A-SLOTH: Tracing Early Stars & Local Observables

• A-SLOTH is a semi-analytical model that reconstructs early star formation and cosmic evolution by evolving baryons along dark matter merger trees.
• It distinguishes between metal-free Pop III and metal-enriched Pop II star formation, incorporating feedback mechanisms and chemical enrichment to model observables.
• Its modular design and multi-observable calibration allow diverse applications from Milky Way archaeology to high-redshift galaxy and reionization studies.

A-SLOTH, short for Ancient Stars and Local Observables by Tracing Halos, is a semi-analytical model of galaxy formation focused on the early Universe and on linking the first stars and galaxies to present-day and high-redshift observables. It is built on dark-matter merger trees generated either from Extended Press–Schechter theory or imported from numerical simulations, and it applies analytical baryonic prescriptions to model gas cooling, star formation, feedback, chemical enrichment, and stellar populations, including both metal-free Population III and metal-enriched Population II stars. Public releases describe it as the first public code that connects the formation of the first stars and galaxies to observables, while emphasizing moderate computational requirements, modularity, and applications ranging from Milky Way stellar archaeology to reionization, transients, compact remnants, and high-redshift galaxy populations (Hartwig et al., 2022, Magg et al., 2022).

1. Origins, scope, and scientific niche

A-SLOTH was developed to address the dynamic-range problem inherent in first-star studies: the relevant physics spans minihaloes of $10^5$–$10^6\,M_\odot$ at $z\gtrsim 20$, atomic-cooling haloes, Milky Way progenitors, and present-day dwarf satellites, across $\sim 13.8$ Gyr of cosmic evolution. In this regime, semi-analytic models are computationally advantageous because they can sit on top of dark-matter-only merger trees and replace unresolved baryonic processes with parametrized prescriptions, making systematic parameter studies feasible where full hydrodynamical calculations remain prohibitive (Magg et al., 2022).

The code’s stated purpose is not generic galaxy formation alone, but the reconstruction of “ancient stars and local observables” from hierarchical halo assembly. That emphasis gives A-SLOTH a distinctive position among semi-analytic models: it is optimized for Population III star formation, the Pop III-to-Pop II transition, the build-up of extremely metal-poor stellar populations, and the use of the Milky Way and its satellites as a fossil record of early-Universe physics. The software description in the Journal of Open Source Software notes that A-SLOTH can “build up a Milky-Way-like galaxy star by star in only a few minutes,” and that its modular design is intended to let users add new physics without constructing a new framework from scratch (Magg et al., 2022).

2. Computational framework and merger-tree architecture

At its core, A-SLOTH evolves baryons along halo merger trees. The original implementations used Extended Press–Schechter trees generated with the Parkinson et al. algorithm, while later applications also ingested trees from numerical simulations such as the Caterpillar Milky Way zooms and a high-resolution $8\,h^{-1}\,\mathrm{cMpc}$ cosmological box. Across these modes, the workflow is consistent: read or generate a merger tree, step through haloes in time order, update gas reservoirs, form stars, apply feedback, track metals, and assemble observables from the resulting stellar and halo histories (Magg et al., 2022, Lipatova et al., 12 Aug 2025, Liu et al., 6 Jun 2025).

Later baryon-cycle implementations make the state variables explicit by evolving cold gas, hot gas, cumulative stellar mass, and unbound outflows. A representative star-formation law used in current A-SLOTH applications is

$$\dot{M}_\star \simeq \eta\,\frac{M_{\rm cold}}{t_{\rm cold,ff}},$$

where $\eta$ is a star-formation efficiency per mean free-fall time, $M_{\rm cold}$ is the cold gas mass, and $t_{\rm cold,ff}$ is the cold-gas free-fall time. In this sense, the model is semi-analytic in the strict SAM sense: halo growth is inherited from the merger tree, while baryonic evolution is encoded in coupled, timestep-based reservoir equations (Liu et al., 6 Jun 2025).

A-SLOTH also samples stars individually in several of its major implementations. That feature is central to its treatment of stochastic feedback, because ionizing luminosities, stellar lifetimes, and supernova events can be attached to discrete stellar masses rather than only to IMF-averaged populations. In Milky Way satellite applications, this is extended to individually sampled Pop II stars; in first-star applications, the code likewise samples individual Pop III stars from an explicit IMF (Hartwig et al., 2022, Chen et al., 2022).

