{"id":"a3b54b70-0eba-4f3e-8542-9b670749453f","arxiv_id":"2412.06492","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Two-temperature magnetically arrested disk simulations with turbulent electron heating are resolution-converged and produce emission similar to the simpler R(beta) parametric model, with a moderate spin of a*=0.5 matching Sgr A* ALMA data best.","lead":"This paper tests whether computer simulations of gas falling into a black hole (specifically Sgr A*) give stable results when the numerical grid is made finer, and compares two ways of modeling how electrons get heated. The two approaches produce nearly the same predicted millimeter-wave images and variability, so current model comparisons to telescope data remain valid.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Resolution convergence claim rests on a single-spin, anisotropic refinement study comparing only integrated statistics; emissivity-weighted electron temperature convergence is not demonstrated.","rationale":"The reader's verdict is CONDITIONAL, and my read does not change that: the paper is transparent, openly discusses the resolution-plateau caveat, and provides a valuable reference simulation set for the EHT community. However, I find the most load-bearing concern to be slightly different from the reader's weakest assumption. The reader focuses on whether numerical dissipation equals a physical turbulent cascade; that is important for the physical interpretation but does not directly threaten the abstract's convergence claim, which is a numerical statement. My concern targets the evidence for that numerical statement directly: the convergence test uses a single spin, non-nested anisotropic grids, and only integrated summary statistics. It does not demonstrate that the physical quantity driving the emission, the emissivity-weighted electron temperature, is converged. If that quantity drifts with resolution while the integrand statistics happen to agree, the central claim would be misleading. The proposed post-processing test is feasible with existing data and would settle the question. Because the paper already acknowledges limitations and the reader's conditional acceptance is appropriate, I retain UNCHANGED rather than moving to ACCEPT or REJECT. The reader identified the subgrid heating assumption as weakest; my concern is related but distinct, hence partial agreement.","tokens_in":17893,"tokens_out":11515,"duration_ms":127320,"concrete_test":"Post-process the four a*=0 convergence runs (and, if snapshots allow, the a*=0.9375 fiducial run) to compute emissivity-weighted electron temperature radial profiles, e.g., <Theta_e> weighted by rho*B and by the local 229 GHz synchrotron emissivity, over the same time interval used in Figure 2. Compare these profiles across the four resolutions. If the profiles vary outside the run-to-run turbulent scatter, the integrated convergence in Figure 2 is an artifact of insensitive observables and the central claim is overstated; if they agree, the convergence claim would be substantially strengthened.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's central claim is that radiative properties of two-temperature MAD models with turbulent electron heating are well converged with respect to grid resolution, 'which has not been demonstrated before.' The evidence in Section 2.4 and Figure 2 is an a*=0 study with four non-nested grids: LR 120x120x128, MR 240x120x128, HR 240x240x128, and HHR 360x120x192. These refinements are anisotropic (MR to HR doubles theta only; MR to HHR increases r and phi but not theta), so no systematic convergence sequence is established. The comparison is limited to distributions of integrated observables M3, RM, LP, and CP at 229 GHz, with only a statement that 86 and 690 GHz behave similarly. Because the electron entropy scheme in ebhlight dissipates numerical truncation error, the electron temperature distribution can be resolution-dependent even when these coarse summary statistics are stable; integrated measures may be insensitive to spatial redistribution of Theta_e (e.g., compact hot spots that cancel in net polarization). No convergence of images, spectral index, or emissivity-weighted electron temperature structure is shown. Furthermore, the single spin a*=0 may not represent high-spin MADs where flux eruptions and jet dynamics are more violent. Thus the unqualified convergence claim rests on narrow evidence, and the observed agreement could be a property of the chosen observables rather than genuine convergence of the electron thermodynamics.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports new 3D GRMHD simulations of two-temperature magnetically arrested disks (MADs) around black holes, using the ebhlight code and the Kawazura et al. (2019) turbulent electron heating prescription. The radiative output is post-processed with ipole to produce multifrequency light curves and polarimetric observables, which are then compared with the standard parametric R(β) electron temperature model, with models including radiative cooling, with a nonthermal f_kappa electron distribution, and finally with ALMA observations of Sgr A*. The principal claims are that the integrated radiative properties (M3, RM, LP, CP) are well converged with respect to grid resolution at a*=0, that the turbulent-heating K models are similar to, and in practice may be indistinguishable from, R(β) models (most closely R10), that radiative cooling has negligible effect at Sgr A* accretion rates, that nonthermal electrons affect mainly circular polarization, and that among the prograde models a*=0.5 at low viewing angles is closest to the ALMA constraints, although no model matches all observables.","tokens_in":18114,"tokens_out":6343,"duration_ms":71229,"significance":"If the convergence claim holds, it materially strengthens the reliability of grid-resolution choices used in EHT-scale model libraries and supports the use of integrated variability and polarization diagnostics for two-temperature MAD models. The claimed closeness of the turbulent-heating K models to the parametric R10 model is also practically