Resolving Galactic binaries in LISA data using particle swarm optimization and cross-validation

The space-based gravitational wave (GW) detector LISA is expected to observe signals from a large population of compact object binaries, comprised predominantly of white dwarfs, in the Milky Way. Resolving individual sources from this population against its self-generated confusion noise poses a major data analysis problem. We present an iterative source estimation and subtraction method to address this problem based on the use of particle swarm optimization (PSO). In addition to PSO, a novel feature of the method is the cross-validation of sources estimated from the same data using different signal parameter search ranges. This is found to greatly reduce contamination by spurious sources and may prove to be a useful addition to any multi-source resolution method. Applied to a recent mock data challenge, the method is able to find $O(10^4)$ Galactic binaries across a signal frequency range of $[0.1,15]$ mHz, and, for frequency $\gtrsim 4$ mHz, reduces the residual data after subtracting out estimated signals to the instrumental noise level.