3. Physical prescriptions: cooling, stellar populations, and feedback

A-SLOTH distinguishes between molecular-cooling minihaloes and atomic-cooling haloes, and it explicitly separates metal-free Population III star formation from metal-enriched Population II star formation. In multiple applications, Pop III formation occurs in gas below a critical metallicity threshold of order $Z \lesssim 10^{-5}\,Z_\odot$, while Pop II star formation takes over once enrichment crosses that threshold. The Pop III IMF is generally represented as a power law,

$10^6\,M_\odot$0

with $10^6\,M_\odot$1, $10^6\,M_\odot$2, and $10^6\,M_\odot$3 treated as tunable parameters in the calibrated framework (Riaz et al., 2022, Hartwig et al., 2024).

The feedback channels are radiative, chemical, and mechanical. Radiative feedback includes ionizing photons and Lyman–Werner suppression of $10^6\,M_\odot$4 cooling; chemical feedback follows stellar yields and the resulting Pop III-to-Pop II transition; mechanical feedback is dominated by supernova-driven gas removal and halo outflows. In the high-redshift galaxy implementation, the outflow prescription is explicitly energy-driven, with mass loading proportional to supernova energy injection divided by halo binding energy. That scaling captures the expected strong suppression of star formation in shallow potential wells and is one of the key control knobs in reproducing both dwarf-galaxy properties and high-redshift metallicities (Hartwig et al., 2022, Liu et al., 6 Jun 2025).

Chemical evolution is a persistent theme across A-SLOTH papers. The code tracks metals through halo growth, accretion, enrichment, and outflows, and later versions incorporate a stochastic description of inhomogeneous mixing by assigning star-forming gas a metallicity offset relative to the halo-average metallicity. This allows the framework to model not only global metallicity growth but also the low-metallicity tails relevant for extremely metal-poor stars, second-generation enrichment signatures, and Population III relic diagnostics (Magg et al., 2022, Liu et al., 6 Jun 2025).

4. Calibration strategy and inferred parameter space

The public-release description states that A-SLOTH was originally calibrated to six observables, including the optical depth to Thomson scattering, the stellar mass of the Milky Way and its satellite galaxies, the number of extremely-metal poor stars, and the cosmic star-formation rate density at high redshift. Those observables were chosen to tie together local stellar archaeology and cosmological reionization-era constraints within a single forward model (Hartwig et al., 2022).

A later full statistical calibration expanded this to a likelihood based on nine independent observables and produced posterior constraints for 11 parameters describing early star formation and feedback. The nine observables span Milky Way-specific and cosmologically representative quantities, and the inferred best-fit model gives a Pop III IMF slope of $10^6\,M_\odot$5, with $10^6\,M_\odot$6 and $10^6\,M_\odot$7. The same calibration also reports posterior medians $10^6\,M_\odot$8, $10^6\,M_\odot$9, $z\gtrsim 20$0, and $z\gtrsim 20$1, while emphasizing that the IMF-generating parameters remain poorly constrained and that the IMF slope could range from log-flat to Salpeter within the allowed posterior volume (Hartwig et al., 2024).

Parameter	Posterior median	Source
$z\gtrsim 20$2	$z\gtrsim 20$3	(Hartwig et al., 2024)
$z\gtrsim 20$4	$z\gtrsim 20$5	(Hartwig et al., 2024)
$z\gtrsim 20$6	$z\gtrsim 20$7	(Hartwig et al., 2024)
$z\gtrsim 20$8	$z\gtrsim 20$9	(Hartwig et al., 2024)
$\sim 13.8$0	$\sim 13.8$1	(Hartwig et al., 2024)
$\sim 13.8$2	$\sim 13.8$3	(Hartwig et al., 2024)
$\sim 13.8$4	$\sim 13.8$5	(Hartwig et al., 2024)

This calibration architecture is one of the reasons A-SLOTH is used as a forward model rather than only as an exploratory toy model. A plausible implication is that the framework’s value lies as much in its parameter posteriors and degeneracy structure as in any single best-fit realization, because multiple later application papers reuse these calibrated parameter ranges to test new observables without abandoning consistency with earlier Milky Way and reionization constraints (Hartwig et al., 2024).

5. Major application domains

A-SLOTH has been extended into a broad suite of application-specific models, all retaining the merger-tree backbone while modifying the baryonic modules or the observables.

