{"id":"d1ba21a0-c548-4fc6-896c-f9a90c406bf0","arxiv_id":"1908.05362","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"The gas velocity width in DLA and GRB host galaxies falls with impact parameter at the slope predicted if metallicity traces local gravitational potential, supporting a local mass-metallicity relation out to 40-60 kpc back to z~3.","lead":"Using quasar and gamma-ray burst sightlines through 30 galaxies, this paper tests whether gas metallicity in a galaxy halo is set by the local gravitational potential rather than by the galaxy's total stellar mass. The measured decline of gas velocity width with distance from the galaxy center matches the predicted value, supporting a local mass-metallicity relation out to 40 to 60 kiloparsecs and back to redshift 3.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claim assumes rather than demonstrates that Δv90 and σem are purely gravitational (Sec. 4.2); non-gravitational gas kinematics with a radial trend could reproduce the measured -0.017 dex/kpc slope without any potential–metallicity link.","rationale":"The paper is transparent about its key assumption, and the empirical gradient is a useful new measurement. My read is that the claim “metallicity follows the local gravitational potential” is not yet established because the observable used to stand in for potential is a gas velocity width with known non-gravitational contributions. The authors' own Section 4.2 admits this assumption. The vrel test in Section 3.3 constrains the cloud geometry but not the force balance. The α0 sensitivity reinforces the conditional nature of the agreement: the slope comparison is only as strong as the adopted central relation, and the literature values of α0 disagree by up to a factor of two. None of this falsifies the claim; it means the correct verdict is CONDITIONAL, exactly as the reader concluded. I therefore recommend UNCHANGED.","tokens_in":29050,"tokens_out":8377,"duration_ms":92526,"concrete_test":"Run a forward-model test with cosmological hydrodynamic simulations (e.g., EAGLE or IllustrisTNG): select mock DLA/sub-DLA sightlines matched to the b, z, and N(HI) distribution of Table 1, compute the true gravitational potential depth at each sightline and the observable Δv90 of neutral gas plus σem from galaxy emission-line kinematics. If the synthetic log(Δv90/σem) versus b slope reproduces the simulated potential slope, the assumption is supported; if non-gravitational motions shift the slope by more than the 0.003 dex/kpc measurement error, the observed agreement can no longer be uniquely attributed to gravity.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the identification of the observed decline in log(Δv90/σem) with b with a decline in gravitational potential. Section 4.2 states the assumption verbatim: “we here assume that σem and Δv90 are purely dictated by gravity.” The empirical result (§4.2, 6σ slope −0.017 ± 0.003 dex/kpc) is a kinematic correlation; the interpretation as potential requires that Δv90 is not controlled by local turbulence, outflows, projection, or the redshift/column-density degeneracy noted in §2.1. The vrel analysis in §3.3 favours distributed sub-clouds, but an ensemble of clouds with outflow or turbulence broadening would also produce a vrel distribution uncorrelated with halo potential, so it does not independently validate the gravity assumption. The comparison slope is also tied to α0: with α0 = 1.46 the prediction is −0.015, but α0 ranges from 0.74 ± 0.21 (Neeleman et al. 2013) to 1.55 ± 0.12 (Ledoux et al. 2006); under α0 = 0.74 the required slope is −0.030, more than 4σ from the measured value. Thus the agreement is presently consistent with a gravitationally dictated potential, but the same data could be produced by non-gravitational kinematics with a radial dependence; the central claim is conditional on an untested physical assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper tests the hypothesis of Arabsalmani et al. (2015) that the mass-metallicity relation of DLA and GRB host galaxies is local, i.e., that metallicity follows local gravitational potential, by predicting that log(Δv90/σem) declines with impact parameter with slope -0.015 dex/kpc, derived as -Γ[M/H]/α0 with Γ[M/H] = -0.022 and α0 = 1.46. The authors compile 21 QSO-DLA/sub-DLA host galaxies with σem, b, Δv90, and N(HI), plus 9 GRB-DLAs, measure ΓΔv90 = -0.017 ± 0.003 dex/kpc (6σ) out to 54 kpc, and find agreement with the prediction. They also analyze the distribution of vrel and conclude that DLA absorbers are ensembles of sub-clouds distributed through the halo. The central interpretation is that both metallicity and Δv90 follow the local gravitational potential from z = 0.16 to z = 