We propose an efficient algorithm to solve positive a semidefinite matrix approximation problem with a trace constraint. Without constraints, it is well known that positive semidefinite matrix approximation problem can be easily solved by one-time eigendecomposition of a symmetric matrix. In this paper, we confirmed that one-time eigendecomposition is also sufficient even if a trace constraint is included. Although an additional binary search is necessary, it is not computationally expensive.
Article
View Positive semidefinite matrix approximation with a trace constraint