Only skip KDE generation if sample variance is "relatively" close to zero
I'm attempting to use PESummary for producing plots for pulsar searches. One of the parameters searched for is the strain amplitude, which is obviously a very small number. Currently, if you try and produce a KDE this line checks if the sample variance is close to zero - for strain samples this is always triggered based on the default NumPy isclose()
values meaning you can't generate KDEs for this parameter. This MR switches the check to set the atol
value to 0, so isclose()
only checks if the variance is "relatively" close to zero rather than absolutely close to zero.
@charlie.hoy - would this change be ok and not break anything else?