{"id":"6977b114-f6e6-41d1-a351-580ad58ea82b","arxiv_id":"1908.11448","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A five-parameter halo occupation model, applied to dark-matter-only simulations with corrected halo masses, reproduces clustering of bright galaxies but not faint ones unless spatial, velocity, and assembly biases are included.","lead":"This paper tests whether a standard galaxy-population recipe (halo occupation distribution) can reproduce the clustering of galaxies in two large hydrodynamic simulations. It finds that correcting the dark-matter halo masses fixes high-luminosity samples, while faint galaxies require extra freedom for spatial, velocity, and assembly biases.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"High-luminosity success hinges on rank-based halo abundance matching for the HMF correction (Section 6, Fig. 8); the paper's environment-dependence check is reported but not shown, leaving the agreement potentially an artifact of the correction method.","rationale":"I agree with the reader that the rank-based abundance matching in Section 6, Figure 8 is the weakest link. The HMF correction is the lynchpin of the paper's main positive result: without it, even the high-luminosity number density is overpredicted by the HOD (Figures 3-4), and the claim that the five-parameter HOD can accurately reproduce all high-luminosity clustering statistics rests entirely on this correction. The correction is derived by ranking haloes independently in the two simulations, which is a statistical rather than physical matching. The paper acknowledges that the correction may not capture the environment dependence of the conditional HMF and reports having checked that an environment-dependent correction changes the results negligibly, but the quantitative results of that check are absent, and it is performed for Illustris only. For EAGLE, where the baryonic effect is stronger and more monotonic (Figure 8), the corresponding test is not described. This makes the high-luminosity success vulnerable to the possibility that the rank mapping is biased. A concrete particle-ID-based rematch test would settle this. I also note the secondary gap, already flagged by the reader, that the low-luminosity conclusion is inferred by removing spatial, velocity, and assembly bias from the hydrodynamic galaxies rather than by fitting an extended HOD; this does not affect my primary concern. Because the paper is a careful and useful test with otherwise plausible results, the reader's CONDITIONAL verdict is appropriate, with the condition being a demonstration of the robustness of the mass-correction scheme. Hence no change to the verdict.","tokens_in":27027,"tokens_out":17255,"duration_ms":167713,"concrete_test":"Compute the true hydro-to-DMO halo correspondence using particle IDs or spatial overlap (as in Springel et al. 2018, whose TNG mass-ratio curve the paper compares against in Table 4) instead of rank matching. Use the resulting per-halo mass ratios (with appropriate binning or smoothing) to correct the DMO halo masses, then repeat the full clustering analysis for the M^-21r samples of Illustris and EAGLE. If the p-values for wp, xi(s), n(N), P0, P1 remain above the 3-sigma level (cf. Table 3 rows 'Halo Mass Function'), the rank-based method is validated; if the p-values drop substantially, the high-luminosity agreement is an artifact of the abundance-matching scheme.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central high-luminosity claim (abstract; Section 6) depends entirely on the rank-based halo abundance matching correction shown in Figure 8. The authors pair haloes in the hydrodynamic and DMO simulations by mass rank, not by physical correspondence, and then multiply each DMO halo mass by M_hydro/M_DMO at that rank. This procedure exactly equates the global HMFs, but it can only yield the correct halo population if the rank ordering between the two simulations is preserved. Baryonic feedback, which the paper itself shows differs strongly between Illustris and EAGLE (Figure 8), is likely environment-dependent, so a DMO halo's rank may not correspond to the hydro halo that evolved from the same initial conditions. The paper states (Section 6) that the conditional HMF in Illustris depends on environment only at very high masses and that using an environment-dependent correction changes the clustering statistics negligibly, but these checks are not shown, and no equivalent check is reported for EAGLE. Because the 'accurate reproduction of all clustering statistics for the high luminosity sample' is demonstrated only after this correction, an error in the rank mapping could artificially inflate the apparent success of the vanilla HOD. The high-luminosity claim is therefore not yet robust to the choice of abundance-matching scheme.