Uniform convergence of: $$f_n(x):={(1-\frac{x^2}{n})^n}$$

(a) In a closed interval $[-L, L]$ as $L>0$.

(b) In $\mathbb{R}$.

My try:

There is a pointwise convergence to $e^{-x^2}$.

Obviously, there is no uniform convergence in $\mathbb{R}$ as $\lim\limits _{x\to \infty }|f_n(x)|=\infty$.$\;$ Thus, $\sup \limits _{x\in\mathbb{R}}|f_n(x)-e^{-x^2}|= \infty$

Can anyone give a direction for a closed interval?


Using the basic inequalities $e^y \geqslant 1 +y$ and $e^{-y} \geqslant 1 - y$ for $0 \leqslant y \leqslant 1$ we can show that for $x^2 < n$,

$$\left(1- \frac{x^2}{n}\right)^n \leqslant e^{-x^2}\leqslant \left(1+ \frac{x^2}{n}\right)^{-n}.$$

Applying the Bernoulli inequality $(1 - x^4/n^2)^n \geqslant 1 - x^4/n,$ we get

$$0 \leqslant e^{-x^2} - \left(1- \frac{x^2}{n}\right)^n = e^{-x^2}\left[1 - e^{x^2}\left(1- \frac{x^2}{n}\right)^{n}\right]\\ \leqslant e^{-x^2}\left[1 - \left(1+ \frac{x^2}{n}\right)^{n}\left(1- \frac{x^2}{n}\right)^{n}\right]\\= e^{-x^2}\left[1 - \left(1- \frac{x^4}{n^2}\right)^{n}\right]\leqslant e^{-x^2}\frac{x^4}{n}.$$

For $x \in [-L,L]$, the RHS is bounded by $L^4/n$, which proves uniform convergence.

  • $\begingroup$ You know that $e^y > 1 +y$. Take $y = x^2/n$ and we get $e^{x^2/n} > 1+x^2/n$ which implies the inequality on the right. $\endgroup$ – RRL Apr 30 '18 at 17:44
  • $\begingroup$ You also have $e^{-y} > 1 - y$ which leads to the left inequality. Do I need to add this to the answer? $\endgroup$ – RRL Apr 30 '18 at 17:46
  • $\begingroup$ I assume you are familiar with $e^y > 1 + y$ which follows from the Taylor expansion for y > 0$. $\endgroup$ – RRL Apr 30 '18 at 17:47
  • $\begingroup$ Thank you very much. No need to add it. $\endgroup$ – Um Shmum Apr 30 '18 at 17:49
  • 1
    $\begingroup$ @Um Shmum: You're welcome. $\endgroup$ – RRL Apr 30 '18 at 17:54

To find $\sup \limits _{x\in\mathbb{[-L,L]}}|f_n(x)-e^{-x^2}|$; supremum will be exist for some $ {x\in\mathbb{[-L,L]}}$.Suppose $x=a$ then $\lim\limits_{n\rightarrow\infty} \sup |f_n(x)-e^{-x^2}|=0$.Hence uniformly convergent

  • $\begingroup$ This just shows pointwise convergence, for $x=a$, unless you can somehow show that it's the same $a$ for all $n\in \mathbb{N}$ $\endgroup$ – Um Shmum Apr 30 '18 at 17:52
  • $\begingroup$ I take that a for which supremum willbe exist and after calculating supremum I take limit $\endgroup$ – Epsilon Delta Apr 30 '18 at 19:15

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.