For $n \in \mathbb{N}$, define the formula, $$f_n(x)= \frac{x}{2n^2x^2+8},\quad x \in [0,1].$$ Prove that the sequence $f_n$ converges uniformly on $[0,1]$, as $n \to \infty$.

I know that the definition says $f_n$ converges uniformly to $f$ if given $\forall \epsilon \gt 0$, $\forall n \geq N$, such that $|f_n(x) - f(x)| \lt \epsilon, \forall n \geq N$ and $\forall x \in [0,1].$

I looked first at the pointwise convergence and found that $$\lim_{n \rightarrow \infty} \frac{x}{2n^2x^2+8} = 0, \forall x \in [0,1].$$

So how do I use this to choose an $n \geq N$ such that $|f_n(x) - f(x)| \lt \epsilon$ ?

Right now, I have

"proof: Let $\epsilon > 0, \exists N \in \mathbb{N}$ such that, n $\geq N \Rightarrow \frac{1}{2n^2+8} \lt \epsilon$,

by $|f_n(x) - 0| = |\frac{x}{2n^2x^2+8}| \leq |\frac{x^2}{2n^2x^2+8}| \leq \frac{1}{2n^2x^2+8} \;\;\;\; \forall x \in [0,1].$

Since $\lim_{n \rightarrow \infty} \frac{x}{2n^2x^2+8} = 0, \forall x \in [0,1]$, $f_n(x)$ will converge uniformly to $0$ on $[0,1]$."

Is this correct? Am I missing something? Is something not correct? I'm unsure about my choice of $N$. Please & thanks!

  • $\begingroup$ Hi Jamie, you seemed to have proved the sequence converges pointwise, but not uniformly. $\endgroup$ – Suugaku Apr 23 '13 at 4:37
  • $\begingroup$ What is missing from my proof to make it uniform? I thought if I proved it pointwise, and then showed that it converges $\forall n \geq N$ and $\forall x \in [0,1]$, then that implies uniform convergence? $\endgroup$ – Jaime Apr 23 '13 at 5:24

related problem: (I), (II), (III). Here is a systematic technique. In order to find $\sup_{0\leq x\leq 1} |f_n(x)-f(x)| $, you need to maximize the function $\Big|\frac{x}{2n^2x^2+8}\Big|$ over the interval $[0,1]$. Now, let

$$ g(x)=\frac{x}{2n^2x^2+8} \implies g'(x) = \frac{4-n^2 x^2}{(2n^2x^2+8)^2}=0 \implies x=\frac{2}{n}$$

gives the max of the function $g(x)$ which is $g(2/n)=1/8n$. You can check this by checking the sign of $g''(x)$ which should be $< 0$. Hence we have

$$ \sup_{0\leq x\leq 1} |f_n(x)-f(x)|= \sup_{0\leq x\leq 1} \Big|\frac{x}{2n^2x^2+8}\Big|=\frac{1}{8n} < \epsilon. $$

  • $\begingroup$ Using your technique, which I have seen in three posts already, how would you apply it to a sequence of functions of this form: $f_n(x)=x^{1+\frac{1}{2n-1}}$ $\endgroup$ – funmath May 5 '18 at 19:06

You can write $$f_n(x)={1\over n} g(n\>x)\qquad(0\leq x\leq 1)$$ with $$g(t)={t\over 2t^2+8}\qquad(0\leq t<\infty)\ .$$ Since $g(t)$ converges to $0$ for $t\to\infty$ we already can say that $g$ is bounded. A quantitative estimate can be obtained as follows: For $t>0$ one has $$0<g(t)={1\over 8}\ {2\over{t\over 2}+{2\over t}}\leq{1\over 8}\ .$$ Therefore we now have $$|f_n(x)|\leq{1\over 8n}\qquad(n\geq1,\ 0\leq x\leq1)\ ,$$ which shows that the $f_n$ converge uniformly to $0$ on $[0,1]$.


Another way to maximize :

$\dfrac{x}{2n^2x^2+8}\leq \dfrac{x}{8nx}=\dfrac{1}{8n}$ where we used the AM-GM inequality :http://en.wikipedia.org/wiki/Inequality_of_arithmetic_and_geometric_means to show

$\dfrac{2n^2x^2+8}{2} \geq \sqrt {2n^2x^2 \cdot 8} \ \ \ \ $ Applying this easy inequality is often a good first way to find a supremum for the function if you have a summation involved. So since the function converges pointwise to $f(x)=0$ We have the following result:

$\lim_{n \to \infty}\sup_{0\leq x\leq 1} |f_n(x)-f(x)|=\lim_{n \to \infty}|\dfrac{1}{8n}|=0$


Pointwise convergence is not enough to say that the function converges uniformly. Here, $$f_n(x)= \frac{x}{2n^2x^2+8},x \in [0,1]$$ has pointwise convergence to $f(x)=0$, so by definition $$|f_n(x)-f(x)|=\bigg|\frac{x}{2n^2x^2+8}-0\bigg|= \bigg|\frac{x}{2n^2x^2+8}\bigg|< \frac{1}{2n}$$ This shows that that the function is not uniformly convergent.

  • $\begingroup$ Please read this tutorial on how to typeset mathematics on this site. $\endgroup$ – N. F. Taussig Jan 19 '16 at 17:00

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.