Balancing bias, baryons, and scale cuts in LSST 3x2pt analysis

arXiv:2606.10679 · astro-ph.CO · Submitted 2026-06-09 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: I'm Vera, and with me are Jocelyn and Subrahmanyan, guest researcher.

Jocelyn: Today's paper: "Balancing bias, baryons, and scale cuts in LSST 3x2pt analysis".

Vera: Stage IV surveys such as LSST will probe deeply into the nonlinear regime,

Jocelyn: First, who's behind it and why it matters.

Paper summary: Vera: Welcome back to the show, everyone! We've got some really exciting work coming in today that tackles some tricky aspects of how we interpret galaxy clustering data from future surveys like LSST.

Jocelyn: That’s right, Vera. We’re talking about a paper titled "Balancing bias, baryons, and scale cuts in LSST three times 2pt analysis <ref:2606.10679#pg0,Balancing bias, baryons, and scale cuts in LSST>." It sounds like they are looking at the real challenges of modeling those small-scale effects that future surveys will uncover.

Subrahmanyan: I've seen the abstract; it points directly to the nonlinear regime where galaxy bias and baryonic feedback start dominating, and how poorly constrained nuisance parameters can cause degeneracies in our cosmological measurements. That’s where a lot of the real physics happens.

Vera: Exactly, Subrahmanyan. The core thesis of this paper is that relying on simple linear models just isn't enough when we push to smaller scales with LSST data; we need a more sophisticated way to handle these systematic effects so our cosmological parameters stay accurate.

Jocelyn: So, what does the authors claim they've done in terms of tackling those systematics, Vera? Are they suggesting a specific approach for modeling galaxy bias and baryonic feedback?

Subrahmanyan: They introduce a hybrid approach by combining the Hybrid Effective Field Theory or HEFT for nonlinear galaxy bias with the BACCO emulator to model baryonic feedback. This combination is what allows them to explore scales down to k = zero point seven h/Mpc, which was previously hard because of the limitations of simpler linear bias models up to k = zero point one h/Mpc (<ref:2606.10679#pg1>).

Vera: That scale extension is significant, Jocelyn; moving from a strict cut at kmax = zero point one h/Mpc to something closer to k = zero point seven h/Mpc opens up a much wider window for what we can learn from LSST data.

Jocelyn: And they use the BACCO emulator, which incorporates seven additional baryonic parameters like the extent of ejected gas and different density profiles, to model how baryonic feedback suppresses the power spectrum on those smaller scales.

Paper summary: Subrahmanyan: The paper shows that by including this complexity, we can quantify the impact of these parameters through a suppression factor called Smm(k, z) = Pmm,DMB(k, z) / Pmm,DMO(k, z), which they found to be accurate at the percent level for scales up to k ≤ zero point seven h/Mpc (<ref:2606.10679#pg2>).

Vera: That quantification of the suppression factor is really helpful because it gives us a concrete way to measure how much that baryonic physics is messing with the dark matter only predictions, and it shows that we can do this while still getting percent-level accuracy on the power spectrum.

Jocelyn: So, when they compared linear bias models to the full HEFT model, what did they find about the resulting constraints on cosmological parameters like m and sigma eight ?

Subrahmanyan: They found that a linear bias model introduces a "bias in the results," causing it to exit the 95th percentile region for scale cuts around k = zero point four h/Mpc, which suggests that sticking to a conservative cut of kmax = zero point one h/Mpc might be wise (<ref:2606.10679#pg1>).

Vera: But then they show that when you include baryonic feedback in the analysis, extending the scale cut from kmax = zero point one h/Mpc all the way up to zero point seven h/Mpc actually yields an FoM gain of about a factor of two, which indicates that adopting a more complex bias model doesn't weaken our cosmological constraints; it seems to improve them (<ref:2606.10679#pg1>).

Jocelyn: That’s really interesting, because it suggests that the complexity we add to model the physics is actually beneficial when interpreting these future surveys. What about testing if some of those higher-order bias parameters can mimic the effect of baryonic suppression?

Subrahmanyan: They tested this by fixing certain higher-order terms, specifically setting b s squared and b grad squared to zero in a "minimal bias model," which they found is consistent with the LIMD Lagrangian bias assumption (<ref:2606.10679#pg1>). They even showed that for certain values of these parameters, the resulting power spectrum closely matches the reference one within uncertainties for multipoles corresponding to kmax ≤ zero point seven h/Mpc beyond which the HEFT approach isn't valid (<ref:2606.10679#pg2>).

Vera: It sounds like they've really shown how these different modeling choices interact, demonstrating that the choice of complexity matters for reaching those deeper scales where LSST will shine.

