Abstract

Constrained iterative algorithms have been applied primarily to single objective applications,¹ where by objective we mean that distribution that is to be reconstructed from partial information or synthesized with desired characteristics. In a recent work by the authors,² Fienup's error-reduction algorithm was extended to multiple objectives, as shown in Fig. 1, and was applied to a specific synthesis problem with two objectives. In this paper, we illustrate an important characteristic of multiple objective iterative algorithms, dependent constraints.

© 1983 Optical Society of America