Origins of Nonlinear Noise
In addition to the main strain channel, where gravitational waves are observed, each LIGO detector has over 10,000 channels which monitor the operation of different subsystems as well as the seismic, acoustic, and electromagnetic environment. This treasure trove of data presents a unique and largely untapped opportunity: How can we best leverage these vast amounts of data to improve our understanding of noise processes in LIGO, and ultimately improve our ability to detect gravitational waves?
In collaboration with Prof. Vagelis Papalexakis of the Department of Computer Science & Engineering, we are developing novel, largely unsupervised, machine learning and data science techniques to model and analyze the vast amounts of data recorded in the LIGO detectors. Our goal is to identify major contributing sources of nonlinear noise and gain insight into the physical nature of their couplings, leading to actionable information to guide the detector commissioning and directly improve the science data.