Paper summary: Jocelyn: Thinking about the bigger picture, this work on "Balancing bias, baryons, and scale cuts in LSST three times 2pt analysis" has implications for how we interpret data from next-generation surveys regarding galaxy evolution <ref:2606.10679#pg0,Balancing bias, baryons, and scale cuts in LSST>.

Subrahmanyan: From a theoretical perspective, it reinforces the idea that to accurately constrain cosmological parameters using galaxy clustering or cosmic shear data at high precision, we absolutely must account for these nonlinear and baryonic effects instead of just relying on simpler linear approximations (<ref:2606.10679#pg2>).

Vera: I think what this means practically is that when we look at the LSST data, we can trust those constraints more even at smaller scales if we use methods like HEFT and BACCO to properly account for the messy physics.

Jocelyn: So, to wrap up this section of our discussion on "Balancing bias, baryons, and scale cuts in LSST three times 2pt analysis," it seems the key message is that sophisticated modeling of both galaxy bias and baryonic feedback is necessary to fully utilize the precision offered by Stage IV surveys <ref:2606.10679#pg0,Balancing bias, baryons, and scale cuts in LSST>.

Subrahmanyan: Indeed, because without properly balancing those two sets of nuisance parameters—bias modeling and feedback—we risk significant degeneracies in our cosmological parameter inferences (<ref:2606.10679#pg1>).

Vera: That’s the essence of it; it’s about making sure the observational data translates into the most accurate picture of cosmology possible, especially when we look at those challenging small scales.

Jocelyn: So, as we move into our next segment, I want to talk about what this all means for detecting neutrino mass using LSST.

Subrahmanyan: That’s a great transition; connecting the modeling challenges here directly to the specific goals of measuring the total neutrino mass M nu is where the real impact lies.

Vera: We'll explore that next, but first, let's just take a moment to think about how much these new modeling techniques can actually help us when we look at those elusive neutrino mass signals.

Jocelyn: I’m ready for that discussion; it feels like the next logical step from tackling the bias and feedback issues they detailed in this paper.

Conclusion: Vera: So, to wrap up our discussion on "Balancing bias, baryons, and scale cuts in LSST three times 2pt analysis," this paper tackles how to use future survey data like LSST when you're dealing with those tricky nonlinear effects <ref:2606.10679#pg0,Balancing bias, baryons, and scale cuts in LSST>.

Jocelyn: It seems the authors are really focused on finding that sweet spot between modeling galaxy bias and accounting for baryonic feedback so the cosmological constraints stay reliable across different scales.

Subrahmanyan: Precisely, they’re moving past simple models to incorporate more complex physics, which is crucial for connecting what we see in the sky to our standard cosmological model predictions.

Vera: They are essentially showing how a careful combination of hybrid modeling and scale selection helps us get better results from these massive future datasets.

Jocelyn: I’m curious about the specific authors; who were they, and what kind of research background do you think they bring to this kind of modeling?

Subrahmanyan: The authors are clearly deep in the weeds of both theoretical modeling and data analysis, which is exactly what you need when dealing with these kinds of systematic uncertainties. Their work sits right at the intersection of simulation techniques and observational cosmology.

Vera: And I think their background is a perfect fit for this topic because they’re bridging the gap between what simulations predict and what we hope to measure from actual telescope data.

Jocelyn: So, looking at the title itself, "Balancing bias, baryons, and scale cuts," it really tells you that the authors are tackling multiple complex problems simultaneously in one analysis.

Subrahmanyan: That balancing act is exactly where the real scientific progress happens; you can't solve all those interconnected physics problems with just one simple equation.

Vera: I feel like this paper is important because it gives us a concrete roadmap for how to approach these high-precision studies using upcoming surveys.

Jocelyn: It certainly seems like a foundation for how we should interpret the results coming from LSST when we look at galaxy clustering and cosmic shear data.

Subrahmanyan: And this has big implications for our understanding of structure formation itself, showing us exactly where the physics needs to be refined in our simulations.

Vera: We’ll explore those deeper implications in a bit, but it’s clear this paper sets a really important standard for how we should proceed with these upcoming observational challenges.

