{"id":"42416432-2f71-4eca-8bab-05affb506372","arxiv_id":"2411.10514","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"In high-resolution MHD simulations of gas-rich galactic disks, magnetic pressure supports dense gas, raises cold gas fractions by up to 40%, and produces a steady-state magnetic field that grows roughly as the square root of gas surface density.","lead":"This paper presents a new suite of high-resolution magnetohydrodynamic simulations of star-forming galactic disks, finding that magnetic fields support the dense gas and reduce violent star formation bursts. A relation between magnetic field strength and gas surface density emerges, suggesting supernova-driven turbulence sets the field in dense gas.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Dynamo saturation is not numerically converged: B100 drops by ~2x with CG cleaning (App. C), magnetic-to-kinetic ratios reach 40% (Fig. 13), and beta_th,100 depends on resolution (Fig. 5), so the B-Sigma normalization is not yet secure.","rationale":"The reader's weakest assumption identified the small-scale dynamo saturation as the load-bearing point, and I agree. The paper's initial-condition insensitivity tests (Sec. 4.1, Table 2) are genuine support: they show B100 forgets B0 within a factor of about 2. However, forgetting initial conditions is necessary but not sufficient for a physical saturation level. Appendix C provides the sharpest available evidence of numerical sensitivity: at fixed resolution, switching to the CG divergence cleaner reduces saturated B100 by a factor of ~2 in M10. Additional resolution dependence appears in Fig. 5 (beta_th,100 lower at higher resolution) and Fig. 13 (magnetic-to-kinetic energy ratio up to 40%, far above the few-percent saturation in other SN-driven SSD simulations). The authors explicitly call for a future study of the small-scale dynamo, so this is not a manufactured concern. Since B100 feeds the abstract's 10-40 microG range and the B-Sigma normalization, and since magnetic support scales as B^2, even a factor-of-2 over-amplification materially changes the quantitative claims while possibly preserving the qualitative picture. The proposed test is targeted and inexpensive given the existing CG implementation: re-running the three highest-Sigma models with CG directly determines whether the M10 factor-of-2 shift is universal. If universal, the normalization and the 'magnetic support' strength are overestimated; if not, the concern is retired. This does not move the reader's CONDITIONAL verdict; it reinforces it. I also note the paper's other strengths: the PRFM consistency check, KS relation agreement, and the HV control runs for pure-hydrodynamical comparison are well executed and survive this concern.","tokens_in":23182,"tokens_out":8462,"duration_ms":84142,"concrete_test":"Extend the constrained-gradient (CG) divergence-cleaning test already applied to M10 in Appendix C to M40, M60, and M80 at mg=10 Mo, computing B100 and the midplane magnetic-to-kinetic energy ratio over 200-500 Myr in both the CG and default schemes. If the CG runs reproduce the factor-of-2 reduction in B100 and bring the energy ratio down from ~40% toward the few-percent level found in other SSD simulations, then the dense-gas field strengths and the B-Sigma normalization are scheme-dependent overestimates and the abstract's 10-40 microG statement needs revision. If the factor-2 shift does not persist at higher Sigma_gas or the energy ratio remains ~40%, the numerical-convergence objection is retired and the dynamo interpretation is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that B100 be a physical quasi-steady state set by SN-driven small-scale dynamo, not by initial conditions or numerics. The paper convincingly shows insensitivity to B0 (Sec. 4.1, Fig. 9, Table 2), but this does not establish physical saturation: Appendix C reports that replacing the default divergence cleaning with the constrained-gradient (CG) method lowers saturated B100 by a factor of ~2 at identical resolution (M10, Fig. 15). Fig. 5 shows a systematic resolution trend in beta_th,100, and Fig. 13 shows midplane magnetic-to-kinetic energy ratios up to 40%, which the authors themselves note is in tension with other SN-driven dynamo simulations (Pakmor et al. 2017; Rieder & Teyssier 2017; Gent et al. 2021, 2023) and 'necessitates a more in-depth future study' (Appendix A). Because magnetic pressure enters as B^2, a factor-2 scheme shift changes beta_th,100 by roughly 4x, directly affecting the 'magnetic support' and 10-40 microG summary claims as well as the normalization of the B-Sigma relation in Fig. 10. The B-Sigma slope may be robust, but the normalization and physical interpretation remain unsettled until convergence is demonstrated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper introduces GHOSDT, a suite of galactic-patch magnetohydrodynamical simulations with star-by-star SN feedback, spanning initial gas