The Nova Synthetic Data Base: A Principal Component/AI Analysis of Novae Synoptic Spectra

TL;DR

This paper introduces the Nova Synthetic Data Base, a comprehensive set of synthetic nova spectra, and presents an AI-based framework for extracting physical nova properties from limited spectral data.

Contribution

It develops a principal component analysis and AI framework to efficiently interpret nova spectra and derive physical parameters using minimal observational data.

Findings

High prediction accuracy for nova physical properties.

Robustness of the method to data noise.

Effective identification of diagnostic spectral lines.

Abstract

The Nova Synthetic Data Base (NSDB) is presented as the first publicly available database of synthetic spectra for classical nova shells, spanning an unprecedented range of physical parameters (e.g., ejecta mass, chemical composition, temperature, and luminosity of the white dwarf) at several post-eruption ages. Generated using detailed 3D photoionization models, this homogeneous database enables a systematic exploration of spectral features in novae. In this work, we introduce a principal component analysis/AI-based framework to derive time-dependent proxies for retrieving the physical properties of novae from limited spectral data. By analyzing the correlations between the eigenspectra and the grid's variables, a reduced set of diagnostic spectral lines is derived, paving the way for robust multiregressor machine-learning algorithms with a minimal effort observational set. The prediction capability of the method is high and robust to data noise. The results establish a proof of concept for the use of model grids combined with physically controlled AI as a tool to interpret novae observations in the context of the large number of events expected from future wide-area surveys.