Application	Representative result	Source
Milky Way satellites and ultra-faints	The $\sim 13.8$6 stellar mass–halo mass relation develops a plateau in the ultra-faint regime; the plateau stellar mass is set by how many stars form before supernovae regulate further star formation	(Chen et al., 2022)
Pop III contribution to high-$\sim 13.8$7 galaxies	Pop III fractions are highest in $\sim 13.8$8–$\sim 13.8$9 haloes; Pop III-dominated galaxies are predicted to be too faint for direct JWST detection	(Riaz et al., 2022)
Multidimensional comparison to MW satellites	Unsupervised clustering applied to A-SLOTH satellites identified mean stellar metallicity as the dominant mismatch between the fiducial model and observations	(Chen et al., 2022)
White dwarf luminosity function	A-SLOTH star-formation histories imply that the faint end of the white dwarf luminosity function can carry a Pop III remnant signature under a bottom-heavy Pop III IMF	(Dzięcioł et al., 2024)
High-$8\,h^{-1}\,\mathrm{cMpc}$0 chemical evolution and IMF constraints	Reproducing the observed mass–metallicity–SFR relation at $8\,h^{-1}\,\mathrm{cMpc}$1–10 favored a Kroupa-like Pop II IMF with $8\,h^{-1}\,\mathrm{cMpc}$2 under the adopted yield and outflow model	(Liu et al., 6 Jun 2025)
Dust in early galaxies	$8\,h^{-1}\,\mathrm{cMpc}$3 ALMA galaxies are consistent with normal dust growth, whereas dust-poor JWST galaxies at $8\,h^{-1}\,\mathrm{cMpc}$4 require radiation-pressure-driven dust clearing during recent highly efficient bursts	(Tsuna et al., 2023)
Balmer-series observability of Pop III	Default Pop III $8\,h^{-1}\,\mathrm{cMpc}$5 fluxes peak near $8\,h^{-1}\,\mathrm{cMpc}$6, remaining 2–3 orders of magnitude below the quoted JWST/NIRSpec threshold without strong lensing	(Lipatova et al., 12 Aug 2025)
Little Red Dots and direct-collapse black holes	DCBH-seeded SMBH populations agree better than stellar-remnant seeds with observed LRD demographics and with the spectral properties of the most extreme LRD candidates	(Jeon et al., 19 Aug 2025)

Taken together, these studies show that A-SLOTH has evolved from a Milky Way archaeology model into a general early-Universe inference platform. Its observables now include halo stars, satellite mass functions, white dwarfs, cosmic SFRD, nebular and dust emission, black-hole populations, and prospective gravitational-wave and transient signatures (Hartwig et al., 2022, Dzięcioł et al., 2024).

6. Limitations, uncertainties, and methodological status

A-SLOTH remains a semi-analytic model, and its public software description explicitly notes that it uses simplified, parametrized baryonic physics rather than resolving small-scale hydrodynamics. Predictions at very low halo masses or extreme metallicities therefore rely on approximate prescriptions and extrapolations, and inference about the first stars remains model-dependent. This limitation is structural, not incidental: the framework gains speed and parameter-space coverage by replacing direct fluid evolution with analytic or statistical subgrid rules (Magg et al., 2022).

Later application papers make the main uncertainties explicit. In the white-dwarf study, the faint-end Pop III signature is sensitive to the assumed Pop III initial–final mass relation, core composition, and the neglect of binary evolution; the paper states directly that adopting a different Pop III IFMR can erase the faint-end hump (Dzięcioł et al., 2024). In the high-redshift chemical-evolution study, the preferred $8\,h^{-1}\,\mathrm{cMpc}$7 result depends on stellar-evolution tracks up to $8\,h^{-1}\,\mathrm{cMpc}$8, metallicity calibrations, a small simulation volume, and phenomenological outflow parameters (Liu et al., 6 Jun 2025). In the dust extension, the conclusions depend on uncertain supernova dust yields, dust geometry, grain properties, and the assumption of tight gas–dust coupling (Tsuna et al., 2023). In the LRD study, DCBH seeding criteria, Bondi-like gas structure, instantaneous BH mergers, and non-self-consistent dust treatment remain important caveats (Jeon et al., 19 Aug 2025).

Even with those limitations, A-SLOTH’s significance lies in the combination of public availability, modularity, and cross-domain calibration. It can function as a forward model for Bayesian inference, a generator of synthetic galaxy populations, a subgrid layer for larger cosmological analyses, and a bridge between first-star physics and local fossil records. In that sense, A-SLOTH is best understood not as a single static model, but as an extensible semi-analytic research program for tracing halo assembly into stellar, chemical, radiative, and compact-remnant observables across cosmic time (Hartwig et al., 2022, Magg et al., 2022, Hartwig et al., 2024).