Think inside the box: cosmic variance and large-scale conformity of high-redshift massive galaxies in the FLAMINGO simulations

TL;DR

The paper demonstrates that cosmic variance substantially elevates the variance in high-redshift mass functions beyond Poisson expectations and that a pronounced large-scale conformity in galaxy formation efficiency extends across tens to ~100 cMpc within JWST-like survey footprints. By leveraging the FLAMINGO $(1\,\mathrm{cGpc})^3$ hydrodynamical simulation, the authors quantify how the most massive systems, especially $M_{\ast,\max}$, trace environmental coherence and influence the inferred SMHMR and stellar baryon content. They show that conformity is footprint-bound, stronger at high redshift, and weakens when including galaxies outside the primary coherence volume, implying observational biases in small JWST fields. The results stress the need to model both CV and footprint effects to robustly interpret early massive galaxy populations and to use large-volume simulations to calibrate expectations for JWST surveys.

Abstract

We use the highest-resolution FLAMINGO hydrodynamical simulation to quantify cosmic variance and large-scale coherence in the evolution of massive galaxies at high redshift. FLAMINGO combines a $(1\,\mathrm{cGpc})^3$ volume with baryonic resolution sufficient to identify ${\gtrsim}\,10^3$ independent JWST-like survey volumes of $(100\,\mathrm{cMpc})^3$, providing unprecedented statistics to characterize the extremes of cosmic variance. At $z\,{\simeq}\,6$, the total variance in the number of haloes with $M_{200}\,{\simeq}\,10^{11.5}\,\mathrm{M_\odot}$ (or $M_\ast\,{\simeq}\,10^{10}\,\mathrm{M_\odot}$) is 2--3 times the Poisson expectation, while this ratio decreases with redshift. Similarly, at $z\,{\gtrsim}\,4$, the variance in the most massive halo per JWST-like field is twice the Poisson prediction. We find a pronounced large-scale \emph{conformity}: in volumes ranked by the stellar mass of their most massive galaxy ($M_{\ast,\mathrm{max}}$), the stellar-to-halo mass relation and star-formation efficiency are coherently elevated or suppressed throughout the full $(100\,\mathrm{cMpc})^3$ volume. When accounting for galaxies outside the volume, this signal persists only to radii $\lesssim 50\,\mathrm{cMpc}$, demonstrating that the detectable conformity is enhanced by the survey footprint. Moreover, $M_{\ast,\mathrm{max}}$ is a better predictor of the volume-wide efficiency of massive galaxies than the total number counts, which mainly trace clustering. Finally, the stellar fraction of the most massive galaxies peaks at $f_\ast\,{=}\,M_\ast\,/\,(M_{200}f_{\rm b,cosmic})\,{\simeq}\,0.2$ at $z\,{\simeq}\,5$, with a narrower dispersion in $f_\ast$ at fixed redshift and stronger redshift evolution than commonly assumed. These results show that both cosmic variance and footprint-confined conformity must be modelled when interpreting early massive galaxy populations in JWST fields.

Figures (13)

• Figure 1: Cumulative halo (left) and stellar (right) mass functions from the FLAMINGO simulation (L1_m8) over $0\,{<}\,z\,{<}\,10$ (dashed line), together with the 95th-percentile scatter (shaded bands) measured across 1,000 independent $(100\,\mathrm{cMpc})^3$ sub-volumes. For comparison, the error bars show the corresponding variance from a Poissonised reference sample, in which galaxy positions were randomised to remove the intrinsic clustering. The lower panels plot the ratio of the dispersions between the fiducial and Poisson cases (only for the bins with more than three objects per subbox). The results highlight that cosmic variance contributes a substantial excess over Poisson noise, with the relative impact increasing towards lower masses and lower redshifts. Also, the dotted lines in the lower right panel show the ratio for halo mass, matched to stellar mass via the mean SMHMR. The cosmic variance is greater for the stellar mass function than for the halo mass function at $M_\ast\,{\lesssim},10^{11}\,{\rm M}_\odot$ and $z\,{\lesssim}\,2$, an effect we attribute to the inclusion of satellites (see text).
• Figure 2: Stellar mass functions, as in Fig. \ref{['fig_MFs']}, but measured in sub-volumes of $3\times10^5\,\mathrm{cMpc}^3$ (left panel) and $10^7\,\mathrm{cMpc}^3$ (right panel).
• Figure 3: Cosmic variance of the most massive halo ($M_{\rm h,\max}$; left) and galaxy ($M_{\ast,\max}$; right) in $(100\,\mathrm{cMpc})^3$ sub-volumes of the FLAMINGO simulation (L1_m8) over $0\,{<}\,z\,{<}\,10$. Circles mark the medians, and vertical error bars show the 68th, 95th, and 99.7th percentiles across 1,000 sub-volumes. Grey squares with error bars indicate the same quantities after adding a Gaussian random scatter of 0.3 dex to mimic observational mass uncertainties (shifted by $\Delta z\,{=}\,0.05$ for clarity). Predictions from Extreme Value Statistics (EVS; Lovell2023) are shown for comparison, with shaded bands denoting the 68th and 99.7th percentile ranges. EVS underestimates the variance in $M_{\rm h,\max}$ because it accounts only for Poisson noise and neglects cosmic variance. This is directly confirmed by the results from the reference Poissonian samples (black crosses with error bars representing the 68th percentile range), which show excellent agreement with the EVS result. The broader dispersion in $M_{\ast,\max}$ for EVS arises from the wider stellar-fraction distribution assumed by Lovell2023 relative to that predicted by FLAMINGO (see Fig. \ref{['fig_fs']}).
• Figure 4: Cumulative distributions of halo mass ($M_{200}$; top row) and stellar mass ($M_{\ast}$; bottom row) for the most massive (solid lines), 5th most massive (dashed lines), and 10th most massive (dot–dashed lines) objects in sub-volumes of $3\times10^5\,\mathrm{cMpc}^3$ (red), $10^6\,\mathrm{cMpc}^3$ (green), and $10^7\,\mathrm{cMpc}^3$ (blue) from the FLAMINGO simulation (L1_m8) at $5\,{<}\,z\,{<}\,10$. When comparing to observations, random scatter should be convolved to mimic mass-estimate uncertainties. For reference, we include the most massive high-$z$ galaxies observed in COSMOS-Web from Shuntov2025. The survey volume roughly corresponds to $4\times10^6\,\mathrm{cMpc}^3$ for $z\,{=}\,5$--9. We only present their samples with JWST MIRI photometry, which results in more robust stellar mass estimates. They also corrected for the Eddington bias in their stellar mass estimates.
• Figure 5: Distribution of the stellar fraction, $f_\ast\,{=}\,M_\ast\,/\,(M_{200}f_{\rm b,cosmic})$, for the most massive halo (left) and galaxy (right) in each of 1,000 $(100\,\mathrm{cMpc})^3$ sub-volumes of the FLAMINGO simulation (L1_m8). Dotted curves show log-normal fits to the simulated distributions (the best-fitting parameters are summarized in Table \ref{['tab_fs']}). For comparison, the grey dashed line indicates a commonly adopted log-normal model with $\mu\,{=}\,0.135$ and $\sigma\,{=}\,1$Lovell2023. Relative to this assumption, the FLAMINGO predictions yield narrower distributions with higher mean stellar fractions.