Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Assume $U$ is a bounded open subset of $\mathbb{R}^N$. Furthermore $\Delta: H^2(U) \subset L^2(U) \to L^2(U)$ is the Laplace operator. My question is:

What is the domain of $\vert \Delta\vert^{1/2}$?

I've seen this notation a lot recently and still don't know what to make of it. First of all the Laplace operator is not a positive operator. I guess, that is what the $\vert \cdot \vert$ is for.

Here is a definition of the root of $-\Delta$, which suggests that the domain could be $L^1 (U)$.

It would be perfect for my cause, if the domain could be taken as $H^2(U)$ again.

share|cite|improve this question
up vote 4 down vote accepted

Strictly speaking, this is not a well-posed question: the definition of an operator, as well as of any function, must include the domain. If we don't know what the domain of $|\Delta|^{1/2}$ is, then we don't know what $|\Delta|^{1/2}$ means.

The fractional Laplacian is a nonlocal operator: we can't find the values of $(-\Delta)^{1/2}u$ on a domain just by looking at the values of $u$ on that domain. This is why the papers dealing with fractional Laplacian normally work on $\mathbb R^n$ rather than on $U$. (If the paper you are reading is an exception to this, please give a reference.) So, in order to find $(-\Delta)^{1/2}u$, we must decide how to extend the function $u$ outside of its original domain $U$.

You could extend by zero outside of $U$, but this may be a bad move. The boundary discontinuity can take $u$ out of all Sobolev spaces, and $(-\Delta)^{1/2} u$ will likely fail to be integrable on $U$. (It will be in $L^2$ locally, though.)

Harmonic extension outside of $U$ works better, and is easy to implement as follows: first take $\Delta u$ (local operator), then extend $\Delta u$ by zero outside of $U$, and finally apply $(-\Delta)^{-1/2}$, which is realized by convolution with integrable kernel $\|x\|^{-n+1}$. If $u\in H^2(U)$, this process yields $(-\Delta)^{1/2} u\in L^2(U)$ (in fact a little better), which apparently is what you wanted.

You may find the nonlocal PDE wiki helpful.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.