Speaker
Description
The earliest chemical enrichment in the Universe was driven by supernovae of Population III stars. Low-mass second-generation stars formed from gas polluted by these ejecta can survive to the present day as extremely metal-poor (EMP) stars in the Galactic halo. Their abundance patterns provide a fossil record of early nucleosynthesis and chemical evolution. However, an observed EMP star may have been enriched either predominantly by a single Population III supernova or by a mixture of several supernovae. Identifying mono-enriched candidates is therefore essential for using stellar abundances to infer the properties of individual first stars.
In this study, I develop a framework to identify mono-enriched candidates by connecting EMP-star abundances from the SAGA Database to Population III supernova yield models. Mock mono- and multi-enriched abundance patterns are constructed from theoretical Pop III yield models with varying progenitor mass and explosion energy, including dilution effects, and support-vector-machine classifiers are trained to distinguish the two cases. Cross-validation tests indicate that four-dimensional abundance-ratio classifiers balance performance and observational applicability, with C/iron-peak ratios emerging as effective diagnostics.
I apply an ensemble of selected classifiers to SAGA EMP-star abundances, propagate observational uncertainties by resampling, and assign mono-likeness scores to individual stars. By examining the performance of classifiers with different abundance-ratio combinations, I identify which elements and abundance ratios are most important for selecting mono-enriched candidates in observed EMP samples. This framework provides a physically motivated way to select stars that may preserve relatively clean chemical signatures of individual Population III supernovae, forming a basis for future yield fitting and inference of first-star properties.
| Category | Theory |
|---|