BlackHoles@Home: Open-Sourcing within NRPy+ 2.0

Each observation of a gravitational wave (GW) is compared against millions of theoretical predictions to perform parameter estimation and extract important scientific insights. Our most reliable GW predictions are built upon catalogs of numerical relativity (NR) compact binary coalescence simulations, which, to date, have always required a computing cluster. Such a large computational expense has limited the catalog size to only about 4,000 in 15 years. Given the vast parameter space of even the simplest (but most commonly observed) GW source—binary black holes (BBHs)—this small GW collection threatens the potential science gains from future GW observations.

BlackHoles@Home is a proposed BOINC project that leverages new NR techniques to fit BBH simulations onto a consumer-grade desktop computer, enabling GW follow-ups and catalogs with unprecedented throughput using volunteer computers. We recently demonstrated that new numerical gridding algorithms enable BlackHoles@Home to model BBH inspirals, mergers, and ringdowns on consumer-grade desktop computers using about 1/100th the amount of memory of the most popular gridding approach in NR: adaptive-mesh refinement. We further find that higher-order GW modes exhibit less noise than any other NR code, and that numerical errors converge cleanly to zero. We present concrete plans for launching the BOINC project, as well as progress toward open-sourcing BlackHoles@Home within the new NRPy+ 2.0 infrastructure.