Institute, founded 2016
Flatiron Institute Center for Computational Astrophysics (CCA)
The Flatiron Institute's Center for Computational Astrophysics in New York builds simulations and open tools. In 2026 its team saw a star vanish.
- Location
- 162 Fifth Avenue, Flatiron District, New York City
- Focus
- Cosmology, Galaxies, Exoplanets, Gravitational waves, Stars, Black holes
- Coordinates
- 40.740, -73.991
The Center for Computational Astrophysics (CCA) is part of the Simons Foundation’s Flatiron Institute in New York City, launched in 2016 and dedicated on September 6, 2017. It focuses on the computing side of astronomy: simulations, statistics, machine learning and the shared software that researchers rely on. Julianne Dalcanton has been its director since September 2021. Its eight groups range from Astronomical Data and Cosmology to Exoplanets, Gravitational Wave Astronomy and Machine Learning X Astrophysics.
Flagship tools
- Open-source software: emcee, MESA, celerite, starry and Astrometry.net, all widely used across astronomy.
- CAMELS: a suite of 4,233 cosmological simulations (2,184 hydrodynamic, 2,049 N-body) built to train machine-learning models.
Recent results
On February 12, 2026, a team led by Flatiron’s Kishalay De reported in Science that the supergiant star M31-2014-DS1 in the Andromeda galaxy vanished without a supernova, collapsing directly into a black hole. A “failed supernova” of this kind is a rare chance to watch a black hole form.
The CCA’s simulation work follows a long tradition in which computer models shaped how we see the cosmos, from the 1997 discovery of a universal dark matter halo profile to the two-stage theory of galaxy formation that modern simulations implement.
What to watch, 2026-2030
Our sources list no dated CCA milestones. The data flood is the thing to watch: the Rubin Observatory collects about 10 terabytes a night, Gaia DR4 arrives on December 2, 2026, and Euclid’s first major release starts in November 2026.
Why scientists are converging here
Astronomy is becoming a computing discipline. On arXiv, instrumentation and methods (astro-ph.IM) is the fastest-growing astro-ph category: listings for January to September 2026 are up 32% on the same months of 2025, against 8.5% to 14.8% for the other five categories (the cause has not been analyzed). In our Research Pulse count, “machine learning” appeared in 306 astro-ph abstracts from April to September 2026, about level with the previous six months: established rather than rising. A center built around shared code and simulations serves cosmology, exoplanets and gravitational-wave work at once.
