Skip to content

Installation Guide

1. Prerequisites

Before installing PEKET, ensure you have the following dependencies correctly set up in your environment:

HTCondor Requirement

gw-setup-pipeline heavily relies on HTCondor to create multiples .sub or .dag files. Make sure HTCondor is installed on your cluster/device.

Dynamic slots are recommended as the main job reserves 16 Gb of RAM, and sub-jobs reserve 4 Gb. (Note: HTCondor is not required if you only plan to use kn-make-grid and kn-ts-loop).

Important: Python Version

Build FIESTA and NMMA from source in the same environment. NMMA explicitly requires Python 3.12.

LSST Utilities (Required for the kn_side)

If you plane to use the KN related cli or utils or the LSST synthetic lightcurve generation utilities (generate_synth_lc_lsst), you will need the Rubin Observatory simulation framework. It is highly recommended to install it via conda-forge to gracefully handle its complex dependencies (rubin_sim is not in the requirements BUT is necessary): 

conda install -c conda-forge rubin_sim
You also need to run 
rs_download_data -d throughputs
To get the throughputs files of the filters needed for the LSST related functions.

2. Source Code Modifications

Ensure compatibility

To ensure full compatibility with PEKET, minor modifications to NMMA and PyMultiNest are required. These modifications are minimal and will not alter the standard functioning of either package.

NMMA Modification 1 (nmma/em/model.py, line 637): Allows generating Bu2019lm lightcurves with inclinationEM instead of KNtheta. 

def generate_lightcurve(self, sample_times, parameters, filters = 'all'):
        parameters = self.parameter_conversion(parameters) # <-- THIS ONE 
        parameters_list = self.em_parameter_setup(parameters)

NMMA Modification 2 (nmma/post_processing/plotting_routines.py, line 160): Explicitly names the argument for corner plots. 

corner_plot(plot_samples.T, labels, limits, outdir=outdir) # <-- Explicitly name the argument

PyMultiNest Modification (pymultinest/analyse.py, line 34): Prevents Fortran reading issues. 

def loadtxt2d(intext): 
    try:
        return numpy.loadtxt(intext, ndmin=2)
    except ValueError:
        # Catch Fortran formatting missing the 'e'
        with open(intext, 'r') as f:
            text = f.read()
        text = re.sub(r'(?<=\d)(-[1-3]\d\d)', r'e\1', text) 
        text = re.sub(r'(?<=\d)(\+[1-3]\d\d)', r'e\1', text)
        try:
            return numpy.loadtxt(StringIO(text), ndmin=2)
        except:
            return numpy.loadtxt(StringIO(text))
    except:
        return numpy.loadtxt(intext)

3. Install PEKET

Once prerequisites and modifications are done, clone and install the package: 

git clone -b main git@github.com:NJamsin/peket.git
cd peket
pip install -e .