2
$\begingroup$

Is there a series of functions $\sum (-1)^n u_n(x)$ that is convergent for Leibniz rule (alternating series test) in some interval but it is not uniformly convergent?

In particular the series must be (for using Leibniz rule)

  • positive $(u_n(x) \geq 0)$
  • decreasing $(u_{n+1}(x)\leq u_n(x))$
  • such that $\lim_{n \to \infty} u_n(x)=0$

Edit: I refer to the following definition: a series of function is uniformly convergent in a subset $S$ if

$$\lim_{n \to \infty} \sup_{x \in S} |\sum_{n \geq 0} u_n(x) -\sum_{n=0}^{N} u_n(x)| =0$$

$\endgroup$
  • 1
    $\begingroup$ Which rule are you referring to, specifically? (I am used to Leibniz's rule being a name for the alternating series criterion) $\endgroup$ – Clement C. Dec 6 '16 at 17:53
  • $\begingroup$ It's $u_n(x),$ not $u(x)_n.$ Please edit. $\endgroup$ – zhw. Dec 6 '16 at 17:58
  • $\begingroup$ @ClementC. I'm guessing he means $\sum (-1)^nu_n(x)$ $\endgroup$ – zhw. Dec 6 '16 at 18:03
  • $\begingroup$ Wow, that's quite a far cry from what's written. $\endgroup$ – Clement C. Dec 6 '16 at 18:04
2
+50
$\begingroup$

Let $$u_n(x)=\frac{x^n}{n}, \quad n\in\mathbb N.$$ Then $u_n: [0,1]\to\mathbb R$ is a decreasing sequence of non-negative functions.

Clearly $$ \sum_{n=1}^\infty u_n(1)=\sum_{n=1}^\infty\frac{1}{n}=\infty, $$ while $$ \sum_{n=0}^\infty (-1)^nu_n(x)=\sum_{n=0}^\infty \frac{(-1)^nx^n}{n}=-\log(1+x), \quad\text{for all $x\in [0,1]$.} $$ In fact, $$ \sum_{k=0}^n (-1)^ku_k(x)=-\log(1+x)-\frac{(-1)^{n+1}\xi^{n+1}}{n+1}, \quad \text{for some $\xi\in (0,x)$}, $$ and thus $$ \bigg|\sum_{k=0}^n (-1)^ku_k(x)+\log(1+x)\,\bigg|\le\frac{1}{n+1}, \quad \text{for all $x\in [0,1]$}, $$ i.e., the series $\,\sum (-1)^nu_n\,$ converges uniformly in $[0,1]$.

Note. That the advantage of this example is that $[0,1]$ is compact.

$\endgroup$
3
$\begingroup$

Let $u_n(x)=e^{-nx}$ where $x\in I=(0,1]$ then

$$R_n(x)=\sum_{k=n+1}^\infty (-1)^k u_k(x)=\frac{(-1)^{n+1} e^{-(n+1)x}}{1+e^{-x}}$$ so $$\Vert R_n\Vert_\infty=\sup_{x\in I}\vert R_n(x)\vert\ge \frac12$$ so the convergence isn't uniform on $I$

$\endgroup$
  • 1
    $\begingroup$ We could also use $u_n(x) = x^n$ on $[0,1)$ - perhaps a bit more elementary. $\endgroup$ – zhw. Dec 9 '16 at 19:39
  • $\begingroup$ is you result true for any decreasing $u_n(x)$ with respect to $x$? $\endgroup$ – DuFong Apr 19 '18 at 3:52
  • $\begingroup$ If $u_n$ is decreasing then we have $$\left\vert\sum_{k=n+1}^\infty (-1)^ku_k(x)\right\vert\le \vert u_{n+1}(x)\vert$$ and in the case where $sup\vert u_n(x)\vert$ tends to 0 we use the above inequality to prove the uniform convergence of the series. @DuFong $\endgroup$ – user296113 Apr 20 '18 at 19:04

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.