Questions about Configuration Settings in CosmoCov

[lingwenkai]

Hi @xfangcosmo,

Thank you for developing CosmoCov, which offers an efficient framework for computing the covariance matrix of two-point statistics. I’m a PhD candidate (Wenkai Ling) from Yunnan University. I’m currently using CosmoCov to estimate the covariance matrix for shear shear two-point correlation functions based on KiDS-DR4 data.

I have a few questions regarding the configuration file:

1. Survey area (area): Should this value correspond to the total observed area including masked regions, or to the effective area after applying a smoothed HEALPix mask? Similarly, when specifying the galaxy number density, should it be computed using the total survey area or the effective area Aeff after masking?

2. Importance of the mask power spectrum: How critical is the inclusion of the mask power spectrum for the accuracy of the covariance estimation?

3. Ellipticity dispersion (sigma_e): Should this be the dispersion of a single ellipticity component (i.e., sigma_{e1}, sigma_{e2} or sigma_{e}/2^0.5 ), or the total ellipticity dispersion (i.e., sigma_{e}^2 = sigma_{e1}^2 + sigma_{e2}^2 )? I noticed that using only the dispersion of a single component (i.e., sigma_{e}/2^0.5) appears to underestimate the covariance. Could you kindly confirm which convention CosmoCov adopts.

Thanks again for making this tool available — it’s been very helpful for my research. I really appreciate your time and guidance!

Best,
Wenkai Ling

[xfangcosmo]

Hi @lingwenkai ,
Glad you find the code useful. To answer your questions,

1. It should be the area where the galaxies are observed. If your mask (or window) takes out some areas in the survey region, then the area is reduced. number density is also effective, because you are not counting galaxies outside of your window.

2. The importance of mask power spectrum depends on the area of your survey. Small survey (and large angular scale in correlation function) cares more about survey geometry, because the boundary becomes significant. This should be used for KiDS survey. For detail you can refer to Troxel et al..

3. Convention is the total shape noise sigma_{e} = sqrt{sigma_{e1}^2 + sigma_{e2}^2}