Computing the nearest Euclidean distance matrix with low embedding dimensions (Q463737)

In many applications like Euclidean embedding of network coordinate systems, sensor network localization, and molecular conformation, the distances are either not known or inaccurate. In such situations the nearest Euclidean distance matrix (EDM) to the known one has to be determined. The authors develop a new method using the majorization approach to tackle this nonconvex optimization problem theoretically and numerically. Numerical results reveal that the semi-smooth Newton conjugate gradient algorithm with the majorization method is the most effective one.