The exercise states:

Let $a,b\in\mathbb{R}$, $a<b$ and let $(C[a,b],\Vert\cdot\Vert)$ denote the vector space of continuous real functions on $[a,b]$ endowed with the uniform norm. Let $C^1[a,b]\subset C[a,b]$ denote the set of continuously differentiable functions and define $$ M_K=\{f\in C[a,b]: |f(x)-f(y)|\leq K|x-y|\quad\mathrm{for\;every\;}x,y\mathrm{\;in\;}[a,b]\}\\ A_K=\{f\in C^1[a,b]: |g'(x)|\leq K \quad\mathrm{for\;every\;}x\mathrm{\;in\;}[a,b]\} $$ Show that $\bar{A_K}=M_K$

I've shown that $M_K$ is closed and therefore, the fact that $A_K\subset M_K$ implies $\bar{A_K}\subset M_K$ but I couldn't prove the other inclusion. My attempt so far was to take advantage of the fact that we are dealing with a metric space and hence it will suffice to show that every $f\in M_K$ can be approximated by a sequence of functions in $A_K$. This is somehow equivalent to show that every Lipschitz function can be approximated by continuously differentiable functions, but this is something I couldn't show.

So...could you please give me some hints to solve this? Thank you in advance.

  • $\begingroup$ Suppose $f \notin \bar{A}_k$ $\endgroup$ – Matt A Pelto Sep 3 '14 at 22:44
  • 2
    $\begingroup$ $f\in M_K$ is absolutely continuous with $|f'|\le K$. Approximate $f'$ in $L^1$ by an $h$ with $|h|\le K$ and set $g(x)=f(a)+\int_a^x h(t)\, dt$. $\endgroup$ – user138530 Sep 4 '14 at 2:06

One approach was pointed out by Christian Remling in a comment: use the fact that Lipschitz functions are absolutely continuous, hence represented as an indefinite integral of an $L^1$ function. Approximate that function by a continuous one, truncating to size if necessary. Then integrate.

Here is another, more direct approach. Extend $f$ to $\mathbb{R}$ by letting $f(x)=f(a)$ for $x<a$ and $f(x)=f(b)$ for $x>b$. This is still a Lipschitz function, same constant. Let $f_\epsilon$ be the convolution of $f$ with a standard mollifier $\phi_\epsilon$. Then

  1. $f_\epsilon$ is $C^\infty$ smooth,
  2. $f_\epsilon\to f$ uniformly as $\epsilon\to0$,
  3. $f_\epsilon$ is $K$-Lipschitz because $$|f_\epsilon(x)-f_\epsilon(y)|\le |f(\cdot )-f(\cdot +y-x)|*\phi_\epsilon\le K|x-y|$$

Hence, $|f'|\le K$ everywhere.


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.