🔭 astrophysics

The FLAMINGO simulations data release

This paper announces the public release of over 2.3 petabytes of data from the FLAMINGO cosmological simulation suite, which includes 22 hydrodynamical and 16 gravity-only simulations designed to study galaxy formation, cosmology, and dark matter through various physical models and resolutions.

Original authors: John C. Helly (ICC, Department of Physics, Durham University), Robert J. McGibbon (Leiden Observatory, Leiden University), Joop Schaye (Leiden Observatory, Leiden University), Matthieu Schaller (Leide

Published 2026-04-28

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: John C. Helly (ICC, Department of Physics, Durham University), Robert J. McGibbon (Leiden Observatory, Leiden University), Joop Schaye (Leiden Observatory, Leiden University), Matthieu Schaller (Leiden Observatory, Leiden University, Lorentz Institute for Theoretical Physics, Leiden University), William McDonald (Leiden Observatory, Leiden University), Joey Braspenning (Leiden Observatory, Leiden University, Max-Planck-Institut fur Astronomie), Jeger C. Broxterman (Leiden Observatory, Leiden University, Lorentz Institute for Theoretical Physics, Leiden University), Emily E. Costello (Astrophysics Research Institute, Liverpool John Moores University), Willem Elbers (ICC, Department of Physics, Durham University), Carlos S. Frenk (ICC, Department of Physics, Durham University), Adrian Jenkins (ICC, Department of Physics, Durham University), Roi Kugel (Leiden Observatory, Leiden University), Ian G. McCarthy (Astrophysics Research Institute, Liverpool John Moores University), Jaime Salcido (Astrophysics Research Institute, Liverpool John Moores University), Marcel P. van Daalen (Leiden Observatory, Leiden University), Bert Vandenbroucke (Leiden Observatory, Leiden University), Tianyi Yang (Astrophysics Research Institute, Liverpool John Moores University)

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Cosmic Lego Set: Explaining the FLAMINGO Data Release

Imagine you wanted to build the most realistic, massive, and detailed model of a bustling city ever seen. You wouldn’t just want a map of the streets; you’d want to know where the people are walking, how the traffic flows, how the power grid works, and even how the trash is collected.

But there’s a problem: building a real city is impossible, and even a tiny model is incredibly expensive and time-consuming to make.

In the world of science, astronomers face this exact problem when trying to understand the Universe. They can’t build a real galaxy, so they build "Cosmological Simulations." This paper describes the release of a massive digital project called FLAMINGO.

1. What is FLAMINGO? (The Ultimate Digital Universe)

Think of FLAMINGO as a super-powered, digital "SimCity" for the entire Universe.

Instead of simulating houses and cars, scientists are simulating billions of years of history, trillions of stars, massive clouds of gas, and mysterious "dark matter" that acts like the invisible scaffolding of everything.

The "FLAMINGO" name is actually an acronym, but you can think of it as a High-Definition Virtual Reality headset for the Cosmos. It’s not just one simulation; it’s a whole suite of 38 different "universes" that scientists have created to test different rules of physics.

2. Why so many different versions? (The "What If?" Machine)

Why would you build 38 different versions of the same thing? Because scientists are playing a giant game of "What If?"

What if gravity was a little stronger?
What if dark matter behaved differently?
What if black holes were more "aggressive" and blew more gas out of galaxies?

By running these different versions, scientists can compare them to the real sky we see through telescopes. If "Universe Version #12" looks exactly like the real sky, but "Universe Version #5" looks totally wrong, scientists know they are on the right track to understanding the true laws of nature.

3. The Scale: A Digital Ocean of Data

The paper mentions that they are releasing 2.3 petabytes of data. To put that in perspective:

If you tried to download this much data on a standard home internet connection, it would take centuries.
It’s like trying to hand someone a single grain of sand from a massive beach.

Because the data is so huge, the researchers didn't just dump it on the internet and say "good luck." They built a special digital buffet. Instead of forcing you to eat the whole ocean, they created a "web service" that lets you pick exactly what you want—like asking for just a single "slice" of a galaxy or a specific "snapshot" of a cluster of stars—so you don't crash your computer.

4. What’s inside the box? (The Ingredients)

The simulation includes all the "ingredients" of a real universe:

The Stars: The bright lights of the cosmos.
The Gas: The invisible "soup" that stars are born from.
The Black Holes: The cosmic vacuum cleaners that influence everything around them.
The Neutrinos: Ghostly particles that fly through everything.
The Dark Matter: The invisible "glue" that holds galaxies together.

5. Why does this matter to you?

We are currently in a "Golden Age" of astronomy. New telescopes (like the Euclid mission) are sending back incredible photos of the deep sky. But a photo is just a picture; it doesn't tell you why things look that way.

