A direct approach to Plateau's problem (Q2012445)

If \(H\subset\mathbb R^{n+1}\) is any closed set, \(\mathcal{P}(H)\) is a class of relatively closed subsets \(K\) of \(\mathbb R^{n+1}\setminus H\), then the formulation of Plateau's problem involves the minimization of the Hausdorff measure, where the relevant minimum is \(m_0=\inf\{\mathcal{H}^n(K)\}\) for \(K\in\mathcal{P}(H)\), and a minimizing sequence \(\{K_j\}\subset\mathcal{P}(H)\) is characterized by the property \(\mathcal{H}^n(K_j)\to m_0\). If \(H\subset\mathbb R^{n+1}\) is closed, \(K\subset\mathbb R^{n+1}\setminus H\), and \(B_{x,r}=\{y;\;|y-x|l<r\}\subset\mathbb R^{n+1}\setminus H\), then the set \((K\setminus B_{x,r})\cup\{\lambda x+(1-\lambda) z\}\), where \(z\in K\cap\partial B_{x,r}\) and \(\lambda\in[0,1]\), is called the cone competitor for \(K\) in \(B_{x,r}\), and any set of the form \((K\setminus B_{x,r})\cup(\partial B_{x,r}\setminus A)\) is called a cup competitor for \(K\) in \(B_{x,r}\), where \(A\) is a connected component of \(\partial B_{x,r}\setminus K\). An element \(K\in\mathcal{P}(H)\) is said to have good comparison property in \(B_{x,r}\) relative to \(\mathcal{P}(H)\) if \(\inf\{\mathcal{H}^n(J)\}\leq\mathcal{H}^n(L)\) whenever \(L\) is the cone competitor or any cup competitor for \(K\) in \(B_{x,r}\), where \(J\in \mathcal{P}(H)\) and \(J\setminus\text{cl}(B_{x,r})=K\setminus\text{cl}(B_{x,r})\). The family \(\mathcal{P}(H)\) is said to be a good class if the set \(K\) has the good comparison property in \(B_{x,r}\) for any \(K\in\mathcal{P}(H)\) and for every \(x\in K\) and \(r\in(0,\text{dist}(x,H))\). In this paper, the authors provide a compactness principle which is applicable to different formulations of Plateau's problem in codimension one and which is exclusively based on the theory of Radon measures and elementary comparison arguments. They show that if \(H\subset\mathbb R^{n+1}\) is closed, \(\mathcal{P}(H)\) is a good class, the infimum \(m_0\) is finite, and \(\{K_j\}\subset\mathcal{P}(H)\) is a minimizing sequence of countably \(\mathcal{H}^n\)-rectifiable sets, then, up to subsequences, the measures \(\mu_j=\mathcal{H}^n\lfloor K_j\) converge weakly\(^*\) in \(\mathbb R^{n+!}\setminus H\) to a measure \(\mu=\theta\mathcal{H}^n\lfloor K\), where \(K=\text{spt}\mu\setminus H\) is a countably \(\mathcal{H}^n\)-rectifiable set and \(\theta\geq 1\).