On the symmetry function of a convex set (Q2467164)

Let \(S\subset R^{n}\) be a convex body. The symmetry of \(S\) about \(x\in S\) is \(\mathbf{sym}(x,S):=\max \{\alpha \geq 0:x+\alpha (x-y)\in S\) for every \(y\in S\}\). The function \(h\left(\cdot,S\right):=\frac{\mathbf{sym}(\cdot,S)}{1+\mathbf{sym}(\cdot ,S)}\) is concave on \(S\), and hence \(\mathbf{sym}(\cdot ,S)\) is logconcave. The authors obtain many results related to this function, including: 1) bounds on the volume of the intersection of \(S\) with a halfspace containg \(x\) in its boundary \(H\), as well as on the volume of \(S\cap H\), in terms of \(\mathbf{sym}(x,S)\); 2) upper bounds on the distance with respect to any norm and on the so-called cross-ratio distance between two points \(x,y\in S\) in terms of \(\mathbf{sym}(x,S)\) and \(\mathbf{sym}(y,S)\); 3) relations between \(\mathbf{sym}(\cdot ,S)\) and \(\mathbf{sym}(\cdot ,P)\) when \(P\) is an approximation of \(S\); 4) bounds on the expected value of \(\mathbf{sym}(X,S)\) when \(X\) is a random vector uniformly distributed on \(S\); 5) a characterization of the symmetry points of \(S\) (a point \(x^{\ast }\in S\) is said to be a symmetry point of \(S\) if \(\mathbf{sym}(x^{\ast },S)=\max_{x\in S}\text\textbf{sym}(x,S)\)). Methods for computing an \(\epsilon \)-approximate symmetry point of a convex polyhedron \(S\) are also provided.