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper tests whether the standard five-parameter HOD of Zheng et al. (2007) can reproduce galaxy clustering in two hydrodynamic simulations, Illustris-2 and EAGLE, for a high- and a low-luminosity threshold sample. For each simulation and sample, the HOD parameters are fitted to the mean central and satellite occupation functions of the hydrodynamic run, and the same parameters are then applied to the DMO counterpart simulation to generate 1000 mock catalogues. Six statistics are measured on both the hydrodynamic galaxies and the mocks: galaxy number density, the projected and redshift-space two-point correlation functions, the group multiplicity function, the void probability function, and a 'singular probability function.' The authors find that the HOD systematically overpredicts the number density in every case, which they trace to baryons shifting the halo mass function to lower masses in both simulations, albeit with different mass dependence (Section 6). After correcting the DMO halo masses by rank-based abundance matching to the hydrodynamic haloes, the vanilla HOD reproduces all clustering statistics for the high-luminosity samples at roughly the 2-3 sigma level.","tokens_in":27299,"tokens_out":20054,"duration_ms":172881,"significance":"If the central high-luminosity result holds, the paper is a useful quantitative demonstration of a systematic that is usually ignored in HOD applications: baryonic feedback shifts the halo mass function in a simulation-dependent way, and the correction must be calibrated rather than assumed. The study has notable strengths. The HOD is fitted only to the occupation function, so the projected and redshift-space correlation functions, group multiplicity function, VPF, and SPF are out-of-sample predictions; the number density, by contrast, is essentially fixed by the HMF correction and the HOD fit, a point the paper should state explicitly. The use of two independent hydrodynamic codes with different numerics and feedback models gives some handle on theoretical systematics, and the suite of six statistics, including the rarely used VPF and SPF, provides diagnostic power beyond the usual two-point clustering. The paper also provides fitting functions for the baryonic mass correction (Table 4) that are directly useful to the community.","major_comments":[{"comment":"The abstract's high-luminosity claim ('After applying a correction to the halo mass function in each simulation, the HOD is able to accurately reproduce all clustering statistics') rests entirely on the rank-based abundance-matching correction of DMO halo masses (Section 6, Figure 8). That correction is only valid if the baryonic mass shift is essentially independent of environment; the manuscript says 'We have examined the conditional HMF in Illustris ... the effect of baryons on the HMF only depends on environment at very high halo masses' and that an environment-dependent correction changes the clustering statistics 'negligibly,' but no such analysis is shown, and no analogous check is reported for EAGLE. Because the correction makes the global HMF (and, together with the HOD fit to the occupation function, the galaxy number density) agree by construction, the nontrivial content of the high-luminosity claim is the spatial and velocity clustering of the corrected halo population. If the rank mapping assigns systematically wrong masses to DMO haloes in particular environments, the agreement in wp(rp) and xi(s) could be an artifact of the correction method rather than evidence for the HOD. I request that the authors (i) show the conditional mass ratio M_hydro/M_DMO in environment bins for both simulations, (ii) repeat the clustering comparison using an alternative correction, e.g., position- or particle-ID-matched haloes or an environment-dependent correction, and (iii) present the halo correlation function comparison that is currently described only in words in Section 6.","section":"Section 6, Figures 7–9"},{"comment":"The low-luminosity conclusion is established by removing effects from the hydrodynamic simulation galaxies rather than by fitting an extended HOD. The abstract's wording ('we find evidence that ... is necessary') is appropriately hedged, but the conclusions should state explicitly that the removal experiments test necessity, not sufficiency: they show that a vanilla HOD cannot match galaxies that exhibit spatial/velocity/assembly bias, but they do not demonstrate that an HOD with free bias parameters can fit the original simulation statistics. In addition, the assembly-bias removal procedure (Section 7.3) swaps galaxies between 'pairs of haloes with similar masses' without specifying the mass tolerance or the construction of the four pairings used to generate the 4000 realizations; the reported p-values could depend on these choices. Please specify the pairing criterion and test sensitivity