search

cosmodesi / desilike

BSD 3-Clause "New" or "Revised" License
20 stars 17 forks source link

Add RIC correction based on multiplicative template. #25

Closed echaussidon closed 7 months ago

echaussidon commented 9 months ago

Hi,

I propose here a multiplicative template correction to deal with the RIC correction that affect strongly, in particular, the large scale modes of the power spectrum. For now, it is only available for the power spectrum and it is completely independent of the rest of the model.

This can be added in TracerPowerSpectrumMultipolesObservable via the key word ric_correction that should a dictionary containing for each multipole the range of validity of the correction and parametrization for the RIC template. The range of validity is derived, for instance, by limited the ratio to the positive values on large scales. Note: Due to statistical errors, the kmin is not the same for the different multipoles.

The template was derived to match an arctan function from the ratio of the mean power spectrum of 100 EZmocks (without the shuffling and with the shuffling). The figure shows how this function match this ratio: Ratio_correction

Finally, this figure shows how the correction (black line) works compare to the fit without the correction (grey line). Note: the values measured from the mean of 100 EZmocks without the shuffling are (fnl, b1) = (4.1, 2.3).
Correction_radial_constraints_loglog

echaussidon commented 7 months ago

The new method to correct the RIC that I propose, does not need to change the desilike code. It will only update the window matrix used in TracerPowerSpectrumObservable().

I closed this PR and delete it