We describe a method performing w-projection using the fast Gauss transform of Strain (1991). We derive the theoretical performance, and simulate the actual performance for a range of w for a canonical array. While our implementation is dominated by
overheads, we argue that this approach could for the basis of a higher-performing algorithms with particular application to the Square Kilometer Array.
Searching for dispersed radio pulses in interferometric data is of great scientific interest, but poses a formidable computational burden. Here we present two efficient, new antenna-coherent solutions: The Chirpolator and The Chimageator. We describe
the equations governing both techniques and propose a number of novel optimizations. We compare the implementation costs of our techniques with classical methods using three criteria: the operations rates (1) before and (2) after the integrate-and-dump stage, and (3) the data rate directly after the integrate-and-dump stage. When compared with classical methods, our techniques excel in the regime of sparse arrays, where they both require substantially lower data rates, and The Chirpolator requires a much lower post-integrator operations rate. In general, our techniques require more pre-integrator operations than the classical ones. We argue that the data and operations rates required by our techniques are better matched to future supercomputer architectures, where the arithmetic capability is outstripping the bandwidth capability. Our techniques are, therefore, viable candidates for deploying on future interferometers such as the Square Kilometer Array.
