Searching for EMRI Signals through a Bayesian Analysis

The Laser Interferometer Space Antenna, LISA, set to launch in 2035 will offer a novel look into gravitational wave sources inaccessible to ground based detectors. One such source is Extreme Mass Ratio Inspirals or EMRIs. Every aspect of detecting EMRIs in LISA presents novel challenges particularly from the perspective of data analysis even in the simplest cases. Waveforms currently are slow to generate resulting in long computations. Furthermore, the EMRI parameter space is high-dimensional and is expected to be multimodal, making it difficult to find the true parameters. Here we explore these challenges in simulated noisy EMRI data with a sampler repurposed from pulsar timing array analyses, PTMCMCSampler, a parallel tempering enabled Markov chain Monte Carlo. Ultimately, EMRIs will be our best test of general relativity yet and a strong understanding of potential biases in the posterior is necessary before LISA is set to fly.