The trace theorem for nice enough domains states that there is a operator $T:H^1(\Omega) \to L^2(\partial \Omega)$ such that $$|Tu|_{L^2(\partial \Omega)} \leq C|u|_{H^1}.$$

My question, is there an expression for the constant $C$? I want to see exactly how it depends on the domain $\Omega.$ This is because I want to see how the constant varies (eg. continuously) if I vary the domain.


2 Answers 2


Such an estimate can be found in Grisvard "Elliptic problems in nonsmooth domains", Theorem

It basically says $$ \delta \|u\|_{L^2(\partial\Omega)}^2 \le \|\mu\|_{C^1(\bar\Omega)} \left(\epsilon^{1/2}\|\nabla u\|_{L^2(\Omega)}^2 + (1+\epsilon^{-1/2}) \|u\|_{L^2(\Omega)}^2 \right) $$ for all $\epsilon\in(0,1)$, $u\in H^1(\Omega)$. The vector field $\mu$ has to be chosen to be $C^1(\bar\Omega,\mathbb R^n)$ such that $$ \mu \cdot \nu \ge \delta $$ on $\partial \Omega$ with $\nu$ the outer normal vector.

This could give you an estimate of the constant for a fixed domain at least. It should help to prove continuity with respect to domain variations as well.


This is an old question, but since I stumbled upon it when wondering the same thing, I'll just add two points:

(1) The trace theorem is actually sharper; you can bound even the $H^{1/2}$ norm (not just the $L^2$ norm) of the trace on $\partial \Omega$ in terms of the $H^1$ norm of the function on $\Omega$.

(2) I found this reference and also this which characterize and approximate the value of the constant in the Sobolev trace theorem. I haven't read either one carefully, but they appear interesting and on point.


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.