Pith. sign in

REVIEW 4 major objections 5 minor 119 references

Beyond the Goldilocks Zone: Identifying Critical Features in Massive Black Hole Formation

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that whether an early atomic-cooling halo can host a direct-collapse black hole depends mainly on its central gas density and radial mass influx, not on Lyman-Werner radiation or proximity to neighboring galaxies.

desk verdict Density and inflow ranking is real but may just trace the starless selection; needs a starless-control test to separate selection from physics. read the letter →

arxiv 2412.08829 v2 pith:HPQK6LVG submitted 2024-12-12 astro-ph.GA astro-ph.CO

Elizabeth Mone , Brandon Pries , John Wise , Sandrine Ferrans This is my paper · ORCID
classification astro-ph.GAastro-ph.CO
keywords directcollapseblackholessupermassiveholeseedsRenaissancesimulationsfeatureimportancemachinelearningclassificationLyman-Wernerradiationatomiccoolinghaloshigh-redshiftgalaxyformation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks what actually sets apart the rare halos that can directly collapse into massive black hole seeds in the early universe. Using the Renaissance simulations' Rarepeak region, the authors compare 35 direct-collapse black hole candidate halos with roughly 4,000 ordinary atomic-cooling halos across 18 halo, core, and environmental properties. They find that the central core's density and radial mass influx are the strongest discriminators after the defining candidacy properties (low metallicity and halo mass), while Lyman-Werner radiation, distance to the nearest galaxy, and tidal environment matter little. The authors conclude that early massive black hole seeding is governed by gas piling up in the halo center, not by the Goldilocks-zone picture in which a nearby galaxy supplies dissociating radiation. If right, this redirects subgrid seeding models toward core density and infall triggers.

What carries the argument

The load-bearing quantity is the spherically averaged radial gas mass influx into a 50 pc sphere around the halo center, defined as $\dot{m} = -4\pi r^2 \rho v_r$, together with the central gas density. These two core quantities, computed as mass-weighted 50 pc averages, carry the argument: they appear at or near the top of every feature ranking once the selection variables (metallicity, halo mass) are removed, and they are physically tied to whether a starless, atomic-cooling halo can concentrate enough gas to collapse directly. The statistical machinery, including Mahalanobis-distance recursive elimination, recursive logistic regression, and random-forest permutation importance with correlation-grouped features, is used to show that the separation is not an artifact of one ranking method.

What would settle it

Re-simulate a large fraction of the 35 candidate halos at sub-parsec resolution and track fragmentation: if a substantial fraction produce ordinary star clusters instead of supermassive stars or a massive seed, then central density and radial inflow are features of candidacy selection, not of DCBH formation.

Watch

Extended reading notes

Core claim

The paper's central claim is that DCBH candidacy is primarily a story of the halo's core: candidate halos are statistically separable from non-candidates mainly through central gas density and radial mass influx, plus rapid recent mass growth, rather than through external Lyman-Werner flux, distance to neighbors, overdensity, or tidal field. Across four ranking schemes, including Z-score distributions, recursive Mahalanobis-distance elimination, recursive logistic regression, and random-forest permutation importance, central density and radial mass influx appear among the top features once the selection-defining variables (metallicity and halo mass) are set aside. The case study of a matched pair of halos shows the candidate with an isothermal ($\rho \propto r^{-2}$) envelope, a collapsing core with inflow rising to 5 solar masses per year, and no prior star formation, whereas the non-candidate has a relic H II region, low central density, and inflows orders of magnitude smaller. Candidate halos also grow dramatically faster: over the final 130 Myr they increase in mass by a factor of about 12 versus about 2 for non-candidates, suggesting rapid halo growth suppresses H2 and sustains infall. The paper therefore argues that the needed cooling suppression comes from dynamical core processes, not from a Goldilocks Lyman-Werner environment.

Load-bearing premise

The 35 candidates are stand-ins for real black hole formation: the simulation cannot resolve the final collapse, so a halo only has to be starless, nearly metal-free, and near the atomic cooling limit to count as a candidate; if many of these halos would actually fragment into ordinary stars, the feature ranking describes the selection criteria rather than the formation of massive black hole seeds.

Editorial extensions

If this is right

  • DCBH subgrid seeding models should be triggered by central density and radial mass influx instead of halo mass or environmental Lyman-Werner flux alone.
  • The absence of a Goldilocks zone means the cosmic number density of DCBHs is not set by the abundance of nearby Lyman-Werner sources, so environments are weak regulators of the heavy-seed channel.
  • Rapid halo growth can supply the hydrogen-suppression mechanism that Lyman-Werner radiation was previously invoked to provide.
  • Because candidates and non-candidates are statistically separable populations with a Mahalanobis distance above 1, a probabilistic formation model built from these features is well posed.
  • The H2 fraction does not need to be a separate input to seeding models, since the starless condition already selects halos with negligible molecular hydrogen.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An extension the paper leaves implicit is that the same dense-core, high-infall signature is the common fuel condition for both direct-collapse seeds and rapid growth of light seeds, so the ranking may transfer beyond DCBH sites to early black hole accretion generally.
  • A testable prediction from this view is that JWST-identified overmassive black hole hosts at z > 6 should preferentially live in rapidly growing halos with dense, inflowing cores and no nearby Lyman-Werner partner, compared with ordinary star-forming galaxies.
  • The feature ranking could be turned into a probabilistic seeding trigger and checked against larger-volume simulations that resolve atomic-cooling halos; if the resulting seed mass function matches future gravitational-wave merger rates, that would support the mechanism as the main heavy-seed channel.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. This paper uses the Rarepeak region of the Renaissance simulations to compare 35 DCBH candidate halos with roughly 4,000 non-candidates at z = 15--24. The authors compute 18 halo, central, and environmental features and rank them with Z-scores, Mahalanobis-distance-based recursive elimination, logistic-regression recursive feature ranking, and random-forest permutation importance, after grouping correlated features. The central claim is that, apart from the candidacy-selection variables (metallicity and halo mass), central density and radial mass influx are the most important discriminators, while Lyman-Werner flux and large-scale environment are not, in tension with the 'Goldilocks zone' picture of DCBH formation.

