“…If the stepsize is too large, the algorithms will diverge easily, while if the stepsize is too small, the algorithms will converge slowly. For the SPARTA, SWF, and Thresholding Wirtinger Flow algorithms, the stepsizes are selected empirically, and the default stepsize functions or stepsize constants are proposed for the measurement vectors that conform to the standard Gaussian random distribution [26,27,28,29,30,31,32,33,34,35]. However, the measurement vectors of PL-TCAI conform to sub-Gaussian random distribution, and the stepsize selection criteria or functions based on the standard Gaussian random distribution are no longer applicable to PL-TCAI.…”