Buggy interface with `lalsimulatin.SimInspiralFD` for time domain models.
For time-domain waveforms, we use lalsimulation.SimInsprialFD
.
@charlie.hoy recently pointed out that that function ignores the requested frequency and instead sets the frequency spacing by assuming the passed maximum frequency is the Nyquist frequency and the number of frequency bins is a power of two (despite the description in the documentation for that function).
The consequence of this is that SimInspiralFD
should only ever be called with maximum frequency / frequency spacing as a power of two.
Proposed fix:
-
hack around unexpected behaviour. Possibly by making sure that the maximum frequency passed is a power-of-two multiple o the frequency spacing. The frequency bounds are already applied by multiplication. -
add a unit test of some kind. Maybe testing for near equality of IMRPhenomX
andIMRPhenomT
?
Edited by Colm Talbot