Significance. If the central claim survives a matched starless control, the result is valuable: it would redirect DCBH formation criteria toward core gas dynamics rather than external radiation and would give concrete feature guidance for subgrid seeding models. The paper has real strengths: it uses multiple complementary ranking methods, groups correlated features, provides a robustness test that relaxes the metallicity criterion (Sec. 4.3), includes an illuminating case study with radial profiles (Sec. 3.3), and states its caveats unusually candidly (Sec. 4.2). The principal weakness is that the two promoted features may be imprints of the starless selection criterion itself, so the quantitative ranking needs a matched control before the main conclusion is supported.

major comments (4)
  1. [§2.2, §3.3, Conclusion item 2] The central conclusion that central density and radial mass influx are the most important features is not controlled against the starless selection criterion. Candidates are starless by construction (Sec. 2.1), while non-candidates are only required to lack metal-enriched stars younger than 20 Myr and therefore may contain older stellar populations. Prior star formation evacuates and heats the central gas, mechanically lowering both central density and radial mass influx; the Sec. 3.3 case study demonstrates this directly, attributing the non-candidate's diffuse core to a star-formation event 41 Myr earlier (Table 1, Fig. 8). The feature ranking may thus be measuring 'remained starless while crossing the atomic cooling limit' rather than 'has the physics to form a DCBH.' The robustness test in Sec. 4.3 relaxes only the metallicity criterion and leaves the starless cut in place, so it cannot resolve this confound. I request a matched analysis using non-candidate halos that are starless, or at minimum a control for time since last star formation, before this conclusion is supported.
  2. [§3.1, Fig. 1, Eq. (6)] The Z-scores quoted in Fig. 1 are not defined as sample-mean distances, carry no uncertainties, and are computed after ad hoc outlier trimming (20--50 halos for most features, up to 300--500 for H2 fraction and temperature). The text interprets values such as 1.1σ and 1.0σ as meaningful separations, but without standard errors, bootstrap confidence intervals, or a significance test, these numbers are not quantitative evidence. Please report the sample-mean version of Eq. (6) with uncertainties, and state whether the trimming affects the reported Z-scores or perform the calculation on the full samples.
  3. [§2.2, §2.1] The construction of the non-candidate sample is incompletely specified: the text says all atomic cooling halos from all 40 outputs are analyzed, but only the candidates are described as deduplicated by formation lineage. If a given halo appears at multiple snapshots, the ~4,000 non-candidates are not independent, which will bias both the statistical comparisons and the machine-learning feature importances. Please state how non-candidates were deduplicated and report the number of unique halos. Relatedly, the paper should reconcile the 35 candidates with the 76 Rarepeak candidates reported in Regan et al. (2020b), since the selection criteria appear similar.
  4. [§2.4.1, §2.4.3, Figs. 3--4] Two methodological issues weaken the quantitative feature-ranking claims. First, the recursive Mahalanobis procedure compares distances after removing features, but the dimension of the distance changes with each removal, and 'distance > 1' is not a statistical significance criterion; the resulting rankings therefore need normalization or a statistical test. Second, the Random Forest permutation ranking is trained on 35 candidates versus ~4,000 non-candidates without apparent class weighting, and the reported 'decrease in accuracy score' is a poor metric under strong class imbalance; the paper notes large error bars but not this imbalance. Please address both points or qualify the affected rankings accordingly.
minor comments (5)
  1. [§2.1] The text states a maximum spatial resolution of 19 comoving pc and then calls this 'parsec-scale resolution'; this is off by an order of magnitude and should be reworded, e.g., to 'tens of parsec scale.'
  2. [§4] The phrase 'observational research research cited above' contains a duplicated word and should be corrected.
  3. [§2.4.1] The statement that 'Mahalanobis distances with a value > 1 are generally considered to show a low similarity' is vague; please replace it with a formal criterion such as a chi-square quantile or a permutation-based p-value.
  4. [Fig. 3] The labels 'Top values' and 'Bottom values' in the stair plots are easy to misread; please clarify in the caption which row corresponds to least-to-most versus most-to-least elimination and what 'top' and 'bottom' mean in each panel.
  5. [§2.2.3] The overdensity definition says ρ̄ is the critical density at the halo redshift, but the surrounding text suggests it should be the mean density in the 15 kpc region; please reconcile the notation.

Circularity Check

1 steps flagged · score 6.0 of 10

Starless selection criterion is never controlled for, so top-ranked density and inflow features partially restate the candidate definition.

  1. fitted input called prediction [Sec. 2.1-2.2 (candidate selection), Sec. 3.3 (case study), Sec. 4.3 (robustness), Sec. 5 Conclusion item 2]
    "the candidates are chosen from atomic cooling halos to be starless and metal-free ... a cut was made on non-candidate halos to remove those containing metal enriched stars younger than 20 Myr ... as being starless is one of our requirements for direct collapse. ... stellar feedback from the prior star formation event in the non-candidate halo evacuated its dense core. ... Excepting properties used for candidacy selection, we identify the central density and radial mass influx ... as the features of most importance in determining capability of hosting a DCBH."

