Measuring the Metallicities of Type Ia Supernova Using Neural Networks

Age and metallicity are believed to be possible environmental factors for affecting the intrinsic scatter of Type Ia supernovae (SNe Ia). But the relationship between SNe Ia spectra and these properties is not well understood. We apply a unique data-driven approach by designing a neural network to estimate progenitor age and metallicity from near-peak spectra. We use 326 near-peak SNe Ia spectra from the Open Supernova Catalog, and obtain age and metallicity estimates from RCSED, a catalog of several galaxy properties, to train our neural network. Finally, we also run our trained model on the ‘twin’ SN 2011fe and SN 2011by to find estimates of their metallicities and compare our estimates with the current literature.