fix-redshift-minus-one-lib_tools-gravwaves

[shreyas-tiruvaskar]

Description

Setting redz=-1.0 for the binaries not merging within the age of the universe causes problems. -1 is used as a flag to identify such binaries, but that redshift value is used in averaging, which can cause troubles.

While converting the redz_final from (91, 81, 101, 5) to (90, 80, 100, 5), i.e. going from bin edges to bin midpoints, the average is taken of the 8 vertex redshift values to obtain an average bin mid-point value. That -1 was creating problem by occasionally causing very small positive midpoint redshift values.

My fix changes the flag value from -1 to -np.inf in librarian/lib_tools.py. So that in averaging, the result can't become accidentally positive.

This, however, creates nan values for those bins in gravwaves.py hs calculation in the function char_strain_sq_from_bin_edges_redz. I fix that by simply applying a filter there.

[shreyas-tiruvaskar]

Even though I have used -np.inf, more robust and safe option would be using np.nan. Although, it would mean more changes in different relevant files. I didn't do it because of time constraints.

[kayhangultekin]

@lzkelley This was the PR I was talking about in our zoom.