reLAISS: A Python Package for Flexible Similarity Searches of Supernovae and Their Host Galaxies

TL;DR

reLAISS is a Python package that enhances similarity searches for supernovae and their host galaxies by combining interpretable light curve features with host galaxy photometry, aiding in the identification of rare events.

Contribution

It introduces a flexible, customizable framework for supernova similarity searches that integrates explosion physics and stellar populations, building on and expanding the original LAISS framework.

Findings

Allows customization of neighbor retrieval and feature weighting.

Incorporates Monte Carlo simulations for better matching.

Provides a publicly available package with tutorials.

Abstract

Discovery rates of supernovae are expected to surpass one million events annually with the Vera C. Rubin Observatory. With unprecedented sample sizes of both common and rare transient types, photometric classification alone will be insufficient for finding one-in-a-million events and prioritizing the 1% of events for spectroscopic follow-up observations. Here, we present reLAISS, a modified framework for similarity searches of supernovae using extracted features of ZTF light curves and Pan-STARRS host galaxy photometry and built on the original LAISS framework. Unlike its predecessor, reLAISS couples interpretable light curve morphology features with extinction-corrected host-galaxy colors to probe both explosion physics and associated stellar populations simultaneously. The library allows users to customize the number of neighbors retrieved, the weight of host and light curve features, and the use of Monte Carlo simulations to ensure relevant matches when features are poorly constrained. We release reLAISS as a pip-installable package with an accompanying reference set of 20,000 features, and a set of tutorials that demonstrate the code's expanded functionality. All source code can be found at https://github.com/evan-reynolds/re-laiss .