Inferring Core-Collapse Supernova Physics with Gravitational Waves

J. Logue, C. D. Ott, I. S. Heng, P. Kalmus, and J. Scargill
arXiv:1202.3256


Abstract


Stellar collapse and the subsequent development of a core-collapse supernova explosion emit bursts of gravitational waves (GWs) that might be detected by the advanced generation of laser interferometer gravitational-wave observatories such as Advanced LIGO, Advanced Virgo, and LCGT. GW bursts from core-collapse supernovae encode information on the intricate multi-dimensional dynamics at work at the core of a dying massive star and may provide direct evidence for the yet uncertain mechanism driving supernovae in massive stars. Recent multi-dimensional simulations of core-collapse supernovae exploding via the neutrino, magnetorotational, and acoustic explosion mechanisms have predicted GW signals which have distinct structure in both the time and frequency domains. Motivated by this, we describe a promising method for determining the most likely explosion mechanism underlying a hypothetical GW signal, based on Principal Component Analysis and Bayesian model selection. Using simulated Advanced LIGO noise and assuming a single detector and linear waveform polarization for simplicity, we demonstrate that our method can distinguish magnetorotational explosions throughout the Milky Way (D < ~10 kpc) and explosions driven by the neutrino and acoustic mechanisms to D < ~2 kpc. Furthermore, we show that we can differentiate between models for rotating accretion-induced collapse of massive white dwarfs and models of rotating iron core collapse with high reliability out to several kpc.


Here we provide numerical and tabulated results of this first Supernova Model Evidence Extractor (SMEE) study. Should you have comments or questions, please contact Josh Logue (j.logue #at# physics.gla.ac.uk) or Christian Ott (cott #at# tapir.caltech.edu).

Principal Components

SMEE decomposes catalogs of waveforms into orthorgonal principal component (PC) vectors via singular value decomposition. Below we provide the PC vector time series data for the four waveform catalogs employed in this study. Each principal component is normalized so that the magnitude of its peak amplitude is unity. The length of each principal component is 3 seconds and the uniform sampling rate is 4096 Hz. We provide 2D matrices of PCs for each GW catalog for download. Each column represents a PC and the first PC is in the leftmost column.

Waveform Catalog PC Data File
Dimmelmeier et al. 2008download
Murphy et al. 2009download
Abdikamalov et al. 2010download
Ott 2009download


Full Results Table

Due to space restrictions, we are unable to include the full results tables in the published/arXiv version of our paper. You may download the table in pdf format from here: download results tables