Low-frequency bug in SEOBNRv4_ROM for BNS
Comparison of SEOBNRv4_ROM / TaylorF2 with SEOBNRv4T_surrogate (as part of SEOBNRv4T_surrogate's review) revealed a low-frequency phasing bug that afflicts SEOBNRv4_ROM waveforms for:
- BNS systems with total-masses between
2-6M_\odot
, - between 20 - 30Hz.
Some more findings:
- See this wiki for figures showing SEOBNRv4_ROM's dephasing as a function of GW frequency for select cases.
- Further evidence here showing mismatches betwen TaylorF2 / SEOBNRv4T_surrogate and SEOBNRv4_ROM with varying upper frequency cutoffs and a flat PSD, and here showing the same with ZDHP PSD.
This can potentially afflict the BNS model SEOBNRv4_ROM_NRTidal. This approximant was used in our re-analysis of GW170817 down to 23Hz.
Ping @michael.puerrer and @ian-harry / @tito-canton .