jharno commented 3 years ago

Infusing IA in SkySim5000

This sprint will infuse IA in mock lensing catalogues based on SkySim5000 (not the official one) and assign intrinsic alignment models based on the projected tidal field tensor.

Contacts: Joachim Harnois-Deraps Day/Time: M-Tu-W-Th Main communication channel: desc-ia (slack) GitHub repo: https://github.com/LSSTDESC/SkySim5000_IA_infusion

Goals and deliverable

1- The lensing catalogues (with the tidal field values) already exist, but need validation (e.g. comparing the shear 2PCF with predictions). Different choices were made in their constructions, which could potentially be improved.

2-The prescription to assign IA based on the tidal field exists but needs to be implemented. Likely the normalisation of the tidal field is off and needs to be determined empirically, e.g. from comparing the measured 2PCF with IA with theoretical predictions.

3-All ingredients to infuse the TATT model are there, but that needs to be coded in a notebook

4-We need to be able to run the cosmoSIS TATT IA modules.

5-I have mocks with the same n(z), but in which the cosmology is varied (25 points in a wCDM space) so we could verify the cosmology dependence of the IA signal.

Resources and skills needed

Needs access to NERSC (ideally cori) to access data and run the existing notebooks Knowledge of TreeCorr and cosmoSIS a big plus, as these will be the main validation tools.

Detailed description

The final goal is to emulate IA as a function of cosmology, and show that we can recover the IA parameters completely from our simulation-infused IA (i.e. without relying on the analytical IA model for the 2PCF). Once that's established, we can apply the method on weak lensing statistics beyond 2PCFs.

Link to the Simulation Data:

cori:/global/homes/j/jharno/IA-infusion/SkySim5000/GalCat/StageIV_nz/GalCat_tomo[TOMO]_All_0.06GpAM_RA_Dec_g1_g2_w_den_Sij.dat

where [TOMO] in [1-5]

mishakb commented 3 years ago

It would be great to let all know when we are meeting. Also, my student Leonel has been reading the paper on the method and would like contribute but not sure if he has a desc github userneme so far. I will ping him

EiffL commented 3 years ago

Notes from monday meeting with Joachim

kappa maps up to z=3 (Born approximation)
shear maps by spherical harmonics
0.6 gal/arcmin^2 with n(z) for 5 tomo bins catalogs with: ra, dec, z, kappa, gamma1, gamma2, s11, s22, s12

Things to do:

[ ] : Validate catalogs with treecorr
[ ] : Compute full sky Cl with either namaster [@EiffL ] or by replicating the field to remove border effects
[ ] : Writing prediction code with CCL (or if CCL doesnt have TATT) [@EiffL, @dlanzieri ]
[ ] : Add infusion of IA to catalog [@jharno]

EiffL commented 3 years ago

Here is my progress report for today, I opened a notebook here: https://github.com/LSSTDESC/SkySim5000_IA_infusion/blob/main/NaMaster.ipynb And should be not too far away from some nice pseudo Cells with NaMaster

jharno commented 3 years ago

Sorry everyone, I accidentally closed the issue, instead of deleting a comment. Our Sprint is still open!

Lhior commented 3 years ago

I have calculated the 3D auto and cross correlation functions for all 5 bins. The PR was approved by @jharno and merged. @mishakb and I are wondering what needs to be done for "validating" the catalogs? [item 1 in the to-do list]