, a great test of Plancks theory comes from evaluating
the integral and comparing numbers against experimental results. Well stop the story of the physics
here and focus on evaluating the integral. Finding an antiderivative is out of the question and a numeric
treatment would be arduous, but we can use series to evaluate the integral explicitly!
2. For x > 0 and s > 1 show that
x
s1
e
x
1
=

n=1
x
s1
e
nx
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The original form of the equation looks familiar to the standard geometric series:

k=0
r
k
=
1
1r
, so we
will adjust it.
x
s1
e
x
1
=
e
1
e
1
x
s1
e
x
1
=
e
x
x
s1
1 e
x
= e
x
x
s1
1
1 e
x
We look at the geometric series for
1
1e
x
=

n=0
e
nx
. Thus:
e
x
x
s1
1
1 e
x
= e
x
x
s1

n=0
e
nx
Since there is no dependency on n, we can bring all the terms into the summation.
=

n=0
e
x
x
s1
e
nx
=

n=0
x
s1
e
(n+1)x
We can then change the summation to start at 1, and change n+1 to n.

n=0
x
s1
e
(n+1)x
=

n=1
x
s1
e
(n)x
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3. A central result in modern integration theory is the Lebesgue Monotone Convergence Theorem (or MCT
for short).
Theorem (MCT). Let g
1
(x), g
2
(x), g
3
(x), . . . be a sequence of measurable functionsthis is a technical
term which includes most functions, such as continuous ones. Fix an interval [a, b], which can be either
bounded or innite. Assume that the sequence is monotone; that is, for each x [a, b],
g
1
(x) g
2
(x) g
3
(x)
Then it follows that
lim
n
_
b
a
g
n
(x) dx =
_
b
a
lim
n
g
n
(x) dx.
Assume this theorem is true and prove this important corollary.
Corollary (MCT for Series). Let f
1
(x), f
2
(x), f
3
(x) . . . be continuous functions dened on an interval [a, b].
Assume that for any x [a, b], each f
n
(x) 0. Then it follows that

n=1
_
b
a
f
n
(x) dx =
_
b
a

n=1
f
n
(x) dx.
Hint: How can an innite series be thought of as a sequence?
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Assuming the monotone convergence theorem, we can say that:
g
n
(x) = f
1
(x) + f
2
(x) + ... + f
m
(x) =
m

i=1
f
i
(x)
Where all f
i
(x) are continuous and dened on the interval [a,b].
We will assume that all f
i
(x) 0 for any x [a, b].
Since g
n
(x) is monotonically increasing:
lim
n
_
b
a
n

i=1
f
i
(x) dx =
_
b
a
lim
n
n

i=1
f
i
(x) dx
lim
n
_
b
a
[f
1
(x) + f
2
(x) + ... + f
n
(x)] dx =
_
b
a

i=1
f
i
(x) dx
lim
n
_
_
b
a
f
1
(x) +
_
b
a
f
2
(x) + ... +
_
b
a
f
n
(x)
_
dx =
_
b
a

i=1
f
i
(x) dx
lim
n
n

i=1
_
b
a
f
i
(x) dx =
_
b
a

i=1
f
i
(x) dx

i=1
_
b
a
f
i
(x) dx =
_
b
a

i=1
f
i
(x) dx
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4. There are many examples of sequences g
n
(x) for which limit and integral cannot be interchanged. Dene
g
n
(x) =
_
n if 0 < x < 1/n
0 otherwise
Show that these functions are not monotone on [0, 1] and that
lim
n
_
b
a
g
n
(x) dx =
_
b
a
lim
n
g
n
(x) dx.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
To show that g
n
(x) is not monotone, we look at g
n
(x) at values n =1, 2 and 3:
g
1
(x) =
_
1 if 0 < x < 1
0 otherwise
g
2
(x) =
_
2 if 0 < x < 1/2
0 otherwise
g
3
(x) =
_
3 if 0 < x < 1/3
0 otherwise
At x = 2/5, g
1
(x) = 1, g
2
(x) = 2, while at g
3
(x) = 0, since g
n
(x) is both increasing and decreasing,
it is clearly not monotone.
Also we can show:
lim
n
_
b
a
g
n
(x) dx =
_
b
a
lim
n
g
n
(x) dx
by evaluating both sides. We take the integral over all real numbers.
lim
n
_

g
n
(x) dx = lim
n
1 = 1
Also
_

lim
n
g
n
(x) dx =
_

0 = 0
1 = 0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5. Dene the gamma function for s > 0 as
(s) =
_

0
t
s1
e
t
dt.
Use integration by parts to show that (s + 1) = s(s).
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
(s + 1) =
_

0
t
s
e
t
From the product rule (fg)

= f

g + g

f we can deduce that


_
b
a
f

g = fg

b
a

_
b
a
g

f
which is known as integration by parts. We set g = t
s
and f

= e
t
which implies g

= st
s1
and f = e
t
.
As a result:
_

0
t
s
e
t
= (e
t
t
s
)