surface densities of 4-100 M_sun pc^-2. The authors report that magnetic fields stabilize high-column-density disks against strong initial collapse, increase cold gas fractions by up to 40%, reduce disk scale height and star-formation burstiness, and produce a steady-state magnetic field in dense gas (n > 100 cm^-3) of 10-40 microG that is insensitive to the initial field strength and roughly follows B proportional to Sigma_gas^0.5. They also compare their Sigma_SFR versus P_DE relation with the PRFM framework of Ostriker and Kim (2022) and find consistency within about 30% for the equilibrium runs.","tokens_in":23417,"tokens_out":4287,"duration_ms":41290,"significance":"If the dynamo saturation is physical, the paper provides a plausible local, turbulence-based origin for magnetic fields in dense gas of high-redshift disk galaxies, with concrete predictions (10-40 microG, B-Sigma_gas dependence) that can be tested against Zeeman and dust-polarization observations. Strengths include the wide range of gas surface densities, simulation times of 0.5-1 Gyr, explicit variation of the initial magnetic field B0 over up to four orders of magnitude, and the external PRFM benchmark against observations and TIGRESS-NCR. The main weakness is that the saturated field strength is not yet shown to be numerically converged: the constrained-gradient test in Appendix C changes B100 by a factor of about two at fixed resolution, and the magnetic-to-kinetic energy ratio reaches 40%, in tension with other SN-driven dynamo simulations as the authors acknowledge. The significance of the central claims is therefore conditional on establishing numerical convergence of the dynamo saturation.","major_comments":[{"comment":"The central claim that B100 is a physical quasi-steady state set by SN-driven small-scale dynamo is not yet established, because the saturated value is scheme-dependent: replacing the default Powell/Dedner cleaning with the constrained-gradient reconstruction reduces B100 by a factor of approximately two at identical resolution in the M10 test (Fig. 15, right column). Since magnetic pressure enters as B^2, this factor-of-two change alters beta_th,100 by roughly a factor of four and directly affects the magnetic-support statements in Sections 3.3.2 and 3.3.3, as well as the normalization of the B100-Sigma_gas relation in Fig. 10. The insensitivity to B0 shown in Section 4.1 is necessary but not sufficient; the authors should demonstrate that the saturation is converged with respect to resolution and divergence-control method, for example by running CG variants for several Sigma_gas,0 values and by a resolution study of B100 and of the magnetic-to-kinetic energy ratio in Fig. 13.","section":"Appendix C, Fig. 15"},{"comment":"The abstract and Section 5 claim an approximately B proportional to Sigma_gas^0.5 relation, but no quantitative fit is reported: the power-law index, normalization, and scatter are not given, and Fig. 10 shows only a visual guide. Given the factor-of-about-two scatter in B100 visible in Table 2 and Fig. 9, the authors should provide a fitted slope with uncertainty, separate fits for the 1-4 M_sun and 10 M_sun resolution samples, and a comparison with observational B-Sigma_gas constraints. Without this, the scaling claim is not testable and its uncertainty cannot be assessed.","section":"Sec. 4.2, Fig. 10"},{"comment":"The resolution trend in beta_th,100 and beta_turb is in the direction of increasing magnetic importance at higher resolution, which means the conclusion that magnetic pressure is dynamically important in dense gas is not yet converged. The paper notes the trend but does not quantify it or demonstrate that it saturates; a convergence study, or at least an explicit statement of the systematic uncertainty on beta_th,100 arising from resolution, is needed before the magnetic-support claim can be considered robust.","section":"Sec. 3.3.2, Fig. 5"}],"minor_comments":[{"comment":"The text says that for the 2D histogram the runs with Sigma_gas,0 >= 40 M_sun pc^-2 are substituted with group HV runs, but the caption of Fig. 3 states that groups H and M are used; please clarify which runs are actually plotted.","section":"Sec. 3.1, Fig. 3"},{"comment":"The caption reads 'the ratio median ratio between the total magnetic and kinetic energy'; the duplicated word should be corrected.","section":"Appendix A, Fig. 13"},{"comment":"The phrase 'we supplement the simulations in Table 1 with an additional set of simulations' is redundant and could be shortened.","section":"Sec. 2.4"},{"comment":"The M60 model is run with m_g = 4 M_sun and is grouped with the 1 M_sun models in the analysis; please state explicitly how this affects the resolution comparisons, since Fig. 10 and Fig. 5 plot the 1-4 M_sun and 10 M_sun samples separately.","section":"Table 1"},{"comment":"The convergence criterion for a quasi-steady magnetic field is described as a visual check with a 200 Myr moving