EFFICIENT COMPUTATION OF NEAR-FIELD DIFFRACTION PATTERNS BY MEANS OF SUBSAMPLED CONVOLUTION

We introduce subsampled convolution and convolution with an aggregate kernel as computationally efficient techniques for discrete convolutio n of the free-space propagation kernel with generalized and rectangula r aperture functions, respectively, in computing near-field diffractio n patterns. We develop fast Fourier transform based algorithms for imp lementing subsampled and aggregate kernel convolution, then demonstrat e the memory and execution time economy of these techniques through co mputational examples.