Thanks to visit codestin.com
Credit goes to www.scribd.com

Scribd
Upload
55 views4 pages

Foss Lecture3 Appendix

This document defines and discusses uniform convergence of functions. Some key points: 1. It defines uniform convergence (u.o.c.) of functions f_n to f on compact sets as the supremum of |f_n(x) - f(x)| approaching 0 as n approaches infinity. 2. It proves this is equivalent to the norm |f_n - f| approaching 0, where norms are defined for functions over bounded intervals. 3. Several lemmas are presented, including one showing uniform convergence is implied by pointwise convergence on a dense subset for non-decreasing, continuous functions. 4. The concept of Lipschitz continuity is introduced, and

Uploaded by

Jarsen21
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
55 views4 pages

Foss Lecture3 Appendix

This document defines and discusses uniform convergence of functions. Some key points: 1. It defines uniform convergence (u.o.c.) of functions f_n to f on compact sets as the supremum of |f_n(x) - f(x)| approaching 0 as n approaches infinity. 2. It proves this is equivalent to the norm |f_n - f| approaching 0, where norms are defined for functions over bounded intervals. 3. Several lemmas are presented, including one showing uniform convergence is implied by pointwise convergence on a dense subset for non-decreasing, continuous functions. 4. The concept of Lipschitz continuity is introduced, and

Uploaded by

Jarsen21
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

1 Appendix: the u.o.c.

convergence
Let f
n
: R R, n = 1, 2, . . ., and f : R R be any functions.
Denition 1.

f
n
u.o.c.
f as n , if, for any compact (bounded and closed) set K,
sup
xK
[f
n
(x) f(x)[ 0 as n .
For any f and for any t > 0, dene a norm |f|
t
:
|f|
t
= sup
|x|t
[f(x)[,
and a norm |f| :
|f| =

m=1
2
m

|f|
m
1 +|f|
m
.
Lemma 1.

f
n
u.o.c.
f |f
n
f| 0 as n .
Indeed,
(). > 0, choose m
0
:

m=m
0
+1
2
m
2
m
0


2
Then choose n
0
:
|f
n
(x) f(x)|
m
0


2m
0
n n
0
.
[= |f
n
f|
m
0

m=1
|f
n
f|
m
+ 2
m
0
m
0


2m
0
+

2
= .
(). Let K be a compact non-empty set. Take m
0
1: K [m
0
, m
0
]. Set
g
n
= f
n
f. Then
|f
n
f| 0 [= 2
m
0

|g
n
|
m
0
1 +|g
n
|
m
0
0 [= |g
n
|
m
0
0 [= sup
xK
[g
n
(x)[ 0.
.
Remark 1.

For |f| to be nite, all |f|


m
have to be nite. This is always the case
if f is, say, continuous or monotone.
Remark 2.

If f is dened on a measurable subset B R only, we let f(x) = 0


x , B.
Remark 3.

Any (right- and/or left-) continuous function is determined by its val-


ues on a dense subset (e.g., by values at all rational points).
1
Lemma 2.

Let all f
n
, n = 1, 2, . . . be non-decreasing functions and f a continuous
function.
Then the following are equivalent:
(a) f
n
u.o.c.
f;
(b) for any dense subset B of the real line, f
n
(x) f(x), x B.
Proof.
( Only!). Take any m and prove that |f
n
f|
m
0. Assume, by the contrary,
that
limsup
n
|f
n
f|
m
= C > 0 (Note: C < !)
[= a subsequence n
l
: |f
n
l
f|
m
C,
[= since [m, m] is a compact set, a sequence x
l
x [m, m] such that
[f
n
l
(x
l
) f(x
l
)[ C/2, l, x
l
[m, m].
Choose > 0 : (i) B;
(ii) [f(x + ) f(x)[ C/4.
Then
f
n
l
(x
l
)f(x
l
) = f
n
l
(x
l
)f
n
l
(x+)+f
n
l
(x+)f(x+)+f(x+)f(x)+f(x)f(x
l
)
and
limsup
l
(f
n
l
(x
l
) f(x
l
)) 0 + 0 + C/4 + 0 = C/4.
Similarly, choose

> 0 : (i)

B;
(ii) f(x

) f(x) C/4.
Then
liminf
l
(f
n
l
(x
l
) f(x
l
)) C/4 we arrive at a contradiction!
.
Lemma 3.

(The triangular scheme).


Let D, d > 0 and f
n
: sup
|x|d
[f
n
(x)[ D n.
Then a subsequence n
r
: rational x,
a limit lim
r
f
nr
(x) := f(x).
Proof. Number all rational points in [d, d] in an arbitrary order: x
1
, x
2
, . . . Start
with the procedure:
Step 1. Since [f
n
(x
1
)[ D n, a subsequence n
l,1
: f
n
l,1
(x
1
) converges
(denote the limit by f(x
1
)).
. . .
2
Step r + 1. Assume we have dened a sequence n
l,r
: f
n
l,r
(x
i
) f(x
i
) i =
1, . . . , r. Since [f
n
l,r
(x
r+1
)[ D l, [= a subsequence n
l,r+1
: f
n
l,r+1
(x
r+1
)
converges (denote the limit by f(x
r+1
)).
. . .
Thus, r = 1, 2, . . ., we dened a sequence n
l,r
: f
n
l,r
(x
i
) f(x
i
) i =
1, . . . , r.
Now: set n
r
= n
r,r
.
Note: r
0
, n
r
, r r
0
is a subsequence of n
l,r
0

[= f
nr
(x
r
0
) f(x
r
0
).
.
Corollary 1.

(I). Assume, in addition, that C < : x, y [d, d],


limsup
n
[f
n
(x) f
n
(y)[ C[x y[.
If f is any limit from Lemma 3, then
[f(x) f(y)[ C[x y[, rational x, y [d, d].
Therefore,
(i) t [d, d], one can dene f(t) = lim
x
l
t
f(x
l
), where
x
l
are rational;
(ii) this limit does not depend on x
l
;
(iii) f(t) is continuous in [d, d] and
[f(t
1
) f(t
2
)[ C[t
1
t
2
[.
(II). If conditions (I) are satised d and if all f
n
are non-
decreasing, then
f
nr
u.o.c.
f.
Problem No 1. Prove the corollary.
Denition 2.

A function f is Lipshitz continuous with parameter C < if x, y


[f(x) f(y)[ C[x y[.
3
Theorem 1.

Let f : R R (or f : R
+
R) be Lipshitz continuous. Then
(a) a measurable set B B(f) R (or R
+
) : (B) = 0 and
x , B, f

(x) = lim
0
f(x + ) f(x)

, [f

(x)[ C.
(b) Let f

(x) = 0 for x B. Then x, t > 0


f(x + t) f(x) =

x+t
x
f

(z)dz.
Denition 3.

Any point x , B is a regular point of f.


Corollary 2.

Let f : R
+
R
+
be Lipshitz continuous, f(0) > 0.Assume > 0 :
regular point t 0, if f(t) > 0, then f

(t) .
Then t
0

f(0)

: f(t) = 0 t t
0
.
Proof.
First, if f(t
0
) = 0, then f(t) = 0 t t
0
, since f(t) f(t
0
) =

t
t
0
f

(z)dz 0.
Second, if f(t) > 0 t
f(0)

, then
f

f(0)

f(0)/
0
()dz + f(0) 0.
.
4

You might also like