Schmidt-Mirsky matrix approximation with linearly constrained singular values (Q1362662)

The author considers the mathematical problem of identifying and computing the closest approximation of a matrix by another matrix whose singular values satisfy any specified set of linear contraints, a generalization of \textit{E. Schmidt}'s theorem [Zur Theorie der linearen und nichtlinearen Integralgleichungen. Math. Ann. 63, 433-476 (1907; JFM 38.3770.02)], generated by \textit{L. Mirsky} [Quart. J. Math. Oxford 11, 50-59 (1960; Zbl 0105.01101)]. The theorem is applied to the determination of the oblate spheroid that most closely fits a specified ellipsoid, which amounts to require that the two largest principal axes have a common length.