I was looking at the question whether $$ \int_{\mathbb{R}^N} \left|f(x+h)-f(x)\right|^p dx \leq \left( |h| \cdot \left\Vert f\right\Vert_{W^{1,p}(\Omega)} \right)^p $$ holds for a function $f\in W^{1,p}(\Omega)$ for $1\leq p < \infty$. In Evans' book on PDE, chapter 5 on Sobolev spaces, difference quotients are defined in section 5.8.2. Then follows theorem 3:

$(i)$ Suppose $1\leq p < \infty$ and $u\in W^{1,p}(U)$. Then for each $V\subset \subset U$ $$\left\Vert D^hu\right\Vert_{L^{p}(V)} \leq C \left\Vert Du\right\Vert_{L^{p}(U)} $$ for some constant $C$ and $0<|h| < \frac{1}{2} \text{dist}(V,\partial U)$.

I am pretty sure that there is no neead to assume the bound on $h$ and $V\subset\subset U$.

My question for this post is: Can I use the above theorem to prove my initial statement?


If $f\in W^{1,p}(\Omega)$, then you cannot write $\int_{\mathbb{R}^N} \left|f(x+h)-f(x)\right|^p dx$ since $f$ is not defined outside $\Omega$. Instead of $\mathbb{R}^N$ you can use the set $$\Omega_h=\{x\in\Omega:\, x+th\in\Omega\text{ for all }t\in [0,1]\}.$$ You need the segment to be contained in $\Omega$ because the proof relies on the fundamental theorem of calculus along that segment. This is why Evans was restricting to the sets $U$ and $V$, to make sure $x+h$ stays inside $\Omega$.

PS: Extending $f$ to be zero outside $\Omega$ will not do because you will lose the property that $f$ is a Sobolev function and won't be able to use the FTC.


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.