The multi-conjugate adaptive optics (MCAO) systems proposed for future giant telescopes will require new, computationally efficient, concepts for wavefront reconstruction due to their very large number of deformable mirror (DM) actuators and wavefront sensor (WFS) measurements. Preliminary versions of such reconstruction algorithms have recently been developed, and simulations of MCAO systems with 9000 or more DM actuators and 33000 or more WFS measurements are now possible using a single desktop computer. However, the results obtained to date are limited to the case of open-loop wavefront
reconstruction, and more work is needed to develop computationally efficient reconstructors for the more realistic case of a closed-loop MCAO system that iteratively measures and corrects time-varying wavefront distortions. In this paper, we describe and investigate two reconstruction concepts for this application. The first approach assumes that knowledge of the DM actuator command vector and the DM-to-WFS influence matrix may be used to convert a closed-loop WFS
measurement into an accurate estimate of the corresponding open-loop measurement, so that a standard open-loop wavefront reconstructor may be applied. The second approach is a very coarse (but computationally efficient) approximation to computing the minimum variance wavefront reconstructor for the residual wavefront errors in a closed-loop AO system. Sample simulation results are presented for both concepts with natural guide star (NGS) AO and laser guide star (LGS) MCAO systems on 8- and 32-meter class telescopes. The first approach yields a stable control loop with closed-loop performance comparable to the open-loop estimation accuracy of the classical minimum variance reconstructor. The second approach is unstable when implemented in a type I servo system.© (2003) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.