to it. A direct check of whether a decorated HOD (e.g., Hearin et al. 2016) with bias parameters can reproduce the low-luminosity clustering would considerably strengthen the paper's central recommendation.","section":"Section 7, Table 3"}],"minor_comments":[{"comment":"The statement that for the M−21r samples 'all clustering statistics are at or better than the 2σ level' after the HMF correction is not consistent with Table 3 for EAGLE: wp(rp) has p=3.64×10^-2 and xi(s) has p=4.07×10^-2, below the conventional two-sided 2σ threshold (p≈0.0455). Please state the σ convention used or reword the claim.","section":"Section 6"},{"comment":"The post-correction number density agreement is close to guaranteed: both the HOD parameters (Section 3.2) and the HMF correction (Section 6) are derived from the same hydrodynamic simulation, so the number-density p-values after correction are a consistency check rather than an independent prediction. The paper should say this when presenting the corrected number-density p-values.","section":"Section 6"},{"comment":"The bullet list in the summary contains a grammatical error: 'It also able to accurately reproduce galaxy number density' should read 'It is also able to ...'.","section":"Section 8"},{"comment":"The caption says 'The values of these points are given in Tables 3'; this should be 'Table 3'.","section":"Figure 9 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper's central claims are interesting and, I believe, likely correct, but the robustness of the rank-based HMF correction needs to be demonstrated rather than asserted. I concur with the stress-test assessment that the unshown environment-dependence check is the key gap; requiring items (i) and (ii) in major comment 1 before acceptance is essential. The paper is within the scope of MNRAS, and the length and presentation are appropriate. The low-luminosity conclusion would be strengthened by an actual decorated-HOD fit, though I regard that as an enhancement rather than a blocker."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take on 1908.11448, Beltz-Mohrmann et al.'s test of the standard five-parameter HOD on Illustris-2 and EAGLE. The reader's conditional verdict is fair. The paper is more systematic than prior HOD tests against hydro sims: it fits the HOD solely to the occupation function, then asks whether the resulting DMO mock catalogs reproduce six clustering statistics measured on the hydro galaxies. The clustering predictions are nontrivial because they are not fit. The main new results: (1) the baryonic halo mass function shift is qualitatively different in the two simulations, EAGLE showing a near-linear mass ratio and Illustris a broken relation; (2) after correcting the DMO HMF, the vanilla HOD works for the bright sample but fails for the faint sample; (3) the faint-sample failures can be decomposed into spatial, velocity, and assembly bias. The Table 4 fitting formulae for the mass corrections are practically useful.\n\nWhere are the soft spots? The high-luminosity claim rests on the rank-based abundance-matching correction of Section 6. Pairing hydro and DMO haloes by mass rank exactly equates the global HMFs, but it only preserves the right halo population if the rank mapping is environment-independent. The paper says an environment-dependent correction changes clustering negligibly and that corrected DMO halo correlation functions agree better with hydro, but those checks are not shown, and no conditional-HMF check is reported for EAGLE. I do not think this sinks the paper: for the bright samples the clustering statistics were already mostly within 2-3 sigma before the correction; the correction mainly fixes number density. Still, the abstract's phrase 'accurately reproduce all clustering statistics' is a bit more fragile than it looks.\n\nThe low-luminosity conclusion is also indirect. Rather than fitting an extended HOD with spatial, velocity, and assembly-bias parameters, they remove those effects from the simulation galaxies and show the vanilla HOD then works. That is a valid diagnostic, but it does not demonstrate that an extended HOD can actually recover the original clustering. The abstract's wording that including such parameters 'is necessary' is stronger than what was shown. Fitting a decorated HOD to the faint samples would nail it.\n\nNo code or data release is provided, which would have helped verify the corrections, but the simulations themselves are public.