Testing waveform models for the LISA and Einstein Telescope gravitational wave detectors

Abstract

[eng] The interest in gravitational waves has greatly increased since the first detection in 2015 [1]. This has given extra momentum to the development of new detectors, which will extend the accessible frequency range, while also improving the sensitivity compared to current detectors. Two of these detectors are studied here and they are the Einstein Telescope (ET) and the Laser Interferometer Space Antenna (LISA). The ET will probe the same frequency range as current detectors, though the band will be widened (to about 1 to 104 Hz) and the sensitivity will be much better. LISA will be an observatory in space and it will look for sources in an entirely new frequency range at about 10−4 to 1 Hz. A wide variety of sources is expected to be found by these detectors. Furthermore, they will have a much higher signal-to-noise ratio, which means that they can find out more information about the sources. These new detectors will have sensitivity curves that are different from current detectors such as the Laser Interferometer Gravitational-Wave Observatory (LIGO), which means that their response to gravitational waves will be different. Some approximations were used for the LISA sensitivity curve and the consequence is that only the lower part of the LISA frequency range was considered in this work. A comparison was made between LISA, the ET and LIGO and it was found that LISA puts more emphasis on the merger, whereas the ET emphasizes the inspiral instead. These different sensitivity curves could mean that the performance of waveform models is different. There are various methods for modelling gravitational waves from binary black holes. The three main families of waveform models that are used in parameter estimation (which means that they need to have a low computational cost) are compared. These are the phenomenological models IMRPhenomXHM [53] and IMRPhenomHM [52], the effective one-body reduced order model SEOBNRv4HM ROM [54] and the hybrid surrogate model NRHybSur3dq8 [55]. All of these models use higher modes of the multipole expansion and are non-precessing. The highest similarities were found between IMRPhenomXHM and NRHybSur3dq8. The overall performance for LIGO and the ET was similar, whereas the results for LISA were not as good. This depends on the choice of how to scale the mass between LIGO and LISA. The other results suggest that the matches for LISA might be better when the full response can be used to include higher frequencies. For both LISA and the ET it is found that the current models would induce systematic errors in parameter estimation.