Abstract
Parameter estimation on gravitational wave signals from compact binary coalescence (CBC) requires the evaluation of computationally intensive waveform models, typically the bottleneck in the analysis. This cost will increase further as low frequency sensitivity in later second and third generation detectors motivates the use of longer waveforms.
We describe a method for accelerating parameter estimation by exploiting the chirping behaviour of the signals to sample the waveform sparsely for portions where the full frequency resolution is not required. We demonstrate that the method can reproduce the original results with a waveform mismatch of ≤ 5 x 10-7 , but with a waveform generation cost up to ∼50 times lower for computationally costly frequency-domain waveforms starting from below 8 Hz.
We describe a method for accelerating parameter estimation by exploiting the chirping behaviour of the signals to sample the waveform sparsely for portions where the full frequency resolution is not required. We demonstrate that the method can reproduce the original results with a waveform mismatch of ≤ 5 x 10-7 , but with a waveform generation cost up to ∼50 times lower for computationally costly frequency-domain waveforms starting from below 8 Hz.
| Original language | English |
|---|---|
| Journal | Classical and Quantum Gravity |
| Volume | 34 |
| Issue number | 11 |
| Early online date | 18 May 2017 |
| DOIs | |
| Publication status | Published - 8 Jun 2017 |