significant because it suggests that simpler electron temperature prescriptions may suffice for current Sgr A* image and light-curve modeling. The paper benefits from long-duration simulations (up to 30,000M), a broad spin range, multifrequency polarized radiative transfer, and a direct comparison with ALMA data. The main weaknesses are that the convergence evidence is narrow and that the K-versus-R(β) similarity is presented through qualitative correlation and distribution comparisons rather than quantitative model-comparison statistics.","major_comments":[{"comment":"The abstract's headline claim that radiative properties are 'well converged with respect to the numerical grid resolution' is not fully supported by the evidence shown. The four convergence runs are all at a*=0, and the refinement sequence is not nested: LR (120x120x128) to MR (240x120x128) doubles only N_r; MR to HR (240x240x128) doubles only N_theta; MR to HHR (360x120x192) increases N_r and N_phi but leaves N_theta unchanged. This is not a systematic convergence sequence. In addition, the reported comparison in Figure 2 is restricted to distributions of integrated quantities at 229 GHz, the 86/690 GHz agreement is asserted only in the text, and no image-space, spectral-index, or emissivity-weighted electron-temperature diagnostic is shown. Since the convergence claim is one of the paper's two central results, it should either be narrowed to 'integrated observables at 229 GHz for a*=0' or supported by additional diagnostics and additional spins.","section":"Section 2.4, Figure 2, Table 1"},{"comment":"The physical interpretation of the K-models depends on an assumption that is stated but not validated in this work: the numerical truncation-error heating in the ideal GRMHD scheme is identified with the turbulent cascade heating that is partitioned between electrons and protons. The manuscript acknowledges that 'the total viscous heating is produced by truncation errors at the numerical grid level' and defers to Ressler et al. (2015) for the justification. This is not necessarily an error, but it is load-bearing for the claim that the K models represent physically motivated turbulent electron heating. The paper should either present a validation or sensitivity test for this identification, or explicitly frame the convergence and K-versus-R(β) results as properties of the numerical dissipation model rather than of the physical turbulent heating scenario.","section":"Section 2.3"},{"comment":"The second central claim, that the K models 'in practice, may be indistinguishable from the R(β) models,' is supported only by qualitative visual comparison of correlations and distributions. Figure 4 shows that K is strongly correlated with R10 for spectral index and CP but only weakly correlated for RM and LP, which the text itself identifies as the observables most sensitive to electron temperature. No quantitative distribution-overlap metric, likelihood, or model-selection statistic is provided. Given that the claim is stated in the abstract, a formal comparison (for example, a two-sample test on the observable distributions or a scoring against the same ALMA data used in Section 3.3) should be reported, or the wording should be weakened to reflect the qualitative agreement.","section":"Section 3.1, Figures 4-5"}],"minor_comments":[{"comment":"The word 'self-consisitent' should be corrected to 'self-consistent'.","section":"Abstract"},{"comment":"The sentence 'Although not shown, this is also true for two other neighboring frequencies of 86 and 690 GHz' is an unsupported parenthetical in a section whose purpose is to establish convergence; the statement should either be removed or the supporting figure should be included.","section":"Section 2.4"},{"comment":"The text refers to 'mixed modified Kerr-Schield logarithmic coordinates'; the correct spelling is 'Kerr-Schild'.","section":"Section 2.1"},{"comment":"The 'Cooling' column contains both 'No' and 'no' entries; capitalization should be made uniform.","section":"Table 1"},{"comment":"The mass-scaling procedure is described as setting M to reproduce an average 229 GHz flux of 2-3 Jy, but the iterative nature of the procedure and the behavior of the 86 GHz flux under this scaling are not described; a sentence clarifying the procedure would improve reproducibility.","section":"Section 2.2"}],"recommendation":"major_revision","confidential_remarks":"This is a competent and useful simulation paper whose main conclusions, if properly qualified, are of interest to the Sgr A* and EHT modeling community. The convergence claim in the abstract is stronger than the evidence, but it can be fixed by narrowing the scope of the claim or adding diagnostics; the K-versus-R(β) comparison would benefit from a quantitative metric. I do not see a reason to reject the paper, provided the authors address the load-bearing concerns in revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a careful, honest simulation study and arguably the reference point for two-temperature MAD modeling that the EHT community has been missing. The genuinely new pieces are the resolution convergence test (first for two-temperature MAD with turbulent heating) and the long 30,000M comparison of the Kawazura et al. turbulent heating model against the parametric R(beta) models using variability and polarimetric observables. The work uses public codes, publishes a complete parameter table, and is transparent about its own caveats. The result that K and R10 look alike is not baked in; it emerges from the runs.