I was trying to compute $\lim\limits_{n\rightarrow\infty} \int_{0}^{1} \frac{e^{-nt}-(1-t)^n}{t} dt$ using Lebesgue's dominated convergence THM, but I can't exactly figure out how to do. I mean, I managed to prove that each the integrand function $f_n(t)$ is less or equal than $g_n(t)=e^{-nt}\sqrt{n}e^\frac{1}{\sqrt{n}}\, \forall n\in\mathbb{N}$. And since we are dealing with positive functions and $\int_{0}^{1} g_n(t)dt\leq\frac{e}{\sqrt{n}}\overset{\mathrm{n\rightarrow\infty}}{\rightarrow}0$, I can deduce that the original limit is $0$.

Now, I was just wondering if anyone is able to show analytically that there exists a function in $\mathcal{L}^1$ which dominates all the $f_n$ in order to apply Lebesgue's dominated convergence THM. I made some attempts, but I failed.

Thanks in advance, a humble half-mathematician.


We have $$\exp(-nt)-(1-t)^n=\exp(-nt)\,\Big(1-\big((1-t)\exp(t)\big)^n\Big)\,.\tag{*}$$ By Bernoulli's Inequaliy, $$\big((1-t)\exp(t)\big)^n\geq 1+n\,\big((1-t)\exp(t)-1\big)\,.$$ Therefore, $$\exp(-nt)-(1-t)^n\leq n\,\exp(-nt)\,\big(1-(1-t)\exp(t)\big)\,.$$ By taking derivative with respect to $n$, we can show that $$n\,\exp(-nt)\leq \frac{1}{\text{e}\,t}\text{ for all }t>0\text{ and positive integers }n\,.$$ Furthermore, we have $$(1-t)\,\exp(t)=1-\sum_{k=1}^\infty\,\left(\frac{k-1}{k!}\right)\,t^k\geq 1-t^2\,\sum_{k=2}^\infty\,\frac{k-1}{k!}=1-t^2$$ for all $t\in[0,1]$.

By (*), we conclude that $$f_n(t):=\frac{\exp(-nt)-(1-t)^n}{t}\leq \frac{1}{\text{e}\,t^2}\,\big(1-(1-t^2)\big)=\frac{1}{\text{e}}$$ for every $t\in (0,1]$ and every positive integer $n$ (where the only equality case is when $n=1$ and $t=1$). Therefore, the constant function $f\equiv \dfrac{1}{\text{e}}$ dominates $f_n$ for all $n=1,2,3,\ldots$.

  • $\begingroup$ Absolutely brilliant. Thanks a lot. $\endgroup$ – piotor Dec 8 '18 at 18:45

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.