Bayesian WIMP Parameter Inference for LXe TPCs

Astrophysical evidence has long indicated the presence of a large percentage of invisible mass in the universe. The presence of this invisible and essentially collision-less mass, known as dark matter, has only been detected so far by its gravitational influence on astrophysical objects. Physicists have hypothesized that the Weakly Interacting Massive Particle (WIMP), a particle with mass in the GeV/c^2 range that interacts with baryonic matter at a weak scale, is a prime candidate for dark matter. If the WIMP hypothesis is correct, experiments on Earth can aim to discover them by looking for rare scatters off of atomic nuclei, which the LUX-ZEPLIN (LZ) experiment aims to do. LZ will use a liquid xenon Time Projection Chamber (LXe TPC) to detect elastic scatters off of xenon atoms caused by impinging WIMPs.

This thesis will introduce a generalized Bayesian statistical framework that is able to infer a WIMP's mass and spin-independent scattering cross section from simulated datasets. The framework will be developed around the LZ detector template, but will efficiently generalize to other similar LXe TPCs. We utilize this framework to generate and reconstruct several simulated experiments to highlight the statistical limitations on parameter inference posed when utilizing LXe TPCs. We also compare the change in reconstruction efficiencies caused by detectors of different masses and different background rates. Our reconstructions indicate that at low event rates, reduced backgrounds do not provide a significant gain in parameter resolution due to limitations posed by the WIMP scattering spectrum. On the contrary, increased detector sizes provide significant gains in statistical resolution. This indicates that for parameter inference purposes, future experiments should prioritize larger tonne-year detectors over reduced backgrounds.