Projects
***Always changing!***
CosmoBench: A Cosmology Benchmark for Geometric Deep Learning
A very interesting project connecting cosmologists at the Center for Computational Astrophysics with graph experts at the Center for Computational Mathematics! Featuring the merger trees from my CAMELS-SAM simulation suite. Accepted to NeurIPS 2025!
Co-leading the LtU Connections working group, to make the first end-to-end data set that goes from initial conditions + cosmology --> galaxies in observable space.
Expanding CAMELS-SAM to explore the galaxy-halo connection
I created 1000 N-body only simulations, ran them through the Santa Cruz SAM, measured lots of galaxy clustering, and fed it all to deep neural networks to try and predict the input parameters! What's next?
The Halo Occupation of LAGER foreground Emission Line Galaxies
While catching z=6.9 LAEs, LAGER also finds a lot of low-redshift galaxies with strong emission in Hα and [OIII]; what can their clustering tell us about their halo occupation?