ΛCDM is still not broken: empirical constraints on the star formation efficiency at z ~ 12-30

TL;DR

This work tests whether JWST-detected ultra-high-redshift galaxies challenge ΛCDM by pairing precise halo growth predictions from the GUREFT N-body suite with a simple empirical link from halo accretion to star formation and UV light. The model maps $\text{SFR} = \epsilon_*(M_h,z)\, f_b \dot{M}_h$ with a flexible, mass- and redshift-dependent efficiency $\epsilon_*(M_h,z) = \dfrac{2\epsilon_0}{(M_h/M_0)^{-\alpha} + (M_h/M_0)^{\beta}}$ and a fixed UV conversion $L_{UV} = \text{SFR}/K_{UV}$, enabling direct comparisons to observed rest-UV LFs across $z \sim 12$–$25$. Under conservative assumptions for $L_{UV}/m_*$ and no dust, the required peak baryon conversion efficiencies are $\sim 0.2$–$0.65$, indicating no fundamental tension with ΛCDM; bursty star formation or a top-heavy IMF would further reduce this requirement. The results are broadly consistent with other studies that favor higher early SF efficiencies or stochastic star formation, and they underscore the importance of improved modeling of star formation physics and spectroscopic confirmation of high-z candidates. Overall, the paper supports ΛCDM as a robust framework for the first galaxies while highlighting key astrophysical uncertainties in interpreting JWST observations.

Abstract

The James Webb Space Telescope continues to push back the redshift frontier to ever earlier cosmic epochs, with recent announcements of galaxy candidates at redshifts of $15 \lesssim z \lesssim 30$. We leverage the recent GUREFT suite of dissipationless $N$-body simulations, which were designed for interpreting observations in the high redshift Universe, and provide predictions of dark matter halo mass functions and halo growth rates for a state-of-the-art cosmology over a wide range of halo masses from $6 < z< 30$. We combine these results with an empirical framework that maps halo growth rate to galaxy star formation rate and then to rest-frame UV luminosity. We find that even if all of the photometrically selected $15 \lesssim z \lesssim 30$ galaxy candidates are real and actually at these extreme redshifts, there is no fundamental tension with $Λ$CDM, nor are exotic explanations required. With stellar light-to-mass ratios similar to those in well-studied lower redshift galaxies, our simple model can account for the observed extreme ultra-high redshift populations with star formation efficiencies that peak at values of 20-65 percent. Bursty star formation, or higher light-to-mass ratios such as are expected for lower metallicity stellar populations or a top-heavy Initial Mass Function, would result in even lower required star formation efficiencies, comparable to values predicted by high resolution numerical simulations of high-surface density star forming clouds.

Figures (9)

• Figure 1: Top panel: HMFs from $z=6$ to 30 (color coded as shown on the figure key) constructed with halos from the gureft simulations and the VSMDPL simulation. Circular symbols show the binned data from the simulations and the results of the fitting function are shown with solid lines. In addition, we show HMFs between $0 \lesssim z \lesssim 5$ from the Bolshoi-Planck simulation (gray dotted lines). Results are only shown for well-resolved halos that contain at least 100 particles. Bottom panel: The fractional differences between the $N$-body HMFs and fitting functions $(\phi_\text{fitted} - \phi_{n-\text{body}})/\phi_{n-\text{body}}$.
• Figure 2: Halo mass accretion rate normalized to halo mass, $(dM_\text{h}/dt)/M_\text{h}$, averaged over a dynamical time, as a function of $M_\text{h}$ across a wide range of redshifts. The data points show results extracted from $N$-body simulations, including VSMDPL (circles), gureft-35 (triangles), and gureft-15 (stars). The solid lines show our fitting functions, which are obtained by fitting to data from $z = 2$ to 18. The results of extrapolating the fitting functions to $z = 30$ are also shown. Both the data points and lines are colour-coded by redshift as shown in the figure key.
• Figure 3: The left panel shows predicted UV LFs between $z=6$ to 30 from our maximal $\epsilon_* = 1$ scenario. The right panel shows the rest-frame FUV magnitudes and SFR as a function of $M_\text{halo}$ between $z=6$ to 30 (also in the $\epsilon_* = 1$ scenario), where the luminosity ranges (and their corresponding halo mass ranges) that are constrained by current observations are highlighted (see Section \ref{['sec:methods:obs']} for details on the observational datasets).
• Figure 4: Simulated ultra-high-redshift rest-frame UV luminosity functions at $z=12$, 14, 17, and 25 assuming maximal $\epsilon_* = 1$ (blue solid line) and the posterior median and 16th and 84th percentile for a variable $\epsilon_*(M_\text{h},z)$ fitted to existing observations (purple line and shaded regions). These results are compared to ultra-high-redshift observational measurements reported by Donnan2024, Adams2024, Perez-Gonzalez2023, Whitler2025, Finkelstein2024, Robertson2024, Casey2024, Castellano2025, Perez-Gonzalez2025, Franco2025, and Weibel2025. The fact that the blue lines are everywhere higher than the observations implies that there is no fundamental tension between $\Lambda$CDM and these observations.
• Figure 5: Median (solid lines) and 16th and 84th percentiles (shaded regions) for the values of the baryon conversion efficiency $\epsilon_{*}$ obtained from our MCMC fitting procedure. Darker lines and shaded regions show the approximate range of halo mass where there are current observational constraints, assuming the median relation for $\epsilon_{*}$. The maximum required values of $\epsilon_{*}$ where there are constraints range from $\sim 20$--60 %, which is not in fundamental tension with $\Lambda$CDM.