Professional Documents
Culture Documents
sp16 mth322 ch03
sp16 mth322 ch03
We have seen that a function f that is the sum of two or more functions will share
certain desirable properties with those functions. For example, our study of continuity,
differentiation, and integration allows us to state if
f = f 1 + f 2 + ... + f n
f 0 = f 10 + f 20 + .... + f n0 .
It is natural to ask whether the corresponding results hold when f is the sum of an
infinite series of functions,
∞
f = f 1 + f 2 + f 3 + ... = ∑ fk.
k =0
Such type of questions lead us to the theory of sequence of functions and series of functions.
If f 1 , f 2 , f 3 , ... are real valued function defined on an interval I of the reals numbers. We
say that { f n } is an infinite sequence of functions on I and ∑∞ k=1 f k or ∑ f k represents the
infinite series of functions on I.
f ( x ) = lim f n ( x ), x ∈ I.
n→∞
Thus if f is the pointwise limit of a sequence of function { f n } define on [ a, b], then to
each ε > 0 and to each x ∈ [ a, b], there correspond an integer m such that
| f n ( x ) − f ( x )| < ε, ∀ n ≥ m. (1)
Suppose that ∑∞ ∞
k=1 f k is a series of functions on I. If the series ∑k=1 f k converges for
every point x ∈ I, and we define
∞
f (x) = ∑ fk, x ∈ I,
k =1
the function f is called the sum or the point-wise sum of the series ∑ f n on I.
Examples:
(1) Consider a sequence { f n ( x )} define by f n ( x ) = x n on [0, 1]. One can note that
lim f n ( x ) = 0, when x ∈ [0, 1) and lim f n (1) = 1. Thus we have
n→∞ n→∞
(
0 if 0 ≤ x < 1,
f ( x ) := lim f n ( x ) =
n→∞ 1 if x = 1.
1 sin nx 1
−√ ≤ √ ≤ √ .
n n n
2
Definition 3: Uniform convergence of sequence of functions
It is clear that every uniformly convergent sequence is pointwise convergent, and the
uniform limit function is same as the pointwise limit function.
The difference between the two concepts is this: In case of pointwise convergence, for
ε > 0 and for each x ∈ [ a, b] there exist an integer N (depending on ε and x both) such
that (1) holds for n ≥ m; whereas in uniform convergence for each ε > 0, it is possible
to find one integer N (depend on ε alone) which will do for all x ∈ [ a, b].
Note: Uniform convergence ⇒ pointwise convergence but not vice-versa.
Also a sequence which is not pointwise convergent cannot be uniformly convergent.
Note: The proof of above result can be seen in [1, p.73]. It is equivalent to the statement;
“A sequence of real numbers is convergent if and only if it is Cauchy sequence”.
3
Theorem 5: Cauchy’s criterion for uniform convergence of sequence
Proof. Let the sequence { f n } uniformly converge on [ a, b] to the limit function f , so that
for a given ε > 0 and for all x ∈ [ a, b], there exist integers m1 , m2 such that
ε
| f n ( x ) − f ( x )| < ∀ n ≥ m1
2
and
f n+ p ( x ) − f ( x ) < ε
∀ n ≥ m2 , p ≥ 1.
2
Let N = max(m1 , m2 ). Then
f n+ p ( x ) − f n ( x ) = f n+ p ( x ) − f ( x ) + f ( x ) − f n ( x )
≤ f n+ p ( x ) − f ( x ) + | f n ( x ) − f ( x )|
ε ε
< + = ε, ∀ n ≥ N, p ≥ 1.
2 2
Conversely, suppose that the given condition (3) holds.
By Cauchy’s general principle of convergence, { f n } converges for each x ∈ [ a, b] to a
limit, say f . Thus the sequence converges pointwise to f . Let us now prove that the
convergence is uniform.
For a given ε > 0, let us choose an integer N such that (3) holds. Fix n, and consider
p → ∞ in (3). This gives us f n+ p → f as p → ∞, so we get
4
Example: Consider a sequence of function { f n }, where
nx
f n (x) = , for all x ∈ R.
1 + n2 x 2
Prove that { f n } is pointwise convergent but not uniformly convergent on an interval
containing 0.
Solution.
nx 1
f ( x ) := lim f n ( x ) = lim 2 2
= lim (by L’Hospital Rule)
n→∞ n→∞ 1 + n x n→∞ 2nx 2
=0 ∀ x ∈ R.
Let m be an integer greater than N such that m1 ∈ [ a, b]. Now if we take n = m and
x = m1 , then we have
nx m · (1/m) 1 1
1 + n2 x2 = 1 + m2 · (1/m2 ) = 2 6< 3 = ε.
