{"id":"940fd9bd-f13a-4ef5-b689-97f5b57684ae","arxiv_id":"2504.19922","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Stacked cluster dynamical masses overestimate the true mean mass by about 20% at the virial radius, and this can be corrected with two simulation-calibrated factors.","lead":"This paper measures how much the standard Jeans equation overestimates masses when many galaxy clusters are stacked together. The authors derive two correction factors that recover the true mass profile, a correction relevant to using stacked clusters to test gravity.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The F1F2 correction is algebraically self-consistent, but the load-bearing step for observational use is the unverified assumption that galaxies trace dark matter and BCGs are freely falling centers.","rationale":"The reader's weakest_assumption is exactly the load-bearing concern: the freely falling center and collisionless galaxy tracers. The paper's central derivation is mathematically correct, and the simulation demonstration is a self-consistency check because F2 uses the true acceleration field. The measured bias is a genuine result, but its applicability to real surveys hinges on the tracer assumption. The paper states this assumption explicitly in a footnote, so it is a known limitation rather than a hidden flaw. The single simulation and the coarse finite-difference estimate of F1 are additional caveats, but they do not invalidate the existence of the bias. Therefore the reader's ACCEPT verdict remains appropriate; no change is needed, but the tracer-transfer issue should be highlighted as the main precondition for observational use.","tokens_in":14505,"tokens_out":29874,"duration_ms":292585,"concrete_test":"Apply a galaxy formation model (semi-analytic or hydrodynamical, e.g. IllustrisTNG or a SAM on the Millennium halo catalogues) to the same cluster sample, select mock galaxies as tracers, and recompute the stacked Jeans mass estimate, F1, F2, and the resulting bias profile. If the bias at r200 changes by more than about 20% relative to the dark-matter-only calibration, or if the sign of the correction reverses in any radial bin, the presented correction factors do not transfer to observed galaxy samples without a tracer-bias correction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The derivation of equations (3)-(8) is internally sound: F2 is constructed from the true gravitational acceleration, so the product F1F2 recovers the true mass profile by algebraic identity, and the simulation demonstration is a consistency check rather than an empirical test. The actual scientific content is the measured bias (1/(F1F2)-1). The place where the central claim could fail for its stated purpose is the transfer to real observations. Section 1 and the footnote explicitly equate galaxies with collisionless dark matter particles and treat the most bound particle (plus 30 h^-1 kpc core average) as a freely falling BCG. Real galaxies have a velocity bias relative to dark matter and may not be in free fall (dynamical friction, feedback, orbital offsets). If so, the stacked pressure tensor n<v_i v_j> and the a0 and F2 terms inferred from dark matter particles will not describe an observed cluster stack, and the calibrated bias factors will not correct observed dynamical masses. This is not a flaw in the simulation result, but it is the weakest link in the paper's claim that the bias 'can be used to correct the dynamical mass estimate' in surveys.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper investigates biases in dynamical mass estimates of stacked galaxy clusters obtained by applying the Jeans equation to the stacked pressure tensor. The authors derive two correction factors: F1 (equation 7), which accounts for the time derivative of the radial momentum density, and F2 (equation 8), which accounts for the acceleration of the cluster centre and for the difference between density-weighted and solid-angle-weighted gravity. Using the Millennium simulation, they stack tens of thousands of haloes in narrow mass bins and in cumulative velocity-dispersion-selected samples, measure the pressure tensor and the correction terms, and report that the naive Jeans estimate is biased high by about 20% at the virial radius and by up to a factor of two for broad mass-range samples. After applying F1F2, the estimated mass profiles match the true mean mass profiles (Figure 1).","tokens_in":14678,"tokens_out":7834,"duration_ms":82994,"significance":"The decomposition of the bias into an out-of-equilibrium term F1 and a weighting/centre-acceleration term F2 is clear and useful. Equations (3)-(8) are derived carefully, and the numerical evaluation in the simulation is a direct measurement of phase-space quantities rather than a fit to the true mass profile, with no free parameters beyond the core radius and the time step. The result that stacked dynamical masses are biased high by tens of percent, with a strong dependence on sample breadth, is important for upcoming cluster surveys and for comparisons of dynamical masses with weak-lensing masses. However, because F2 is defined so that F1F2 algebraically recovers the true mass, the agreement in Figure 1 is a consistency check rather than an independent validation. The principal quantitative content is the measured bias factor 1/(F1F2) and its