Reconstructing wavefront phase from measurements of its slope, an adaptive neural network based approach

Abstract We present a neural network based method to integrate wave front shapes from its slopes. Tests were done in order to verify and compare its performance with other state of the art finite-difference least squares methods, proving that better precision can be achieved having representative data to train our model, even when there exist noise in the data. Also, a simple simulation of the Gran Telescopio de Canarias telescope was carried out, finding that the results provided by our method ranged from  ≈ 5 to  ≈ 19 times better than other ones in this case. Finally, the source code of our method has been incorporated into a public repository with the objective that it can be tested and used by any other researcher.