Phase retrieval for sub-Gaussian measurements

Generally, the phase retrieval problem can be viewed as reconstruction of a function/signal from only magnitude linear measurements. These measurements be, for example, Fourier transform density function. Computationally is very challenging. Many algorithms are based on i.i.d. Gaussian random However, remain one few classes In this paper, we develop an efficient algorithm sub-Gaussian frames. We provide general condition and modified spectral initialization. algorithm, first obtain good approximation solution through initialization, there use Wirtinger Flow to solve solution. prove that converges global minimizer linearly.