dependence on sample selection, and those estimates currently lack error bars and an explicit test of the galaxy-tracer assumption.","major_comments":[{"comment":"Equation (8) defines F2 as the ratio of the true enclosed mass to the density-weighted radial acceleration, so the product F1F2 recovers M(<r) by construction when the individual terms are measured exactly. The dotted curves in Figure 1 therefore verify the numerical implementation but do not independently confirm the physical model of the bias. The statement in Section 3.3 that the recovery is 'numerical justification of our reasoning' should be tempered, and the primary result should be presented as the measured bias factor 1/(F1F2) and its sample dependence.","section":"Section 2, Eq. (8); Section 3.3"},{"comment":"No uncertainties are reported for F1, F2, or the inferred bias factors. F1 is estimated from a single finite difference between a=1 and a=1.04, and F2 is described in Section 3.2 as 'quite noisy' beyond 3r200. Since the central quantitative claims (for example, the ~20% bias at r200 and the factor-of-two bias for cumulative samples) are calibrations intended for observational use, the paper should provide bootstrap or jackknife errors over haloes, and a convergence check on the time step Δa and on the angular binning used in Appendix A3.","section":"Section 3 and Appendix A2"},{"comment":"The correction factors are evaluated from dark matter particles, with the most bound particle plus a 30 h^-1 kpc core average used as a proxy for a freely falling BCG. The abstract and Section 4 state that the bias can be used to correct dynamical mass estimates in surveys, but real galaxy tracers are not necessarily collisionless dark matter particles: galaxies can have a velocity bias relative to dark matter, and BCGs may be affected by dynamical friction or feedback-induced offsets. The transfer of F1F2 from this simulation to an observed cluster stack is therefore an untested assumption. The applicability claim should either be restricted to dark-matter tracers or the calibration should be repeated with subhalo or galaxy tracers.","section":"Section 1, footnote 1; Appendix A1"}],"minor_comments":[{"comment":"The footnote ends with the truncated phrase 'between the m.'; it should be completed (for example, 'between the mass distribution and the galaxies').","section":"Footnote 1"},{"comment":"The title contains a stray space in 'non-s tatic', and the Introduction contains the typo 'deliever'. Both should be corrected.","section":"Title and Introduction"},{"comment":"The caption uses the typos 'Mesti' and 'Mcorct'; these should be written as, for example, 'M_est' and 'M_corct'.","section":"Figure 1 caption"},{"comment":"Appendix A2 refers to '30 h^-1 Mpc cores', but the core radius defined in Appendix A1 is 30 h^-1 kpc; this appears to be a unit error.","section":"Appendix A2"},{"comment":"The text mixes units by quoting r < 0.1 h^-1 Mpc in one place and r200 in nearby statements; please define the normalization consistently throughout.","section":"Section 3.3"},{"comment":"The acknowledgments contain 'a part for the first one', which should presumably read 'apart from', and 'we have made necessary revision', which should be plural.","section":"Acknowledgments"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. The paper identifies a real systematic in stacked dynamical mass estimates and gives a correction that works in the simulation. The genuinely new bit is the F2 factor, which accounts for the accelerating cluster center and the mismatch between mass-weighted and equal-weighted acceleration. This is something the spherical one-halo treatments like Falco et al. (2013) miss.\n\nThe derivation is sound. Equations (3) through (8) are clear, and the numerical check in Figure 1 is convincing: applying F1F2 brings the estimated mass onto the true profile. That is the central claim, and it holds.\n\nWhat the simulation actually calibrates is the size of the bias: about 20% at the virial radius for narrow mass bins, and up to a factor of two for cumulative samples. The stress-test note has a point here—since F2 is built from the true acceleration, the recovery of the true mass is partly by construction. But the quantitative bias curves are genuine simulation output, not algebra, and those are what would be used.\n\nThe main soft spot is the transfer to observations. The paper explicitly models galaxies as collisionless dark matter particles and takes the most bound particle plus a 30 h^-1 kpc core average as a freely falling BCG. Real galaxies have velocity bias, dynamical friction, and feedback; if those move galaxies relative to dark matter, equation (3) does not describe an observed stack and the calibrated factors do not immediately apply. That is not a flaw in the simulation result, but it is the load-bearing assumption behind 'use this to correct dynamical mass estimates' in surveys. The paper flags it but should push it harder.\n\nMinor issues: one cosmology, one halo definition, no error bars on the bias curves, and a coarse single-step difference for the F1 time derivative. These are all modest and set the precision, not the existence, of the effect.