3.15.","tokens_in":29354,"tokens_out":6161,"duration_ms":56643,"significance":"If correct, this is a significant unification: it turns the absorption-line Δv90-[M/H] relation and the emission MZ relation into two projections of a local potential-metallicity relation, extending local MZ results to z ~ 3 and to impact parameters well beyond galaxy disks. The paper's strengths are the careful homogeneous reprocessing of heterogeneous archival data, the explicit falsifiable prediction, and the new direct measurement of ΓΔv90; the vrel analysis also provides independent information on absorber substructure. The claim, however, is conditional on the assumption that Δv90 and σem are purely gravitational, which is not independently established, and on the adopted slope α0; the quantitative agreement is therefore less secure than the abstract suggests.","major_comments":[{"comment":"The central claim that the measured slope of log(Δv90/σem) versus b represents a gravitational potential gradient rests on the assumption stated verbatim in §4.2: 'we here assume that σem and Δv90 are purely dictated by gravity.' The paper offers no independent test of this assumption. The vrel analysis in §3.3 favours a model of independent sub-clouds spread along the sightline, but an ensemble of clouds whose velocity dispersion is set by turbulence or outflow kinematics with a radial dependence would also produce a vrel distribution uncorrelated with the halo potential; it therefore does not validate the gravity-only interpretation. Without a discriminating test (for example, comparing low-ion and high-ion kinematics, checking for a correlation with star-formation or outflow indicators, or contrasting with simulations that include feedback), the 6σ slope is a kinematic correlation whose identification with the potential well is plausible but unproven. The title and abstract overstate the certainty of the physical interpretation.","section":"§4.2; supporting analysis in §3.3"},{"comment":"The predicted slope is not parameter-free: it is computed as ΓΔv90 = -Γ[M/H]/α0 using α0 = 1.46 and Γ[M/H] = -0.022 from previous fits (Møller et al. 2013; Christensen et al. 2014), and the present test sample overlaps the sample used to derive Γ[M/H]. The sensitivity to α0 is severe: with the alternative value α0 = 0.74 ± 0.21 (Neeleman et al. 2013), which the paper itself cites in §4.1, the required slope is -0.030, about 4.3σ away from the measured -0.017 ± 0.003. The agreement in §3.1 and §4.2 is therefore partly inherited from the adopted prior chains. I ask the authors to present the predicted slope as a function of α0 over the full published range, and to quantify how much of the 'confirmation' depends on that choice.","section":"§4.1, Eq. (6); §3.1"},{"comment":"The claimed 6σ detection of ΓΔv90 rests on the combined sample that includes GRB-DLAs, all assigned the same median impact parameter b = 1.0 kpc, and on the average log(Δv90/σem) of nine GRB sightlines. The QSO-DLA-only fits shown in Fig. 2 give slopes of -0.011 and -0.020 with no quoted significance, so it is unclear whether the slope is detected at high significance without the GRB-DLA anchor. Because GRB-DLAs probe a different geometry along the sightline and are subject to different selection effects, the authors should report the slope, uncertainty, and significance for the QSO-DLA sample alone, and should discuss how excluding the b = 86 kpc object (which changes the low-redshift fit in Fig. 6) affects the claim that the slope extends to 40-60 kpc.","section":"§4.2 and Fig. 2; §2.1"},{"comment":"The conclusion that there is no redshift evolution, and hence that the relation holds 'since z=3', is weakly supported. The sample has a degeneracy between redshift and N(HI): the 11 low-redshift hosts are mostly sub-DLAs and the 10 high-redshift hosts are mostly DLAs (§2.1). In Fig. 6 the low- and high-redshift slopes agree only after excluding the single b = 86 kpc object, and with 10 high-redshift objects the test has low power. 'No evidence for redshift dependence' is a defensible statement, but the abstract's 'since z = 3' should be softened to 'over the sampled redshift range' unless a more powerful joint fit in b, z, and N(HI) is added.","section":"§3.4 and §2.1"}],"minor_comments":[{"comment":"The text refers to 'fit a line to the data points of log(Δv90/σem) vs. b in Fig. 3', but the linear fit is shown in Fig. 2; Fig. 3 shows projected profiles. The cross-reference should be corrected.","section":"§4.2"},{"comment":"The caption says 'Predicted slope and average GRB-DLA are included as in Fig. 1'; it should refer to Fig. 2.","section":"Fig. 3 caption"},{"comment":"The sentence 'In Fig. B1(a) vi plot the standard Δv90-[M/H] relation' contains a typo: 'vi' should be 