Maximization of quadratic forms expressed by distance matrices (Q2641682): Difference between revisions

A real symmetric matrix \(C\) is called nonnegative, \(C\geq0\), if \((Cx,x)\geq0\) for all \(x\in\mathbb{R}^n\). A matrix \(A=(a_{ij})\) with \(a_{ii}=0\) and \(a_{ij}=a_{ji}\geq0\) for \(i\neq j\) is called a distance matrix. If \(x^t=(x_1, \ldots, x_n)\) then \(v=\frac{1}{2}(Ax,x)=\sum_{1\leq i<j\leq n}a_{ij}x_ix_j\) is a quadratic form. Let \(F=(e_{ij})\) where \(e_{ij}=1\) for all \(i,j\). A matrix \(A\) is said to be \(F\)-bounded above if there is some \(\lambda>0\) such that \(A-\lambda F\geq0\). Put \(\lambda_A=\inf \{\lambda\mid A-\lambda F\geq0,\;\lambda>0\}\). The authors investigate properties of distance matrices, in particular they obtain conditions for a distance matrix \(A\) to be \(F\)-bounded above; also for \(\sum_{i=1}^n x_i=k\) they show that \(v_{max}=\frac{1}{2}\lambda_Ak^2\).