Intrinsic alignment of disks and ellipticals across hydrodynamical simulations

This study systematically compares intrinsic alignment signals of disk and elliptical galaxies across three hydrodynamical simulations (TNG300, Horizon-AGN, and EAGLE) at redshifts $z=0$ and $z=1$, revealing generally positive correlations that are robust to shape definitions but sensitive to sub-grid physics and morphological selection criteria, particularly in the Horizon-AGN simulation.

The Gist

Imagine the universe as a giant, cosmic dance floor. For decades, astronomers have been trying to figure out how the dancers (galaxies) move and orient themselves.

Usually, we think of galaxies as just floating randomly. But, they actually have a secret: they tend to line up with each other, like dancers in a conga line or soldiers in a formation. This is called Intrinsic Alignment (IA).

Why does this matter? Because when we look at the universe to measure its secrets (like dark energy), we use a technique called "weak lensing." It's like looking at the universe through a slightly warped glass. If the galaxies are already lined up on their own (intrinsic alignment), it messes up our view of the warping glass, making our measurements of the universe's expansion inaccurate.

The Great Debate: The "Disk" Dancers

The main problem this paper tackles is a disagreement among scientists about how one specific type of galaxy—disk galaxies (the flat, spinning ones like our Milky Way)—align themselves.

• The Observations: When astronomers look at the real sky, disk galaxies seem to have no strong alignment. They look random.
• The Simulations: When scientists run computer simulations of the universe, they get conflicting results. Some simulations say disks align one way (radially, like spokes on a wheel), others say the opposite (tangentially, like a spinning top), and some say nothing at all.

It's like three different chefs trying to bake a cake. One says it should be chocolate, one says vanilla, and one says it should be blue. They are all using different recipes, which makes it hard to know which cake is the "real" one.

The Investigation: A Taste Test

The authors of this paper decided to stop arguing and start comparing. They took three of the most popular "recipes" (computer simulations) used by the scientific community: TNG300, EAGLE, and Horizon-AGN.

They decided to bake the cakes using the exact same ingredients and measuring tools for all three. They didn't just look at the final cake; they looked at the batter, the oven temperature, and the cooling process.

Here is what they found, translated into everyday language:

1. The "Shape" of the Galaxy Matters

In these simulations, you can define a galaxy's shape in two ways:

• The "Simple" Shape: Looking at the whole galaxy, including its fuzzy, messy outer edges.
• The "Reduced" Shape: Ignoring the fuzzy edges and focusing only on the bright, dense center.

The Analogy: Imagine looking at a dandelion.

• The Simple shape is the whole fluffy seed head.
• The Reduced shape is just the green stem and the tight cluster of seeds in the middle.

The paper found that how you define the shape changes the result. If you look at the whole dandelion (simple), the alignment is usually positive (they line up nicely). If you look only at the tight center (reduced), things get weird.

2. The "Spin" vs. The "Color"

The scientists tried to sort the galaxies into "Disks" and "Ellipticals" (round, football-shaped galaxies) using two different methods:

• Method A (Kinematics): How fast are they spinning? (Like sorting dancers by how fast they spin).
• Method B (Color): Are they blue (young, star-forming) or red (old, dead)? (Like sorting dancers by their outfit color).

The Findings:

• TNG300 and EAGLE: These simulations were very consistent. No matter how you sorted them, the disks and ellipticals lined up positively. They were good team players.
• Horizon-AGN: This simulation was the rebel. When they used the "Reduced Shape" (focusing on the center) and sorted by "Spin" (how fast they rotate) at a specific time in the universe's history, the disks actually lined up in the opposite direction (negative alignment).

3. The "Ghost" in the Machine

Why did Horizon-AGN show this weird negative alignment?

The authors realized that the "Spin" method (|v/σ|) was accidentally picking up a lot of low-mass, small galaxies. The "Color" method (red vs. blue) was picking up high-mass, big galaxies.

They did a clever trick: they took the small galaxies from the "Spin" group and mathematically "re-weighted" them to pretend they were big galaxies.

