🔭 astrophysics

Impact and interplay of $Λ$ CDM analysis choices for LSST cosmic shear

This paper forecasts that while LSST cosmic shear has the potential to constrain $S_8$ more than five times better than current surveys, this gain is severely compromised by uncertainties in baryon feedback and photometric redshift calibration, necessitating improved methodologies to realize the survey's full scientific potential.

Original authors: N. C. Robertson, C. Heymans, J. Zuntz, P. Burger, C. D. Leonard, I. G. McCarthy, J. G. Paine, J. Salcido, N. Šarčević, M. Schaller, J. Schaye, M. P. van Daalen

Published 2026-03-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: N. C. Robertson, C. Heymans, J. Zuntz, P. Burger, C. D. Leonard, I. G. McCarthy, J. G. Paine, J. Salcido, N. Šarčević, M. Schaller, J. Schaye, M. P. van Daalen

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 and normal matter. For decades, astronomers have tried to map this web by looking at how light from distant galaxies gets stretched and twisted as it passes through the web's gravity. This stretching is called "cosmic shear." It's like looking at a funhouse mirror: the distortion tells us about the shape and strength of the invisible object (the dark matter) causing the distortion.

The Rubin Observatory (LSST) is about to become the most powerful "funhouse mirror" ever built. It will take pictures of billions of galaxies over ten years. The goal? To figure out how the universe is expanding and what "Dark Energy" is doing.

However, this paper by Robertson and colleagues is a reality check. It says: "We have a great camera, but the picture might be blurry because of two main problems: 'Baryon Feedback' and 'Redshift Calibration.' If we don't fix these, our amazing camera won't give us the answers we need."

Here is a breakdown of the paper's findings using simple analogies:

1. The "Baryon Feedback" Problem: The Rowdy Party Guests

Imagine the universe is a quiet library (Dark Matter). But inside the library, there are rowdy party guests (normal matter, or "baryons," like stars and gas). These guests are throwing a party in the center of the library (inside galaxy clusters).

The Issue: The party guests are so loud and energetic (due to supermassive black holes blowing gas out) that they kick the furniture around. They push the quiet library patrons (dark matter) away. This changes the shape of the library.
The Paper's Finding: In the past, scientists thought these party guests only messed up the furniture in the very center of the room. But new simulations (called FLAMINGO) show they are actually kicking furniture around the entire library, even in the quiet corners.
The Consequence: If we try to map the library's shape without accounting for the party guests, our map will be wrong.
The Solution: To get an accurate map, we have to either:
- Ignore the noisy corners: Throw away the data from the most crowded parts of the library (cutting out small-scale data). This makes the map accurate but less detailed (like looking at a map from far away).
- Model the party: Try to mathematically guess how rowdy the guests are. But we don't know exactly how loud they are, so we have to guess wildly. This "guessing" adds so much uncertainty that it cancels out the benefit of having more data.

Result: When the team accounted for this uncertainty, the error in their most important measurement (called S8, which tells us how "clumpy" the universe is) doubled. It's like trying to weigh a feather while standing on a trampoline; the trampoline (baryon feedback) makes the measurement much harder.

2. The "Redshift Calibration" Problem: The Blurry Glasses

To map the 3D universe, astronomers need to know not just where a galaxy is on the sky, but how far away it is. They estimate distance by looking at the galaxy's color (Redshift).

The Issue: For the first year of the survey (Y1), they have a perfect reference library of real spectroscopic data (like a high-resolution dictionary) to check their color estimates. But for the final 10 years (Y10), the galaxies are so faint and far away that no one has taken a high-resolution "dictionary" of them yet.
The Analogy: Imagine you are trying to identify birds in a forest.
- Year 1: You have a guidebook with clear photos of every bird species. You can identify them easily.
- Year 10: You are looking at birds in a thick fog. You have to guess their species based on a blurry silhouette and a different, less reliable guidebook.
The Paper's Finding: The team found that if they use the current "blurry" methods for Year 10, the uncertainty in distance is so high that Year 10 provides almost no better results than Year 1. The extra data is useless because we don't know how far away the objects are.

3. The "Perfect Storm" Scenario

The authors ran a simulation of what happens if we combine these two problems:

We don't know exactly how the "party guests" (baryons) are messing up the dark matter.
We don't know exactly how far away the galaxies are in the deep survey.

