A smooth filament origin for distant prolate galaxies seen by JWST and HST

TL;DR

The study tackles why many early galaxies appear prolate at $z>3$ and whether this can constrain the nature of dark matter. It compares hydrodynamical simulations for Cold DM (CDM), Warm DM (WDM), and Wave/Fuzzy DM ($\\psi$DM) against JWST CEERS and HST CANDELS morphologies, using MVEE to quantify 3D and projected shapes and a KDE-based likelihood framework to compare with observations. The results show that WDM and $\\psi$DM naturally produce elongated, prolate galaxies via smooth, long filaments during the first ~0.5 Gyr, while CDM tends to yield spheroidal morphologies with early subhalo merging; log-likelihoods decisively prefer WDM/$\\psi$DM over CDM across datasets. This filament-origin scenario suggests a new constraint on the small-scale power spectrum of dark matter and motivates targeted searches for an early filament era, offering a path to test DM models with JWST data.

Abstract

The initial gravitational collapse of Dark Matter and gas forms a universal filamentary network where the first galaxies form, with shapes and sizes that depend on the choice of Dark Matter. Claims from deep space imaging surveys that elongated galaxies predominate at $z > 3$ are examined here by comparison with detailed hydrodynamical simulations of Cold Dark Matter (CDM), Warm Dark Matter (WDM), and Wave/Fuzzy Dark Matter, $ψ$DM. For CDM and WDM we have sufficient volume, $10^{3}\,\mathrm{Mpc/h}^{3}$, to generate galaxies with stellar masses $> 10^{9}\,M_{\odot}$ at $z > 2$, allowing comparison with the CEERS and CANDELS surveys. We find the observed tendency towards elongated, prolate-shaped young galaxies is well matched by WDM, from material accreted along smooth filaments during the first $\simeq 500\,\mathrm{Myr}$, with little dependence on stellar mass. This contrasts with CDM, where the stellar morphology is mainly spheroidal, formed from merging of fragmented filaments. For CDM, several subhalos are predicted to be visible, whereas for WDM and $ψ$DM, early merging is rare. Our findings show how the shapes and sizes of early galaxies are sensitive to the smoothness of the underlying filament network, providing a new constraint on the nature of dark matter.

Figures (11)