    The candidate set is defined by the starless criterion; the non-candidate pool only removes stars younger than 20 Myr, so it retains halos with older stellar populations. Older star formation injects feedback that evacuates and heats the central gas, and the paper's own case study attributes the non-candidate's diffuse core to a star-formation event 41 Myr earlier. Density and radial mass influx are therefore downstream physical manifestations of the starless part of the selection rule, not independent predictors. The Sec. 4.3 robustness test relaxes only the metallicity cut ('specifically removing the metal-poor requirement'), leaving the starless cut in place, so the top-ranked non-selection features remain statistically forced by the selection.

full rationale

The paper is unusually transparent about two of its explicit selection criteria: metallicity and halo mass are both acknowledged to be top-ranked features, and the authors rerank without the metallicity cut in Sec. 4.3. The partial circularity lies with the third criterion, starless, which is never entered as a feature and never relaxed. Because candidates are starless by construction while non-candidates may contain older stars, and because older star formation is shown in Sec. 3.3 to evacuate the core and suppress both density and radial mass influx, the headline result that these two central quantities are the most important 'non-candidacy' features is substantially forced by the selection rule. This is partial, not total, circularity: density and inflow are not themselves selection variables, and the paper's environmental conclusions (e.g., low importance of Lyman-Werner flux and distance) are not artifacts of the selection. The self-citations to Wise et al. 2019 and Regan et al. 2020c for validating three candidates are empirical follow-up simulations, so they do not by themselves make the argument circular. The missing matched starless-control analysis is the key reason the central claim cannot yet be read as an independent physical discovery.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The load-bearing assumptions are simulation fidelity and the candidacy proxy. The free parameters are hand-chosen analysis apertures and selection thresholds, not fitted constants; they are disclosed in the text but not propagated into uncertainties.

free parameters (4)
  • Central property aperture = 50 pc
    Mass-weighted averages of central features are taken in a 50 pc sphere around the halo center (Sec. 2.2.1). The 50 pc scale is chosen by hand and directly sets density, radial mass flux, velocities, H2 fraction, and other core quantities; a different aperture would change feature values and rankings.
  • Overdensity aperture = 15 kpc
    Environmental overdensity uses a 15 kpc region chosen as approximately the turnaround radius for a 3 x 10^10 Msun halo at z=8 (Sec. 2.2.3). This hand-set scale determines the environmental overdensity feature.
  • Non-candidate young-star exclusion age = 20 Myr
    Non-candidate halos with metal-enriched stars younger than 20 Myr are removed (Sec. 2.2). This shapes the comparison population and removes internal Lyman-Werner peaks from young stars.
  • Candidate metallicity threshold = Z < 10^-4 Zsun
    The low-metallicity condition defines candidacy (Sec. 2.1). Since metallicity is also used as a feature, this threshold strongly drives the feature rankings and is the target of the robustness test in Sec. 4.3.
assumptions (4)
  • domain assumption The Renaissance Rarepeak simulation faithfully models the gas, radiation transport, H2 chemistry, star formation, and feedback relevant to atomic cooling halos.
    All 18 features are measured from these simulation outputs (Sec. 2.1). Errors in radiation transport, H2 chemistry, star formation, or feedback would propagate directly into the candidate selection and feature ranking.
  • ad hoc to paper Atomic cooling halos that are starless, metal-poor (Z < 10^-4 Zsun), and near the atomic cooling mass are valid DCBH candidates.
    This is the working definition of a DCBH host in Sec. 2.1 and Sec. 4.3. The simulation cannot resolve the final collapse, so the paper's central claim is conditional on this proxy.
  • domain assumption ROCKSTAR halo catalogs and consistent-trees merger trees correctly identify halos, their centers, masses, and progenitors.
    Feature definitions in Sec. 2.2 rely on these catalogs; systematic errors would affect central 50 pc averages, growth rates, and the closest-galaxy calculation.
  • domain assumption The virial mass-temperature relation of Eq. (1) from Fernandez et al. (2014) holds for halos at z=15 to 24.
    This relation is used to select atomic cooling halos and to define candidacy. If its normalization is wrong, the candidate sample shifts in mass and temperature.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Beyond the Goldilocks Zone: Identifying Critical Features in Massive Black Hole Formation." pith.science (2026). https://pith.science/paper/HPQK6LVG

@misc{pith2026241208829,
  author       = {Pith},
  title        = {Pith review of: Beyond the Goldilocks Zone: Identifying Critical Features in Massive Black Hole Formation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HPQK6LVG}},
  note         = {Machine review of arXiv:2412.08829}
}
read the original abstract

Most galaxies, including the Milky Way, host a supermassive black hole (SMBH) at the center. These SMBHs can be observed out to high redshifts (z>=6) if the accretion rate is sufficiently large. However, we do not fully understand the mechanism through which these black holes form at early times. The heavy (or direct collapse) seeding mechanism has emerged as a probable contender in which the core of an atomic cooling halo directly collapses into a dense stellar cluster that could host supermassive stars that proceed to form a BH seed of mass ~10^5 M_sun. We use the Renaissance simulations to investigate the properties of 35 DCBH candidate host halos at z=15-24 and compare them to non-candidate halos. We aim to understand what features differentiate halos capable of hosting a DCBH from the general halo population with the use of statistical analysis and machine learning methods. We examine 18 halo, central, and environmental properties. We find that DCBH candidacy is more dependent on a halo's core internal properties than on exterior factors such as Lyman-Werner flux and distance to closest galaxy; our analysis selects density and radial mass influx as the most important features (outside candidacy establishing features). Our results concur with the recent suggestion that DCBH host halos neither need to lie within a "Goldilocks zone" nor have a significant amount of Lyman-Werner flux to suppress cooling. This paper presents insight to the dynamics possibly occurring in potential DCBH host halos and seeks to provide guidance to DCBH subgrid formation models.

Figures

Figures reproduced from arXiv: 2412.08829 by the authors.