The Result: The precision of the Year 10 survey barely improves on Year 1. It's like buying a Ferrari (the Rubin telescope) but driving it with the parking brake on (systematic errors).

4. The Good News: If We Fix It...

The paper ends with a hopeful note. It says: "If we can solve these two problems, the Rubin Observatory will be a game-changer."

If we develop better ways to understand the "party guests" (baryon feedback) using new observations.
If we develop better ways to measure distances for faint galaxies (improving redshift calibration).

Then, the Year 10 survey could be five times more powerful than any current survey. It would finally give us the precision needed to solve the biggest mysteries in physics, like the nature of Dark Energy.

Summary

This paper is a call to action. It tells the scientific community:

"We have built the most powerful telescope ever, but we are currently limited by our lack of knowledge about how galaxies interact with their environment and how to measure their distances. If we don't fix these 'housekeeping' issues, we will waste the potential of this incredible machine. But if we do fix them, we will unlock the secrets of the universe."

The Bottom Line: The telescope is ready, but the software (our understanding of the physics) needs an upgrade before we can take the final picture.

1. Problem Statement

The Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) and the Euclid satellite are poised to deliver unprecedented statistical power in weak gravitational lensing (cosmic shear) surveys. However, realizing their full potential requires accurate modeling of astrophysical and data-related systematics. The paper addresses two critical challenges:

Baryon Feedback: Non-gravitational processes (primarily Active Galactic Nuclei feedback) redistribute matter, suppressing the non-linear matter power spectrum on small scales ( $k > 0.1 h \text{Mpc}^{-1}$ ). Recent hydrodynamical simulations (FLAMINGO) suggest this suppression is significant on larger scales than previously forecasted, potentially biasing cosmological parameter recovery if not modeled correctly.
Photometric Redshift Calibration: While LSST Year 1 (Y1) depths can be calibrated using existing spectroscopic samples (e.g., C3R2), the deeper LSST Year 10 (Y10) depths will lack representative spectroscopic samples. This forces a reliance on cross-correlation techniques, which currently offer significantly lower precision (an order of magnitude worse than requirements), potentially degrading cosmological constraints.

The authors aim to define an analysis framework that can accurately recover unbiased $\Lambda$ CDM parameters (specifically $S_8$ ) in the presence of these systematics, contrasting with previous forecasts that often assumed idealized conditions or focused on $w_0w_a$ CDM dark energy constraints.

2. Methodology

The authors generate mock cosmic shear observables for LSST Y1 and Y10 using a 5-bin tomographic convergence power spectrum ( $C_\ell$ ) within the COSMOSIS pipeline.

Mock Data Generation:
- Baryon Feedback: Mocks are created using the FLAMINGO suite of hydrodynamical simulations. The fiducial mock uses the 'L1_m9' simulation, while 'weak' ('fgas+2 $\sigma$ ') and 'strong' ('Jet_fgas-4 $\sigma$ ') scenarios are used to test robustness.
- Systematics: The analysis includes nuisance parameters for shear calibration ( $\sigma_m$ ), redshift bias ( $\delta_z$ ), and intrinsic galaxy alignments (IA).
Analysis Framework:
- Models: The non-linear matter power spectrum is modeled using HMCODE2020 (with a free parameter $\Theta_{\text{AGN}}$ to modulate feedback) and an alternative emulator, SP(K) (trained on the ANTILLES suite).
- Scale Cuts: To mitigate high- $k$ modeling uncertainties, the authors employ redshift-dependent scale cuts defined by $K_{\text{max}}$ , where $\ell_{\text{max}}^{ij} = K_{\text{max}} \chi_{ij}$ .
- Priors: They test three priors for $\Theta_{\text{AGN}}$ $Θ_{AGN}$ :
  1. Uninformative: Wide top-hat range.
  2. FLAMINGO-informed: Constrained to the range of FLAMINGO simulations.
  3. DESxROSAT-informed: Tight Gaussian prior based on external cross-correlation data.
- Intrinsic Alignment: The fiducial model uses the redshift-dependent Non-linear Linear Alignment (NLA-z) model, with a test using the more complex TATT model.
Success Metric: The primary metric is the accurate recovery of the composite parameter $S_8 = \sigma_8 \sqrt{\Omega_m/0.3}$ . Success is defined as recovering the "truth" value within 0.5 $\sigma$ of the test analysis's credible interval, accounting for "projection bias" (offsets caused by priors and parameter degeneracies).

