We announce the public release of SNID-SAGE (SuperNova IDentification – Spectral Analysis and Guided Exploration), a framework for supernova spectral classification and analysis.
SNID-SAGE builds on the original SNID methodology, retaining its cross-correlation approach while introducing a new similarity metric and combining information from multiple template matches through a clustering-based approach in redshift space. This allows more robust redshift and classification estimates than relying on a single best-matching template, and provides quantitative measures of the quality and confidence of each solution.
SNID-SAGE includes a curated and upgradeable template library of approximately 6000 spectra, spanning 15 template categories: AGN, CV, GAP, Galaxy, II, Ia, Ib, Ibn, Ic, Icn, KN, LFBOT, SLSN, Star, and TDE. In addition to classification, SNID-SAGE provides tools for interactive spectral analysis, including emission and absorption line identification with velocity measurements, and integrates an AI-based assistant that produces natural-language summaries to support interpretation of the results while remaining separate from the numerical inference.
A full description of the method and its validation is given in Stoppa & Smartt (2026):
https://arxiv.org/abs/2603.28741
The code and documentation are publicly available at:
https://github.com/FiorenSt/SNID-SAGE
https://fiorenst.github.io/SNID-SAGE/
SNID-SAGE can be installed via:
pip install snid-sage
The main application can then be launched with:
snid-sage
The default template library can be accessed and managed using:
snid-sage-templates
which provides utilities for uploading, updating, and inspecting the spectral templates.
We encourage the community to contribute additional supernova spectra to expand and improve the template library. Template-matching classification methods such as SNID-SAGE directly benefit from increased template diversity and coverage, leading to more robust and accurate redshift and classification estimates, particularly for rare or underrepresented transient types. Authors interested in contributing spectra are invited to contact the corresponding author at fiorenzo.stoppa@physics.ox.ac.uk. Spectra can be submitted by email, including the original spectral files together with the phase relative to maximum light (in days) and the host-galaxy redshift. The phase information can be provided either in the file names or in an accompanying CSV file. Contributed spectra will be reviewed and incorporated into future releases with appropriate attribution.
Comments