Ottavia Truttero, Maria Tsedrika, Joe Zuntza, Alkistis Pourtsidoua, Nikolina Sarˇceviˇc´c´

Institute for Astronomy, University of Edinburgh · Higgs Centre for Theoretical Physics, School of Physics and Astronomy, University of Edinburgh · Department of Physics, Duke University · Institut de Física d’Altes Energies (IFAE), The Barcelona Institute of Science and Technology

astro-ph.CO

Submitted: 2026-06-09

Updated: 2026-10-05

Comments: 18 pages, 15 figures, Published in the Open Journal of Astrophysics

Code: https://github.com/LSSTDESC/forecasting

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 78/100

The gist: Stage IV surveys such as LSST will probe deeply into the nonlinear regime, where systematic effects from galaxy bias and baryonic feedback become dominant and poorly constrained nuisance parameters

Key concepts

Nonlinear Galaxy Bias
This describes how galaxies cluster differently than dark matter on small scales. Simple linear models work only at large scales (k < 0.1 h/Mpc), but more complex models like HEFT are needed to accurately describe this behavior down to smaller scales, which is crucial for probing the nonlinear regime of galaxy surveys.
Baryonic Feedback
This refers to the physical processes where gas within dark matter halos affects their structure. The BACCO emulator models these effects by simulating how gas is ejected and retained in halos, allowing researchers to quantify how these baryonic physics modify the dark matter power spectrum on small scales.
Hybrid Effective Field Theory (HEFT)
HEFT is a perturbative approach used to model nonlinear galaxy bias. It combines early-time gravitational evolution with later evolution solved via N-body simulations, enabling accurate modeling of galaxy clustering across a wider range of scales than simple linear approximations allow.
Scale Cut ($k_{max}$)
The scale cut defines the smallest spatial scales (represented by wavenumber k) that are included in the analysis. The paper investigates how changing this cut, from 0.1 h/Mpc to 0.7 h/Mpc, impacts cosmological constraints and the Figure of Merit (FoM).

Terminology

Summary

Stage IV surveys such as LSST will probe deeply into the nonlinear regime, where systematic effects from galaxy bias and baryonic feedback become dominant and poorly constrained nuisance parameters can lead to degeneracies. This work presents a 3 × 2pt analysis for LSST Y1 and Y10 data using the BACCO emulator for modeling both hybrid-effective field theory (HEFT) for nonlinear galaxy bias and the baryonification mechanism for baryonic feedback, aiming to find a balance between model complexity and scale cuts while addressing parameter degeneracies.

The Gist

A linear bias model delivers percent-level, unbiased constraints on omegam and σ8 only up to kmax = 0.1 h/Mpc, but pushing to smaller scales requires a perturbative approach, and the specific measured value of the total neutrino mass Mν is not robust across equally plausible mock scenarios: the inferred Mν can be significantly biased by adopting the minimal bias model.

Modeling Systematics

The analysis addresses two primary systematic effects that become dominant on small scales: nonlinear galaxy bias and baryonic feedback. The paper utilizes a hybrid approach to model these complexities, moving beyond simple linear approximations.

  1. For nonlinear bias, the authors compare several models:

  2. A simple linear bias model is valid only up to k ∼ 0.1 h/Mpc; pushing further requires a perturbative approach such as the Hybrid Effective Field Theory (HEFT) approach, which allows modeling down to scales of k = 0.7 h/Mpc.

  3. The HEFT approach combines the relation between galaxy overdensity and gravitational potential perturbatively at early times with subsequent evolution under gravity solved via numerical N-body simulations, significantly increasing the accessible range of scales to k ≤ 0.7 h/Mpc.

  4. The analysis also explores simplifying bias models by fixing certain higher-order terms; for instance, fixing bs2 and b∇2 to zero results in a minimal bias model, which is consistent with the local-in-matter-density (LIMD) Lagrangian bias assumption.

Baryonic Feedback Implementation

To account for baryonic feedback effects on scales smaller than kmax ≃ 0.2 h/Mpc, the authors employ the BACCO emulator, which includes baryonic effects in the nonlinear dark matter-only (DMO) power spectrum through a baryonification algorithm.

  1. BACCO is a neural network-based emulator that incorporates seven additional baryonic parameters: "the extent of the ejected gas (η), the density profiles of hot gas in haloes (θinn, Minn, θout), the fraction of gas retained in haloes of a given mass (Mc, β), and the characteristic halo mass scale for central galaxies (M1,z0,cen)."

  2. The net impact is quantified by the suppression factor: Smm(k, z) = Pmm,DMB(k, z) / Pmm,DMO(k, z).

  3. A specific approximation used is the Zennaro24 approach: Pgg(k) = Pgg,DMO(k), Pgm(k) = p Smm(k) Pgm,DMO(k), which has been found to be accurate at the percent level for scales k ≤ 0.7 h/Mpc.

