TY - JOUR
T1 - Towards rapid parameter estimation on gravitational waves from compact binaries using interpolated waveforms
AU - Smith, R. J.E.
AU - Cannon, K.
AU - Hanna, C.
AU - Keppel, D.
AU - Mandel, I.
PY - 2013/6/4
Y1 - 2013/6/4
N2 - Accurate parameter estimation of gravitational waves from coalescing compact binary sources is a key requirement for gravitational-wave astronomy. Evaluating the posterior probability density function of the binary's parameters (component masses, sky location, distance, etc.) requires computing millions of waveforms. The computational expense of parameter estimation is dominated by waveform generation and scales linearly with the waveform computational cost. Previous work showed that gravitational waveforms from nonspinning compact binary sources are amenable to a truncated singular value decomposition, which allows them to be reconstructed via interpolation at fixed computational cost. However, the accuracy requirement for parameter estimation is typically higher than for searches, so it is crucial to ascertain that interpolation does not lead to significant errors. Here we provide a proof of principle to show that interpolated waveforms can be used to recover posterior probability density functions with negligible loss in accuracy with respect to noninterpolated waveforms. This technique has the potential to significantly increase the efficiency of parameter estimation.
AB - Accurate parameter estimation of gravitational waves from coalescing compact binary sources is a key requirement for gravitational-wave astronomy. Evaluating the posterior probability density function of the binary's parameters (component masses, sky location, distance, etc.) requires computing millions of waveforms. The computational expense of parameter estimation is dominated by waveform generation and scales linearly with the waveform computational cost. Previous work showed that gravitational waveforms from nonspinning compact binary sources are amenable to a truncated singular value decomposition, which allows them to be reconstructed via interpolation at fixed computational cost. However, the accuracy requirement for parameter estimation is typically higher than for searches, so it is crucial to ascertain that interpolation does not lead to significant errors. Here we provide a proof of principle to show that interpolated waveforms can be used to recover posterior probability density functions with negligible loss in accuracy with respect to noninterpolated waveforms. This technique has the potential to significantly increase the efficiency of parameter estimation.
UR - http://www.scopus.com/inward/record.url?scp=84879100250&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84879100250&partnerID=8YFLogxK
U2 - 10.1103/PhysRevD.87.122002
DO - 10.1103/PhysRevD.87.122002
M3 - Article
AN - SCOPUS:84879100250
SN - 1550-7998
VL - 87
JO - Physical Review D - Particles, Fields, Gravitation and Cosmology
JF - Physical Review D - Particles, Fields, Gravitation and Cosmology
IS - 12
M1 - 122002
ER -