\n\nCitation pattern looks fair, with proper engagement of Falco et al. (2013) and the earlier CNOC/Wojtak work. Who gets value: anyone doing stacked dynamics for modified-gravity tests via dynamical-versus-lensing mass comparison, and anyone interpreting stacked velocity dispersions from DESI or Euclid. It deserves serious refereeing; I would accept after a revision that adds error bars and states the galaxy-tracer caveat prominently.","headline":"A careful simulation study that identifies a real ~20% systematic in stacked dynamical masses and offers a workable correction, though the transfer to observed galaxies is the main caveat.","tokens_in":15242,"tokens_out":3359,"would_cite":true,"duration_ms":31844,"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":"The paper claims that multiplying a stacked-cluster Jeans mass estimate by a product of two correction factors, $F_1$ and $F_2$, recovers the true mean mass profile without bias, correcting a typical 20 percent overestimate at the virial…","keywords":["galaxy clusters","Jeans equation","stacked cluster samples","dynamical mass bias","correction factors F1 and F2","N-body simulations","weak lensing comparison","cluster infall"],"falsifier":"Take a real stacked cluster sample with measured velocity dispersions and weak-lensing masses; if, after applying the simulation-calibrated $F_1F_2$ correction, the ratio of dynamical mass to lensing mass at $r_{200}$ differs from unity by more than the combined measurement error, then the correction factors are incomplete or the freely-falling-centre assumption fails. A complementary simulation test is to rerun the same stacking analysis in a hydrodynamical simulation where the central galaxy experiences dynamical friction and baryonic feedback; a residual bias in the corrected mass profile would show that equation (3) needs an extra term.","tokens_in":14249,"feed_emoji":"🌌","tokens_out":10268,"duration_ms":87015,"temperature":0.7,"pith_summary":"Astronomers stack many galaxy clusters to measure an average mass profile, applying the Jeans equation to the combined velocity dispersion. This paper argues that such dynamical mass estimates are systematically too high, not only because of the known velocity-anisotropy problem but because a stacked system is not in equilibrium and because observed velocities are measured relative to cluster centres that are themselves accelerating. Using a large cosmological N-body simulation, the paper measures two correction factors, $F_1$ and $F_2$, and shows that multiplying the naive Jeans mass by $F_1F_2$ recovers the true mean mass profile without residual bias. The uncorrected bias is roughly 20 percent at the virial radius and can reach a factor of two for cumulative samples spanning a wide mass range. This matters because stacked dynamical masses are now being compared with weak-lensing masses to test gravity on cluster scales.","feed_headline":"Two factors fix the 20% bias in stacked cluster masses","feed_subtitle":"Jeans-equation masses for stacked clusters run high; multiplying by the two corrections recovers the true mean profile.","key_machinery":"The load-bearing object is the generalized Jeans equation for velocities measured in the free-falling rest frame of the cluster centre: $\\partial_t n\\langle v_i\\rangle + \\partial_j n\\langle v_i v_j\\rangle + n a_{0i} + n\\partial_i \\Phi = 0$. From it the paper forms the mass estimator $M(<r) = -r^2\\tilde{g}(r) F_1(r)F_2(r)/G$. The factor $F_1 = 1 + \\hat{r}_i\\partial_t n\\langle v_i\\rangle / (\\hat{r}_i\\partial_j n\\langle v_i v_j\\rangle)$ corrects for the neglected rate of change of momentum density, i.e. infall; it is evaluated by differencing two simulation outputs separated in time. The factor $F_2$ is the ratio of the isotropically surface-averaged radial gravity to the galaxy-density-weighted radial gravity, absorbing both the nonsphericity of individual haloes and the non-zero acceleration $a_0$ of the centre; it is evaluated with spherical grids of massless test particles placed around each halo.","core_discovery":"The central discovery is that the usual Jeans-equation mass estimate of a stacked cluster sample is biased high because two terms are neglected: the time-rate of change of radial momentum density, captured by $F_1$, and the difference between density-weighted and surface-averaged gravity, which includes the acceleration of the cluster centre, captured by $F_2$. The paper derives the corrected mass formula $M(<r) = -r^2 \\tilde{g}(r) F_1(r) F_2(r)/G$ from the generalized Jeans equation $\\partial_t n\\langle v_i\\rangle + \\partial_j n\\langle v_i v_j\\rangle + n a_{0i} + n\\partial_i \\Phi = 0$, where $a_0$ is the acceleration of the freely falling centre. In the simulation, applying $F_1F_2$ to the pressure-gradient estimate brings it onto the true mean mass profile at all radii shown. The two effects are distinct: $F_1$ matters mainly outside the virial radius and grows with halo mass, while $F_2$ is already non-negligible inside $r_{200}$ and is amplified when the stacked sample contains a wide range of halo masses.","pith_inferences":["Inference: the same two-correction scheme should apply to other stacked dynamical probes such as caustic or virial-theorem mass estimates, but the numerical values of $F_1$ and $F_2$ would need to be re-calibrated for each estimator.","Inference: applied to real survey stacks, the correction factors cannot be measured directly and must be taken from simulations, so the calibration inherits any cosmological-model