/MGCosmoMC

This is the official MGCosmoMC repository.

Primary LanguageHTML

MGCosmoMC

Modified Growth with CosmoMC

This is the official repository for the MGCosmoMC package. It implements the patch MGCAMB into the popular Markov Chain Monte Carlo engine CosmoMC. This version is upgraded to be compatible with the latest CosmoMC(v1.3.2).

Table of contents

1. Introduction

MGCosmoMC allows to set constraint on Modified Growth (MG) - Dark Energy (DE) scenarios using cosmological data. It is a patch for the popular code CosmoMC, and as such it follows all the installing procedures and running procedures. We refer the reader to the official CosmoMC webpage and the ReadMe for the instructions.

Citing MGCosmoMC

If you use MGCosmoMC for your scientific work, please cite the following papers:

as well as the original CAMB paper and CosmoMC paper.

2. How to install

To install MGCosmoMC simply run on your terminal:

git clone https://github.com/sfu-cosmo/MGCosmoMC
cd MGCosmoMC
make cosmomc

Follow these instructions for a step-by-step guide to install CosmoMC.

As a general rule, if you are able to install CosmoMC, then you will be able to install MGCosmoMC.

3. How to run

Before running MGCosmoMC set your model parameters in params_CMB_MG.ini. Pick a MG_flag to choose which model you are going to analyze. For a structure of the models see the MGCAMB page.

GRtrans set the scale factor at which MG is switched on. We suggest to set it larger or equal than 0.001.

Since in the MG formalism there is no prescription to build a non-linear P(k), we suggest to set the flag use_nonlinear = F. To do so, some data requires a proper cut to eliminate the nonlinear scales. The cuts on DES 1YR dataset is described in Sec. 5

4. Installing Planck 2018

MGCosmoMC could work with Planck 2018 likelihood, which needs to be installed separately in advance. Please use the Planck 2018 likelihood with MGCosmoMC by following the procedure illustrated in this Section.

In the following, /$$$ is meant to be replaced by the path specific to your installation.

  1. Add these lines to your .bashrc (but with your customized paths instead of` /$$$):
export PYTHONPATH=/$$$/MGCosmoMC/python:$PYTHONPATH
export PLC_PATH=/$$$/Planck2018/baseline/plc_3.0/
source /$$$/Planck2018/plc_3.0/plc-3.01/bin/clik_profile.sh

and run

source /.bashrc
  1. Create a symbolic link to the Planck likelihoods:
ln -s $PLC_PATH MGCosmoMC/data/clik_14.0
  1. Now change into the MGCosmoMC directory and run make
cd MGCosmoMC
make
  1. Now try running the standard CosmoMC test using the test_planck.ini file:
./cosmomc test_planck.ini

This will print some stuff on the screen, testing all the Planck likelihoods and any other likelihoods specified in the test_planck.ini and test.ini files. Then it will likely stop with some error message, e.g.

Test likelihoods done, total logLike, chi-eq =    2540.646   5081.292
Expected likelihoods,  total logLike, chi-eq =    2625.485   5250.970
 ** Likelihoods do not match **

In any case, as long as it can run the Planck 2018 likelihoods, you are OK and all other procedures are the same as you would do before the update.

5. DES 1YR dataset

Since there is no MG counterpart of Halofit, nonlinear corrections should be turned off when using MGCosmoMC. Datasets probing nonlinear scales should be used with care and with proper cuts (to avoid nonlinear scales). For the DES 1YR dataset we provide three cuts of the nonlinear regime: soft, standard and aggressive. Choose one of them in DES_1YR_final.dataset . Also, be sure to set wl_use_nonlinear = F and wl_use_Weyl = T in DES.ini.

The weak lensing likelihood is modified to use the Weyl potential: the new implementation is in wl.f90, while the default code is kept in wl_std.f90

The method to cut the data is described in our paper and it is based on this DES paper.

The code used to generate this cuts can be found in this repository

Aggressive Cut

The aggressive cut is obtained by setting Delta Chi^2 = 1 . The shaded regions in the plots below are removed:

Standard Cut

The starndard cut is obtained by setting Delta Chi^2 = 5 . The shaded regions in the plots below are removed:

Soft Cut

The soft cut is obtained by setting Delta Chi^2 = 10. The shaded regions in the plots below are removed:

6. Known bugs

  • (Fixed) The reading issue of flag muSigma_par for DES paramterization. Thanks to Kushal Lodha.

7. Authors List

Main Developers:

Original Code Developers:

Repo created and maintained by Zhuangfei Wang. If you find any bugs in the code, please contact Zhuangfei Wang at zhuangfei_wang@sfu.ca .