3. Key Contributions

Re-evaluation of Scale Cuts: The paper demonstrates that the original DESC-SRD proposal of analyzing scales up to $\ell_{\text{max}} = 3000$ is insufficient to mitigate baryon feedback bias in a $\Lambda$ CDM context.
Quantification of Redshift Calibration Impact: It provides a concrete forecast showing that the degradation in redshift calibration precision expected for Y10 (due to the lack of deep spectroscopy) will completely negate the statistical gains from increased survey depth.
Comparison of Feedback Models: It rigorously compares the standard HMCODE2020 approach with the more flexible SP(K) emulator, highlighting that even flexible models fail to recover unbiased constraints without tight external priors.
Distinction between $\Lambda$ CDM and Dark Energy Forecasts: The authors clarify why their results (showing large error degradation) differ from recent studies (e.g., Boruah et al. 2024) that found minimal impact; the difference lies in the target parameter ( $S_8$ vs. $w_p$ ) and the analysis scope ( $\Lambda$ CDM vs. $w_0w_a$ CDM with 3 $\times$ 2pt).

4. Key Results

Baryon Feedback and $S_8$ Precision:
- When accounting for current uncertainties in baryon feedback, the error on $S_8$ almost doubles compared to an analysis that neglects the effect or ignores scale cuts.
- A "scale-cut only" approach (removing small scales) reduces bias but triples the error on $S_8$ , rendering the results comparable to current precursor surveys (DES, KiDS).
- Using an uninformative prior on feedback parameters effectively acts as a severe scale cut, doubling the error.
- Using tight external priors (e.g., from DESxROSAT) restores precision but introduces significant bias ( $\sim 0.7\sigma$ ) if the theoretical model (HMCODE2020) does not perfectly match the underlying physics (FLAMINGO).
Y1 vs. Y10 Constraints:
- Despite LSST Y10 having double the source density and deeper magnitude limits, the final constraints on $S_8$ barely improve upon Y1.
- This "information plateau" is driven by the degradation in redshift calibration precision ( $\delta_z \sim 0.015$ for Y10 vs. $\sim 0.002$ for Y1) and the marginalization over baryon feedback.
- In an "idealized" scenario (perfect redshift calibration and no scale cuts), Y10 constraints would be roughly twice as tight as Y1, but this is not feasible with current methodology.
Model Robustness:
- The analysis is robust to variations in feedback strength (weak vs. strong) when using the FLAMINGO-informed prior and $K_{\text{max}} = 0.5$ ( $\ell_{\text{max}} \approx 700$ ).
- Switching to the SP(K) emulator does not significantly improve constraints without tight priors on its parameters; an "all-scale" SP(K) analysis still fails to meet the 0.5 $\sigma$ accuracy target due to model mismatch.
- The choice between NLA-z and TATT intrinsic alignment models has no significant impact on $S_8$ constraints in this specific mock setup (where the truth matches NLA-z).
Joint Constraints:
- While baryon feedback biases $S_8$ significantly in a pure $\Lambda$ CDM analysis, this bias has a relatively small impact on joint constraints for dark energy parameters ( $w_0, w_a$ ) when combined with CMB and other external probes. This explains the discrepancy with Boruah et al. (2024).

5. Significance and Conclusion

This paper serves as a critical "reality check" for the LSST Dark Energy Science Collaboration. It argues that the community must prioritize:

Developing methodology and observations to constrain baryon feedback models (e.g., via X-ray or SZ cross-correlations) to allow for the use of smaller scales without bias.
Enhancing photometric redshift calibration for deep surveys (Y10) to match the precision of shallower surveys, likely through improved cross-correlation techniques.

Without these advancements, the statistical power of LSST's final decade of data will be wasted, with $S_8$ constraints offering little improvement over current surveys. The authors conclude that if tight priors on baryon feedback and redshift calibration can be achieved, LSST cosmic shear could deliver constraints on $S_8$ more than five times as constraining as existing surveys. The study emphasizes that understanding systematics in the $\Lambda$ CDM subspace is a prerequisite for safely combining datasets to constrain evolving dark energy models.