The mean curvature of a Lipschitz continuous manifold (Q2568757)

For an open set \(U\in\mathbf R^{n+1}\) a function \(H\in L^ 1(U)\) is said to be a variational mean curvature of a given set \(E\subset U\) if \(E\) locally minimizes the functional \(\mathcal F_ H(F) = \int\limits_ U|D\varphi_ F|+\int\limits _{U\cap F}\!H(x)\, dx\). If \(H\) is a variational mean curvature of \(E\), \(\partial E\) is a smooth manifold in a neighborhood of a point \(x\in \partial E\cap U\), and \(H\) is a continuous function at \(x\), then \(H(x)\) is the classical mean curvature of \(\partial E\) at \(x\). If \(H\) is a variational mean curvature of \(E\) and \(H\in L^ p(U)\) with \(p > n + 1\), then there is the decomposition \(\,\partial E\cap U=\Sigma_ r\cup \Sigma _ s\), where the so-called regular subset \(\Sigma_ r\) is an \(n\)-dimensional \(C^ {1,\alpha}\) manifold, \(\Sigma_ s\) is a closed subset of \(\,\partial E\cap U\), and \(\mathcal H_ k(\Sigma_ s)=0\). On the other hand, the mean curvature of non-regular manifolds can be defined in a different way by using the methods of Geometric Measure Theory. Let \(M\subset U\) be an \(n\)--dimensional Lipschitz--continuous manifold such that \(\left|\int\limits _ M\text{div}_ MX\,d\,\mathcal H_ n\right|\leq K\|X\|_\infty\), where \(X\in C^1_ 0(U,\mathbf R^{n+1})\) and \(K>0\). There exists an \((n+1)\)-dimensional vector valued Radon measure on \(M\), denoted by \(\overarrow{\mathbf H}=({\mathbf H}_ 1,\dots,{\mathbf H}_ {n+1})\), such that \(\int\limits _ M\text{div}_ MX\,d\,\mathcal H_ n=-\int\limits_ MX\!\bullet\!d\overarrow{\mathbf H}=-\sum\limits_ {j=1}^ {n+1}X_ jd\,{\mathbf H}_ j\). The measure \(\overarrow{\mathbf H}\) is called the mean curvature measure of \(M\) in \(U\). In this paper, the authors present some connections between the mean curvature in a distributional sense and the mean curvature in a variational sense for several classes of non-smooth sets. They prove the existence of the mean curvature measure of \(\partial E\) using the concept of variational mean curvature. That is, under suitable assumptions, the mean curvature measure of \(\,\partial E\cap U\) is the weak limit (in the sense of measures) of the mean curvatures of a sequence of regular \(n\)-dimensional manifolds \(M_ j\) convergent to \(\partial E\). The manifolds \(M_ j\) are closely related to the level surfaces of the variational mean curvature \(H_ E\) of E. It is shown that if \(\,\partial E\cap U\) is locally the graph of a function \(f\in C^{1,\alpha}\) that is a weak supersolution of the minimal surface equation or \(E\) is a convex set, then there exists an \((n+1)\)-dimensional vector valued Radon measure on \(\overarrow{\mathbf H}\) such that \(\int \limits_ M\text{div}_ MX\,d\,\mathcal H_ n=-\int\limits_ MX\!\bullet\!d\overarrow{\mathbf H}\), where \(M=\partial E\).