consider the folloing sum

$$ \sum_{n=0}^\infty a_ne^{-\alpha_nx} $$

for every n $ 0 < \alpha_n < \alpha_{n+1}$

it is also given that for $x_0 \in \Bbb{R}$ the sum converges

prove that the sum converges uniformly in the interval $[x_0,\infty]$

  • $\begingroup$ Weierstrass and the fact that $e^{-\alpha_n x}\leq e^{-\alpha_n x_0}$ $\endgroup$ – Antoine Aug 11 '15 at 18:49
  • 2
    $\begingroup$ @Antoine Only if $a_n\ge 0$ $\endgroup$ – David C. Ullrich Aug 11 '15 at 19:57
  • 2
    $\begingroup$ You can do this using summation by parts. en.wikipedia.org/wiki/Summation_by_parts $\endgroup$ – David C. Ullrich Aug 11 '15 at 20:09
  • $\begingroup$ @hermes Why is $a_n \geq 0$ presumed? (I missed that and this may be the reason that you have a down vote) $\endgroup$ – Antoine Aug 12 '15 at 14:05
  • 1
    $\begingroup$ Best not to use $\alpha_n$ and $a_n$ in the same question, when you can avoid it. $\endgroup$ – Robert Israel Aug 13 '15 at 0:44

First we prove it is true for $x_0=0$. The given condition becomes $\sum_{k=0}^n a_k$ converges. The idea is to use partial summation (used in proving Abel theorem) to get an estimate of Cauchy sum.

Let $b_n=\sum_{k=m}^n a_k$. So by Cauchy Criterion, $|b_n|<\epsilon$ for $n,m>N$.

We have \begin{align} \sum_{k=m}^n a_ke^{-\alpha_kx}&=\sum_{k=m}^n (b_k-b_{k-1})e^{-\alpha_kx} \\ &=\sum_{k=m}^n b_ke^{-\alpha_kx} -\sum_{k=m}^n b_{k-1}e^{-\alpha_kx} \\ &=\sum_{k=m}^{n-1} b_k(e^{-\alpha_kx}-e^{-\alpha_{k+1}x})+b_ne^{-\alpha_nx}\tag{$b_{m−1}=0$} \end{align} Since $\alpha_n>0, \:\alpha_n \uparrow$, $e^{-\alpha_kx}-e^{-\alpha_{k+1}x}\geqslant0\:$ for all $k>0$ and $x\in[0,\infty)$.

Since $-\epsilon<b_k<\epsilon$ for all $k>m$ $$ |b_k(e^{-\alpha_kx}-e^{-\alpha_{k+1}x})|<\epsilon(e^{-\alpha_kx}-e^{-\alpha_{k+1}x}) $$ So for all $n,m>N-1$ and $x\in[0,\infty)$, there is \begin{align} \left|\sum_{k=m}^n a_ke^{-\alpha_kx}\right|&\leqslant\sum_{k=m}^{n-1} |b_k(e^{-\alpha_kx}-e^{-\alpha_{k+1}x})|+|b_ne^{-\alpha_nx}| \\ &\leqslant\sum_{k=m}^{n-1} \epsilon(e^{-\alpha_kx}-e^{-\alpha_{k+1}x})+\epsilon e^{-\alpha_nx} \\ &=\epsilon \:(e^{-\alpha_{m}x}-e^{-\alpha_{n}x}+e^{-\alpha_{n}x}) \\ &=\epsilon \:e^{-\alpha_{m}x} \\ &\leqslant\epsilon \end{align} So by Cauchy Criterion, it uniformly converges on $[0,\infty)$. Finally we can prove it on $[x_0,\infty)$ by replace $x$ with $x-x_0$.

| cite | improve this answer | |
  • $\begingroup$ didnt get this part $\sum_{k=m}^n b_ke^{-\alpha_kx} -\sum_{k=m}^n b_{k-1}e^{-\alpha_kx} = \sum_{k=m}^{n-1} b_k(e^{-\alpha_kx}-e^{-\alpha_{k+1}x})+b_ne^{-\alpha_nx}$ arent you missing $ + b_{m-1}e^{-\alpha_mx} $? $\endgroup$ – Daniel Katzan Aug 13 '15 at 7:45
  • $\begingroup$ also didnt get this equality $\sum_{k=m}^{n-1} \epsilon(e^{-\alpha_kx}-e^{-\alpha_{k+1}x})+\epsilon e^{-\alpha_nx} =\epsilon \:e^{-\alpha_{m}x} $ $\endgroup$ – Daniel Katzan Aug 13 '15 at 7:46
  • $\begingroup$ By definition $b_{m−1}=0$. 2nd question left is telescope series. Read the proof again. $\endgroup$ – hermes Aug 13 '15 at 8:49

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.