Let $(g_n(x))$ a sequence of function such that $g_n :\mathbb{R}_+\to \mathbb{R}$ and $g_n(x)=\dfrac{x}{n^2}e^{-\frac{x}{n}}$

The target is to show $(g_n(x))$ converges uniformly.

Attempt :

  1. I show the pointwise convergence.

For $x=0$, $\quad g_n(x)=0$ $\quad \forall n\in \mathbb{N}^*$

For $x\in ]0,+\infty),$ $\quad g_n(x)\underset{n\to+\infty}{\longrightarrow}0$

  1. I show the uniform convergence to $f:x\to f(x)=0\qquad$

    $\underset{x\in\mathbb{R}_+}\sup\big\{|g_n(x)-f(x)|\big\}<\varepsilon \iff \underset{x\in\mathbb{R}_+}\sup\bigg\{\bigg|\frac{x}{n^2}e^{-\frac{x}{n}}\bigg|\bigg\}<\varepsilon$

$g_n'(x)=\dfrac{1}{n^2}\dfrac{(1-x)e^{x/n}}{e^{2x/n}}$ thus $g_n$ is increasing on $[0,1]$ and decreasing on $[1,+\infty]$ as $g_n(0)=0$ and $g_n(x)\underset{x\to +\infty}{\longrightarrow}0$ then $\underset{x\in\mathbb{R}_+}\sup\bigg\{\bigg|\frac{x}{n^2}e^{-\frac{x}{n}}\bigg|\bigg\}=\dfrac{1}{n^2e^{1/n}}\underset{n\to +\infty}{\longrightarrow}0$

Hence $\boxed{g_n\overset{\text{unif.}}{\longrightarrow}f}$

Is that correct, and are there other methods more efficient ??

  • 2
    $\begingroup$ Alternatively notice that $h(y)=y e^{-y}$ is a bounded function and therefore $g_n(x) = 1/n h(x/n)\le C/n\to 0$ with $C>0$ some constant independent of $x$. $\endgroup$ – J.R. May 25 '18 at 21:09

You can shorten it. Consider $f : [0, +\infty\rangle \to \mathbb{R}$ defined as $f(t) = te^{-t}$.

$f$ is continuous and $\lim_{t\to+\infty} f(t) = 0$ and so $f$ is bounded on $[0, +\infty\rangle$, i.e. there exists $M > 0$ such that $f(t) \le M, \forall t \ge 0$.

Now we have

$$g_n(x) = \frac1n \cdot \frac{x}{n}e^{-\frac{x}n} = \frac1n f\left(\frac{x}n\right) \le \frac{M}n \xrightarrow{n\to\infty} 0$$ uniformly in $x \in [0, +\infty\rangle$.

Hence $g_n \xrightarrow{\text{unif.}} 0$.


It is correct, but computing $g_n'$ is easier if you do it like this:$$g_n'(x)=\frac1{n^2}\left(x'e^{-\frac xn}+x\left(e^{-\frac xn}\right)'\right)=\frac1{n^2}\left(e^{-\frac xn}-\frac xne^{-\frac xn}\right).$$


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.