GRAVITATIONAL-WAVE DATA ANALYSIS FROM EARTH TO SPACE: COMPUTATIONAL AND THEORETICAL CHALLENGES

Abstract

The Laser Interferometer Space Antenna (LISA) is a joint NASA-ESA deepspace mission consisting of three drag-free spacecraft, separated by 5 x 10 km and following an Earth-trailing solar orbit; LISA will use picometer interferometry to measure the modulations in the inter-spacecraft distances induced by gravitational waves (GWs) of frequency between 10~ and 1 Hz. LISA will seek to detect GWs from binary systems of compact stellar objects in our galaxy, from the inspiral and merger of binaries of massive and supermassive black holes, and from the capture of compact stellar objects into the supermassive black holes at the center of galaxies; it will also set new constraints on background GWs from the early universe, and naturally (but perhaps most interestingly) it will be sensitive to GWs from previously unknown sources within its frequency band. LISA is expected to start collecting data around 2013 (two years after its launch); thus, its deployment will trail by at least ten years the beginning of science operations for ground-based GW interferometers such as LIGO, GEO, Virgo, and TAMA. At the time of writing, these experiments have begun generating result papers with upper limits on the distribution of sources and GW events; more importantly, they can rely on highly developed technical literature and individual expertise on the characterization and operation of detectors, the handling and analysis of data, and the astrophysics and waveform modeling of sources. It follows that the theoretical activity on LISA is likely to be modeled, at least initially, on the tools and techniques developed for the ground-based detectors (henceforth, GBDs). LISA and GBDs have remarkable similarities, but crucial differences. They look at the same (or similar) sources from different windows in the frequency spectrum (10-l Hz for LISA, 10-10 Hz for GBDs): consequently, the LISA data set will consist of ~ 10 samples/year, while the GBDs are now collecting a rather more daunting ~ 10 samples/year. Thus, the search for continuous, quasimonochromatic sources of unknown position (e.g., galactic binaries for LISA, and spinning, asymmetric neutron stars for GBDs), will be much less computationally taxing for LISA than it is for GBDs. This asymmetry is partially offset for chirping sources such as inspiraling binaries, which will transit slowly through the LISA frequency band, while they move rapidly (in hours or minutes) in and out of the