Baryonic Feedback across Halo Mass: Impact on the Matter Power Spectrum

TL;DR

This work uses hybrid halo-replacement techniques within IllustrisTNG to quantify how baryonic feedback redistributes matter across halo masses and radii, and how this redistribution suppresses the nonlinear matter power spectrum up to $k\sim30\,h\mathrm{Mpc}^{-1}$. The key finding is that group-scale halos with $\log_{10}(M_{200\mathrm{m}}/h^{-1}M_\odot)\in[13,14]$ dominate the suppression (about $\sim10\%$ locally, contributing ~60% of the total from halos $M_{200\mathrm{m}}\gtrsim 10^{12}\,h^{-1}M_\odot$), while lower- and higher-mass halos contribute only a few percent each. Replacing matter within $\alpha R_{200\mathrm{m}}$ and enforcing mass conservation demonstrates that the total suppression is nearly additive across halo-mass bins, and that the relevant redistribution happens mostly within a few halo radii with additional mass moved to about $1.5\,R_{200\mathrm{m}}$. The results connect to observable weak-lensing signatures (group-galaxy lensing) and SZ measurements, motivating joint modeling via emulators that predict both the matter power spectrum and halo–matter correlations in the presence of baryonic physics for unbiased cosmological inference on nonlinear scales.

Abstract

Upcoming weak-lensing surveys will probe the matter distribution at a few percent level on nonlinear scales ($k>1\,{\rm h\,Mpc}^{-1}$) where baryonic feedback from galaxy formation modifies the clustering of matter. Using the IllustrisTNG hydrodynamical simulations, we quantify the mass and radial dependence of baryonic suppression of the matter power spectrum by selectively replacing halos in the collisionless run with their full-physics counterparts. We find that group-scale halos with $\log M_{\rm 200m}/h^{-1}M_\odot \in[13, 14]$ dominate the suppression, contributing a large fraction of the total reduction in power at $k\sim2-30\,h\,{\rm Mpc}^{-1}$. The suppression is smaller on either side of this mass bin. Correctly reproducing the full suppression of the power spectrum requires accounting for matter redistribution (while enforcing mass conservation) beyond the virial radius of each halo. We show that the same group-scale regime produces the most detectable deviations in group-galaxy lensing, making stacked group lensing a powerful observational test of feedback models together with SZ measurements. Our results motivate emulators that jointly predict the matter power spectrum and halo-matter correlations including baryonic effects, enabling unbiased cosmological inference from small scales.

Figures (5)

• Figure 1: The ratio of the full-physics and gravity-only power spectrum $P/P_{\rm{GO}}$ for each non-overlapping one-dex mass bin. The three panels show three different replacement radii, i.e. the radii out to which matter is redistributed due to baryonic feedback. The gray curve is the power spectrum ratio for the case where all halos above $10^{12}h^{-1}M_{\odot}$ are replaced at once. The black curve is the ratio for the power spectrum computed on the entire full-physics simulation.
• Figure 2: Left: The summation of the power spectrum ratio $P/P_{\rm{GO}}$ for two adjacent half-dex halo mass bins compared to the power spectrum computed from joint replacement of the two bins. Right: The same, but for two non-adjacent half-dex bins. The dotted blue curve is the sum of the red and green curves, and the solid orange curve is from the power spectrum computed with both half-dex bins replaced at once. The two curves are nearly identical in both cases, indicating that there are no significant cross-correlations between mass bins
• Figure 3: Left: The mean weak lensing signal profile of baryons for each mass bin, i.e. the sum of the gas, star, and black hole lensing signals. Center: The total lensing profiles for each halo mass bin from the full-physics case with estimated shape noise and sampling error mimicking a DES-like survey, plotted alongside the lensing profiles from the gravity-only simulation. Note that these first two plots are multiplied by $R/R_{\rm{200m}}$ for visual clarity. Right: The residual profile, i.e. the difference between the gravity-only and full-physics weak lensing signals divided by the gravity-only signal for each mass bin.
• Figure 4: The histogram of mass ratios between corresponding halos in the full-physics and gravity-only simulations. Full-physics halos are seen to generally have a slightly lower $M_{\rm{200m}}$ relative to their gravity-only simulation counterparts because baryonic feedback processes eject gas ejected out beyond $R_{\rm{200m}}$.
• Figure 5: $P/P_{\rm{GO}}$ for each non-overlapping one-dex mass bin for three different mass conservation methods. Solid curves correspond to halo-by-halo redistribution of the mass deficit to surrounding gravity-only simulation particles. Dashed curves correspond to halo-by-halo boosting of the replaced full-physics particles. Dotted curves correspond to no mass conservation at all.