Galactic merger-history clustering
Galactic merger-history clustering was part of his work as a research assistant at NARIT, the National Astronomical Research Institute of Thailand, in 2026. It is a pipeline that reconstructs a galaxy's history of mergers from large catalogues of stars, a field known as galactic archaeology.
The idea
When a small galaxy is torn apart and absorbed by a larger one, its stars spread out across the sky. But they keep traces of their shared origin in how they move and what they are made of, so stars from the same merger stay clumped together in those properties. Finding the clumps recovers the mergers.
Method
The pipeline clusters stars with HDBSCAN clustering, a density-based clustering algorithm. It suits this problem for three reasons:
- it doesn't need to be told how many groups to find;
- it finds clusters of different densities;
- it labels scattered stars as noise instead of forcing them into a group.
The work is in Python and runs on HPC and PBS job orchestration infrastructure, because the stellar datasets are large.
Status
The results are not published yet: a paper on this work is in preparation. Until it is out, this page describes the method only.
Related
- Intro to DBSCAN, his course on the algorithm HDBSCAN extends.
- Paper Scout, whose example run maps the literature on exactly this question: how galaxy merger histories are reconstructed from stellar data.
- Faint stellar structure detection, the other half of his NARIT work.