Figure 1
Figure 1. Probability distribution comparisons for candidate (orange) vs. non-candidate (blue) halo populations. The his￾tograms display binned probabilities and kernel density estimate functions. Individual feature Z-scores are provided in the upper left-hand corners of each plot. The plots show regions of peak probability with outliers removed; in the case of most plots be￾tween 20-50 (non-candidate only) halos are removed … view at source ↗
Figure 2
Figure 2. The left panel displays a hierarchy linkage map between features based on the Spearman correlation coefficient distance (e.g., Yu & Hutson 2024). The groups are numbered from left to right in the diagram and designated by colors with Group 1 in pink, Group 2 in yellow, Group 3 in green, Group 4 in blue, and Group 5 in purple; these numbers and colors will be used throughout the rest of this paper. The right panel sh… view at source ↗
Figure 3
Figure 3. Stair plots displaying the effect on the Mahalanobis score as features are progressively removed. The top row shows recursive elimination of least important features and contain the most important values in the lower left corners. The bottom row shows recursive elimination of the most important features with the least important labeled in the top right. In the left column features are removed individually, and in th… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The left panel displays the Gini importances (Breiman et al. 1984; Nembrini et al. 2018) of each feature using a random forest classifier model with a higher score indicating more importance. The right panel shows box plots for the decrease in accuracy score of the mod…
Figure 5
Figure 5. Figure 5: Information gain produced by each feature group where the groups are as follows: (1) Gas and Dark Mat￾ter Spin Parameter, (2) Tidal Field t1, Distance to closest galaxy, LW Flux, and Overdensity, (3) Growth Rates (dM/dt and dM/dz), (4) Temperature, Radial Velocity, Met…
Figure 6
Figure 6. Figure 6: Comparison of normalized halo mass with respect to time for non-candidate halos (blue) and candidate halos (orange) with a final redshift of z = 15.0. Both samples are normalized to the final median DCBH mass. The lines rep￾resent the median values of each population w…
Figure 7
Figure 7. Figure 7: A visual comparison of a DCBH candidate halo (top, Mvir = 6.28 × 107 M⊙, LW flux = 2.42 × 10−21 cgs) and a non-candidate halo (bottom, Mvir = 6.19 × 107 M⊙, LW flux = 2.35 × 10−21 cgs). Slices are through the halo center for three main halo variables: density (left), t…
Figure 8
Figure 8. Figure 8: Radial profiles for density (top left), temperature (top right), H2 fraction (bottom left), and radial mass flux (bottom right) for a DCBH candidate (orange) and non-candidate (blue). Dashed lines in the radial mass flux profile represent radii of mass outflux. date ha…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

119 extracted references · 10 canonical work pages

  1. [1]

    2024, , 627, 281, 10.1038/s41586-024-07053-4

    Abuter, R., Allouche, F., Amorim, A., et al. 2024, , 627, 281, 10.1038/s41586-024-07053-4

    doi:10.1038/s41586-024-07053-4 2024

  2. [2]

    L., Zackrisson, E., et al

    Agarwal, B., Johnson, J. L., Zackrisson, E., et al. 2016 a , , 460, 4003, 10.1093/mnras/stw1173

    doi:10.1093/mnras/stw1173 2016

  3. [3]

    2016 b , , 459, 4209, 10.1093/mnras/stw929

    Agarwal, B., Smith, B., Glover, S., Natarajan, P., & Khochfar, S. 2016 b , , 459, 4209, 10.1093/mnras/stw929

    doi:10.1093/mnras/stw929 2016

  4. [5]

    C., Volonteri, M., & Rees, M

    Begelman, M. C., Volonteri, M., & Rees, M. J. 2006, , 370, 289, 10.1111/j.1365-2966.2006.10467.x

    arXiv 2006

  5. [6]

    S., Wechsler, R

    Behroozi, P. S., Wechsler, R. H., & Wu, H.-Y. 2013, , 762, 109, 10.1088/0004-637x/762/2/109

    doi:10.1088/0004-637x/762/2/109 2013

  6. [7]

    K., Blecha, L., Torrey, P., et al

    Bhowmick, A. K., Blecha, L., Torrey, P., et al. 2022, , 510, 177, 10.1093/mnras/stab3439

    doi:10.1093/mnras/stab3439 2022

  7. [8]

    2024 a , , 531, 4311, 10.1093/mnras/stae1386

    ---. 2024 a , , 531, 4311, 10.1093/mnras/stae1386

    doi:10.1093/mnras/stae1386 2024

  8. [9]

    2024 b , , 533, 1907, 10.1093/mnras/stae1819

    ---. 2024 b , , 533, 1907, 10.1093/mnras/stae1819

    doi:10.1093/mnras/stae1819 2024

Show all 119 references
  1. [10]

    A., & Parker, R

    Blaylock-Squibbs, G. A., & Parker, R. J. 2023, , 519, 3643, 10.1093/mnras/stac3683

    2023 doi

  2. [11]

    D., Natarajan, P., et al

    Bogd \'a n, \'A ., Goulding, A. D., Natarajan, P., et al. 2024, NatAs, 8, 126, 10.1038/s41550-023-02111-9

    2024 doi

  3. [12]

    Bosman, S. E. I., \'A lvarez-M \'a rquez, J., Colina, L., et al. 2024, NatAs, 8, 1054, 10.1038/s41550-024-02273-0

    2024 doi

  4. [13]

    2024, Aeos : Star-by-Star Cosmological Simulations of Early Chemical Enrichment and Galaxy Formation, 10.48550/arXiv.2410.16366

    Brauer, K., Emerick, A., Mead, J., et al. 2024, Aeos : Star-by-Star Cosmological Simulations of Early Chemical Enrichment and Galaxy Formation, 10.48550/arXiv.2410.16366

    2024 doi arXiv

  5. [14]

    2001, ML, 45, 5, 10.1023/a:1010933404324

    Breiman, L. 2001, ML, 45, 5, 10.1023/a:1010933404324

    2001 doi

  6. [15]

    H., Olshen, R

    Breiman, L., Friedman, J. H., Olshen, R. A., & Stone, C. J. 1984, Classification And Regression Trees, 1st edn. (Routledge), 10.1201/9781315139470

    1984 doi

  7. [16]

    2003, , 596, 34, 10.1086/377529

    Bromm, V., & Loeb, A. 2003, , 596, 34, 10.1086/377529

    2003 doi

  8. [17]

    2019, JOSS, 4, 1636, 10.21105/joss.01636

    Brummel-Smith, C., Bryan, G., Butsky, I., et al. 2019, JOSS, 4, 1636, 10.21105/joss.01636

    2019 doi

  9. [18]

