Installation & Dependencies#

This guide provides complete instructions for installing streamobs and its dependencies.

Quick Start#

# 1. Clone the repository
git clone https://github.com/LSSTDESC/streamobs.git
cd streamobs

# 2a. Set environment variables
export PYTHONPATH=${PWD}:${PYTHONPATH}
export PATH=${PWD}/bin:${PATH}

# 2a. alternatively you can pip install by running the following int the base directory
pip install -e . 

# 3. Install ugali FROM GITHUB (not from PyPI -- see the warning below)
pip install git+https://github.com/DarkEnergySurvey/ugali.git

# 4. Download required data files
python bin/download_data.py

Data Download#

streamobs needs external data files (maglim maps, dust map, completeness, photometric errors). Use the downloader and refer to the Data page for details.

# Download required data (default location: data/)
python bin/download_data.py

# Useful options
python bin/download_data.py --list          # Show what's installed
python bin/download_data.py --force         # Re-download/overwrite
python bin/download_data.py --data-dir DIR  # Custom install location

For troubleshooting and data structure, see StreamObs Data Files.

Dependences#

Required Python packages:

ugali (install from GitHub — see below)
numpy
scipy
pandas
matplotlib
astropy
gala
healpy
healsparse

Installing `ugali`#

Warning

Install ugali from GitHub, not from PyPI.

pip install git+https://github.com/DarkEnergySurvey/ugali.git

pip install ugali gives you release 1.8.0, which is many commits behind and predates support for the current CMD 3.8 isochrone file format. With that version the isochrone files are read against the wrong columns — mass_init picks up the logAge column and becomes constant — so the mass PDF is all zeros and sampling fails with:

RuntimeWarning: invalid value encountered in divide
ValueError: Probabilities contain NaN

The fix for this landed on main after the 1.8.0 tag, so a git install is required until a newer release is published.

Optional Dependencies#

skyproj — sky-projection plots (plotting.plot_stream_in_mask); installable via the plotting extra: pip install -e ".[plotting]"

Installing with pip/conda#

Note

Package installation via pip/conda is planned for future releases.

Installation & Dependencies

Contents