论文部分内容阅读
In phase retrieval,one aims to recover a signal from magnitude measurements.In this talk,we show how to recover the signal by a nonconvex optimization method.The initialization scheme is based on a spectral method; then a specific type of gradient descent is applied iteratively.This algorithm is guaranteed to converge to the original signal with fast convergence rate.Large scale simulations are also employed to verify the effectiveness of our method.