Spectroscopic r-Process Abundance Retrieval for Kilonovae (SPARK): Abundances and Features in the Early, Blue Ejecta of GW170817

Nicholas Vieira ( Université McGill )

Freshly-synthesized r-process elements in kilonovae ejecta imprint billions of absorption lines on optical spectra, as observed in the GW170817 binary neutron star merger. These spectral features encode insights on the physical conditions of the r-process and the origins of the ejecta material, but identifying specific lines and inferring the abundance pattern is computationally challenging. We introduce Spectroscopic r-Process Abundance Retrieval for Kilonovae (SPARK), a framework to perform Bayesian inference on kilonova spectra with the goals of (1) inferring elemental abundance patterns, and (2) identifying individual absorption features in early-time, optically-thick spectra. SPARK inputs an atomic line list and abundances from nuclear network calculations into the TARDIS Monte Carlo radiative transfer code, and performs fast Bayesian inference on observed kilonova spectra by training a Gaussian process surrogate model for the approximate posteriors of key kilonova ejecta parameters. We use the spectrum of GW170817 at 1.5 days post-merger to perform the first inference on a kilonova spectrum. We recover the previous identification of Strontium lines and also identify absorption by Yttrium at ∼3500 Å. Our inference shows that the early blue ejecta had a relatively high electron fraction, hot entropy pattern, and considerable velocity, resulting in an abundance pattern with no lanthanide elements. Our approach will enable computationally-tractable inference using spectra of the large number of kilonovae expected to be discovered through multi-messenger gravitational wave observations over the next few years.