'we'.","section":"Fig. B1 caption"},{"comment":"The phrase 'steep log scale slope of -0.015 dex/kpc' is awkward and could be rephrased as a 'logarithmic slope of log(Δv90/σem) with impact parameter of -0.015 dex/kpc'.","section":"Abstract"},{"comment":"Several vrel values are listed without errors (e.g., 0152-2001, 1436-0051A, 1228-1139); a note on how these are treated in the Fig. 5 analysis would improve reproducibility.","section":"Table 1"}],"recommendation":"major_revision","confidential_remarks":"To the editor: this is a genuinely interesting and potentially important paper, but the 'local potential' interpretation hinges on an untested physical assumption and on the adopted value of α0; the abstract is more assertive than the internal evidence strictly supports. The direct measurement of ΓΔv90 and the careful data compilation are strengths, and the issues raised are addressable with robustness tests and a moderated interpretation. Fig. 3 and §4.3 lean on the companion paper by Christensen et al. (2019), so the referee should be aware of that dependency when assessing completeness."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe genuinely new thing here is a clean observational test: predict that log(Δv90/σem) declines with impact parameter at roughly -0.015 dex/kpc if the velocity width traces local potential, and then measure a slope of -0.017±0.003 dex/kpc (6σ) out to ~50 kpc. That measurement is new, and the sample compilation is careful—plenty of reprocessing details in the appendices, and the data presentation is transparent. The vrel analysis favoring distributed sub-clouds over a single bulk-moving cloud is a nice extra.\n\nCredit where due: the prediction is falsifiable, the sample spans z=0.16 to z=3.15, and the authors are honest about what they assume. The direct measurement of Γ_Δv90 is genuinely independent of the earlier samples used to build the prediction, and it agrees.\n\nThe soft spots are real but proportionate. The load-bearing step is the assumption, stated verbatim in Section 4.2, that σem and Δv90 are purely gravitational. The measured gradient is a kinematic correlation; interpreting it as a potential gradient depends on that assumption, and non-gravitational gas with a radial trend could mimic the slope. Second, the predicted slope uses α0=1.46, which is fitting-method dependent. With α0=0.74 from Neeleman et al. (2013), the prediction becomes -0.030, more than 4σ from the measured value. That sensitivity deserves a direct address. Third, the sample is small (21 QSO-DLA + 9 GRB) with a known redshift–NHI degeneracy, and the overlap with the sample used to derive Γ[M/H] means the cancellation is not fully independent. The authors acknowledge most of this, which helps.\n\nOn balance, the central argument holds up as a conditional claim: if the gas kinematics are gravitational, the data confirm the local-potential picture out to large radius. The title's 'since z=3' is a bit broad given the degeneracies, but the redshift coverage is genuinely wide.\n\nThis paper deserves a serious referee. It is exactly the kind of archival, hypothesis-driven work that moves the subfield forward, even if the grand interpretation needs care. I would cite it for the measured gradient, and I'd bring it to a reading group focused on CGM physics.","headline":"A careful, falsifiable test of the local-potential hypothesis that delivers a new 6σ gradient measurement, but the interpretation leans on an untested gravitational assumption.","tokens_in":29917,"tokens_out":1789,"would_cite":true,"duration_ms":17540,"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":"This paper tests and confirms that gas metallicity tracks the local gravitational potential rather than total stellar mass, with the relation unchanged from z=0.16 to z=3.15.","keywords":["mass-metallicity relation","damped Lyman-alpha absorbers","gamma-ray burst hosts","circumgalactic medium","gravitational potential","metallicity gradients","galaxy evolution","high-redshift galaxies"],"falsifier":"Measure spatially resolved rotation curves or stellar kinematics for the DLA/sub-DLA hosts and compare the absorption-line width at each impact parameter with the independently determined circular velocity at that radius; if $\\Delta v_{90}$ does not scale with the local potential depth, the central claim would fail. A simpler check is to extend the sample beyond 60 kpc and see whether the $\\log(\\Delta v_{90}/\\sigma_{\\rm em})$ gradient and the metallicity gradient flatten together as the prediction requires.","tokens_in":28791,"feed_emoji":"🔭","tokens_out":11461,"duration_ms":102324,"temperature":0.7,"pith_summary":"The mass-metallicity relation is usually read as a global link between a galaxy's total