average; it would be more reproducible to specify a quantitative threshold, for example on the time derivative of the moving average relative to its mean.","section":"Sec. 2.3"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and reports a substantial new simulation suite. My main concern is the numerical convergence of the small-scale dynamo saturation, which is load-bearing for the B-Sigma normalization, the magnetic-support claims, and the physical interpretation. The CG and energy-partition tests in the appendices show a sensitivity that the authors themselves acknowledge. I would support publication if the authors add a focused convergence study (resolution and divergence-cleaning variants across several surface densities) and a quantitative fit to the B-Sigma relation. If those tests show that the factor-of-two scheme sensitivity persists, the paper would need to be reframed as reporting a preliminary or resolution-dependent result rather than a converged physical prediction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about this paper. First, it is a competent, honest simulation study that makes a real extension of the star-by-star galactic patch approach into the gas-rich, high-column-density regime where pure hydro runs blow themselves apart. Second, the headline B-Sigma relation is on shakier ground than the paper's tone suggests: the slope is probably fine, but the normalization shifts by about a factor of two when you change the divergence-cleaning scheme, and the magnetic-to-kinetic energy ratio reaches 40%, which the authors themselves concede is in tension with other SN-driven dynamo simulations.\n\nWhat is actually new: GHOSDT takes the Hu et al. (2021) setup, adds ideal MHD, and runs gas surface densities from 4 to 100 Msun/pc2. The ensemble shows convincingly that B100 is insensitive to the initial field strength over four orders of magnitude, and that magnetic pressure is dynamically important in dense gas at high column density: cold gas fraction up by 40%, disk scale height down by up to a factor of two, burstiness reduced. The agreement with the PRFM framework of Ostriker & Kim (2022) is a useful external benchmark. These are real results, and the paper earns its place.\n\nThe paper's best feature is its honesty. The resolution trend in beta_th,100 is shown in Fig. 5, the CG-cleaning test lowering B100 by a factor of two is right there in Appendix C, and the 40% magnetic-to-kinetic ratio is discussed candidly in Appendix A as needing future work. A reader who wants to see the caveats can find them without digging. That said, those caveats are load-bearing for the normalization claim, not cosmetic. Because magnetic pressure enters as B^2, a factor-of-two scheme shift changes beta by roughly four. So the statement that B in dense gas is set by SN-driven turbulence to 10-40 microgauss is directionally supported but numerically not yet settled.\n\nThe other soft spots are minor by comparison: no simulation data or config files are released, and the B-Sigma power law is presented as approximate without a quantified fit or scatter. Both are fixable in a revision.\n\nWho this is for: the galactic patch and ISM simulation community, and observers working on CO and [CII] tracers in high-redshift galaxies, where the magnetic support conclusion matters. I'd bring it to reading group, and I'd cite it for the qualitative result, but I'd be careful quoting the B-Sigma normalization.\n\nRecommendation: send it to peer review. The referee should push for convergence tests on the dynamo saturation and a quantified B-Sigma fit, but the paper deserves the time.","headline":"Strong simulation paper with a likely-robust B-Sigma slope and a normalization that is not yet converged; worth refereeing seriously, but the dynamo saturation tests need to be addressed before the 10-40 microgauss claim is quoted.","tokens_in":24030,"tokens_out":2553,"would_cite":true,"duration_ms":25089,"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":"In gas-rich disks, supernova-driven turbulence, not initial conditions, sets the magnetic field of dense star-forming gas to 10-40 microgauss, following roughly B proportional to the square root of gas surface density.","keywords":["interstellar medium","magnetohydrodynamics","small-scale turbulent dynamo","supernova feedback","galactic patch simulations","Kennicutt-Schmidt relation","magnetic fields in galaxies","cold gas fraction"],"falsifier":"Rerun a fixed-surface-density case such as M40 or M60 with progressively more accurate magnetic-field treatment: the constrained-gradient divergence cleaning already tested in Appendix C, higher resolution, and possibly non-ideal resistivity, then check whether the saturated dense-gas field stays near 10--40 microgauss; the paper already reports that divergence cleaning alone lowers $B_{100}$ by a factor of about two, so a continued decline would refute convergence. Observationally, Zeeman or dust-polarization