\n\nAll in all, this is a solid and worthwhile paper. The flaws are in unshown robustness checks and indirect inference, not in the core logic. I would send it to a serious referee with a request to show the environment-dependent checks and, if feasible, fit an extended HOD to the faint samples. I would bring it to our reading group and would likely cite it.","headline":"A systematic and useful HOD stress test whose main claims hold up, though the key mass-correction robustness check is not shown and the faint-sample conclusion is indirect.","tokens_in":27812,"tokens_out":5243,"would_cite":true,"duration_ms":53569,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A simple five-parameter galaxy-halo recipe works for bright galaxies after correcting halo masses for baryons, but faint galaxies need extra bias parameters.","keywords":["halo occupation distribution","galaxy clustering","hydrodynamic simulations","halo mass function","assembly bias","satellite spatial bias","satellite velocity bias","counts-in-cells statistics"],"falsifier":"Compare the rank-based mass ratios shown in Figure 8 with ratios obtained by matching haloes between the hydrodynamic and dark-matter-only runs through their particle IDs; if the two mass-ratio curves disagree, the correction is not isolating a pure baryonic mass shift, and the reported clustering agreement could be a coincidence. A second check is to recompute the correction separately in high- and low-density regions and see whether a single multiplier fits both.","tokens_in":26846,"feed_emoji":"🌌","tokens_out":14139,"duration_ms":115205,"temperature":0.7,"pith_summary":"This paper asks whether the standard five-parameter halo occupation distribution (HOD) — the usual recipe for placing galaxies inside dark matter haloes — can reproduce the galaxy clustering of full hydrodynamic simulations. The answer it argues is yes for bright galaxies, and only with extra parameters for faint ones. Baryonic feedback shifts the halo mass function to lower masses, so applying an HOD calibrated on a hydrodynamic run to a dark-matter-only run produces too many galaxies. Once the dark-matter-only halo masses are rank-matched to undo that shift, the five-parameter HOD reproduces every clustering statistic tested for the bright sample in both simulations. The faint sample requires additional spatial, velocity, and assembly-bias freedom; the paper shows that removing those biases from the simulation one by one restores agreement.","feed_headline":"Five-parameter HOD matches bright galaxies after mass fix","feed_subtitle":"Baryons lower halo masses; correcting that lets the HOD fit bright samples, while faint samples need extra bias terms.","key_machinery":"The engine is the \"vanilla\" HOD: central galaxies are placed with probability $\\langle N_{\\mathrm{cen}}\\rangle = \\frac{1}{2}[1+\\mathrm{erf}((\\log M-\\log M_{\\min})/\\sigma_{\\log M})]$, and satellites are Poisson draws with mean $\\langle N_{\\mathrm{sat}}\\rangle = \\langle N_{\\mathrm{cen}}\\rangle((M-M_0)/M_1)^\\alpha$. The second piece is a rank-based mass correction: dark-matter-only haloes are sorted by mass and assigned the mass of the hydrodynamic halo at the same rank, giving each dark-matter-only halo a multiplier that recovers the hydrodynamic halo mass function. The third piece is a diagnostic protocol: replacing satellites' positions and velocities with random dark-matter particle values removes spatial and velocity bias, while swapping galaxies between haloes of similar mass removes assembly bias; the improvement in agreement after each operation identifies which missing ingredient the baseline model needs.","core_discovery":"The paper's central claim is that the standard five-parameter HOD is not inherently inadequate, but it is usually applied to the wrong haloes. In the hydrodynamic runs, baryonic feedback lowers halo masses relative to the dark-matter-only runs, shifting the halo mass function to lower masses, so the same best-fitting HOD populates too many high-mass haloes and overproduces galaxies. The paper corrects this by rank-matching dark-matter-only haloes to hydrodynamic haloes and multiplying each mass by the ratio of the two, which makes the mass functions agree. After that correction, the five-parameter HOD reproduces the projected correlation function, the redshift-space correlation function, the group multiplicity function, the void probability function, the singular probability function, and the number density for the bright galaxy sample in both simulations. The faint sample does not agree on most of these statistics until satellite spatial bias, satellite and central velocity bias, and assembly bias are successively removed from the simulation galaxies; the paper reads this as evidence that faint samples need HOD parameters for those biases, while bright samples do not. The exact baryonic correction is simulation-dependent, which the authors take as a warning that any fixed correction carries systematic uncertainty.","pith_inferences":["A direct test of the correction is to run the same rank-matching on a third hydrodynamic simulation with different feedback physics; the spread in the required mass-ratio curves would show how much of the bright-sample success is simulation-specific.","The