Simulation and Data Analysis for LISA. Instrumental Modeling, Time-Delay Interferometry, Noise-Reduction Performance Study, and Discrimination of Transient Gravitational Signals

The Laser Interferometer Space Antenna is a European Space Agency mission that aims to measure gravitational waves in the 10^-3 Hertz range. Three spacecraft are placed in a quasi-equilateral triangular formation whose barycenter trails the Earth on its heliocentric orbit. Laser beams are exchanged to monitor pico-metric variations between the test masses due to gravitational waves. Because various instrumental noise sources couple to the measurements, several data processing techniques are used to reduce them before we can extract gravitational-wave signals. We propose a realistic instrumental model to study these noise-reduction algorithms. We investigate how the main noise sources appear in the measurements and work out their residuals in almost noise-free combinations. To validate these results, we develop a flexible numerical simulation tool that aims to generate realistic measurements: LISANode propagates noise time series between the spacecraft and in the optical benches, all the way down to the phasemeters and the on-board computers. It also computes the response to gravitational waves. LISANode is capable of executing the main noise-reduction algorithms, including the computation of Time-Delay Interferometry laser noise-free combinations, as well as clock-calibrated combinations. As we account for instrumental and numerical imperfections, noises do not exactly vanish in the final combinations. In particular, we study the performance hit of the constellation flexing and the on-board data processing on the laser-noise reduction. We model the flexing-filtering coupling and propose a technique to mitigate this effect. Moreover, we derive exact clock-noise calibration expressions. Simulations are used to validate these results and confirm that the dominant sources of noise can be reduced to the required levels. We also investigate machine-learning techniques to discriminate between instrumental glitches and transient gravitational signals. Analytic studies show that both appear differently in noise-free combinations and experiments suggest that some neural networks are capable of distinguishing between them. (author)