🔭 astrophysics

Intrinsic alignments in the FLAMINGO simulations with two-point statistics

Using the FLAMINGO hydrodynamic simulations, this study constrains intrinsic alignment models for Luminous Red Galaxies, demonstrating that a mass-dependent TATT model (TATT-M) is strongly preferred over the standard NLA model and showing that alignment parameters are robust against variations in AGN and supernova feedback.

Original authors: A. Herle, N. E. Chisari, H. Hoekstra, D. Navarro-Gironés, M. Schaller, J. Schaye

Published 2026-01-23

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: A. Herle, N. E. Chisari, H. Hoekstra, D. Navarro-Gironés, M. Schaller, J. Schaye

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

Imagine the universe as a giant, invisible web made of dark matter. Galaxies are like beads strung along this web. For decades, astronomers have been trying to measure the shape of this web by looking at how the light from distant galaxies gets stretched and distorted as it travels to us. This distortion is called "weak gravitational lensing," and it's our best tool for understanding dark energy and the expansion of the universe.

However, there's a problem. The galaxies themselves aren't perfect spheres; they are often stretched out like rugby balls. Because they form in the same gravitational tides that shape the web, they tend to line up with each other naturally. This is called Intrinsic Alignment (IA).

Think of it like this: If you are trying to measure how a strong wind (dark matter) is bending a flag (light from a galaxy), you need to know if the flag is already crumpled because of how it was sewn together (intrinsic alignment). If you don't account for the crumple, you'll think the wind is stronger or weaker than it actually is. This "crumple" is a major source of error for upcoming giant sky surveys like Euclid and LSST.

To fix this, the authors of this paper used a massive, super-detailed computer simulation called FLAMINGO. Instead of just watching the universe happen, they built a digital universe with over 4.9 million galaxies to study exactly how these "crumples" happen.

Here is what they discovered, broken down simply:

1. The "Crumple" Models

Astronomers use mathematical recipes to predict how galaxies align. The paper tested two main recipes:

NLA (The Simple Recipe): This assumes galaxies align in a straightforward way based on the local gravity. It's like saying, "If the wind blows from the left, the flag points left."
TATT (The Complex Recipe): This adds extra ingredients. It accounts for how galaxies spin (torquing) and how their shapes are influenced by the density of the surrounding matter. It's like saying, "The flag points left, but it also twists because of the spin of the air, and its shape changes if the air is thick or thin."

The Result: Both recipes worked well, but the complex one (TATT) was better at explaining the data, especially on smaller scales. It could describe the "crumple" with less error than the simple one.

2. The "Size Matters" Discovery

The authors found that the strength of this alignment depends heavily on the mass of the galaxy's home (its dark matter halo).

Analogy: Imagine a small twig in a stream versus a giant log. The log is much more likely to be dragged and aligned by the current than the twig.
The Finding: Heavier galaxies align much more strongly than lighter ones. The team found a clear mathematical rule (a power law) that describes this: as the mass goes up, the alignment strength goes up.

3. A New, Smarter Recipe (TATT-M)

Because they discovered that mass is the key, they created a new, upgraded version of the complex recipe called TATT-M.

How it works: Instead of treating the alignment of every galaxy as a mystery with many unknown variables, TATT-M uses the galaxy's mass to predict most of the details. It's like having a recipe that says, "If you have a 5-pound chicken, you need 2 cups of flour; if you have a 10-pound chicken, you need 4 cups." You don't need to guess the flour amount; the weight tells you.
The Benefit: This new model is much more precise. It reduces the "guesswork" for astronomers, allowing them to get tighter, more accurate measurements of the universe's expansion. The data strongly preferred this new, mass-dependent model over the old, simpler ones.

4. Does the "Kitchen" Matter? (Feedback)

In these simulations, "feedback" refers to the energy explosions from stars (supernovae) and black holes (AGN) that blow gas around and change how galaxies form. It's like a chef changing the heat or adding different spices.

The Question: Do these explosions change how the galaxies align?
The Finding: Surprisingly, the answer is mostly no. The authors tested different "kitchen settings" (stronger explosions, different types of black hole jets). They found that as long as they compared galaxies of the same mass, the alignment was the same.
The Catch: The only time the alignment seemed to change was because the explosions changed the mass of the galaxies. If a galaxy lost mass because of a strong explosion, it aligned less. But the explosion itself didn't twist the galaxy's shape directly; it just made the galaxy smaller.

Summary

This paper is like a master chef testing a new recipe for a giant banquet. They used a massive digital kitchen (FLAMINGO) to bake 4.9 million digital galaxies. They found that:

Galaxies naturally line up with the cosmic web, which can mess up our measurements of the universe.
Heavier galaxies line up more strongly than lighter ones.
A new recipe (TATT-M) that accounts for mass is the most accurate way to predict this alignment.
Explosions from stars and black holes don't directly twist the galaxies; they only affect alignment by changing the galaxies' mass.

This work gives astronomers the "instruction manual" they need to clean up their data for the next generation of telescopes, ensuring they can see the true shape of the universe without the "crumple" of intrinsic alignment getting in the way.

Technical Summary: Intrinsic Alignments in the FLAMINGO Simulations with Two-Point Statistics

Problem Statement
Intrinsic alignments (IA) of galaxies constitute a major astrophysical systematic contaminant for next-generation weak gravitational lensing surveys such as Euclid and the Legacy Survey of Space and Time (LSST). IA induces correlations in galaxy shapes that mimic the cosmic shear signal, potentially biasing cosmological parameter estimation if not modeled correctly. While observational data has improved, understanding the limitations of current IA modeling strategies and establishing sensible priors requires large-scale hydrodynamic simulations. These simulations must possess sufficient volume to match survey scales and sufficient resolution to resolve small-scale effects influencing the tidal field. Furthermore, the dependence of IA on baryonic physics (feedback) remains an open question, as existing alignment models do not currently account for feedback variations.