• The Result: Even when they made the small galaxies look like big ones, the weird negative alignment didn't go away.

The Conclusion: It wasn't the size of the galaxies causing the weird alignment. It was the physics inside the simulation (the "sub-grid physics," which is like the secret sauce the chefs use to simulate star formation and black holes). The way Horizon-AGN handles the inner cores of galaxies is different from the others, leading to that strange negative signal.

The Big Takeaway

This paper is like a detective story that solves a mystery by standardizing the investigation.

1. Consistency is King: You can't compare apples to oranges. If you want to know if a simulation is right, you have to measure it the exact same way as the others.
2. The "Blue" Galaxy Myth: Many scientists thought "blue" (disk) galaxies didn't align and could be ignored to fix lensing errors. This paper shows that blue galaxies DO align, and sometimes quite strongly. Ignoring them might actually make our measurements of the universe worse, not better.
3. Physics Matters: The way a simulation models the tiny, inner details of a galaxy (the "sub-grid physics") can completely change the result, even if the galaxies look the same on the outside.

In short: The universe is a complex dance. Sometimes the dancers line up, sometimes they don't. This paper helped us realize that if we want to understand the dance, we need to watch the whole group, use the same camera for everyone, and realize that the "inner steps" of the dancers are just as important as the "outer steps."

Technical Summary

Here is a detailed technical summary of the paper "Intrinsic Alignment of Disks and Ellipticals Across Hydrodynamical Simulations" by M. L. Van Heukelum and N. E. Chisari.

1. Problem Statement

Intrinsic Alignments (IA) refer to the spatial correlations between galaxy positions and shapes caused by gravitational interactions with the large-scale structure, rather than gravitational lensing. IA is a major systematic contaminant in weak gravitational lensing surveys (e.g., Euclid, LSST), which aim to measure cosmological parameters like $\Omega_m$ and $\sigma_8$.

A significant open problem in the field is the inconsistency in the measured IA signal for disk galaxies. Previous studies using different hydrodynamical simulations (e.g., IllustrisTNG, Horizon-AGN, EAGLE) and methodologies have reported conflicting results:

• Some find positive (radial) alignments.
• Some find negative (tangential) alignments.
• Some find null signals.

These discrepancies arise from varying sample selections (morphological definitions), simulation physics, and measurement techniques. This lack of consensus hinders the ability to model and mitigate IA in observational data.

2. Methodology

The authors conduct a consistent, side-by-side comparison of three widely used hydrodynamical simulations: IllustrisTNG (TNG300), EAGLE, and Horizon-AGN.

• Data & Redshifts: Measurements are performed at $z=0$ and $z=1$.
• Shape Definitions: Galaxy shapes are calculated using two inertia tensor definitions:
  1. Simple Inertia Tensor: Standard weighting ($I_{k,l} = \sum m_n r_k r_l$).
  2. Reduced Inertia Tensor: Down-weights the outskirts of galaxies ($I_{k,l} = \sum m_n r_k r_l / r^2$).
• Morphological Selection: To split galaxies into "Disks" and "Ellipticals," the authors use four distinct kinematic and photometric variables, remeasured consistently across simulations where possible:
  • $|v/\sigma|$: Ratio of rotational velocity to velocity dispersion.
  • $\kappa_{rot}$: Ratio of rotational kinetic energy to total kinetic energy.
  • $r-i$ Color: Photometric color (SDSS-like).
  • Bulge-to-Total Ratio (BTR): Available only for TNG300.
• Sample Selection Strategy: To ensure fair comparison despite different simulation resolutions and box sizes, the authors use abundance matching. They fix the elliptical fraction to 10% of the total resolved galaxy sample for each simulation, rather than using fixed threshold values for the variables.
• Statistics: The study measures the projected correlation functions using the quadrupole moment ($\tilde{\xi}_{g+,2}$), which offers a higher signal-to-noise ratio than the standard projected correlation function ($w_{g+}$). They also analyze the "disks around ellipticals" correlation, a specific configuration often cited in literature for producing negative signals.