0
+
_

0
st
s1
e
t
e
t
t
s
= 0 @ 0 and , which leaves us with:
_

0
t
s
e
t
=
_

0
st
s1
e
t
_

0
t
s
e
t
= s
_

0
t
s1
e
t
(s + 1) = s(s)
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6. Evaluate (1) and explain why (n) = (n 1)! follows for positive integers n.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
(1) =
_

0
t
11
e
t
dt =
_

0
e
t
dt = e
t

0
= 1
To show (n) = (n 1)!, we prove by induction.
We assume (n) = (n1)!. From problem 5, we know: (n+1) = n(n) = n(n1)! = n! = ((n+1) 1)!
Thus for all positive integers n, (n) = (n 1)!.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Lets think of the p series as a function of p; dene the Riemann zeta function
(p) =

n=1
1
n
p
.
This function is ubiquitous in number theory and seemingly random areas of science such as quantum
mechanics. Well only need one important fact about it.
7. We already know that (1) diverges; now lets nd two more values of . Euler came up with the following
formula for the sine function:
x
x
3
6
+
x
5
120
+ = sin x = x
_
1
x
2

2
__
1
x
2
4
2
__
1
x
2
9
2
_

Compare the coecients of x
3
and x
5
in these expressions to evaluate (2) and (4).
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
We know on the left the coecient of x
3
is
1
6
. On the right, we can see that x
3
terms are only made
when the outermost x is multiplied by an x
2
term. Looking at all x
3
terms on the left and right, we get
the equality:

1
6
x
3
= (
x
3

2

x
3
4pi
2

x
3
9
2
)
(x
3
)
1
6
= (x
3
)(
1

2
)(
1
1
2
+
1
2
2
+
1
3
2
)
1
6
= (
1

2
)(2)
(2) =

2
6
(x
x
3

2
)(1
x
2
4
2
)(1
x
2
9
2
) =
= (x
x
3
4
2

x
3

2
+
x
5
4
4
)(1
x
2
9
2
) . . .
= x
x
3
4
2

x
3

2
+
x
5
4
4
+
x
3
9
2
+
x
5
36
4
+
x
5
9
4

x
7
36
6
. . .
1

4
(
1
4
+
1
36
+
1
9
) =
1

4
(
1
1
2
2
2
+
1
1
2
3
2
+
1
2
2
3
2
+ . . . )
(2)
2
= (

1
n
2
)
2
=

n,m=1
1
n
2
m
2
=

n<m
+

n=m
1

4
+

n>m
x
5
120
=
x
5

4
(
1
1
2
2
2
+
1
1
2
3
2
+
1
2
2
3
2
+ . . . ) =
_
(
2
(2) (4))
2
_
1

4
60
=
2
(2) (4)
(4) = (2
2
)

4
60
=

4
36


4
60
=

4
90
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8. Use the MCT for series to evaluate
_

0
x
s1
e
x
1
dx
in terms of the gamma function and the Riemann zeta function. Be sure to check the hypotheses of the
theorem! What is the value of the integral in the Stefan-Boltzmann law?
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
We have shown in problem 2 that
x
s1
e
x
1
=

n=1
x
s1
e
nx
, and by applying the Monotone Convergence
Theorem for Series from problem 3 to the initial integral, we obtain the following expression
_

0
x
s1
e
x
1
dx =
_

0

n=1
x
s1
e
nx
dx =

n=1
_

0
x
s1
e
nx
dx
We set u = nx, and du = ndx
Furthermore, we know that (nx)
s1
= x
s1
n
s1
, x
s1
=
(nx)
s1
n
s1
.
We also multiply by a form of 1,
n
n
, which gives us:

n=1
_

0
x
s1
e
nx
dx =

n=1
1
n
1
n
s1
_

0
n(nx)
s1
e
nx
dx =

n=1
1
n
s
_

0
(u)
s1
e
u
du
and given that (p) =

n=1
1
n
p
, and that (s) =
_

0
x
s1
e
x
dx, we have
_

0
x
s1
e
x
1
dx =

n=1
1
n
s
_

0
(u)
s1
e
u
du = (s)(s)
In order to evaluate the integral in the Stefan-Bolzmann law, we simply use the expression we obtained
above, and set s = 4
_

0
x
s1
e
x
1
dx =
_

0
x
3
e
x
1
dx = (4)(4)
We have shown in problem 7 that (4) =

4
90
, and we proved in problem 6 that (n) = (n 1)!, therefore,
the expression becomes
_

0
x
3
e
x
1
dx = (4)(4) =

4
90
3! =

4
15
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
I am morally obligated to mention the Riemann hypothesis, which is maths biggest unsolved problem. Using
formulas such as
(p) = 2(2)
p1
sin(p/2)(1 p)(1 p)
can be dened for any complex numberexcept 1 of course. For technical reasons, people care about the zeroes
of . Here is the Riemann hypothesis: for complex numbers a + bi with 0 < a < 1 is it TRUE or FALSE that
(a + bi) = 0 = a = 1/2
Millions of dollars and eternal glory await the person who can solve this problem.

You might also like