1
$\begingroup$

Consider a continuous function $f:[0, \infty)\rightarrow\mathbb R$ such that $\lim_{x\rightarrow\infty} f(x)=1$. I would like to prove $$ \lim_{n\rightarrow\infty}\frac{1}{n!}\int_0^\infty f(x)e^{-x}x^n\ dx = 1. $$

I first thought of using Lebesgue's DCT, but soon realized the improper integral may not converge absolutely because $f$ is not guaranteed to be positive. I am at a loss as to how to prove this.

I would be happy if you could help me solve the problem (I also welcome a partial solution, so that I can fill in its gap).

$\endgroup$
  • $\begingroup$ This is not true. Consider $f(x)=1$ $\endgroup$ – Norbert Aug 25 '13 at 11:23
  • $\begingroup$ @Norbert I fixed a typo. $\endgroup$ – Semo Ponume Aug 25 '13 at 11:26
3
$\begingroup$

You can prove this without the dominated convergence theorem. $f$ is continuous and has a limit at $\infty$, so there exists an $M>0$ such that $|f(x)|<M$ for all $x\geq 0$. Now, let $\varepsilon>0$. Then, there exists $N>0$ such that, for all $x\geq N$, you have that $|f(x)-1|\leq\varepsilon$. So, you have that $$\left|\frac{1}{n!}\int_0^{\infty}f(x)e^{-x}x^n\,dx-1\right|\leq\left|\frac{1}{n!}\int_0^Nf(x)e^{-x}x^n\,dx\right|+\left|\frac{1}{n!}\int_N^{\infty}f(x)e^{-x}x^n\,dx-1\right|\leq$$$$\frac{1}{n!}\int_0^NMe^{-x}N^n\,dx+\left|\frac{1}{n!}\int_N^{\infty}(f(x)-1)e^{-x}x^n\,dx\right|+\frac{1}{n!}\int_0^Ne^{-x}x^n\,dx,$$ where we used that $\int_0^{\infty}e^{-x}x^n\,dx=n!$. So, this is less than or equal to $$\frac{MN^n}{n!}(1-e^{-N})+\frac{1}{n!}\varepsilon\int_0^{\infty}e^{-x}x^n\,dx+\frac{N^n(1-e^{-N})}{n!},$$ which is less than or equal to $\varepsilon$ at the limit as $n$ goes to $\infty$.

$\endgroup$
  • $\begingroup$ Thank you for your helpful answer. I would be more than happy if you could explain the intuition behind this solution, or how you came up with it. $\endgroup$ – Semo Ponume Aug 25 '13 at 12:08
  • $\begingroup$ What you need to do is to show that the first term goes to zero. For $x$ near infinity, there is no problem, since $f$ is close to $1$. The problem comes from intervals of the form $[0,N]$, where you have no information about $f$, and it could be arbitrarily large there. So, you split the integral in two parts: from $0$ to $N$, and from $N$ to $\infty$. N will be chosen such that $f$ is $\varepsilon$-close to $1$ if $x\geq N$, and the rest follows. $\endgroup$ – detnvvp Aug 25 '13 at 12:11

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.