A parallel algorithm for the unbalanced orthogonal Procrustes problem (Q1180610)

Let \(A\in \mathbb{R}^{l\times m}\) and \(B\in \mathbb{R}^{l\times n}\) be two given matrices. The orthogonal Procrustes problem is to find the best least squares approximation of the matrix \(B\), with \(AQ\), among all possible matrices \(Q\in \mathbb{R}^{m\times n}\), with orthonormal columns. That is to say that the Procrustes problem is to find \(Q\) that achieves \(\min\| AQ-B\|\), where \(Q^ TQ=I\). A new algorithm for this problem is presented which is simpler than the existing one, and which uses the recursive singular value decomposition updates. The proposed algorithm can be efficiently implemented in a parallel environment. By using numerical experimental results, the author shows that the new algorithm has favorable convergence results.