Prediction of wavefronts in adaptive optics to reduce servo lag errors using data mining

Abstract

Servo lag errors in adaptive optics lead to inaccurate compensation of wavefront distortions. An attempt has been made to predict future wavefronts using data mining on wavefronts of the immediate past to reduce these errors. Monte Carlo simulations were performed on experimentally obtained data that closely follows Kolmogorov phase characteristics. An improvement of 6% in wavefront correction is reported after data mining is performed. Data mining is performed in three steps (a) Data cube Segmentation (b) Polynomial Interpolation and (c) Wavefront Estimation. It is important to optimize the segment size that gives best prediction results. Optimization of the best predictable future helps in selecting a suitable exposure time.