Currently hera_pspec does a two-step normalization, first to divide out the volume of the survey, and then to convert from "observer units" (e.g., volumes in Hz * Sr) to cosmological units.
If we fix #328 , then the response matrices will already have correct cosmological units and the normalization can be done in one go.
Ideally we want to get the covariance units from the meta-data of the dataset and cosmological conversions from astropy. This should then always give us the right units for things.
@nkern Makes the good point that there are reasons to keep the scalar variable as something that's computed and stored, just in case people want to (approximately) change the cosmology after the fact.
Currently
hera_pspec
does a two-step normalization, first to divide out the volume of the survey, and then to convert from "observer units" (e.g., volumes in Hz * Sr) to cosmological units.If we fix #328 , then the response matrices will already have correct cosmological units and the normalization can be done in one go.
Ideally we want to get the covariance units from the meta-data of the dataset and cosmological conversions from astropy. This should then always give us the right units for things.