    L., Norman, M

    Bryan, G. L., Norman, M. L., O'Shea, B. W., et al. 2014, , 211, 19, 10.1088/0067-0049/211/2/19

    2014 doi

  10. [19]

    S., Dekel, A., Kolatt, T

    Bullock, J. S., Dekel, A., Kolatt, T. S., et al. 2001, , 555, 240, 10.1086/321477

    2001 doi

  11. [20]

    J., Saxena, A., Cameron, A

    Bunker, A. J., Saxena, A., Cameron, A. J., et al. 2023, , 677, A88, 10.1051/0004-6361/202346159

    2023 doi

  12. [21]

    J., Katz, H., Rey, M

    Cameron, A. J., Katz, H., Rey, M. P., & Saxena, A. 2023, , 523, 3516, 10.1093/mnras/stad1579

    2023 doi

  13. [22]

    Chen, R.-C., Dwei, C., Huang, S.-W., & Caraka, R. E. 2020, JBD, 7, 10.1186/s40537-020-00327-4

    2020 doi

  14. [23]

    2021, , 502, 700, 10.1093/mnras/stab061

    Chon, S., Hosokawa, T., & Omukai, K. 2021, , 502, 700, 10.1093/mnras/stab061

    2021 doi

  15. [24]

    2024, , 530, 2453, 10.1093/mnras/stae1027

    Chon, S., Hosokawa, T., Omukai, K., & Schneider, R. 2024, , 530, 2453, 10.1093/mnras/stae1027

    2024 doi

  16. [25]

    2020, , 494, 2851, 10.1093/mnras/staa863

    Chon, S., & Omukai, K. 2020, , 494, 2851, 10.1093/mnras/staa863

    2020 doi

  17. [26]

    R., & Shirokov, A

    Dalal, N., White, M., Bond, J. R., & Shirokov, A. 2008, , 687, 12, 10.1086/591512

    2008 doi

  18. [27]

    J., et al

    Dattathri, S., Natarajan, P., Porras-Valverde, A. J., et al. 2024, The redshift evolution of the M_ BH - M_ * scaling relation: new insights from cosmological simulations and semi-analytic models, 10.48550/arXiv.2410.13958

    2024 doi arXiv

  19. [28]

    2019, , 486, 2827, 10.1093/mnras/stz937

    Dav \'e , R., Angl \'e s-Alc \'a zar, D., Narayanan, D., et al. 2019, , 486, 2827, 10.1093/mnras/stz937

    2019 doi

  20. [29]

    De Maesschalck, R., Jouan-Rimbaud, D., & Massart, D. L. 2000, Chemom. Intell. Lab. Syst., 50, 1, 10.1016/s0169-7439(99)00047-7

    2000 doi

  21. [30]

    2018, MPLB, 32, 1850385, 10.1142/s0217984918503852

    Dhiman, G., & Kumar, V. 2018, MPLB, 32, 1850385, 10.1142/s0217984918503852

    2018 doi

  22. [31]

    Di Matteo, T., Croft, R. A. C., Feng, Y., Waters, D., & Wilkins, S. 2017, , 467, 4243, 10.1093/mnras/stx319

    2017 doi

  23. [32]

    Dijkstra, M., Haiman, Z., Mesinger, A., & Wyithe, J. S. B. 2008, , 391, 1961, 10.1111/j.1365-2966.2008.14031.x

    2008

  24. [33]

    G., Mahabal, A

    Donalek, C., Djorgovski, S. G., Mahabal, A. A., et al. 2013, in 2013 IEEE International Conference on Big Data (IEEE), 35--41, 10.1109/BigData.2013.6691731

    2013

  25. [34]

    2024, What is the origin of the JWST SMBHs?, 10.48550/arXiv.2410.24224

    Ellis, J., Fairbairn, M., Urrutia, J., & Vaskonen, V. 2024, What is the origin of the JWST SMBHs?, 10.48550/arXiv.2410.24224

    2024 doi

  26. [35]

    L., Haiman, Z., & Li, M

    Fernandez, R., Bryan, G. L., Haiman, Z., & Li, M. 2014, , 439, 3798, 10.1093/mnras/stu230

    2014 doi

  27. [36]

    J., Labb \'e , I., Zitrin, A., et al

    Furtak, L. J., Labb \'e , I., Zitrin, A., et al. 2024, , 628, 57, 10.1038/s41586-024-07184-8

    2024 doi

  28. [37]

    2000, Commun

    Geun Kim, M. 2000, Commun. Stat. - Theory Methods, 29, 1511, 10.1080/03610920008832559

    2000 doi

  29. [38]

    C., Fragione, G., Kremer, K., & Rasio, F

    Gonz \'a lez Prieto, E., Weatherford, N. C., Fragione, G., Kremer, K., & Rasio, F. A. 2024, , 969, 29, 10.3847/1538-4357/ad43d6

    2024 doi

  30. [39]

    D., Greene, J

    Goulding, A. D., Greene, J. E., Setton, D. J., et al. 2023, , 955, L24, 10.3847/2041-8213/acf7c5

    2023 doi

  31. [40]

    W., & Wise, J

    Grace, J., O'Shea, B. W., & Wise, J. H. 2020, RNAAS, 4, 93, 10.3847/2515-5172/ab9e78

    2020 doi

  32. [41]

    A., & Pacucci, F

    Guia, C. A., & Pacucci, F. 2024, RNAAS, 8, 153, 10.3847/2515-5172/ad530c

    2024 doi

  33. [42]

    2002, ML, 46, 389, 10.1023/a:1012487302797

    Guyon, I., Weston, J., Barnhill, S., & Vapnik, V. 2002, ML, 46, 389, 10.1023/a:1012487302797

    2002 doi

  34. [43]

    2016, , 463, 529, 10.1093/mnras/stw1924

    Habouzit, M., Volonteri, M., Latif, M., Dubois, Y., & Peirani, S. 2016, , 463, 529, 10.1093/mnras/stw1924