measurements of molecular clouds across galaxies spanning a decade in gas surface density would test the predicted $B \\propto \\Sigma_{\\rm gas}^{0.5}$ trend directly.","tokens_in":22922,"feed_emoji":"🧲","tokens_out":17474,"duration_ms":148180,"temperature":0.7,"pith_summary":"This paper presents GHOSDT, a suite of high-resolution magnetohydrodynamic simulations of a one-kiloparsec patch of a gas-rich star-forming disk, spanning gas surface densities from 4 to 100 solar masses per square parsec. The authors set out to show that magnetic fields shape the interstellar medium of such disks: magnetic pressure supports the dense gas, and the field strength in star-forming gas is set locally by supernova-driven turbulence rather than by initial conditions or cosmic processes. They find a quasi-steady dense-gas field of 10--40 microgauss that follows roughly $B \\propto \\Sigma_{\\rm gas}^{0.5}$, insensitive to a four-orders-of-magnitude range of seed fields. If the claim holds, magnetic fields in star-forming gas become a predictable function of local column density, and any simulation of gas-rich, high-redshift disks must include them to keep the disk from blowing itself apart.","feed_headline":"Supernovae, not cosmic seeds, set gas-rich disks' magnetic fields","feed_subtitle":"Simulations link dense-gas field strength to gas surface density, yielding 10-40 microgauss and reshaping the cold ISM.","key_machinery":"The mechanism that carries the argument is the supernova-driven small-scale turbulent dynamo embedded in a self-regulated feedback cycle: supernovae inject kinetic energy, turbulence stretches and amplifies a weak seed field, and magnetic energy saturates at a level set by the rate of turbulent injection, which is in turn set by the gas surface density through the star formation rate. The numerical machinery is the GHOSDT suite itself, built on the Gizmo meshless finite-mass method with ideal MHD and Powell plus Dedner divergence cleaning, time-dependent hydrogen chemistry, a Jeans-based stochastic star formation recipe over a Kroupa IMF, and star-by-star supernova and photoionization feedback. The load-bearing test inside the machinery is dynamo convergence: the saturated dense-gas field must be independent of the value and geometry of the initial field, and the paper argues it is, to within a factor of about two across four orders of magnitude in seed strength.","core_discovery":"On the paper's own terms, the central discovery is that a magnetized, self-regulated interstellar medium settles into a quasi-steady state in which the magnetic field of dense star-forming gas is set by the gas surface density. Varying the initial field strength by up to four orders of magnitude, and replacing the uniform seed field with a randomized one, changes the saturated field in gas denser than $100\\;\\mathrm{cm}^{-3}$ by less than a factor of about two. The saturated field follows approximately $B \\propto \\Sigma_{\\rm gas}^{0.5}$, rising from roughly $7\\;\\mu$G at $\\Sigma_{\\rm gas}\\approx 4\\;M_\\odot\\,\\mathrm{pc}^{-2}$ to roughly $30\\;\\mu$G at $\\Sigma_{\\rm gas}\\approx 60\\;M_\\odot\\,\\mathrm{pc}^{-2}$, with 10--40 microgauss across the suite, comparable to fields measured in Galactic molecular clouds. The authors interpret this as the signature of a supernova-driven small-scale dynamo, in which turbulence stretches and amplifies a weak seed field. The amplified field feeds back on the gas: relative to pure-hydrodynamical runs at the same column density, it raises the cold gas fraction by up to 40 percent, halves the disk scale height, smooths the burstiness of star formation, and lets a quasi-steady disk form where hydrodynamics alone blows most of the gas out of the box.","pith_inferences":["Editorial extension: if the dense-gas field is set locally by surface density within a few hundred megayears, galaxy-scale and cosmological simulations could adopt a local magnetic-field closure rather than evolving a global seed field, since initial memory is erased on that timescale.","Editorial extension: the paper's Appendix C reports that a more accurate divergence-cleaning scheme lowers the saturated dense-gas field by a factor of about two, so the quoted 10--40 microgauss values are best read as an upper envelope until convergence of the $\\nabla\\cdot B$ treatment is demonstrated.","Editorial extension: if the dynamo interpretation survives, magnetic-field observations of molecular clouds become an indirect measure of turbulent energy injection, effectively a probe of the local supernova rate in galaxies that are too distant to resolve into individual supernovae."],"forward_implications":["At surface densities above roughly $40\\;M_\\odot\\,\\mathrm{pc}^{-2}$, magnetic fields are required for a quasi-steady star-forming disk to exist; pure-hydrodynamical runs blow out most of their gas unless the initial conditions are deliberately made turbulent.","The dense-gas magnetic field