SALTShaker

An open-source training framework for the SALT3 spectrophotometric model of Type Ia supernovae.

Overview

SALTShaker trains the SALT3 (Spectral Adaptive Light-curve Template 3) model used for standardizing Type Ia supernovae in cosmological distance measurements. The framework:

• Trains spectrophotometric models from combined photometry and spectroscopy data
• In conjunction with SNANA provides a complete pipeline from simulation through cosmological fitting
• Supports custom survey configurations and filter systems

SALT3 is the successor to SALT2 (Guy et al., 2007; Guy et al., 2010) and has been used in major cosmological analyses including Dovekie, DES5YR, and UNITY.

Features

• Model Training: Train SALT3 models on your own photometric and spectroscopic datasets
• Validation Suite: Comprehensive tools to validate trained models against simulations
• Full Pipeline: End-to-end workflow from BYOSED simulations to cosmological parameter estimation
• Flexible Configuration: Extensive configuration options for regularization, priors, and optimization
• Multi-Survey Support: Built-in support for major surveys with K-correction files included
• JAX Backend: GPU-accelerated optimization using JAX

Installation

From PyPI (recommended)

pip install saltshaker-sn

From Source

git clone https://github.com/djones1040/SALTShaker.git
cd SALTShaker
pip install -e .

Dependencies

Core dependencies include: numpy, scipy, astropy, jax, sncosmo, emcee, iminuit, matplotlib, pandas

See setup.py for the complete list.

Quick Start

Download Example Training Data

trainsalt -g

This downloads the public SALT3 training dataset to $OSTRICH_INITFILES.

Train a Model

trainsalt -c /path/to/training.conf

Or use the included example configuration:

cd examples/SALT3TRAIN_K21_PUBLIC
trainsalt -c SALT3_training.conf

Use SALT3 for Light Curve Fitting

Once trained (or using the published model), fit supernova light curves with sncosmo:

import sncosmo

data = sncosmo.load_example_data()
model = sncosmo.Model(source=sncosmo.SALT3Source(outputdir))
res, fitted_model = sncosmo.fit_lc(
    data, model,
    ['z', 't0', 'x0', 'x1', 'c'],
    bounds={'z': (0.3, 0.7)}
)
sncosmo.plot_lc(data, model=fitted_model, errors=res.errors)

Documentation

Full documentation is available at saltshaker.readthedocs.io

• Installation Guide
• Getting Started
• Training Configuration
• Data Formats
• Pipeline Usage

Model Downloads

• Latest SALT3 model: Available in the documentation
• Training data: Included with trainsalt -g or from documentation

Citation

If you use SALTShaker or the SALT3 model, please cite:

Primary SALT3 paper:

Kenworthy et al., 2021, ApJ, 923, 265K arXiv:2104.07795

Additional Publications

• Pierel et al., 2021 - BYOSED simulation framework
• Pierel et al., 2022 - Near-infrared SALT3 extension
• Dai et al., 2023 - Pipeline validation
• Jones et al., 2023 - Host-mass dependent model

Contributing

We welcome contributions! Please:

1. Report bugs and request features via GitHub Issues
2. Submit pull requests for bug fixes and enhancements
3. See the documentation for development setup

License

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

Authors

• David Jones (dojones@hawaii.edu)
• D'Arcy Kenworthy
• Rick Kessler
• Mi Dai
• Justin Pierel

Acknowledgments

SALTShaker builds on the original SALT/SALT2 training code developed by J. Guy. Development has been supported by NASA, DOE, and the Gordon and Betty Moore Foundation.