    2016 doi

  35. [44]

    J., & Loeb, A

    Haiman, Z., Rees, M. J., & Loeb, A. 1997 a , ApJ, 476, 458, 10.1086/303647

    1997 doi

  36. [45]

    1997 b , ApJ, 484, 985, 10.1086/304386

    ---. 1997 b , ApJ, 484, 985, 10.1086/304386

    1997 doi

  37. [46]

    N., & Basu, S

    Hirano, S., Machida, M. N., & Basu, S. 2023, , 952, 56, 10.3847/1538-4357/acda94

    2023 doi

  38. [48]

    M., Zitlau, R., Seitz, S., & Weller, J

    Hoyle, B., Rau, M. M., Zitlau, R., Seitz, S., & Weller, J. 2015, , 449, 1275, 10.1093/mnras/stv373

    2015 doi

  39. [49]

    Hunter, J. D. 2007, CSE, 9, 90, 10.1109/mcse.2007.55

    2007 doi

  40. [50]

    2020, , 58, 27, 10.1146/annurev-astro-120419-014455

    Inayoshi, K., Visbal, E., & Haiman, Z. 2020, , 58, 27, 10.1146/annurev-astro-120419-014455

    2020 doi

  41. [51]

    M., et al

    Juod z balis, I., Maiolino, R., Baker, W. M., et al. 2024, , 636, 594, 10.1038/s41586-024-08210-5

    2024 doi

  42. [52]

    2020, , 494, 2200, 10.1093/mnras/staa639

    Katz, H., Ramsoy, M., Rosdahl, J., et al. 2020, , 494, 2200, 10.1093/mnras/staa639

    2020 doi

  43. [53]

    2017, , 467, 4739, 10.1093/mnras/stx126

    Kaviraj, S., Laigle, C., Kimm, T., et al. 2017, , 467, 4739, 10.1093/mnras/stx126

    2017 doi

  44. [54]

    2014, in 2014 Science and Information Conference (The Science and Information Organization), 372--378, 10.1109/SAI.2014.6918213

    Khalid, S., Khalil, T., & Nasreen, S. 2014, in 2014 Science and Information Conference (The Science and Information Organization), 372--378, 10.1109/SAI.2014.6918213

    2014

  45. [55]

    2017, , 466, 4826, 10.1093/mnras/stx052

    Kimm, T., Katz, H., Haehnelt, M., et al. 2017, , 466, 4826, 10.1093/mnras/stx052

    2017 doi

  46. [56]

    2024, , 534, 3916, 10.1093/mnras/stae2380

    Kiyuna, M., Hosokawa, T., & Chon, S. 2024, , 534, 3916, 10.1093/mnras/stae2380

    2024 doi

  47. [57]

    S., & Glover, S

    Klessen, R. S., & Glover, S. C. O. 2023, , 61, 65, 10.1146/annurev-astro-071221-053453

    2023 doi

  48. [58]

    2021, in 25th International Conference on Pattern Recognition (IEEE), 9318--9325, 10.1109/icpr48806.2021.9413090

    K \"o nig, G., Molnar, C., Bischl, B., & Grosse-Wentrup, M. 2021, in 25th International Conference on Pattern Recognition (IEEE), 9318--9325, 10.1109/icpr48806.2021.9413090

    2021

  49. [59]

    E., Bogd \'a n, \'A ., Natarajan, P., et al

    Kov \'a cs, O. E., Bogd \'a n, \'A ., Natarajan, P., et al. 2024, , 965, L21, 10.3847/2041-8213/ad391f

    2024 doi

  50. [60]

    2020, , 498, 5652, 10.1093/mnras/staa2276

    Kroupa, P., Subr, L., Jerabkova, T., & Wang, L. 2020, , 498, 5652, 10.1093/mnras/staa2276

    2020 doi

  51. [61]

    L., Finkelstein, S

    Larson, R. L., Finkelstein, S. L., Kocevski, D. D., et al. 2023, , 953, L29, 10.3847/2041-8213/ace619

    2023 doi

  52. [62]

    A., Schleicher, D

    Latif, M. A., Schleicher, D. R. G., Schmidt, W., & Niemeyer, J. C. 2013, , 436, 2989, 10.1093/mnras/stt1786

    2013 doi

  53. [63]

    C., Vijayan, A

    Lovell, C. C., Vijayan, A. P., Thomas, P. A., et al. 2021, , 500, 2127, 10.1093/mnras/staa3360

    2021 doi

  54. [64]

    Lupi, A., Quadri, G., Volonteri, M., Colpi, M., & Regan, J. A. 2024 a , , 686, A256, 10.1051/0004-6361/202348788

    2024 doi

  55. [65]

    2024 b , , 689, A128, 10.1051/0004-6361/202451249

    Lupi, A., Trinca, A., Volonteri, M., Dotti, M., & Mazzucchelli, C. 2024 b , , 689, A128, 10.1051/0004-6361/202451249

    2024 doi

  56. [66]

    2024 a , , 627, 59, 10.1038/s41586-024-07052-5

    Maiolino, R., Scholtz, J., Witstok, J., et al. 2024 a , , 627, 59, 10.1038/s41586-024-07052-5

    2024 doi

  57. [67]

    2024 b , , 630, E2, 10.1038/s41586-024-07494-x

    ---. 2024 b , , 630, E2, 10.1038/s41586-024-07494-x

    2024 doi

  58. [68]

    2024 c , , 691, A145, 10.1051/0004-6361/202347640

    Maiolino, R., Scholtz, J., Curtis-Lake, E., et al. 2024 c , , 691, A145, 10.1051/0004-6361/202347640

    2024 doi

  59. [69]

    A., Yue, M., Eilers, A.-C., et al

    Marshall, M. A., Yue, M., Eilers, A.-C., et al. 2024, GA-NIFS & EIGER: A merging quasar host at z = 7 with an overmassive black hole, 10.48550/arXiv.2410.11035

    2024 doi

  60. [70]

