Speeding Up Blind Gamma-Ray Pulsar Searches with GPUs

We present results on the feasibility of a blind search for isolated and binary pulsars using Fermi. A blind search with Fermi data provides essentially uniform coverage of the whole sky, yielding unbiased pulsar samples for further radio followup and population analyses.

We implemented a highly optimized GPU-based phase-modulation search to greatly accelerate (by almost 50 times) blind searches for isolated gamma-ray pulsars. Even at its most efficient, a complete search over one gamma-ray source would take on the order of a few days -- expensive, but feasible utilizing a GPU cluster.

Finally, we used the optimized algorithm to search for gamma-ray pulsars in simulated data to test its detection limits. For photon-based searches we determined that roughly 4000 or more photons are required for the phase-modulation search to detect a typical MSP in 3 years of data. Unfortunately this is brighter than any of the Fermi unassociated sources likely to be pulsars.