Theorem 7
lim f n ( x ) = f ( x ), x ∈ [ a, b]
n→∞
and let
Mn = sup | f n ( x ) − f ( x )| .
x ∈[ a,b]
5
Proof. Let f n → f uniformly on [ a, b], so that for a given ε > 0, there exists an integer N
such that
| f n ( x ) − f ( x )| < ε, ∀ n ≥ N, ∀ x ∈ [ a, b]
⇒ Mn := sup | f n ( x ) − f ( x )| < ε, ∀ n ≥ N.
x ∈[ a,b]
⇒ Mn → 0 as n → ∞.
| Mn − 0| < ε, ∀ n ≥ N,
⇒ Mn < ε ∀ n ≥ N,
that is
sup | f n ( x ) − f ( x )| < ε, ∀ n ≥ N,
x ∈[ a,b]
⇒ | f n ( x ) − f ( x )| < ε, ∀ n ≥ N, ∀ x ∈ [ a, b],
⇒ f n → f unifromly on [ a, b].
6
Question: Prove that the sequence { f n }, where
nx
f n (x) =
1 + nx2
is uniformly convergent on any interval I.
Solution. Here the pointwise limit
x
f ( x ) := lim f n ( x ) = lim =0 ∀ x ∈ R.
n→∞ n→∞ 1 + nx 2
Now let
x
Mn = sup | f n ( x ) − f ( x )| = sup
.
x∈ I 1 + nx2
x∈ I
x
If we take g( x ) = , then
1 + nx2
(1 + nx2 ) · 1 − x · 2nx 1 + nx2 − 2nx2
g0 ( x ) = =
(1 + nx2 )2 (1 + nx2 )2
1 − nx2
= .
(1 + nx2 )2
Put g0 ( x ) = 0, we get
1 1
1 − nx2 = 0 ⇒ nx2 = 1 ⇒ x2 = ⇒ x = ±√ .
n n
7
2. Show that the sequence { x n } is not uniformly convergent on [0, 1].
3. Show that the sequence {exp(−nx )} is not uniformly convergent on [0, k ], k > 0.
| f n ( x )| ≤ Mn for all n.
This gives that ∑ f n is uniformly convergent on [ a, b]. Also from (5), one can conclude
that ∑ f n is absolutely convergent on [ a, b].
8
Remark: The converse of above theorem is not true, i.e. non-convergence of ∑ Mn does
not imply anything as for as ∑ f n is concerned.
cos nθ
Example: Consider the series ∑ for all θ ∈ R. Since we have
np
cos nθ
∑ ≤ 1.
p
n np
We know that ∑ n1p is convergent for p > 1. Hence we conclude that the given series is
uniformly convergent on any interval in R.
Exercise: Prove that the following series are uniformly convergent for all real x.
Proof. Since f n → f uniformly on I, for given ε > 0, there exists an integer N such that
ε
| f n ( x ) − f ( x )| < , ∀ n ≥ N, ∀ x ∈ I. (6)
3
As we have given, each f n is continuous at x0 , there is a δ > 0 such that
ε
| f n ( x ) − f n ( x0 )| < , whenever | x − x0 | < δ. (7)
3
Now for all x ∈ I and all n ≥ N such that | x − x0 | < δ, we have
| f ( x ) − f ( x0 )| = | f ( x ) − f n ( x ) + f n ( x ) − f n ( x0 ) + f n ( x0 ) − f ( x0 )|
≤ | f ( x ) − f n ( x )| + | f n ( x ) − f n ( x0 )| + | f n ( x0 ) − f ( x0 )|
ε ε ε
< + + = ε, by using (6) and (7).
3 3 3
This conclude that f is continuous at x0 .
Corollary 9
9
Theorem 10: Uniform convergence and integration
Since f n → f uniformly on [ a, b], for all ε > 0, there exists an integer N such that
ε
| f n ( x ) − f ( x )| < ∀ n ≥ N, ∀ x ∈ [ a, b],
b−a
this gives
ε
max | f n ( x ) − f ( x )| < ∀ n ≥ N.
x ∈[ a,b] b−a
Thus for n ≥ N, expression (9) leads us to
Z b Z b
ε
f n ( x ) dx − f ( x ) dx ≤ (b − a) · = ε,
a a
b−a
Disclaimer: Most of the contents in these notes are taken from [1]. These notes are made
for students and they are encourage to read the book. Also see some other useful books
in references.
10
References
[1] S.C. Malik and S. Arora, Mathematical analysis, New Age International, 1992.
[3] Robert Gardner Bartle, and Donald R. Sherbert, Introduction to real analysis, Wiley
New York, 1992.
[4] , Brian S. Thomson, and Judith B. Bruckner, and Andrew M. Bruckner, Elementary
real analysis, ClassicalRealAnalysis. com, 2008.
11