[SPRINT] Forecasting constraints to gravity with the MG emulator

[gvalogia]

Forecasting constraints to gravity with the MG emulator

The goal of this hack is to perform Fisher forecasting of constraints to the f(R) Hu-Sawicki MG model by the Rubin Observatory, utilizing the matter power spectrum emulator mgemu.

Contacts: @gvalogia Day/Time: Flexible throughout the week, Asynchronous & Synchronous EST Main communication channel: desc-tjp-beyondwcdm and/or zoom as needed. GitHub repo: This link

Goals and deliverable

The main goals to achieve for this week:

• Predict 3D galaxy power spectrum in MG for a Rubin Observatory-like galaxy sample

• Extract covariance matrix for the fiducial cosmology, using existing large set of COLA simulations.

• Perform tomographic analysis for a realistic range of z's and produce Fisher constraints to f(R) model.

Stretch and follow-up goals are:

• Repeat the same analysis above for angular galaxy power spectrum.

• Repeat the same analysis above for lensing power spectrum and produce combined constraints.

• Repeat the same analysis above for DESI/Euclid-like 3D redshift space power spectrum and produce joint constraints.

Resources and skills needed

Basic coding skills (python and maybe C) and interest in learning about cosmological tests of gravity. Familiarity with Fisher forecasting for surveys of the LSS.

Detailed description

This objective is a direct follow-up of our recent paper on a matter power spectrum emulator for f(R) gravity. The emulator is available in this github repository Details about these tasks can be found at the corresponding project page.

[c-d-leonard]

Hey @gvalogia I'm interested to help with this!

[gvalogia]

Hey @gvalogia I'm interested to help with this!

Awesome, I'll get in touch with you!

[nesar]

I wrote a small script to combine CCL's wCDM P(k) and MG emulator's power spectrum enhancement. Pushed a notebook here. The relevant function is pmg(Om, h, ns, s8, fR0, n, z)

The enhancement factor is multiplied by Cosmic Emu's p(k) predictions. Some more validation needs to be done, but I have the CCL parameters set to that of COLA sims as far as I know. Here is a plot of nonlinear DM power spectra for LCDM and MG.

And this is the output from pmg(Om, h, ns, s8, fR0, n, z) for varying fR0 parameter.

[gvalogia]

Awesome, thanks @nesar . In the meantime, I wrote a script that first evaluates the galaxy power spectra from the simulated COLA matter P(k)s, and then produces the corresponding covariance matrices. I pushed the python code and the covariances for 10 z bins (Y10 sample) and 5 z bins (Y1 sample) here As for what the correlation matrix for the MG galaxy P(k) looks like, I added a sample plot for z=0.2 in slide 6 here

[nesar]

This is great! so for forecasting all we need is a wrapper around P_MG(k, z) to get P_gg(k). Looks like you already have that implemented while computing the fiducial values and covariance matrix above.

[gvalogia]

Yes indeed, I can easily do that combing the script I wrote yesterday, that calculates galaxy bias and shot noise, with your function pmg. In fact, I'm thinking that this is a good thing to work on today

[gvalogia]

I just pushed an updated version of the notebook Galaxy_pmg, that evaluated the bias and shot noise factor for an LSST-like sample, and eventually the power spectrum Pgg. The notebook also loads the pre-calculated covariance matrices from the 2000 sims of the fiducial cosmology.

[gvalogia]

We have the first Fisher plots, for the Y10 sample!