Gravitational Waves from Core Collapse Supernovae

The detection of gravitational waves (GWs) from core collapse supernovae (CCSNe) could provide astrophysical properties such as a likely explosion mechanism. Due to their relativity small GW emission, any improvement in recovering more energy from a generic supernova signal could be critical to discovering a GW from a CCSNe occurring in our galaxy. Here we start to evaluate possible techniques for recovering more information from CCSNe GW signals using X-Pipeline, an algorithm designed to search for gravitational wave bursts (GWBs). X-Pipeline uses data from multiple interferometers, clusters of "loud" time-frequency pixels, and statistical properties of these pixels to differentiate GWBs from other sources of loud noise transients in the LIGO data. The pixels on the final time-frequency map are grouped using the "nearest neighbor" method. With this method, the loudest pixels are grouped with their "neighbors", other loud pixels adjacent to them, in order to form an event. Because of the time-frequency distribution of CCSNe, the default settings or the clustering method of "nearest neighbor" itself may not be the best clustering model for finding GW signals from CCSNe. We varied the connectivity of the clustering method to find the method best suited to recovering a CCSNe. The default connectivity is 8, which corresponds to a 3x3 grid with the center pixel omitted. If the connectivity is increased, then there can be a greater distance between "neighboring" pixels. We injected 3 GW signals from simulated CCSNe at a distance of 10kpc into realistic aLIGO noise and calculated how much energy was recovered at grid sizes of 3x3, 5x5, 7x7, and 9x9. This same procedure was done to 10 examples of 3 noise transients found in real LIGO data. It was found that the 3 CCSN signals all show similar increases in energy recovered. Therefore similar signals are more likely to be detected when using larger connectivity. However the three types of noise transience also increase in energy recovery, especially at larger grid sizes. We are exploring if the increase in recovered energy from the noise transience, corresponds to them being more likely to fail coherent consistency tests.