Is $C^\infty(\mathbb{R})\subset C_b(\mathbb{R})$ dense? I.e. is any continuous bounded function $f:\mathbb{R}\to\mathbb{R}$ the uniform limit of smooth functions?

On any bounded interval this is true since by Stone-Weierstrass polynomials are smooth and dense in the continuous functions. I guess the result is still true for $\mathbb{R}$ but I don't know how to prove it. Since I don't have any integrability conditions on $f$ the technique of convoluting with mollifiers is probably not applicable.

  • $\begingroup$ What about convoluting the approximating smooth functions on intervals $[n,n+1]$? $\endgroup$ – Berci May 19 '13 at 10:06

Given continuous $f$ and $\epsilon>0$, find for each $n\in\mathbb Z$ a smooth approximation $g_n\colon [n-1,n+1]\to\mathbb R$ with $\left\lVert f|_{[n-1,n+1]}-g_n\right\rVert_\infty<\epsilon$. Glue these together with a smooth partition of one.

For example: Start with a smooth function $\phi\colon[0,1]\to [0,1]$ with $\phi(x)=1$ for $x\le\frac13$ and $\phi(x)=0$ for $x\ge \frac23$. Then $$g(x):=\phi(x-\lfloor x\rfloor)g_{\lfloor x\rfloor}(x)+(1-\phi(x-\lfloor x\rfloor))g_{\lceil x\rceil}(x)$$ is smooth (the trick is that $g(x)=g_n(x)$ if $|x-n|<\frac13$) and at each $x$ we have $|f(x)-g(x)|<\epsilon$ because it is a convex combination. Therefore, $\lVert f-g\rVert_\infty<\epsilon$.

  • 1
    $\begingroup$ So boundedness is not actually needed? $\endgroup$ – Amr Dec 19 '15 at 20:35

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.