The suppression of the matter power spectrum: strong feedback from X-ray gas mass fractions, kSZ effect profiles, and galaxy-galaxy lensing

TL;DR

This work constrains baryonic feedback's impact on the non-linear matter power spectrum by jointly analyzing kSZ profiles from SDSS/DESI+ACT and X-ray gas mass fractions from eROSITA and HSC-XXL, anchored by galaxy-galaxy lensing masses. Using the baryonification framework and BCemu, the authors map gas distributions to power suppression, finding a robust $P(k)/P_{ m DM\,Only}$ suppression of $10 \pm 2\%$ at $k=1~h~\mathrm{Mpc}^{-1}$, with HSC-XXL data preferring weaker suppression ($5 \pm 4\%$). The joint fit favors stronger feedback than in fiducial hydrodynamical simulations, such as FLAMINGO and BAHAMAS, and reveals tensions between datasets that motivate further cross-calibration of X-ray and kSZ measurements. The results provide data-driven priors for weak lensing cosmology, enabling future surveys (e.g., LSST) to recover small-scale power without compromising cosmological constraints. Overall, the study demonstrates that multi-wavelength gas observables offer a powerful path to calibrate baryon feedback and unlock small-scale cosmic shear information.

Abstract

Baryon feedback redistributes gas relative to the underlying dark matter distribution and suppresses the matter power spectrum on small scales, but the amplitude and scale dependence of this effect are uncertain. We constrain the impact of baryon feedback on the matter power spectrum by jointly analysing X-ray gas mass fractions from the eROSITA and HSC-XXL samples and SDSS/DESI+ACT kinetic Sunyaev-Zel'dovich (kSZ) effect profiles; the samples are characterised with galaxy-galaxy lensing and together span group and cluster masses at $0<z<1$. Using the baryonification framework, our joint eROSITA and kSZ model gives precise constraints on the suppression of the matter power spectrum: $10 \pm 2\%$ at $k=1~h~\mathrm{Mpc}^{-1}$. The inferred gas profiles are more extended and the power suppression is stronger than predicted by the fiducial models of recent hydrodynamical simulation suites, including FLAMINGO and BAHAMAS. The HSC-XXL gas mass fractions, which the fiducial simulations were calibrated to reproduce, prefer more moderate power suppression than the kSZ and eROSITA data: $5 \pm 4\%$ at $k=1~h~\mathrm{Mpc}^{-1}$. With a simulated LSST Year 1 weak lensing analysis, we demonstrate a framework for next-generation surveys: calibrating feedback models with multi-wavelength gas observables to recover the small-scale statistical power of cosmic shear.

Figures (12)

• Figure 1: The impact of the baryonification model parameters on the suppression of the matter power spectrum relative to a dark matter universe, $P(k)/P_{\rm DM~only}(k)$ (top), the stacked kSZ effect profile as a function of angular radius for a $\log_{10}[M_{500}/\mathrm{M}_\odot~h^{-1}]=13.1$ halo at $z=0.3$ (middle), and the hot gas mass fraction within $R_{500}$ as a function of halo mass (bottom). Each panel varies one baryonification parameter, within the prior bounds (Table \ref{['tab:priors']}). For the leftmost column, $M_{\rm c}$ is reported in units of $\mathrm{M}_\odot~h^{-1}$.
• Figure 2: Injection recovery test: the baryonification model successfully predicts the suppression of the matter power spectrum, $P(k) / P_\mathrm{DM~Only}(k)$, from mock kSZ and gas fraction observations, and vice versa. The fits to mock kSZ and gas fraction observations are shown in blue, and the fits to $P(k) / P_\mathrm{DM~Only}(k)$ are shown in pink; the shaded bands represent the $16$ and $84$th percentiles. The results for the fiducial (L1_m9) FLAMINGO simulation (solid black line) are shown in the top row, and for the strongest feedback ($f_\mathrm{gas}~-8\sigma$) FLAMINGO simulation (dashed black line) in the bottom row. For each fit, we show $P(k) / P_\mathrm{DM~Only}(k)$ (left), the normalized hot gas radial density profile for a group mass halo (centre), the kSZ signal for the same representative halo mass at $z=0.3$ and $0.75$ (upper right), and the hot gas mass fraction as a function of halo mass at $z=0$ (lower right). The mock kSZ and gas fraction observations are shown as black circles, with uncertainties of $15\%$ and $5\%$ respectively; the uncertainties are smaller than the markers.
• Figure 3: The joint fit to the kSZ and eROSITA X-ray measurements (blue), alongside fits to the kSZ (yellow), eROSITA (green), and HSC--XXL (purple) data separately. The top row compares the X-ray gas mass fraction measurements with the X-ray-only fits, and the lower two rows compare the kSZ effect profiles with the kSZ-only fit; the joint fit is presented in all panels. The shaded bands represent the $16$ and $84$th percentiles of the model fits. For the kSZ measurements, we restrict our fits to the innermost data points ($<3'$), as indicated by the vertical shaded bars; at these scales the data is most constraining and least impacted by uncertain modelling (Appendix \ref{['app:two_halo']}). For each kSZ observation, we demarcate $R_{500}$ as a vertical dotted line. The lower right panel presents the mean halo mass and redshift of each observable. The dotted lines show the mean growth histories of halos, binned by their redshift zero mass.
• Figure 4: Constraints from SDSS/DESI$+$ACT kSZ effect profiles and X-ray gas fractions on the radial density profile of a group mass halo (left), the hot gas mass fraction as a function of halo mass (centre), and the suppression of the total matter power spectrum (right). Top: the results of the joint kSZ and eROSITA X-ray fit (blue), alongside fits to the kSZ (gold), eROSITA (green), and HSC--XXL (purple) data separately. We note that the kSZ-only fit is prior dominated at cluster masses, while the X-ray fits are prior dominated at group masses. The shaded bands correspond to the $16$ and $84$th percentiles. In the leftmost panel, the range of comoving radii probed by each observation is indicated by the horizontal shaded bars. In the centre panel, the observations' mean halo masses are shown as vertical lines (dashed for kSZ, dot-dashed for eROSITA, and dotted for HSC--XXL). Bottom: the joint kSZ and eROSITA fit alongside a selection of recent hydrodynamical simulations: BAHAMAS McCarthy2017, SIMBA Dave2019, FLAMINGO Schaye2023, FABLE/XFABLE Henden2018Bigwood2025XFABLE.
• Figure 5: The constraints on the suppression of the matter power spectrum $P(k) / P_\mathrm{DM~Only}(k)$ from our joint fit to kSZ effect profiles and eROSITA X-ray gas mass fractions (blue), compared to constraints from the literature. Left: constraints derived (in part) from kSZ effect measurements Schneider2022bigwood2024Kovac25. Right: constraints from pre-eROSITA X-ray and/or tSZ measurements Grandis2024LaPosta2025Dalal2025Pandey2025.