This package calculates various scaling relations from cluster catalogs. While only appropriate for fitting power laws, it uses a methodology developed by Brandon Kelly for a fully Bayesian approach that incorporates correlated and heteroscedastic measurement errors, intrinsic scatter, selection effects, censored data, and gaussian mixture modeling for the covariates.

For more information on the Kelly method, read his 2007 paper, check out his original IDL implementation linmix_err, or see the python port linmix by Josh Meyers, which was used in this work.

First, you will need conda.

Once you have that, clone this repository into a local directory, and switch to that directory:

git clone https://github.com/sweverett/CluStR
cd CluStR

Next, create a conda environment using the supplied environment.yml:

conda env create -f environment.yml

Now activate the newly-created conda environment:

conda activate clustr

Whenever you are finished, deactivate the conda environment with

conda deactivate clustr

First you will need to clone the repo by moving to the desired local directory and using

git clone https://github.com/sweverett/CluStR

Besides some standard packages like numpy, scipy, and matplotlib that can be aquired through common distributions or pip, CluStR requires the following python packages be installed:

• astropy
• corner
• PyPDF2
• linmix
• rpy2

Note that astropy is now included in Anaconda.

You can look at these links for details, or simply paste the following into terminal:

pip install astropy
pip install corner
pip install pypdf2

The simplest way to get linmix is to clone the repo and install using the given setup file:

git clone https://github.com/jmeyers314/linmix.git
cd linmix
python setup.py install

Now you should be ready to use CluStR!

Most parameters are set in the config.yml file. Here you can set the cosmology, default regression method, plotting options, and most importantly any desired flags. There are three possible flag types: bool, cutoff, and range. For each, you must specify the exact catalog column name you want to make cuts along with the flag type and, if a cutoff or range, the corresponding cut values. All name:value pairs must be separated by a colon.

There are two important things to note that might be unclear:

• Setting a flag type and cut value does not mean the cut will be used! A flag is set to be used in the actual method call - see Example Use below. This allows you to set many flag parameters without having to change the config file everytime you want to use a different combination of flags.

• While it may seem counter-intuitive at first, the flag parameters are set to what data you want to keep, not remove - i.e. set what redshift range you want your clusters to be found in rather than what ranges you want to remove. I found this to eliminate some mental gynmastics when setting cuts but may feel awkward for bools.

Here is an example for each flag type:

<column_name>_bool_type: <True/False>

To only include clusters that are not within r500 of a chip edge, use

edge_exclude_r500_bool_type: False

In other words - only use data that is not flagged with edge_exclude_r500.

<column_name>_cut_type: <above/below>

<column_name>_cut:

To analyze clusters whose redshift is below 1.0, use

redshift_cut_type: below
redshift_cut: 1.0

<column_name>_range_type: <inside/outside>

<column_name>_range_min:

<column_name>_range_max:

To only use clusters with redshift between 0.3 and 0.5, use

redshift_range_type: inside
redshift_range_min: 0.3
redshift_range_max: 0.5

This system may seem inefficient, but allows for quite a bit of flexibility in selecting interesting data subsets.

python clustr.py <catalog.fits> <config.yml>

CluStR has four mandatory inputs: An appropriate cluster FITS file, the covariate variable (x-axis), the response variable (y-axis), and a configuration file. The available columns for axis variables (on the right) and their corresponding input labels (on the left) are:

• lambda : lambda (richness)
• l500kpc : 500_kiloparsecs_band_lumin
• lr2500 : r2500_band_lumin
• lr500 : r500_band_lumin
• lr500cc : r500_core_cropped_band_lumin
• t500kpc : 500_kiloparsecs_temperature
• tr2500 : r2500_temperature
• tr500 : r500_temperature
• tr500cc : r500_core_cropped_temperature

To plot the scaling relation between r2500 temperature and richness, use

python clustr.py <catalog.fits> lambda tr2500 config.yml

The output file will be named <default_prefix>r2500_temperature-lambda.pdf, where you can set the default prefix in config.yml.

Additionally there are other optional arguments: A filename prefix (-p). As described in the Config File section, flag paramters are set in config.yml but are only used if set to True.

This project is licensed under the MIT License - see the LICENSE.md file for details