The cone of distance matrices (Q2639130): Difference between revisions

An \(n\times n\) matrix \(D=(d_{ij})\) is called a Euclidean distance matrix if there are n points \(P_ 1,...,P_ n\) in r-dimensional real space for which \(d_{ij}=\| P_ i-P_ j\|^ 2.\) The least dimensional r for which such points exist is called the embedding dimension of D. The present paper presents another characterization of the Euclidean distance matrices, analyzes the facial structure of this cone of matrices, and solves a special case of the problem of finding the Euclidean distance matrix which minimizes the distance to the set of nonnegative symmetric matrices with zero diagonal.