A question about uniform convergence

Join the discussion
Ask a follow-up here, or get your own question answered by working scientists, mathematicians and engineers — people, not an autocomplete.
Real named experts · corrections over time · the nuance an AI answer skips
3 replies · 2K views
Artusartos
Messages
236
Reaction score
0

Homework Statement



For question 25.15 in this link:

http://people.ischool.berkeley.edu/~johnsonb/Welcome_files/104/104hw9sum06.pdf

I have some questions about pointwise convergence and uniform convergence...

Homework Equations





The Attempt at a Solution



Our textbook says that a function is pointiwise convergent if [tex]lim_{n \rightarrow \infty} f_n(x) = f(x)[/tex], and it is uniformly convergent if [tex]lim_{n \rightarrow \infty} [sup{|f_n(x) - f(x)|}] = 0[/tex]

So can't we just use this for the proof of this question?

Since [tex]lim_{n \rightarrow \infty} f_n(x) = f(x) = 0[/tex] for this question, we have...

[tex]lim_{n \rightarrow \infty} [sup{|f_n(x) - f(x)|}] =lim_{n \rightarrow \infty} [sup{|f_n(x) - 0|}] =lim_{n \rightarrow \infty} [sup{|f_n(x)|}][/tex]

Since we know that [tex]lim_{n \rightarrow \infty} f_n(x) = 0[/tex], then [tex]lim_{n \rightarrow \infty} [sup{|f_n(x)|}]=0[/tex]

Thanks in advance
 
Last edited by a moderator:
Physics news on Phys.org
No, that is incorrect.

Specifically, this step is wrong:

Artusartos said:
Since we know that [tex]lim_{n \rightarrow \infty} f_n(x) = 0[/tex], then [tex]lim_{n \rightarrow \infty} [sup{|f_n(x)|}]=0[/tex]

See example (c) at page 2 of the following pdf: www.math.ubc.ca/~feldman/m321/dini.pdf
This is an example of functions such that [itex]\lim_{n\rightarrow \infty} f_n(x) = 0[/itex] for all x, but such that [itex]\lim_{n\rightarrow +\infty} \sup_{x\in [a,b]} f_n(x)\neq 0[/itex].

You really do need the hypothesis of nonincreasing!
 
micromass said:
No, that is incorrect.

Specifically, this step is wrong:



See example (c) at page 2 of the following pdf: www.math.ubc.ca/~feldman/m321/dini.pdf
This is an example of functions such that [itex]\lim_{n\rightarrow \infty} f_n(x) = 0[/itex] for all x, but such that [itex]\lim_{n\rightarrow +\infty} \sup_{x\in [a,b]} f_n(x)\neq 0[/itex].

You really do need the hypothesis of nonincreasing!

So if they had told us what the function really was, we would just substitute n=1 and see if the limit would equal zero right (since it is nonincreasing). But since they didn't give us the funciton, we cannot use this theorem, right?
 
Artusartos said:
Since we know that [tex]lim_{n \rightarrow \infty} f_n(x) = 0[/tex], then [tex]lim_{n \rightarrow \infty} [sup{|f_n(x)|}]=0[/tex]

This is not generally true; for example if [itex]f_n(x) = x^n[/itex] on [itex][0,1][/itex]. Here [itex]f_n(x)[/itex] is continuous for every [itex]n[/itex], [itex]f_n(x)[/itex] is non-increasing for every x, and [itex]f_n[/itex] converges pointwise to
[tex]f(x) = \left\{ \begin{array}{r@{\quad}l}<br /> 0 & x \neq 1 \\<br /> 1 & x = 1<br /> \end{array}\right.[/tex]
But the [itex]f_n(x)[/itex] are continuous for each [itex]n[/itex], so for all [itex]n[/itex] there are points near [itex]x = 1[/itex] where [itex]|f_n(x) - f(x)| = |f_n(x)|[/itex] is arbitrarily close to 1. So [itex]\sup |f_n(x) - f(x)| = 1[/itex] for all [itex]n[/itex], and convergence is not uniform.

(This is not a counterexample to the proposition you were asked to prove, because [itex]f(x) \neq 0[/itex] for all [itex]x \in [0,1][/itex].)
 
Last edited: