## Interpolation of derivatives

If $U$ is an open interval of $\mathbb{R}$ and $f : U \to \mathbb{R}$ is an $L^2(U)$ function with second derivative $f'' \in L^2(U)$ (in the weak sense), is $f \in W^{1,1}(U)$?

EDIT: Removed false inequality.

-
Sorry, but I can't seem to fix the TeX... – John H Jan 22 2012 at 1:05
I think linear functions might break your inequality what with the zero second derivative and non-zero first derivative. – BSteinhurst Jan 22 2012 at 2:31
I fixed your LaTeX, but don't immediately see how to fix your inequality – Yemon Choi Jan 22 2012 at 2:45
You are totally right, that inequality is false. – John H Jan 22 2012 at 2:51

I assume $U$ is a finite open interval, else the assertion is clearly false (let $f(x)=x$).

Then a standard estimate shows that $f'$ is bounded, and thus in $L^1(U)$, whence $f$ is in the Sobolev space $W^{1,1}(U)$ (in fact in $W^{1,p}(U)$ for all $p$).

Fix some $x_0 \in U$, and write $$\left|f'(x) - f'(x_0)\right| = \left| \int_{x_0}^x f''(y) dy \right| \leq \| f'' \|_2 \phantom. |x-x_0|^{1/2},$$ using Cauchy-Schwarz in the last step. Since $\| f'' \|_2$ is a finite constant and $|x-x_0|$ is bounded, so is $\left|f'(x) - f'(x_0)\right|$, and we are done.

This kind of argument is of course well-known, and probably predates Sobolev himself, but is easier to write up than to look up. A reader better versed in the literature may be able to supply a canonical reference.

-
This argument also shows that $f'$ is of Hölder class $C^{0, 1/2}$ and in particular is continuous. So the conclusion is that $f$ belongs to $C^{1, 1/2}$, and in particular is continuously differentiable. – Phil Isett Jan 22 2012 at 4:31