stellar mass and the metal content of its gas. This paper argues that the underlying relation is local: at any point along a sightline through a galaxy halo, the gas metallicity is set by the depth of the gravitational potential at that radius, not by the galaxy's total mass. To test this, the authors show that the absorption-line width $\\Delta v_{90}$ must decline outward at about $-0.015$ dex kpc$^{-1}$ in log space, exactly compensating the measured metallicity gradient. Compiling 21 DLA/sub-DLA host galaxies plus 9 gamma-ray burst hosts, they find the predicted decline out to 40-60 kpc and no change from $z=0.16$ to $z=3.15$. If right, the classical relation is a projection of a more fundamental local potential-metallicity relation.","feed_headline":"Gas metallicity has traced galaxy gravity since z=3","feed_subtitle":"Halo gas velocity widths fall with radius at the rate that keeps metallicity tied to the local potential from z=0.16 to z=3.15.","key_machinery":"The load-bearing object is a cancellation identity: with metallicity gradient $\\Gamma_{[M/H]}=-0.022$ dex kpc$^{-1}$ and the slope $\\alpha_0=1.46$ of the $\\Delta v_{90}$-metallicity relation, the hypothesis requires $\\Gamma_{\\Delta v_{90}}=\\Gamma_{[M/H]}/\\alpha_0\\approx-0.015$ dex kpc$^{-1}$. The observable is $\\log(\\Delta v_{90}/\\sigma_{\\rm em})$ versus $b$, where $\\Delta v_{90}$ is the velocity width containing 90 percent of the low-ion absorption optical depth and $\\sigma_{\\rm em}$ is the velocity dispersion of the host's emission lines; normalizing by $\\sigma_{\\rm em}$ removes the leading mass dependence. In the paper's equation 7 the two measured gradients cancel, leaving the central relation with a scatter that must arise from something other than impact parameter.","core_discovery":"The paper's central claim is that both metallicity and the absorption velocity width $\\Delta v_{90}$ follow the local gravitational potential in galaxy halos, and that this has been true since at least $z\\approx3$. The quantitative result is a decline in the normalized width $\\log(\\Delta v_{90}/\\sigma_{\\rm em})$ with impact parameter $b$ at $\\Gamma_{\\Delta v_{90}}=-0.017\\pm0.003$ dex kpc$^{-1}$, consistent with the value $-0.015$ dex kpc$^{-1}$ that makes the impact-parameter terms cancel in the $\\Delta v_{90}$-metallicity relation. That cancellation means galaxies observed at random impact parameters are shifted along the same underlying relation rather than off it, so the observed mass-metallicity relation is not biased by unknown sightline radii. The paper finds no dependence of the slope on redshift, stellar mass, or $N({\\rm H\\,I})$, and interprets the lack of correlation between the absorber-host velocity offset and $\\Delta v_{90}$ as evidence that DLA systems are ensembles of independent clouds spread through the halo rather than single bulk-moving clouds.","pith_inferences":["Inference: A direct test of causality would be to compare resolved metallicity maps of nearby galaxies with maps of the reconstructed gravitational potential from stellar kinematics; the local relation predicts that metallicity should track the potential at least as tightly as it tracks stellar mass surface density.","Inference: If the two gradients are coupled, then expanding absorber samples beyond $b\\approx60$ kpc should show both the velocity-width and metallicity gradients flattening together; a decoupling there would indicate that the inner-CGM cancellation is not universal.","Inference: The paper's interpretation assumes the absorption width is a clean dynamical tracer, but comparing $\\Delta v_{90}$ with an independent halo-mass estimate (for example from rotation or lensing) would show whether the width tracks the local potential or the total mass; if it tracks total mass, the local-potential reading needs revision."],"forward_implications":["The mass-metallicity relation can be read as a projection of a local potential-metallicity relation, so absorption-selected samples need no impact-parameter correction to recover the underlying relation.","The relation has been stable from $z=0.16$ to $z=3.15$, meaning the local potential-metallicity link was already in place when the universe was roughly one-fifth of its present age.","Because the impact-parameter terms cancel, the residual scatter in the $\\Delta v_{90}$-metallicity relation is not caused by random sightline radii; it must have another physical origin.","The velocity offset between absorption and emission being uncorrelated with $\\Delta v_{90}$ favors the picture in which DLA complexes are ensembles of independent clouds spread along the pencil beam, with the total velocity width set by the halo potential.","If the potential gradient flattens at large