    R., Zwick, L., & Di Matteo, T

    Mayer, L., Capelo, P. R., Zwick, L., & Di Matteo, T. 2024, , 961, 76, 10.3847/1538-4357/ad11cf

    2024 doi

  61. [71]

    M., Regan, J

    McCaffrey, J. M., Regan, J. A., Smith, B. D., et al. 2024, A HEAVY SEED BLACK HOLE MASS FUNCTION AT HIGH REDSHIFT - PROSPECTS FOR LISA, 10.48550/arXiv.2409.16413

    2024 doi arXiv

  62. [72]

    A., & Prole, L

    Mehta, D., Regan, J. A., & Prole, L. 2024, OJAp, 7, 10.33232/001c.126629

    2024 doi

  63. [73]

    \'A ., Ru \'i z, R., Garc \'i a-Torres, M., & Sarro, L

    Montero, M. \'A ., Ru \'i z, R., Garc \'i a-Torres, M., & Sarro, L. M. 2010, Feature Selection Applied to Data from the Sloan Digital Sky Survey, Lecture Notes in Artificial Intelligence (Springer Berlin Heidelberg), 611--620, 10.1007/978-3-642-13022-9_61

    2010 doi

  64. [74]

    Musa, A. B. 2013, IJMLC, 4, 13, 10.1007/s13042-012-0068-x

    2013 doi

  65. [75]

    2024, , 683, A58, 10.1051/0004-6361/202347724

    Nabizadeh, A., Zackrisson, E., Pacucci, F., et al. 2024, , 683, A58, 10.1051/0004-6361/202347724

    2024 doi

  66. [76]

    2024, , 960, L1, 10.3847/2041-8213/ad0e76

    Natarajan, P., Pacucci, F., Ricarte, A., et al. 2024, , 960, L1, 10.3847/2041-8213/ad0e76

    2024 doi

  67. [77]

    R., & Wright, M

    Nembrini, S., K \"o nig, I. R., & Wright, M. N. 2018, Bioinform., 34, 3711, 10.1093/bioinformatics/bty373

    2018 doi

  68. [78]

    2016, , 459, 1432, 10.1093/mnras/stw725

    Pacucci, F., Ferrara, A., Grazian, A., et al. 2016, , 459, 1432, 10.1093/mnras/stw725

    2016 doi

  69. [79]

    2024, , 964, 154, 10.3847/1538-4357/ad3044

    Pacucci, F., & Loeb, A. 2024, , 964, 154, 10.3847/1538-4357/ad3044

    2024 doi

  70. [80]

    J., Whalen, D

    Patrick, S. J., Whalen, D. J., Latif, M. A., & Elford, J. S. 2023, , 522, 3795, 10.1093/mnras/stad1179

    2023 doi

  71. [81]

    2011, JMLR, 12, 2825

    Pedregosa, F., Varoquaux, G., Gramfort, A., et al. 2011, JMLR, 12, 2825. 1201.0490

    2011 arXiv

  72. [82]

    2022, Supporting AI-Explainability by Analyzing Feature Subsets in a Machine Learning Model, Studies in Health Technology and Informatics (IOS Press), 109--113, 10.3233/shti220406

    Plagwitz, L., Brenner, A., Fujarski, M., & Varghese, J. 2022, Supporting AI-Explainability by Analyzing Feature Subsets in a Machine Learning Model, Studies in Health Technology and Informatics (IOS Press), 109--113, 10.3233/shti220406

    2022 doi

  73. [83]

    F., Baumgardt, H., Hut, P., Makino, J., & McMillan, S

    Portegies Zwart, S. F., Baumgardt, H., Hut, P., Makino, J., & McMillan, S. L. W. 2004, , 428, 724, 10.1038/nature02448

    2004 doi

  74. [84]

    R., Regan, J

    Prole, L. R., Regan, J. A., Glover, S. C. O., et al. 2024, , 685, A31, 10.1051/0004-6361/202348903

    2024 doi

  75. [85]

    A., Haiman, Z., Wise, J

    Regan, J. A., Haiman, Z., Wise, J. H., O'Shea, B. W., & Norman, M. L. 2020 a , OJAp, 3, 10.21105/astro.2006.14625

    2020 arXiv

  76. [86]

    A., Visbal, E., Wise, J

    Regan, J. A., Visbal, E., Wise, J. H., et al. 2017, NatAs, 1, 0075, 10.1038/s41550-017-0075

    2017 doi

  77. [87]

    A., & Volonteri, M

    Regan, J. A., & Volonteri, M. 2024, OJAp, 7, 10.33232/001c.123239

    2024 doi

  78. [88]

    A., Wise, J

    Regan, J. A., Wise, J. H., O'Shea, B. W., & Norman, M. L. 2020 b , , 492, 3021, 10.1093/mnras/staa035

    2020 doi

  79. [89]

    A., Wise, J

    Regan, J. A., Wise, J. H., Woods, T. E., et al. 2020 c , OJAp, 3, 10.21105/astro.2008.08090

    2020

  80. [90]

    Reinoso, B., Schleicher, D. R. G., Fellhauer, M., Klessen, R. S., & Boekholt, T. C. N. 2018, , 614, A14, 10.1051/0004-6361/201732224

    2018 doi

  81. [91]

    Sakurai, Y., Yoshida, N., & Fujii, M. S. 2019, , 484, 4665, 10.1093/mnras/stz315

    2019 doi

  82. [92]

    S., & Hirano, S

    Sakurai, Y., Yoshida, N., Fujii, M. S., & Hirano, S. 2017, , 472, 1677, 10.1093/mnras/stx2044

    2017 doi

  83. [93]

    2021, , 506, 613, 10.1093/mnras/stab1737

    Sassano, F., Schneider, R., Valiante, R., et al. 2021, , 506, 613, 10.1093/mnras/stab1737

    2021 doi

  84. [94]

    A., Bower, R

    Schaye, J., Crain, R. A., Bower, R. G., et al. 2015, , 446, 521, 10.1093/mnras/stu2058

