Factored form descent: a practical algorithm for coherence retrieval

Opt Express. 2013 Mar 11;21(5):5759-80. doi: 10.1364/OE.21.005759.


We formulate coherence retrieval, the process of recovering via intensity measurements the two-point correlation function of a partially coherent field, as a convex weighted least-squares problem and show that it can be solved with a novel iterated descent algorithm using a coherent-modes factorization of the mutual intensity. This algorithm is more memory-efficient than the standard interior point methods used to solve convex problems, and we verify its feasibility by reconstructing the mutual intensity of a Schell-model source from both simulated data and experimental measurements.

Publication types

  • Research Support, Non-U.S. Gov't