Measuring Chemical Likeness of Stars with Relevant Scaled Component Analysis

Abstract Identification of chemically similar stars using elemental abundances is core to many pursuits within Galactic archeology. However, measuring the chemical likeness directly limited by systematic imprints imperfect synthetic spectra in abundance derivation. We present a novel data-driven model that capable identifying from alone. call this relevant scaled component analysis (RSCA). RSCA finds mapping stellar representation optimizes recovery known open clusters. By design, amplifies factors variation and minimizes those nonchemical parameters, such as instrument systematics. The resultant can therefore be used for precise measurements similarity between stars. validate 185 cluster 22 clusters Apache Point Observatory Evolution Experiment survey. quantify our performance reference set 151,145 field find identifies siblings more effectively than stellar-abundance measurements. Using RSCA, 1.8% pairs are birth siblings, compared 2.3% when labels. almost all information leveraged fits into two-dimensional basis, which we link [Fe/H] α -element abundances. conclude tagging their remains prohibitive. has noticeable gain, approach poised benefit larger data sets improved algorithm designs.