Analysis Methodology and Results

The analysis employs a 3 × 2pt statistic combining galaxy clustering (G) and cosmic shear (γ). The methodology involves testing how baryonic and bias parameters affect the power spectra, quantified by the Figure of Merit (FoM) and Figure of Bias (FoB).

  1. When comparing linear bias models to the full HEFT model, a linear bias model introduces a bias in the results, exiting the 95th percentile region for scale cuts at around k = 0.4 h/Mpc, suggesting a conservative cut of kmax = 0.1 h/Mpc might be recommended.

  2. In scenarios with baryonic feedback included, extending the scale cut from kmax = 0.1 h/Mpc to 0.7 h/Mpc yields an FoM gain of approximately a factor of two, demonstrating that our cosmological constraints are not weakened by adopting a bias model more complex than linear – in fact, they benefit from it.

  3. The paper tests if higher-order bias parameters can mimic baryonic suppression: for a combination of suitably chosen values of bs2 and b∇2, the resulting power spectrum closely matches the reference one within uncertainties for multipoles corresponding to kmax ≤ 0.7 h/Mpc beyond which the HEFT approach is no longer valid.

Impact on Neutrino Mass Detection

The study investigates whether LSST can clearly detect a non-zero total neutrino mass (Mν).

Improvements for AI systems

Here are specific improvements to AI systems based on the insights derived from this scientific paper:

The following improvements focus on enhancing cosmological inference, specifically in Stage IV surveys like LSST, by improving how models handle nonlinear physics (galaxy bias and baryonic feedback).

  1. Enhance Cosmological Parameter Estimation Robustness:

  2. Improve Scale-Dependent Physics Modeling:

  3. Develop Advanced Nuisance Parameter Handling:

  4. Enable Robust Detection of Beyond-ΛCDM Physics (e.g., Neutrino Mass):

This improved AI system can perform the following specific tasks:

  1. Perform high-precision cosmological parameter estimation for Dark Energy surveys by incorporating a sophisticated, scale-dependent model for galaxy bias and baryonic feedback, significantly reducing systematic biases in parameters like matter density and the growth rate.

  2. Accurately model the interplay between baryonic physics (ejected gas, halo profiles) and galaxy bias to accurately predict the shape of cosmic power spectra (both galaxy-galaxy and galaxy-matter). This allows for disentangling whether observed spectral deviations are due to physical baryon effects or modeling errors in the bias parameters.

  3. Quantify the potential degeneracy between higher-order bias terms (like quadratic and third derivatives) and baryonic suppression effects, allowing researchers to test whether a specific observed signal is better explained by a complex model (HEFT) or a simplified one (minimal bias).

  4. Provide robust forecasts for future surveys like LSST Y1 and Y10, indicating the necessary scale cuts required to achieve unbiased constraints on cosmological parameters when using different modeling philosophies (e.g., linear vs. perturbative bias).

  5. Assess the sensitivity of detecting exotic physics, such as total neutrino mass, by quantifying how much uncertainty in galaxy bias modeling or baryonic feedback can mimic or mask a true signal at specific scales (e.g., distinguishing between the effect of massive neutrinos and higher-order bias terms).

Abstract

Stage IV surveys such as LSST will probe deeply into the nonlinear regime, where systematic effects from galaxy bias and baryonic feedback become dominant and poorly constrained nuisance parameters can lead to degeneracies. In this work we present a 3x2pt analysis for LSST Y1 and Y10 data using the BACCO emulator for modelling both the hybrid-effective field theory (HEFT) for nonlinear galaxy bias and the baryonic feedback using the baryonification mechanism. We aim to find a balance between model complexity and scale cuts, with particular attention to parameter degeneracies and baryonic feedback effects on the galaxy-matter and galaxy-galaxy power spectra. First, we find that a linear bias model delivers percent-level unbiased constraints on Ω m and σ 8 only up to k max=0.1 h/ Mpc, but pushing to smaller scales requires a perturbative approach. Second, we compare HEFT with a minimal bias variant with fixed higher-order terms, and find that the latter is unbiased in Λ CDM even at k max=0.7 h/ Mpc. We show that higher-order bias can mimic baryonic suppression, but baryons cannot reproduce the full range of higher-order bias behaviour within the parameter range allowed by the BACCO baryonification model. Third, we find that a detection of the total neutrino mass M ν is possible for both Y1 and Y10 for k at least 0.5 h/ Mpc, at least when photo- z uncertainties and related nuisance parameters are precisely known. However, the specific measured value is not robust across equally plausible mock scenarios: the inferred M ν can be significantly biased by adopting the minimal bias model. The entire analysis is conducted with a new independent, open source pipeline MGL that we present for the first time in this work.

Sources

Related papers