    2015 doi

  85. [95]

    2024, , 687, A283, 10.1051/0004-6361/202347187

    Scholtz, J., Witten, C., Laporte, N., et al. 2024, , 687, A283, 10.1051/0004-6361/202347187

    2024 doi

  86. [96]

    T., Haiman, Z., & Wise, J

    Scoggins, M. T., Haiman, Z., & Wise, J. H. 2023, , 519, 2155, 10.1093/mnras/stac3715

    2023 doi

  87. [97]

    2024, , 531, 851, 10.1093/mnras/stae1226

    Shimizu, T., Oogi, T., Okamoto, T., Nagashima, M., & Enoki, M. 2024, , 531, 851, 10.1093/mnras/stae1226

    2024 doi

  88. [98]

    2015, , 452, 575, 10.1093/mnras/stv1340

    Sijacki, D., Vogelsberger, M., Genel, S., et al. 2015, , 452, 575, 10.1093/mnras/stv1340

    2015 doi

  89. [99]

    D., & Lang, M

    Smith, B. D., & Lang, M. 2019, JOSS, 4, 1881, 10.21105/joss.01881

    2019 doi

  90. [100]

    D., Wise, J

    Smith, B. D., Wise, J. H., O'Shea, B. W., Norman, M. L., & Khochfar, S. 2015, , 452, 2822, 10.1093/mnras/stv1509

    2015 doi

  91. [101]

    Suazo, M., Prieto, J., Escala, A., & Schleicher, D. R. G. 2019, , 885, 127, 10.3847/1538-4357/ab45eb

    2019 doi

  92. [102]

    H., et al

    Sun, Y., Lyu, J., Rieke, G. H., et al. 2025, , 978, 98, 10.3847/1538-4357/ad973b

    2025 doi

  93. [103]

    2021, , 653, A154, 10.1051/0004-6361/202037698

    Trebitsch, M., Dubois, Y., Volonteri, M., et al. 2021, , 653, A154, 10.1051/0004-6361/202037698

    2021 doi

  94. [104]

    2017, , 470, 1121, 10.1093/mnras/stx1160

    Tremmel, M., Karcher, M., Governato, F., et al. 2017, , 470, 1121, 10.1093/mnras/stx1160

    2017 doi

  95. [105]

    J., Smith, B

    Turk, M. J., Smith, B. D., Oishi, J. S., et al. 2011, , 192, 9, 10.1088/0067-0049/192/1/9

    2011 doi

  96. [106]

    C., Schleicher, D

    Vergara, M. C., Schleicher, D. R. G., Escala, A., et al. 2024, , 689, A34, 10.1051/0004-6361/202449967

    2024 doi

  97. [107]

    P., Lovell, C

    Vijayan, A. P., Lovell, C. C., Wilkins, S. M., et al. 2021, , 501, 3289, 10.1093/mnras/staa3715

    2021 doi

  98. [108]

    E., et al

    Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, NatMeth, 17, 261, 10.1038/s41592-019-0686-2

    2020 doi

  99. [109]

    2010, , 18, 279, 10.1007/s00159-010-0029-x

    Volonteri, M. 2010, , 18, 279, 10.1007/s00159-010-0029-x

    2010 doi

  100. [110]

    2012, Science, 337, 544, 10.1126/science.1220843

    ---. 2012, Science, 337, 544, 10.1126/science.1220843

    2012 doi

  101. [112]

    2021, NatRP, 3, 732, 10.1038/s42254-021-00364-9

    Volonteri, M., Habouzit, M., & Colpi, M. 2021, NatRP, 3, 732, 10.1038/s42254-021-00364-9

    2021 doi

  102. [113]

    H., & Abel, T

    Wise, J. H., & Abel, T. 2011, , 414, 3458, 10.1111/j.1365-2966.2011.18646.x

    2011

  103. [114]

    H., Regan, J

    Wise, J. H., Regan, J. A., O'Shea, B. W., et al. 2019, , 566, 85, 10.1038/s41586-019-0873-4

    2019 doi

  104. [115]

    H., Turk, M

    Wise, J. H., Turk, M. J., Norman, M. L., & Abel, T. 2012, , 745, 50, 10.1088/0004-637x/745/1/50

    2012 doi

  105. [116]

    2011, , 412, 2603, 10.1111/j.1365-2966.2010.18080.x

    Wolcott-Green, J., & Haiman, Z. 2011, , 412, 2603, 10.1111/j.1365-2966.2010.18080.x

    2011

  106. [117]

    E., Patrick, S., Whalen, D

    Woods, T. E., Patrick, S., Whalen, D. J., & Heger, A. 2024, , 960, 59, 10.3847/1538-4357/ad054a

    2024 doi

  107. [118]

    H., Norman, M

    Xu, H., Wise, J. H., Norman, M. L., Ahn, K., & O'Shea, B. W. 2016, , 833, 84, 10.3847/1538-4357/833/1/84

    2016 doi

  108. [119]

    2003, , 592, 645, 10.1086/375810

    Yoshida, N., Abel, T., Hernquist, L., & Sugiyama, N. 2003, , 592, 645, 10.1086/375810

    2003 doi

  109. [120]

    Yu, H., & Hutson, A. D. 2024, Commun. Stat. - Theory Methods, 53, 2141, 10.1080/03610926.2022.2121144

    2024

  110. [121]

    a mer, N., Sonnenburg, S., & R \

    Zien, A., Kr \"a mer, N., Sonnenburg, S., & R \"a tsch, G. 2009, in Machine Learning and Knowledge Discovery in Databases, Vol. 2, Joint European Conference on Machine Learning and Knowledge Discovery in Databases (Springer Berlin Heidelberg), 694--709, 10.1007/978-3-642-04174-7_45

    2009 doi

  111. [122]

    Zwick, L., Mayer, L., Haemmerl \'e , L., & Klessen, R. S. 2023, , 518, 2076, 10.1093/mnras/stac3204

    2023 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.