radius, the metallicity gradient must flatten the same way to keep the cancellation, predicting a coupled flattening beyond about 60 kpc."],"supporting_citations":[{"why":"It proposes that metallicity follows the local gravitational well depth and supplies the gamma-ray-burst comparison that motivates the test.","marker":"Arabsalmani et al. (2015)"},{"why":"It provides the average metallicity gradient used to derive the predicted velocity-width slope.","marker":"Christensen et al. (2014)"},{"why":"It establishes the velocity-width-metallicity relation and its slope, the basis for the cancellation ratio.","marker":"Ledoux et al. (2006)"},{"why":"It calibrates the velocity-width relation to the stellar-mass-metallicity relation, supplies the adopted slope and redshift evolution, and gives the scatter-fitting method used here.","marker":"Møller et al. (2013)"},{"why":"It supplies an independent measurement of the velocity-width-metallicity slope and redshift dependence, used for comparison.","marker":"Neeleman et al. (2013)"},{"why":"It reports the independent metallicity gradient and contributes host-galaxy measurements to the compiled sample.","marker":"Rhodin et al. (2018)"},{"why":"It gives the median gamma-ray-burst impact parameter used to place the GRB-DLA average point.","marker":"Lyman et al. (2017)"},{"why":"It supplies the velocity-width and emission-dispersion measurements for the GRB-DLA sample used as the small-impact-parameter anchor.","marker":"Arabsalmani et al. (2018)"},{"why":"It identifies the AGN-dominated DLA host that tests whether outflow-driven absorption still follows the gravitational potential.","marker":"Rudie et al. (2017)"}],"fun_headline_variants":["Galaxy halos keep metallicity tied to gravity since z=3","Metallicity rides galactic gravity for 11 billion years","Gas metallicity obeys local gravity in halos out to 60 kpc","Steep halo velocity slope locks metallicity to potential","From z=3 to now: metallicity follows local gravity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the absorption width $\\Delta v_{90}$ measures the depth of the local gravitational potential and the emission width $\\sigma_{\\rm em}$ measures the central potential; if those widths are instead set by turbulence, outflows, or line-of-sight geometry, the observed decline with radius would not prove that metallicity follows gravity.","fun_headline_variants_meta":{"raw":{"variants":["Galaxy halos keep metallicity tied to gravity since z=3","Metallicity rides galactic gravity for 11 billion years","Gas metallicity obeys local gravity in halos out to 60 kpc","Steep halo velocity slope locks metallicity to potential","From z=3 to now: metallicity follows local gravity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000216,"raw_usage":{"total_tokens":1496,"prompt_tokens":1073,"completion_tokens":423,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":689,"completion_tokens_details":{"reasoning_tokens":334}},"tokens_in":689,"tokens_out":423,"duration_ms":4358,"temperature":1.0,"reasoning_tokens":334,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:15:45.715683+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure spatially resolved rotation curves or stellar kinematics for the DLA/sub-DLA hosts and compare the absorption-line width at each impact parameter with the independently determined circular velocity at that radius; if $\\Delta v_{90}$ does not scale with the local potential depth, the central claim would fail. A simpler check is to extend the sample beyond 60 kpc and see whether the $\\log(\\Delta v_{90}/\\sigma_{\\rm em})$ gradient and the metallicity gradient flatten together as the prediction requires.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It provides the average metallicity gradient used to derive the predicted velocity-width slope."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It establishes the velocity-width-metallicity relation and its slope, the basis for the cancellation ratio."},{"cited_title":"M., Prochaska , J","cited_arxiv_id":null,"evidence_quote":"It supplies an independent measurement of the velocity-width-metallicity slope and redshift dependence, used for comparison."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It reports the independent metallicity gradient and contributes host-galaxy measurements to the compiled sample."},{"cited_title":"D., Levan , A","cited_arxiv_id":null,"evidence_quote":"It gives the median gamma-ray-burst impact parameter used to place the GRB-DLA average point."},{"cited_title":"C., Newman , A","cited_arxiv_id":null,"evidence_quote":"It identifies the AGN-dominated DLA host that tests whether outflow-driven absorption still follows the gravitational potential."}],"review_version":1}