Rapid validation of gravitational wave candidates with machine learning  (Dr. Jess McIver , University of Columbia, LIGO deputyperson)


Understanding gravitational wave (GW) detector noise at the time of a candidate event is critical for GW analyses at all latencies, from rapidly generated candidate alerts to GW source catalogs and population studies. GW skymaps, tests of general relativity, and resolution of subtle, low-SNR effects encoded in GW signatures (e.g. spin precession and eccentricity) are especially vulnerable to transient noise present in the data of one or more detectors at the same time as a GW signal. I will give an overview of the challenges in evaluating data quality at the time of a GW candidate as well as past and current “event validation” approaches for LIGO and Virgo data. I will highlight a subset of machine learning methods deployed for the latest LIGO-Virgo-KAGRA observing run (O4) to help automate this process for LIGO-Virgo researchers to field a higher expected rate of GW detections. I will discuss how these improvements will benefit a wide variety of downstream GW analyses.

Coordinate ZOOM:


ID riunione: 842 6172 1878

Passcode: 611379