Inferring the shapes of low-mass galaxies at z>6 from FIREboxHR. Where are the disks?
Abstract
Context. The three-dimensional shapes of galaxies provide clues to their formation and evolution. Inferring intrinsic shapes from projected sizes and axis ratios requires assumptions about the population distribution of shape parameters. Aims. We test commonly adopted assumptions in Bayesian shape inference for high-redshift galaxies. Methods. We analyze FIREboxHR galaxies with log(M⋆/M⊙)=6.5−9.5 at z∼6−15. We measure intrinsic shapes from stellar particles using the minimum covariance determinant (MCD) estimator. We also create mock JWST/NIRCam images, measure half-light radii and axis ratios, and infer intrinsic distributions from the joint distribution of projected axis ratio q and size log a. Results. Low-mass galaxies are predominantly prolate, with a substantial spheroidal component. Intrinsic ellipticity E and triaxiality T are well described by Johnson SB distributions, while intrinsic size A is approximately log-normal. We therefore model (logit(E),logit(T),log A) with a multivariate normal distribution and allow general correlations. This model recovers the particle-measured shape distributions more faithfully than a multivariate normal model in (E,T,log A). Triaxiality-related parameters remain difficult to constrain from projected data without large samples or informative priors. Conclusions. Bayesian forward modeling can recover intrinsic shape distributions when its population model adequately describes the underlying galaxies. An unsuitable distribution can bias inferences about morphological evolution at high redshift.
Related
- Expanding the scope of dark siren cosmology: Inferring the population properties of gravitational wave-hosting galaxies
- Inferring magnetic field strengths in high-redshift lensing galaxies from Faraday rotation measure observations
- Ubiquitous nuclear disks and bars embedded in massive galaxies in the cosmic morning
- Internal structure, not environment, shapes metallicity gradients in low-mass galaxies from MaNGA
- Where Galaxies Go to Die: The Environments of Massive Quiescent Galaxies at 3 < z < 5
BibTeX
@misc{geng2026inferring,
title = {{Inferring the shapes of low-mass galaxies at $z>6$ from FIREbox$^HR$. Where are the disks?}},
author = {Haojie Geng and Robert Feldmann and Jorge Moreno and Xuejian Shen and James Bullock},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.26969},
eprint = {2610.26969},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.26969},
}
RIS
TY - UNPB
TI - Inferring the shapes of low-mass galaxies at z>6 from FIREbox^HR. Where are the disks?
AU - Haojie Geng
AU - Robert Feldmann
AU - Jorge Moreno
AU - Xuejian Shen
AU - James Bullock
PY - 2026
DA - 2026/10/18
PB - arXiv
DO - 10.48550/arxiv.2610.26969
UR - https://arxiv.org/abs/2610.26969
AB - Context. The three-dimensional shapes of galaxies provide clues to their formation and evolution. Inferring intrinsic shapes from projected sizes and axis ratios requires assumptions about the population distribution of shape parameters. Aims. We test commonly adopted assumptions in Bayesian shape inference for high-redshift galaxies. Methods. We analyze FIREboxHR galaxies with log(M_⋆/M_⊙)=6.5−9.5 at z∼6−15. We measure intrinsic shapes from stellar particles using the minimum covariance determinant (MCD) estimator. We also create mock JWST/NIRCam images, measure half-light radii and axis ratios, and infer intrinsic distributions from the joint distribution of projected axis ratio q and size log a. Results. Low-mass galaxies are predominantly prolate, with a substantial spheroidal component. Intrinsic ellipticity E and triaxiality T are well described by Johnson SB distributions, while intrinsic size A is approximately log-normal. We therefore model (logit(E),logit(T),log A) with a multivariate normal distribution and allow general correlations. This model recovers the particle-measured shape distributions more faithfully than a multivariate normal model in (E,T,log A). Triaxiality-related parameters remain difficult to constrain from projected data without large samples or informative priors. Conclusions. Bayesian forward modeling can recover intrinsic shape distributions when its population model adequately describes the underlying galaxies. An unsuitable distribution can bias inferences about morphological evolution at high redshift.
ER -