Halo Occupation Distribution Optimization Routine

Introduction
Dependencies
Usage
Configuration parameters
Co-Developers
Acknowledgements

Introduction

HOD Optimization Routine (HODOR) is an adaptable python code that fits and HOD model on halo catalogues. Currently, there are two HOD implemented models: the vanila 5-parameters one and the 9-parameters one from Contreras (1301.3497). Additionaly, the velocity dispersion along the line-of-sight is added and cannot be removed (on ToDo list) The code has been conceived to by highly adaptable to different data-sets, different clustering (2PCF and P(k)) codes and different HOD models.

If you use this tool in research work that results in publications, please cite the following paper:

Variu et al. 2023 arXiv:2307.14197.

Dependencies

The dependencies of this code are as follows:

One can replace pyfcfc or pypowspec by different clustering codes. In compute_2pcf_old.py file there is different ways to compute the 2PCF, e.g. corrfunc or halotools. Similarly for P(k), in compute_pspec_old_1.py and compute_pspec_old_2.py.

All packages are installed in a docker image that one can pull from here. For a tutorial on how to use docker images on the NERSC super-computer, one can read this. However, this batch-script is already using the docker image on NERSC.

Usage

After downloading the code, one has to fill in the configuration file and run main.py:

python main.py --config <CONFIG FILE>

By default, the code will run PyMultiNest. However a short test is also provided in the same main file. As an initial check, if the chi2 computation is working and provides reasonable values, a large chunk of the code should work properly. Then, one can try the code together with a minimizer.

The following file

python main_cat.py

is conceived to create the catalogues based on the best-fitting parameters.

Currently, the 'model' catalogues (e.g. FastPM) in hod_pack/data.py should be and HDF5 file with the following data-set:

['halo/PID', 'halo/X', 'halo/Y', 'halo/Z', 'halo/VX', 'halo/VY', 'halo/VZ', 'halo/ID', 'halo/Rvir', 'halo/Rs', 'halo/Mvir'].

However, the format of the file is not important, as long as the necessary fields are provided. If another format is required, one should modify accordingly the function

compute_halocat()

in the hod_pack/data.py file.

If one wants to use different clustering codes, one should modify accordingly the hod_pack/compute_2pcf.py hod_pack/compute_pspec.py.

If one wants to implement additional HOD models, one should maintain the same structure as the ones in hod_pack/hod_models.py and then one should modify accordingly the

compute_model_instance()

function in the hod_pack/hod_models.py file.

If one wants to use a different optimizer, one can simply add it in the hod_pack/optimizers.py as the MultiNest example, and remove the unnecessary ones.

...To be developped ...

Configuration parameters

All configuration parameters should be given in the configuration file. An example configuration is configs/config.ini.

Co-Developers

Dr. Cheng Zhao

Acknowledgements

I thank Dr. Shadab Alam, Dr. Chia-Hsun Chuang and Daniel Felipe Forero-Sánchez for their suggestions.

zdplayground/HODOR