A characterization of Newtonian functions with zero boundary values (Q663335)

From MaRDI portal
scientific article
Language Label Description Also known as
English
A characterization of Newtonian functions with zero boundary values
scientific article

    Statements

    A characterization of Newtonian functions with zero boundary values (English)
    0 references
    0 references
    0 references
    0 references
    14 February 2012
    0 references
    Let \(X=(X,d,\mu)\) be a metric measure space, where \(d\) is the metric and \(\mu\) is a doubling Borel regular outer measure. One can define first-order Sobolev spaces \(N^{1,p}\), \(1\leq p<\infty\), in this general setting, but additional assumptions such as a Poincaré inequality are usually required in order to develop a satisfactory theory. The present paper concerns the subspace \(N^{1,p}_0(\Omega)\subset N^{1,p}(\Omega)\) of Sobolev functions with zero boundary values. Here \(\Omega\) is an open bounded set and \(N^{1,p}_0(\Omega)\) consists of all functions in \(N^{1,p}(\Omega)\) whose extension of zero yields a function in \(N^{1,p}(X)\). While this definition is very natural in the Sobolev theory, one would like to know to what extent it is related to the pointwise vanishing of \(u\) on the boundary of \(\Omega\). The answer is given by the following result. Theorem. In addition to the above, assume that \(X\) supports a strong relative isoperimetric inequality. A function \(u\in N^{1,p}(\Omega)\) belogns to \(N^{1,p}_0(\Omega)\) if and only if \[ \lim_{r\to 0}\frac{1}{\mu(B(x,r))}\int_{B(x,r)\cap \Omega} |u| \,d\mu= 0 \] for \(p\)-quasievery \(x\in \Omega\). The term \(p\)-quasieverywhere refers to vanishing \(p\)-capacity of the exceptional set. A metric space \(X\) supports a strong relative isoperimetric inequality if there exists positive constants \(C\) and \(\lambda\) such that for any ball \(B\subset X\) and any Borel set \(E\subset X\) one has \[ \min \big\{\mu(B\cap E),\mu (B\setminus E)\big\} \leq \text{Cr}\mathcal{H}(\lambda B\cap \partial^*E). \] Here \(r\) is the radius of \(B\), \(\partial^*E\) is the measure-theoretic boundary of \(E\), and \(\mathcal{H}\) is the codimension-one Hausdorff measure. It is shown in the paper that the strong relative isoperimetric inequality implies the \((1,1)\)-Poincaré inequality. Another interesting result of the paper is the following Lusin-type approximation theorem. Theorem. If \(X\) supports a strong relative isoperimetric inequality and \(u\in N^{1,p}(X)\), then for every \(\epsilon>0\) there is an open set \(E_\epsilon\) of \(p\)-capacity less than \(\epsilon\), and a continuous function \(u_\epsilon\) that agrees with \(u\) on \(X\setminus E_{\epsilon}\) and satisfies \(\|u-u_\epsilon\|_{N^{1,p}(X)}<\epsilon\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    Sobolev functions on metric spaces
    0 references
    bounded variation
    0 references
    isoperimetric inequality
    0 references
    capacity
    0 references
    0 references
    0 references