Natural kind clustering of presolar silicon carbides and its astrophysical implications
Abstract
Presolar stardust grains are found in meteorites and predate the formation of our Solar System. Their unusual isotopic compositions record several major astrophysical processes such as nucleosynthesis, stellar evolution, and galactic chemical evolution. Silicon carbides are the most studied presolar grains, due to their easy extraction during acid dissolution and their relatively large size in the Murchison meteorite. Their isotopic compositions indicate origins in both winds of evolved relatively low mass stars and supernova explosions of massive stars. Their carbon and nitrogen isotopic compositions have been previously used to classify them into different populations. Here, we investigated a natural kind clustering of SiC grains using cluster analysis and a database containing grain size, carbon, nitrogen, silicon, aluminum and titanium isotopic compositions. We tested different algorithms of cluster analysis, including k-means, hierarchical clustering, and mclust, and using different combinations of attributes on SiC grains in order to overcome data scarcity for some attributes. Robust statistical tests were used to guide the clustering process. Overall, derived clusters are in agreement with previously defined SiC groups but also highlight several discrepancies between our model output and previous classification. Our natural kind clustering of SiC grains will be compared with astrophysical models, to better understand stellar and galactic chemical evolution.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.V11C..02B
- Keywords:
-
- 1099 General or miscellaneous;
- GEOCHEMISTRY;
- 1916 Data and information discovery;
- INFORMATICS;
- 1942 Machine learning;
- INFORMATICS;
- 3699 General or miscellaneous;
- MINERALOGY AND PETROLOGY