FLAMINGO provides the instruction manual. It gives astronomers a way to say, "Ah, I see that weird cluster of galaxies in the telescope; according to my FLAMINGO simulation, that happened because a black hole exploded there a billion years ago!"

In short: FLAMINGO is a massive, digital laboratory that allows us to experiment with the laws of physics on a scale so large, it's almost impossible to wrap our heads around.

Technical Summary: The FLAMINGO Simulations Data Release

1. Problem Statement

As observational cosmology enters an era of unprecedented precision (e.g., through the Euclid and LSST surveys), there is an increasing need to model baryonic effects on large-scale structures. While Dark Matter-only (DMO) simulations are computationally efficient, they fail to capture non-linear processes like radiative cooling, star formation, and Active Galactic Nucleus (AGN) feedback, which redistribute matter on scales $\lesssim 10$ Mpc.

The scientific community faces a "resolution vs. volume" trade-off: high-resolution simulations (e.g., IllustrisTNG, EAGLE) are limited to small volumes and cannot provide statistics for rare objects like galaxy clusters, while large-volume simulations often lack the baryonic physics necessary to predict observables like the Sunyaev-Zel’dovich (SZ) effect or weak lensing convergence.

2. Methodology

The FLAMINGO (Full-hydro Large-scale structure simulations with All-sky Mapping for the Interpretation of Next Generation Observations) project addresses this by providing a suite of large-volume hydrodynamical simulations.

Simulation Framework: All simulations were performed using the open-source Swift code, employing a 4th-order fast-multipole method for gravity and Smoothed Particle Hydrodynamics (SPH) for gas. Neutrinos are explicitly simulated as particles.
Subgrid Physics & Calibration: The fiducial model includes radiative cooling, star formation, chemical enrichment, supernova feedback, and two types of AGN feedback (thermal winds and kinetic jets). Crucially, the subgrid parameters were calibrated using a Gaussian process emulator and Monte Carlo Markov Chain (MCMC) searches to match the observed $z=0$ galaxy stellar mass function (from the GAMA survey) and cluster gas fractions.
Suite Composition: The release includes 22 hydrodynamical simulations and 16 DMO simulations. The suite varies:
- Resolution: Three levels ( $m8, m9, m10$ ).
- Baryonic Physics: Ten variations of feedback to test the galaxy-halo connection and cluster gas fractions.
- Cosmology: Nine different cosmologies, including $\Lambda$ CDM variations, neutrino mass variations, decaying dark matter, and evolving dark energy.
Data Access Infrastructure: To manage the massive data volume, the authors implemented a web service using Apache Tomcat and the MessagePack serialization format. This allows users to "stream" specific datasets or slices of HDF5 files via a Python client (hdfstream), bypassing the need to download entire multi-terabyte files.

3. Key Contributions (Data Products)

The release provides over 2.3 petabytes of data, categorized into:

Snapshots: Full particle snapshots at various redshifts, plus "reduced" (massive haloes only) and "downsampled" (1% random particles) versions.
Catalogues: Halo and galaxy catalogues produced via the HBT-HERONS and SOAP structure finders.
Lightcones:
- HEALPix Maps: All-sky projected maps (at $N_{side} = 16384$ and $4096$) for mass, star formation rate, weak lensing, CMB lensing, thermal/kinetic SZ, X-ray emission, and the Cosmic Infrared Background (CIB).
- Particle/Halo Lightcones: Detailed properties of particles and haloes crossing the observer's past lightcone.
Power Spectra: 3-D auto- and cross-spectra for various species (CDM, gas, stars, neutrinos, etc.) computed at 123 redshifts.

4. Results and Discussion

Observational Consistency: The fiducial model successfully reproduces the $z=0$ galaxy stellar mass function, black hole mass relations, and cluster scaling relations. It also aligns well with CMB lensing and X-ray/cosmic shear cross-correlations.
Tensions: The simulations reflect the known $S_8$ tension between cosmic shear and thermal SZ power spectra.
Numerical Convergence: The authors note that comparing different resolutions is complex because the subgrid calibration is resolution-dependent; thus, different resolutions effectively represent different physical models.
Identified Bugs: The paper transparently documents known issues, such as the "repositioning" of black holes in gas-poor galaxies and specific star-formation threshold errors in primordial gas, providing guidance on how to interpret these effects.

5. Significance

The FLAMINGO data release is a major contribution to Open Science. By providing a large-volume, hydrodynamically consistent suite that includes neutrinos and diverse cosmologies, it serves as a vital "virtual laboratory." It enables researchers to:

Quantify baryonic uncertainties in upcoming large-scale structure surveys.
Study the evolution of rare, massive objects like galaxy clusters.
Test non-standard cosmological models (e.g., decaying dark matter) within a realistic baryonic framework.