Total positive curvature of hypersurfaces with convex boundary (Q2503412)

Let \(\Sigma\) be a compact \(C^0\) hypersurface with boundary \(\partial\Sigma\) in \(\mathbb{R}^n\), which is \(C^1\)-immersed on a neighborhood of \(\partial\Sigma\) (it is not required that \(\Sigma\) be locally embedded away from \(\partial\Sigma\)). A hyperplane \(\Pi\subset\mathbb{R}^n\) is called a restricted support hyperplane of \(\Sigma\) at a point \(p\), if \(p\in\Pi\cap\Sigma\), \(\Sigma\) lies on one side of \(\Pi\), and \(\Pi\) is tangent to \(\Sigma\) when \(p\in\partial\Sigma\). An outward normal of \(\Pi\) is a normal vector to \(\Pi\) which points towards a side of \(\Pi\) not containing \(\Sigma\). If \(\Pi\) is a restricted support hyperplane for an open neighborhood \(U_p\) of \(p\) in \(\Sigma\), then \(\Pi\) is called a restricted local support hyperplane; furthermore, \(p\) is a locally strictly convex point of \(\Sigma\), or \(p\in\Sigma^+\), provided that \(\Pi\cap U_p= \{p\}\). The total positive curvature \(\tau^+\) of \(\Sigma\) is defined as the algebraic area of the unit normals to restricted local support hyperplanes of \(\Sigma\) at points of \(\Sigma^+\), where by area, we mean the \((n-1)\)-dimensional Hausdorff measure. It remains to set \({\mathbf c}_n:= \text{area}({\mathbf S}_n)\), and recall that the inward conormal at \(p\in\partial\Sigma\) is a unit normal vector of \(\partial\Sigma\) at \(p\) which is tangent to \(\Sigma\) and points inside \(\Sigma\). Theorem. Let \(\Sigma\) be a compact \(C^0\)-hypersurface in \(\mathbb{R}^n\) which is \(C^1\) immersed on a neighborhood of its boundary \(\partial\Sigma\). Suppose that \(\partial\Sigma\) lies on the boundary of a convex set \(C\subset\mathbb{R}^n\), and at each point \(p\in\partial\Sigma\), the inward conormal \(\sigma(p)\) is an outward unit normal to a support hyperplane of \(C\). Then \(\tau^+(\Sigma)\geq {{\mathbf c}_{n-1}\over 2}\). Equality holds if and only if {\parindent=8mm \begin{itemize}\item[(i)]\(\partial\Sigma\) lies in a hyperplane \(\Pi\),\item[(ii)]\(\sigma(p)\bot\Pi\) for all \(p\in\partial\Sigma\),\item[(iii)]\(\Sigma\) lies strictly on one side of \(\Pi\), and \item[(iv)]every restricted local support hyperplane of \(\Sigma\) at each point of \(\Sigma^+\) is a restricted support hyperplane of \(\Sigma\) \end{itemize}}