A point balance algorithm for the spherical code problem (Q5942312): Difference between revisions

A spherical code \(P_N^n\) is a finite subset of cardinality \(N\) of the unit sphere \(S^n\). Let \(\omega(s,P_N^n)=\sum_{i<j} \|x_i-x_j\|^{-s}\) if \(s \neq 0\) and \(\omega(0,P_N^n)=\sum_{i<j} \ln \|x_i-x_j\|\). A spherical code of \(N\) points is called extremal if it maximizes (minimizes resp.) \(\omega(s,P_N^n)\) for \(s<0\) (for \(s \geq 0\) resp.). In the paper under review, the authors propose an approach for finding extremal codes by investigating subcodes for certain optimality conditions. To do this, they introduce the notion of an \(L\)-balanced spherical code and present an algorithm in the case \(L=1\). Numerical results are presented for \(s=1\) (the Fekete problem) in small dimensions.