bias-removal protocol could be inverted into a calibration: inject known spatial, velocity, and assembly-bias amplitudes into dark-matter-only mocks and check whether the proposed HOD extensions recover the injected values.","The rank-based correction assumes a single global mass shift, so it likely misses environment-dependent baryonic effects at low masses; a natural extension is a correction that also depends on halo concentration or local density.","The paper's division between bright and faint samples suggests that cosmological constraints from bright galaxy clustering are less affected by uncertain baryonic feedback, so future surveys may want to prioritize bright samples for HOD-based cosmology."],"forward_implications":["Bright galaxy samples can be modelled with the standard five-parameter HOD, provided the dark-matter-only halo mass function is first corrected for the baryonic shift.","Faint galaxy samples will require extended HODs that include spatial, velocity, and assembly-bias parameters; vanilla HOD fits to faint samples will inherit systematic bias.","Galaxy number density is a sharp diagnostic: a dark-matter-only HOD run that matches correlation functions but overproduces number density is probably missing the baryonic halo-mass correction.","Because the baryonic correction differs between the two simulations, HOD analyses of real surveys should repeat with several plausible corrections to quantify the systematic uncertainty.","The void probability function is especially sensitive to assembly bias, so surveys should measure counts-in-cells statistics in addition to correlation functions when constraining extended HOD parameters."],"supporting_citations":[{"why":"Supplies the five-parameter vanilla HOD (central error-function occupation plus satellite power law) that is the model under test.","marker":"Zheng et al. (2007)"},{"why":"Provides the mock-based fitting methodology and survey-like mock catalogues used to judge how detectable the HOD discrepancies are.","marker":"Sinha et al. (2018)"},{"why":"Describes the Illustris hydrodynamic run and its baryonic effects on the halo mass function that motivate the correction.","marker":"Vogelsberger et al. (2014a)"},{"why":"Describes the EAGLE hydrodynamic simulation used as the second test suite.","marker":"Schaye et al. (2015)"},{"why":"Earlier measurement that EAGLE hydrodynamic haloes are less massive than their dark-matter-only counterparts, supporting the mass-function shift.","marker":"Schaller et al. (2015)"},{"why":"Explains how stellar and AGN feedback in EAGLE expel baryons and lower halo masses, giving physical content to the mass-correction step.","marker":"Desmond et al. (2017)"},{"why":"Shows that undetected assembly bias biases HOD parameters inferred from clustering, motivating the assembly-bias removal test.","marker":"Zentner et al. (2014)"},{"why":"Introduces decorated HODs with assembly-bias parameters, the extension direction the faint-sample results point toward.","marker":"Hearin et al. (2016)"}],"fun_headline_variants":["HOD works for bright galaxies after halo mass fix","Faint galaxies need extra bias terms in HOD","Mass-corrected HOD matches bright; faint needs bias","Baryons shift halo masses; HOD fix works for bright"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole post-correction result rests on assuming that matching dark-matter-only haloes to hydrodynamic haloes by mass rank pairs the same halo population, so the correction only shifts halo masses and does not accidentally erase environmental or clustering information; if that pairing is wrong, the bright-sample agreement could be an artifact.","fun_headline_variants_meta":{"raw":{"variants":["HOD works for bright galaxies after halo mass fix","Faint galaxies need extra bias terms in HOD","Mass-corrected HOD matches bright; faint needs bias","Baryons shift halo masses; HOD fix works for bright"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000766,"raw_usage":{"total_tokens":3437,"prompt_tokens":1026,"completion_tokens":2411,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":642,"completion_tokens_details":{"reasoning_tokens":2343}},"tokens_in":642,"tokens_out":2411,"duration_ms":15383,"temperature":1.0,"reasoning_tokens":2343,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:15:08.948166+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the rank-based mass ratios shown in Figure 8 with ratios obtained by matching haloes between the hydrodynamic and dark-matter-only runs through their particle IDs; if the two mass-ratio curves disagree, the correction is not isolating a pure baryonic mass shift, and the reported clustering agreement could be a coincidence. A second check is to recompute the correction separately in high- and low-density regions and see whether a single multiplier fits both.","supporting_citations":[],"review_version":1}