3
$\begingroup$

Let $\Omega$ be a subset of either $\mathbb{R}^n$ ($n\geq 3$, if it matters) or of a compact manifold. In either case, we'll call the manifold $M$. Let $V_i\subset V_{i+1} \subset \Omega$ be an increasing sequence of subsets. Assume that $\cup V_i = \Omega$ up to a set of measure zero. Let $$A(\Omega) = \{f\in W^{1,2}_c(M): f|_{M\setminus \Omega} = 0\},$$ or, in other words, $A(\Omega)$ are the $W^{1,2}$ functions with compact support that are zero except on $\Omega$.

I want to know under what conditions $\cup A(V_i)$ is dense in $A(\Omega)$. If it is easier, we can assume $\Omega$ and/or the $V_i$ are open, or similar, but I am interested in the case where they may only be measurable.

If $\Omega$ is open and the $V_i$ are $\Omega\setminus B_{1/i}$, where $B_{1/i}$ is a ball of radius $1/i$, then it is easy to show that $\cup A(V_i)$ is dense in $A(\Omega)$ by using a cutoff function. The same kind of argument likely works for general $V_i$ that are $\Omega$ minus some open set with "nice" boundary, but I'm not sure about anything more general than that.

$\endgroup$
3
  • $\begingroup$ Do you want the set $\Omega\setminus\bigcup_iV_i$ to have zero measure or zero capacity? Capacity should be a more relevant concept in situations like this. $\endgroup$ Commented Sep 25, 2014 at 18:43
  • $\begingroup$ Zero measure. The point of that is that a set of measure zero won't change what $A(V_i)$ is, since I'm only dealing with weakly differentiable functions. The important property is that integration over $\Omega$ or $\cup V_i$ should give the same value. $\endgroup$ Commented Sep 26, 2014 at 17:04
  • $\begingroup$ If $\Omega$ is the unit ball in $\mathbb R^n$ and $V_i=\{x\in\mathbb R^n;|x|<1-1/i,|x_n|>1/i\}$, then the closure of $\bigcup_iA(V_i)$ seems to be $W^{1,2}_0(\Omega_+)\oplus W^{1,2}_0(\Omega_-)$, where $\Omega_\pm=\{x\in\Omega;\pm x_n>0\}$. The sets $V_i$ fill $\Omega$ up to a hyperplane, but that's not enough. This seems to be (at least vaguely) related to removable sets for Sobolev spaces. $\endgroup$ Commented Sep 26, 2014 at 17:35

0

You must log in to answer this question.

Browse other questions tagged .