dependence of the simulated halo population; mismatched concentrations would shift the correction.","Inference: the gravitational-redshift signal in stacked clusters is sensitive to the same centre acceleration and density weighting, so it may carry a similar mass-dependent bias that could be isolated with the $F_2$-style weighting comparison."],"forward_implications":["Stacked dynamical mass profiles from surveys should be multiplied by $F_1(r)F_2(r)$ before comparison with weak-lensing profiles; in the simulation this restores the true mean mass profile.","The uncorrected Jeans mass is biased high by about 20 percent at $r_{200}$ for narrow mass-selected samples, and by up to a factor of two for cumulative samples selected above a velocity-dispersion threshold.","Because $F_1$ grows with halo mass, the bias depends on sample selection; a mass estimate quoted without the stacking selection is incomplete.","Rescaling individual clusters by their velocity dispersions before stacking would largely remove the wide-mass-range part of the bias, as the paper notes for cumulative samples."],"supporting_citations":[{"why":"Supplies the cosmological N-body simulation whose haloes, particles and output pairs are stacked to measure pressure gradients and correction factors.","marker":"Springel et al. 2005"},{"why":"Supplies the friends-of-friends linking-length method used to define the halo sample.","marker":"Davis et al. 1985"},{"why":"Supplies the subhalo catalogue and most-bound-particle definition adopted as the BCG proxy for halo centres.","marker":"Springel et al. 2008"},{"why":"Provides the earlier Jeans-equation treatment of stacked clusters with radial infall used to cross-check the size of the F1 bias.","marker":"Falco et al. 2013"},{"why":"Provides the earlier measurement of the infall region around haloes whose peak radius the paper compares with its own F1 result.","marker":"Cuesta et al. 2008"},{"why":"Supplies the HEALPix angular pixelisation used to place the spherical grids of test particles that measure the isotropically averaged acceleration for F2.","marker":"Górski et al. 2005"},{"why":"Demonstrates the stacked-cluster redshift-space technique and the BCG velocity subtraction that motivates the centre-rest-frame formulation.","marker":"Wojtak et al. 2011"}],"fun_headline_variants":["Two corrections kill the 20% cluster mass bias","Stacked cluster masses: two factors fix the 20% error","F1 and F2: the duo that fixes stacked cluster masses","Bias in stacked clusters? Two terms restore true mass"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the observable cluster centre---the most bound particle of the most massive subhalo, averaged over a $30\\,h^{-1}\\,\\mathrm{kpc}$ core---is a freely falling reference frame and that galaxies behave as collisionless dark-matter particles; if real brightest cluster galaxies are dragged by dynamical friction or feedback, or if galaxies are biased tracers of the velocity field, the generalized Jeans equation no longer describes the stack and the calibrated $F_1F_2$ values will not transfer to observations.","fun_headline_variants_meta":{"raw":{"variants":["Two corrections kill the 20% cluster mass bias","Stacked cluster masses: two factors fix the 20% error","F1 and F2: the duo that fixes stacked cluster masses","Bias in stacked clusters? Two terms restore true mass"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000174,"raw_usage":{"total_tokens":1297,"prompt_tokens":974,"completion_tokens":323,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":590,"completion_tokens_details":{"reasoning_tokens":253}},"tokens_in":590,"tokens_out":323,"duration_ms":3188,"temperature":1.0,"reasoning_tokens":253,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T05:39:55.498598+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a real stacked cluster sample with measured velocity dispersions and weak-lensing masses; if, after applying the simulation-calibrated $F_1F_2$ correction, the ratio of dynamical mass to lensing mass at $r_{200}$ differs from unity by more than the combined measurement error, then the correction factors are incomplete or the freely-falling-centre assumption fails. A complementary simulation test is to rerun the same stacking analysis in a hydrodynamical simulation where the central galaxy experiences dynamical friction and baryonic feedback; a residual bias in the corrected mass profile would show that equation (3) needs an extra term.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the cosmological N-body simulation whose haloes, particles and output pairs are stacked to measure pressure gradients and correction factors."},{"cited_title":"A., Wojtak R., Hansen S","cited_arxiv_id":null,"evidence_quote":"Provides the earlier Jeans-equation treatment of stacked clusters with radial infall used to cross-check the size of the F1 bias."},{"cited_title":"J., Prada F., Klypin A., Moles M., 2008, , 389, 385","cited_arxiv_id":null,"evidence_quote":"Provides the earlier measurement of the infall region around haloes whose peak radius the paper compares with its own F1 result."},{"cited_title":"H., Hjorth J., 2011, , 477, 567","cited_arxiv_id":null,"evidence_quote":"Demonstrates the stacked-cluster redshift-space technique and the BCG velocity subtraction that motivates the centre-rest-frame formulation."}],"review_version":1}