Abstract

The errors caused by the inherent time delays in an adaptive-optics control loop could be overcome to a large extent by short-term prediction of the incident wavefront. Conversely, the bandwidth requirements of the loop could be relaxed. Furthermore, it is of interest to efficiently use all of the information available in the wavefront when making estimates of future values. In [1,2,3] Jorgenson and Aitken have described evidence suggesting that wavefront distortions should be treated as the output of a nonlinear dynamical system, and not simply as a filtered random process. Schwartz et al [4] have made similar suggestions.

© 1995 Optical Society of America