Wave optics simulations of speckle fields can be particularly prone to aliasing when using the angular spectrum propagator. We examined several methods of mitigating aliasing, which include larger guard bands, output plane windowing, absorbing boundaries, and an alternative propagator known as the sinc-basis propagator. We compared the angular spectrum propagator to the sinc-basis propagator and found that, while absorbing boundaries greatly assisted the angular spectrum propagator, the sinc-basis propagator always achieved lower root-mean-squared errors for a given array size due to the non-periodicity of the sinc basis. We examined the computation time as a function of the number of pixels and the root-mean-squared error associated with each of the propagators. Direct comparisons on same array size configurations primarily indicated that the relative wall clock time between the two methods depended highly on the core count of the machine. For all machines tested at the same pixel number, the sinc-basis propagator was generally faster up to a machine dependent pixel threshold, after which the angular spectrum propagator was faster. For machines with more parallelization, this threshold was higher and the speed-up of the sinc-basis propagator relative to the angular spectrum method was larger. It was found that the sinc-basis propagator usually has comparable to shorter computation times than the angular spectrum method to achieve the same threshold error in simulations on the computers tested.
|