• Figure 1: Illustrative comparison of high-z galaxies detected by JWST and simulated galaxies. The first row shows representative $3^{\prime\prime}\times3^{\prime\prime}$, observed examples of the prolate class reproduced from RefPandya:2024, SE++ catalog. The next two rows show the predicted stellar appearance of simulated galaxies for the two different classes of dark matter from our hydro-simulations. For each column the simulated galaxy locates at the same position in the simulation volume for a fair comparison of all two DM models with matching initial conditions. In each model, we identify a simulated galaxy with similar stellar mass and project it to best match the observed data. Note also that we compare the same span of redshift. We have matched the observed orientation but maintained the size and stellar density range equally between the simulated galaxies.
• Figure 2: 2D & 3D stellar morphology of simulated galaxies compared with observations $3<z<8$. All time frames between $z = 3$ and $z = 6$ are shown, where the color scale indicates stellar mass $M_{*}$ (top) and redshift (bottom), with overlapping points offset slightly for clarity. Note that the redshifts ($z$) and stellar masses ($M_{*}$) of the simulated halos match those of the observations. The left panels present WDM results, while the right panels show galaxy shapes predicted for CDM. Top row: The black boundaries define 3D ellipsoids as oblate, spheroidal, or prolate, following RefsWel:2014Zhang:2019. Colour scales indicate the stellar number weighted age of the stars in each simulated galaxy, determined at each time frame of the simulations. The shaded areas represent the distribution of JWST based measurements from RefPandya:2024. Red data points represent independently calculated ellipticities of the dark matter halos by RefMocz:2020. Each point represents a galaxy, although in some cases the same galaxy is shown at multiple redshifts. In total, the sample includes as few as two distinct galaxies at $z = 6$ and up to 18 at $z = 3$. The solid and dashed brown contours represent the predicted distribution for CDM from TNG50 Pillepich:2019 and the green ones from VELA Zhang:2019. Middle row: Helpful representation of the classification of ellipsoid shapes, as depicted by RefPandya:2024. Bottom row: Projected semi-axis ratio, $b/a$, versus projected semi-major axis $a$, for comparison with the observations represented by shaded areas for z $>$3 Pandya:2024. This shows the larger spread towards smaller b/a with WDM, similar to the data when the simulated galaxies are young and lower mass, as indicated by the color bar. The number of data points here is several times larger than in the upper panel because we show three orthogonal projections for each galaxy.
• Figure 3: 2D & 3D Stellar morphology of simulated galaxies at $2<z<3$. The points represent simulated galaxies of WDM and CDM for comparison with contours representing the observed CEERS and CANDELS surveys, where the redshift ($z$) and stellar mass ($M_{*}$) range of the simulations match the observed ranges. Top row: The black boundaries classify 3D ellipsoids as oblate, spheroidal, or prolate, following the criteria of RefsWel:2014Zhang:2019. The red shaded area represents the distribution of measurements based on JWST data from RefsPandya:2024 for CEERS galaxies with $z = 2-3$. The solid and dashed brown contours represent the predicted distribution for CDM from TNG50 Pillepich:2019 and the green ones from VELA Zhang:2019. Each point in the figure represents a galaxy, although in some cases the same galaxy is shown at multiple redshifts. In total, the sample includes 18 distinct galaxies at $z = 3$ and up to 25 at $z = 2$. The solid and dashed black contours represent the CDM hydro-simulation predictions of TNG50 Pillepich:2019. Middle row: Projected ($b/a$) vs. semi-major axis ($a$) for comparison with the red shaded area for $z = 2-3$ of the CEERS survey Pandya:2024. The colored data points correspond to the three orthogonal projections of each simulated galaxy. The sample includes projections of 18 galaxies at $z = 3$ and up to 25 halos at $z = 2$. Bottom row: Similar to the middle row, where the red contours now represent the CANDELS survey VDW:2014 galaxies with $z = 2-3$.
• Figure 4: Projected semi-axis ratio, $b/a$, vs. projected semi-major axis shown for the stellar distribution of each galaxy.Left panel: results for the WDM simulation, including all three orthogonal projections for $2 < z < 6$ (as presented in Figure \ref{['Fig:compnew']}). We show galaxies enclosed within the “banana”-shaped region (yellow-shaded area and red contour) predicted for a prolate population of triaxial ellipsoids in projection, assuming a fixed 3D major axis of $\log A/{\rm kpc} = 0.5$ and a Gaussian spread of 0.03 dex, following Ref. Pandya:2024. The color of each point indicates the redshift, as in Figure \ref{['Fig:compnew']}. Right panel: same as in the left panel, but for galaxies in the CDM simulation. All other conventions are identical.
• Figure 5: Redshift dependence of the projected axis ratio $b/a$ compared with CEERS (black Pandya:2024 and orange Kartaltepe:2023 points with error bars) and CANDELS (purple VDW:2014 data points) across the full redshift range covered by our simulations. Left panel: results from the $10\,{\rm Mpc}/h$ and $5\,{\rm Mpc}/h$ simulations, corresponding to virial and stellar mass ranges of $M_{200}=10^{8}$--$10^{12}\,M_{\odot}$ and $M_{\star}=10^{7}$--$5\times10^{10}\,M_{\odot}$, respectively. Right panel: results from the higher-resolution $1.7\,{\rm Mpc}/h$ simulation by RefMocz:2020, covering halo and stellar mass ranges of $M_{200}=10^{7}$--$5\times10^{10}\,M_{\odot}$ and $M_{\star}=5\times10^{6}$--$5\times10^{8}\,M_{\odot}$, respectively. Error bars denote the $1\sigma$ dispersion around the median values. Note that the higher-redshift CEERS data (last two orange points, $z>5$) correspond to relatively small galaxies for which PSF smoothing may bias the observed $b/a$ toward rounder shapes, possibly leading to an overestimation of the mean values.