Acta Optica Sinica, Volume. 36, Issue 12, 1201001(2016)
Performance of Stochastic Parallel Gradient Descent Algorithm in Coherent Combination
Stochastic parallel gradient descent algorithm (SPGD), a phase control method based on directly optimizing performance index, has good applicability in adaptive optics. The algorithm involves two variable parameters: gain coefficient and random perturbation amplitude, whose values have great influence on the convergence of the algorithm. The requirements of the parameter values for the convergence of SPGD algorithm are discussed, and the value range of parameters are analyzed by combining with the principle of the algorithm. Furthermore, a large number of simulations are conducted to analyze all the gain coefficients and the amplitude of random disturbance, which can ensure the convergency of bilateral SPGD. Meanwhile, the lower limit of the random disturbance amplitude is obtained, and the reason for its existence and the lower limit value are also analyzed theoretically and experimentally. With the existence of phase noise in coherent combination, the range of parameters for the algorithm to be convergent is analyzed with different phase corrector parameters.
Get Citation
Copy Citation Text
Li Xingke, He Yuntao. Performance of Stochastic Parallel Gradient Descent Algorithm in Coherent Combination[J]. Acta Optica Sinica, 2016, 36(12): 1201001
Category: Atmospheric Optics and Oceanic Optics
Received: Jun. 15, 2016
Accepted: --
Published Online: Dec. 14, 2016
The Author Email: Xingke Li (xkli123@126.com)