Methodology
The authors utilize the FLAMINGO suite of hydrodynamical simulations, specifically a 2.8 Gpc $^3$ box and a suite of 1 Gpc $^3$ runs with varied feedback implementations (thermal vs. jet AGN feedback, and variations in supernova feedback strength). The study focuses on a sample of $\approx$ 4.9 million Luminous Red Galaxy (LRG) analogs at redshift $z=0$ , selected from subhalos with bound dark matter masses $>10^{12} M_\odot$ and $>300$ stellar particles.

The authors measure two-point statistics:

Position-Position Correlation ( $w_{gg}$ ): Tracing galaxy clustering.
Position-Shape Correlation ( $w_{g+}$ ): Tracing the intrinsic alignment signal.

These measurements are performed using the generalized Landy-Szalay estimators and the IACorr code. The authors jointly model the clustering and alignment signals using non-linear galaxy bias for $w_{gg}$ and two IA models for $w_{g+}$ :

Non-Linear Alignment (NLA): A linear alignment model extended to quasi-linear scales using the non-linear matter power spectrum, characterized by amplitude $A_1$ and bias $b_1$ .
Tidal Alignment and Tidal Torquing (TATT): A more complex model incorporating linear and quadratic terms, characterized by amplitudes $A_1$ , $A_2$ , and $A_{1\delta}$ .

The analysis employs a Bayesian approach using the nautilus nested sampler to constrain parameters. The authors also investigate the dependence of these parameters on halo mass and test the sensitivity of the alignment signal to variations in AGN and supernova feedback by comparing different FLAMINGO simulation runs.

Key Contributions and Results

Model Performance: Both NLA and TATT models provide good fits to the FLAMINGO data. The TATT model yields smaller residuals at small scales ( $r < 10$ Mpc/ $h$ ) compared to NLA, allowing for a lower minimum scale cut ($5$ Mpc/ $h$ vs. $10$ Mpc/ $h$ ) while maintaining a reduced chi-squared ( $\chi^2_{red}$ ) comparable to NLA. The residuals between the models and data are within a few percent over the scales considered ($5-100$ Mpc/ $h$ ).
Comparison with Observations: The measured NLA amplitude ( $A_1$ ) from the FLAMINGO sample shows good agreement with various observational studies of LRGs (e.g., KiDS, DES, SDSS) across a range of luminosities. The simulation extends the measured range to higher luminosities than previously probed by observations.
Mass Dependence: The authors confirm that the alignment amplitude $A_1$ scales with halo mass as a power law. Crucially, they present the first measurement of the mass dependence for the higher-order TATT parameters ( $A_2$ and $A_{1\delta}$ ). They find that $A_2$ is roughly a constant multiple of $A_1$ , and the ratio $A_{1\delta}/A_1$ follows a power-law dependence on halo mass.
Introduction of TATT-M: Based on the empirical relations found between TATT parameters and halo mass, the authors introduce a mass-dependent TATT model, TATT-M. This model reduces the TATT parameter space by expressing $A_2$ $A_{2}$ and $A_{1\delta}$ $A_{1 δ}$ as functions of $A_1$ $A_{1}$ and halo mass (derived from fitting the halo sample).
- Bayesian model comparison shows that TATT-M is very strongly preferred over the standard NLA model for the central galaxy sample.
- The use of TATT-M significantly tightens the constraints on $A_1$ (reducing the error bar by a factor of $\approx 3.7$ ) compared to the standard TATT model, due to the informative priors derived from the simulation's halo sample.
Feedback Sensitivity: The authors test the sensitivity of IA parameters to baryonic feedback using the FLAMINGO feedback variations.
- Initial comparisons of raw $A_1$ values across feedback runs showed variations, particularly for runs with shifted stellar mass functions ('M*-σ') and strong jet feedback.
- However, the authors demonstrate that these variations are primarily driven by changes in the mean halo mass of the selected galaxy sample resulting from the fixed stellar particle cut, rather than a direct physical effect of feedback on the alignment mechanism itself.
- When the analysis is binned by halo mass to remove the mass dependence, all feedback variations (including thermal, jet, and varying supernova strengths) yield consistent alignment amplitudes and slopes within statistical uncertainties. The paper concludes that, within the resolution limits of FLAMINGO, feedback does not significantly alter the alignment signal beyond its effect on the galaxy stellar mass distribution.

Significance
This work provides critical guidance for the modeling of intrinsic alignments in upcoming cosmological surveys. By validating the NLA and TATT models against high-resolution hydrodynamic data, the authors confirm that these models are sufficient for two-point statistics analysis down to scales of $5$ Mpc/ $h$ . The introduction of the TATT-M model offers a pathway to improve the constraining power of future weak lensing analyses by utilizing simulation-derived priors on higher-order alignment terms. Furthermore, the results suggest that for LRG-like samples, the impact of baryonic feedback on intrinsic alignments is largely mediated through changes in galaxy stellar mass, implying that current modeling strategies may be robust against uncertainties in feedback physics, provided mass dependencies are correctly accounted for.

1. The "Crumple" Models

2. The "Size Matters" Discovery

3. A New, Smarter Recipe (TATT-M)

4. Does the "Kitchen" Matter? (Feedback)

Summary

More like this