3. Key Contributions

• Unified Framework: The first study to apply a single, consistent methodology (shape calculation, variable remeasurement, and abundance matching) across TNG300, EAGLE, and Horizon-AGN.
• Resolution of Discrepancies: The paper isolates the specific conditions (simulation, shape definition, redshift, and morphological proxy) under which negative disk alignment signals appear.
• Mass vs. Physics Analysis: By re-weighting samples to match stellar mass distributions, the authors disentangle the effects of galaxy mass from sub-grid physics on IA signals.

4. Key Results

A. General Trends (Disks and Ellipticals around All Galaxies)

• Sign Consistency: In TNG300 and EAGLE, all IA signals (both disks and ellipticals around all galaxies) are positive or consistent with zero.
• Horizon-AGN Exception: In Horizon-AGN, signals are generally positive or null, except for a specific case: disks around ellipticals defined by $|v/\sigma|$ using reduced shapes at $z=1$, which yields a negative (tangential) signal.
• Amplitude: Elliptical signals consistently have higher amplitudes than disk signals.
• Robustness: The main trends (sign and scale dependence) are robust across redshift ($z=0$ vs $z=1$) and shape definitions (simple vs. reduced), though reduced shapes generally yield lower amplitudes due to the down-weighting of outer regions.

B. The "Negative Disk" Signal in Horizon-AGN

The negative signal previously reported in Horizon-AGN is not a universal feature but a specific artifact of:

1. Reduced Shapes: The signal is negative only when using the reduced inertia tensor.
2. Morphological Definition: It occurs specifically when disks are selected via a threshold in $|v/\sigma|$.
3. Redshift: It appears at $z=1$ but is absent or positive at $z=0$.
4. Physical Origin: The authors suggest this tangential alignment arises in the inner regions of galaxies (probed by reduced shapes), likely driven by sub-grid physics rather than large-scale tidal fields.

C. Stellar Mass vs. Sub-grid Physics

• Mass Correlation: Morphological definitions based on $|v/\sigma|$ and $\kappa_{rot}$ tend to select lower-mass galaxies, while $r-i$ and BTR select higher-mass galaxies. Since IA amplitude generally scales with mass, these definitions naturally produce different signal strengths.
• Re-weighting Experiment: The authors re-weighted the $|v/\sigma|$ and $\kappa_{rot}$ elliptical samples in TNG300 to match the stellar mass distribution of the $r-i$ sample.
  • Result: The re-weighted signals decreased in amplitude compared to the original unweighted signals.
  • Conclusion: The underlying stellar mass distribution is not the sole driver of the signal differences. The remaining discrepancy implies that sub-grid physics (e.g., feedback mechanisms, star formation models) plays a dominant role in determining IA amplitudes on non-linear scales ($r \lesssim 3$ Mpc/h).

5. Significance and Implications

• For Cosmology: The study cautions against the assumption that "blue" (disk) galaxies have negligible intrinsic alignments. The results show that blue galaxies often exhibit positive IA signals, challenging strategies that simply exclude blue galaxies to mitigate IA in weak lensing surveys.
• For Simulation Physics: The divergence in results between simulations (particularly Horizon-AGN vs. TNG/EAGLE) highlights the sensitivity of IA to sub-grid physics models. The negative signal in Horizon-AGN suggests that specific feedback or gas physics implementations in that simulation may produce tangential alignments in galaxy cores that are not present in others.
• Methodological Guidance: Future observational studies and simulations must be extremely careful regarding:
  • The definition of galaxy shapes (simple vs. reduced).
  • The morphological proxy used (kinematic vs. photometric).
  • The redshift evolution of these proxies.

In summary, the paper demonstrates that while IA signals are generally robust and positive in modern simulations, specific methodological choices (reduced shapes + kinematic selection at high redshift) can artificially induce negative signals. This underscores the critical need for consistent definitions and a deeper understanding of sub-grid physics to accurately model IA for precision cosmology.