becomes a predicted function of local gas surface density, roughly $B \\propto \\Sigma_{\\rm gas}^{0.5}$, directly comparable to field measurements in molecular clouds across galaxies.","Because magnetic support raises the cold gas fraction by up to 40 percent and halves the disk scale height, interpreting cold-gas tracers such as CO, [C I], and [C II] in gas-rich galaxies requires MHD modeling rather than hydrodynamics alone.","Time-averaged star formation stays on the observed Kennicutt--Schmidt relation, and the disks sit in vertical pressure equilibrium consistent with the pressure-regulated, feedback-modulated theory of star formation.","Star formation becomes less bursty when magnetic fields are present, so the observed burstiness of gas-rich galaxies carries information about the magnetization of their interstellar medium."],"supporting_citations":[{"why":"Supplies the parent framework of chemistry, cooling, star-by-star star formation, and supernova feedback that GHOSDT extends to magnetohydrodynamics.","marker":"HSvD21"},{"why":"Provides the pressure-regulated, feedback-modulated theory of star formation whose dynamical-equilibrium and feedback-yield predictions the simulations are shown to satisfy.","marker":"OK22"},{"why":"The TIGRESS MHD galactic-patch simulations, the main previous framework for magnetized high-surface-density ISM patches, used as the comparison for the Kennicutt-Schmidt relation.","marker":"Kim et al. 2020a"},{"why":"Supplies the TIGRESS-NCR fit to the star formation rate versus midplane gas weight relation and the feedback-yield baselines against which the GHOSDT results are compared.","marker":"Kim et al. 2024"},{"why":"A supernova-driven small-scale dynamo simulation whose few-percent magnetic-to-kinetic saturation value forms the tension the authors flag against their own up-to-40 percent ratio.","marker":"Gent et al. 2021"},{"why":"Provides the observed magnetic field--density power-law relation for dense gas against which the simulated B--n relation is compared.","marker":"Crutcher et al. 2010"},{"why":"Supplies the ideal-MHD module and the Powell plus Dedner divergence-cleaning scheme used to evolve the magnetic field in Gizmo.","marker":"Hopkins & Raives 2016"},{"why":"Supplies the star-by-star photoionization and supernova feedback prescription adopted by GHOSDT.","marker":"Hu et al. 2017"},{"why":"Justifies the thermal supernova energy injection scheme at 1-10 solar-mass resolution by showing it captures the correct terminal momentum of remnants.","marker":"Steinwandel et al. 2020"}],"fun_headline_variants":["SN-driven dynamo sets magnetic fields in gas-rich disks","Gas density, not seed fields, controls disk magnetism","Supernova turbulence imprints field on dense gas","Magnetic fields in disks: a supernova signature","Simulations show SN-driven fields reshape cold gas"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central claim depends on the saturation level of the simulated small-scale dynamo being physical and numerically converged; the authors themselves flag that magnetic energy reaches up to 40 percent of kinetic energy, well above the few percent seen in comparable dynamo simulations, and call for a dedicated follow-up study.","fun_headline_variants_meta":{"raw":{"variants":["SN-driven dynamo sets magnetic fields in gas-rich disks","Gas density, not seed fields, controls disk magnetism","Supernova turbulence imprints field on dense gas","Magnetic fields in disks: a supernova signature","Simulations show SN-driven fields reshape cold gas"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000201,"raw_usage":{"total_tokens":1490,"prompt_tokens":1167,"completion_tokens":323,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":783,"completion_tokens_details":{"reasoning_tokens":248}},"tokens_in":783,"tokens_out":323,"duration_ms":4010,"temperature":1.0,"reasoning_tokens":248,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T19:39:24.472801+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Rerun a fixed-surface-density case such as M40 or M60 with progressively more accurate magnetic-field treatment: the constrained-gradient divergence cleaning already tested in Appendix C, higher resolution, and possibly non-ideal resistivity, then check whether the saturated dense-gas field stays near 10--40 microgauss; the paper already reports that divergence cleaning alone lowers $B_{100}$ by a factor of about two, so a continued decline would refute convergence. Observationally, Zeeman or dust-polarization measurements of molecular clouds across galaxies spanning a decade in gas surface density would test the predicted $B \\propto \\Sigma_{\\rm gas}^{0.5}$ trend directly.","supporting_citations":[{"cited_title":"P., Moster , B","cited_arxiv_id":null,"evidence_quote":"Justifies the thermal supernova energy injection scheme at 1-10 solar-mass resolution by showing it captures the correct terminal momentum of remnants."}],"review_version":1}