2
$\begingroup$

If $\sum_{n=0}^\infty|a_n|<\infty$, prove that $\sum_{n=0}^\infty a_nx^n$ converges uniformly for $x\in[0,1]$.

Here's my attempt: Let $x\in[0,1]$. Then,

$$\begin{align}\sum_{n=0}^\infty a_nx^n & \le\sum_{n=0}^\infty|a_nx^n|\\ & = \sum_{n=0}^\infty|a_n||x^n|\\ &\le \sum_{n=0}^\infty|a_n||1|^n \\ & = \sum_{n=0}^\infty|a_n| <\infty, \tag{$*$}\end{align}$$ which by the Weierstrass $M$-test, we have that $\displaystyle \sum_{n=0}^\infty a_nx^n$ converges uniformly on $[0,1]$.

Here's my question: Have I correctly used the Weierstrass $M$-test? The conditions of the Weierstrass $M$-test are:

Let $\sum_{k=1}^\infty u_k$ be a series of real-valued functions on $E$. If there exists positive $M_1,M_2,\ldots$ with $\sum_{k=0}^\infty M_k <\infty $ such that $\sum_{k=0}^\infty u_k \ll \sum_{k=0}^\infty M_k$, then $\sum_{k=0}^\infty u_k$ converges uniformly on $E$.

I don't necessarily know that the $|a_n|$ are necessarily positive. However, I do know that they are non-negative. Thus, can I not use the Weierstrass $M$-test here? And I'm not entirely sure that I showed the dominated part entirely, but I've shown that the series $\displaystyle \sum_{n=0}^\infty a_nx^n$ converges for all $x\in[0,1]$. I suppose I just have a hard time distinguishing between pointwise convergence and uniform convergence.

$\endgroup$
  • $\begingroup$ I don't know if you know this theorem, but I think you should : if a series of real-valued functions converges normally, then it converges uniformly. But here $||x\to a_n x^n||_{\infty, [0;1]} = |a_n|$ $\endgroup$ – Max Mar 14 '17 at 20:19
4
$\begingroup$

You should not use any theorem. For $|x| \le 1$ : $$f_k(x) = \sum_{n=0}^k a_n x^n, \qquad f(x) = \sum_{n=0}^\infty a_n x^n$$ Since $\sum_{n=0}^\infty |a_n| < \infty$, the series for $f(x)$ converges absolutely and is well-defined.

Also, again for $|x| \le 1$ : $$|f(x)-f_k(x)| =| \sum_{n=k+1}^\infty a_n x^n| \le \sum_{n=k+1}^\infty |a_n x^n| \le \sum_{n=k+1}^\infty |a_n |$$

where $\lim_{k \to \infty}\sum_{n=k+1}^\infty |a_n | =0$.

Q.e.d. this is the definition of the uniform convergence.

$\endgroup$
  • 1
    $\begingroup$ In other words, don't use Weierstrass M, reprove Weierstrass M whenever you want to use it. $\endgroup$ – zhw. Mar 14 '17 at 22:23
  • 1
    $\begingroup$ @zhw. Sarcasm ? :D Well yes, at least the first few months you are studying real/complex/functional analysis (I wanted to show to the OP there is nothing difficult here, just need to look at the good things) $\endgroup$ – reuns Mar 14 '17 at 22:26
  • 1
    $\begingroup$ Sarcasm, maybe, but that's exactly what you did here. Also, you did use a theorem or two along the way. $\endgroup$ – zhw. Mar 14 '17 at 22:54
3
$\begingroup$

You aren't really using WM correctly. What you wrote was, essentially, if $M_n\ge 0$ for each $n,$ with $\sum_n M_n < \infty,$ and if $\sum |f_n(x)| \le \sum_n M_n$ on $E,$ then $\sum_n f_n$ converges uniformly on $E.$ That is false (and indeed it is a mistake I see on MSE often, even by skilled people of high reputation).

WM actually says: If for each $n$ we have $|f_n(x)|\le M_n$ for $x \in E,$ and if $\sum_n M_n <\infty,$ then $\sum_n f_n$ converges uniformly on $E.$ In your problem, you would take $E=[0,1],$ $f_n(x) = a_nx^n,$ and $M_n= |a_n|.$ That should lead to a correct proof for you.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.