an astronomer on a quest for new star clusters

header credit: European Space Agency / Gaia

Who am I?

Based at the University of Heidelberg, Germany, I’m an astronomer looking for and researching open clusters in data from the Gaia satellite — with clustering algorithms, machine learning and statistics.

I’m passionate about programming, equality in science, and science communication.


I use machine learning and statistics to push the boundaries of Gaia data.

How many open clusters are in the Milky Way?

Open clusters form when a gas cloud condenses into a group of a few hundred to a few thousand stars. They’re hugely useful in many areas of astronomy, since stars with similar properties (age, composition, etc.) can be studied at the same time.

It’s estimated that our galaxy contains around 100,000 open clusters, but we only currently know of approximately 2000 in our immediate stellar neighbourhood. How many more can be found with new data and new techniques?

In my first paper, I tested three different algorithms and found that HDBSCAN seems to be the best at retrieving open clusters in data from Gaia. Soon, we’ll apply this method to Gaia’s latest data release, EDR3 (out December 2020).

What even is an open cluster?

One of the biggest challenges I’ve had so far in my research is answering a seemingly fundamental question. Hundreds of new open clusters have already been reported since the first data release from ESA’s Gaia satellite, but no clear method exists in the literature to fully quantify whether or not a candidate object is real.

In my first paper, I developed a density-based test to statistically quantify how dense a candidate object appears in data from Gaia — a sort of “signal to noise ratio” for star clusters. In the future, we’d also like to extend existing machine learning approaches in the literature to classify clusters based on their photometry (the brightnesses and colours of member stars.)