Simulations of high order curvature adaptive optics

摘要

CFHT recently considered an upgrade of its curvature adaptive optics system to perform high dynamic range imaging. This requires high Strehl ratios and very stable Point Spread Functions, which is usually achieved with high order AO systems providing a dense sampling and correction across the pupil. Such systems are conventionally thought of as Shack-Hartman/piezostack devices because, in theory the scaling law for noise propagation goes as N.log(N), where N is the number of actuators. However, it is difficult to demonstrate this behavior in practice, and as systems get more complex, their efficiency drops to lower levels than expected. Simulations of such systems are usually optimistic because the fail to take the complexity of a real system's flaws (alignment errors, imperfect lenslet array, etc.) into account. One of the reasons why curvature adaptive optics looses its efficiency advantage for large numbers of degrees of freedom is because of its noise propagation properties. The exact crossover with Shack-Hartman systems in terms of performance and efficiency remains to be determined, but cheap (and may be more reliable) high dynamic range systems on smaller telescopes could conceivably use curvature technology. When the number of degrees of freedom remains low, realistic Monte Carlo simulations are possible and provide a more accurate benchmark, as the computing time remains reasonnable. Monte Carlo simulations of such a system have been performed in view of upgrading PUEO, the CFHT adaptive optics bonnette, as a test bench for high dynamic range imaging, and for visible adaptive optics (Lai et al, these proceedings [4839-78]). Simulations show that 104 subapertures/electrode are sufficient to produce a Strehl ratio of 92in K band (S(I) = 5821 However, because PSF stability is so critical and the noise propagation is crucial to the performance, we are hoping to test this system on different simulation codes to confirm these results and expand on them using specificity of each code to complement the other. As this is work in progress, this paper will concentrate on the road ahead and the advantages and drawbacks of each code for each specific task.