pycbc
pycbc copied to clipboard
Fake strain seed is easy to forget in inference inj workflow
If you don't set a fake-strain-seed
in the [data]
section of the inference config file when running the inference inj workflow, the seed will never be set. If you are generating fake noise, the result will be every injection (and every detector) will use the same exact realization of noise, since the default in the strain module is to use a seed of 0. See lines 236-245 of the pycbc_make_inference_inj_workflow
.
Suggested fix: add a explicit option in a workflow config section that specifies to make fake strain.