Multispherical Euclidean distance matrices (Q977468): Difference between revisions

An \(n \times n\) symmetric and nonnegative matrix with zero diagonal entries \(D=(d_{ij})\) is called a Euclidean distance matrix (EDM), if there exist \(n\) points \( {\mathbf x}_1, \dots, {\mathbf x}_n \in {\mathbb R}^r\) for some \(r\) such that \(d_{ij} = \|{\mathbf x}_i - {\mathbf x}_j\|^{2}\). If the points generating \(D\) lie on \(k(<n)\) spheres each centered at the origin, then an EDM is called multispherical. A theorem from the work of \textit{T. L. Hayden, J. Lee, J. Wells} and \textit{P. Tarazaga} [Linear Algebra Appl. 247, 203--216 (1996; Zbl 0861.15028)] with necessary and sufficient conditions for an EDM to be multispherical is presented and the block structure of different types of the multispherical matrix \(D\) is discussed. The information following from the position of the origin of coordinates when it is placed at the center of the spheres is studied under the condition that the origin multiple representation property is satisfied. The vectors connected with the origin of coordinates generate a number of supporting hyperplanes for a family of multispherical matrices and also allow to build these vectors in the null space of~\(D\).