: Setup automated tool to evaluate shear bias and compare against metacal equivalent

[b-remy]

No description provided.

[b-remy]

This issue aims to track the development of the Metacalibration baseline script.

So far we only input the the postage stamp, the PSF and the noise level to perform metacal.

In order to work on simulations from #1 , the simulations are saved in fits files so that it also keeps track of the noise level and the true shear in the data. I'm not aware of the good practices to create fits file so, I'm open to advices :-) The idea is that both our probabilistic inference code and metacal load the same file containing all the information needed.

[b-remy]

Pointer that I am starting to work on this issue.

• Code to make simulation is in this notebook
• experiments will be based on metacal.py script

[b-remy]

In order to have an order of magnitude, we don't necessarily need to run metacal on our galaxies. Exponential galaxies will work as long as we respect the condition of S/N < 10, which should be the cut for LSST for instance. So I will just run the metacal.py with this condition.

python metacal.py --noise=.8e-2 --ntrial=10000 
10000/10000 100%

S/N: 10.0131
R11: 0.358962
m: -0.646497 +/- 0.978565 (99.7% conf)
c: -0.00133305 +/- 0.00973926 (99.7% conf)

Results involving more galaxies will follow

[b-remy]

With 10^5 galaxies

python metacal.py --noise=.8e-2 --ntrial=100000

S/N: 10.0157
R11: 0.358477
m: -0.00334721 +/- 0.309179 (99.7% conf)
c: 3.61223e-05 +/- 0.00308977 (99.7% conf)

So we are not there yet. Might need 10^7, 10^8 according to Sheldon and Huff, 2017