\n\nThe soft spots are real but not disqualifying. The convergence claim in the abstract is slightly stronger than the evidence. The test is a single spin a*=0, the four grids are not a nested sequence (refinement is anisotropic and inconsistent), and the comparison is limited to distributions of integrated observables. No convergence of images, spectral index, or emissivity-weighted electron temperature structure is shown. The author does flag the resolution-plateau possibility in Section 2.4, so this is a caveat that should move up into the abstract rather than a hidden flaw. The second soft spot is the load-bearing assumption, stated explicitly, that numerical grid-scale dissipation can stand in for the physical turbulent cascade that heats electrons. This is justified by reference to Ressler et al. 2015 but not validated here. That is normal practice in this field, but it does mean the electron temperatures are model-dependent in a way the paper does not quantify. I would also soften \"indistinguishable from R(beta) models\": the correlations with R10 are strong for some observables but weak for RM and LP, and the current EHT best-bet R160 is not reproduced by any turbulent heating model. The author discusses this possibility but the abstract overstates.\n\nNet: this paper deserves a serious referee. The claims are checkable, the code is public, the limitations are mostly acknowledged, and the comparison data are clearly described. The main request to the author should be to qualify the convergence claim and to show at least one higher-spin convergence test or state clearly that it remains to be done.\n\nI would not desk-reject this. Send it out.","headline":"A careful, honest two-temperature MAD study whose headline convergence claim is somewhat stronger than the evidence supports; still worth a serious referee.","tokens_in":18720,"tokens_out":2615,"would_cite":true,"duration_ms":27543,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Two-temperature magnetically arrested disk simulations around a black hole produce millimeter emission that is converged with grid resolution and practically indistinguishable from simple R(beta) electron-temperature models.","keywords":["Supermassive black holes","Magnetohydrodynamical simulations","Low-luminosity active galactic nuclei","Plasma physics","two-temperature accretion flows","magnetically arrested disks","Sgr A*","millimeter polarimetry"],"falsifier":"Run one of the zero-spin MAD models with roughly an order of magnitude more grid points in each direction and compare the distributions of the modulation index, rotation measure, and polarization fractions over the same time interval; if they move outside the run-to-run scatter among the four current resolutions, the claimed convergence plateau is false.","tokens_in":17611,"feed_emoji":"🕳️","tokens_out":11473,"duration_ms":107187,"temperature":0.7,"pith_summary":"This paper revisits three-dimensional general-relativistic magnetohydrodynamic simulations of magnetically arrested disks around black holes, tracking electron temperatures through a turbulent-heating prescription rather than through a fitted proton-to-electron temperature ratio. It aims to establish that the millimeter emission properties of these two-temperature models are numerically converged at currently used grid resolutions, and that the models are observationally similar to the simpler parametric $R(\\beta)$ electron-temperature prescription. If true, simulated images and variability statistics used to interpret black hole observations are stable against grid resolution, and the simpler parametric electron model remains adequate for current Sgr A* millimeter data. The paper also finds that radiative cooling and a nonthermal electron distribution change the millimeter emission only weakly, and that comparing the models with Sgr A* observations favors lower viewing angles and moderate black hole spin.","feed_headline":"Black hole emission models pass the resolution test","feed_subtitle":"Turbulent electron heating does not outperform the simpler R(beta) recipe for Sgr A* millimeter data.","key_machinery":"The central object is a two-temperature magnetically arrested disk simulation in which the electron temperature is evolved from an entropy equation with a subgrid heating split; the split, taken from a turbulent-cascade prescription, sends most dissipation to protons in high-$\\beta$ plasma and to electrons in low-$\\beta$ plasma. The electron-temperature model is compared against the parametric $R(\\beta)$ prescription, in which the proton-to-electron temperature ratio is a function of plasma $\\beta$ with limiting values $R_{\\rm low}$ and $R_{\\rm high}$. The argument is carried by post-processing the flows with polarized synchrotron radiative transfer to produce light curves and image metrics, notably the total-intensity modulation index, rotation measure, and fractional linear and circular polarization. The close similarity of these observable distributions between the turbulent-heating models and the R10 parametric models is the evidence for practical indistinguishability.","core_discovery":"The central claim is that the radiative output of two-temperature magnetically arrested disk (MAD) simulations around a black hole, with electrons heated by a prescribed turbulent cascade, is well converged at the grid resolutions used in current event-horizon-scale model libraries: total-intensity variability, rotation measure, and linear and circular polarization do not shift systematically as the grid is refined. The paper further claims that these self-consistent two-temperature models are in practice indistinguishable from the parametric $R(\\beta)$ models with $R_{\\rm high}=10$ in most observables, even though the spatial map of $T_p/T_e$ is not the same as R10. Radiative cooling and nonthermal electron distribution functions have only weak effects on the millimeter emission. When the models are scaled to Sgr A*, none reproduces all observed properties: the models are too variable and too optically thin, an external Faraday screen is needed to match the rotation measure, and among prograde models spin $a_*=0.5$ best recovers circular polarization at both observed frequencies.","pith_inferences":["A practical consequence left implicit in the paper: if the turbulent-heating and R10 models are indistinguishable for current observables, then the remaining mismatch with Sgr A* data is probably not in the electron-temperature law but in other missing physics such as reconnection heating, resistivity, or anisotropic pressure.","A natural next numerical test is to run the same convergence and comparison pipeline with a reconnection-based heating partition; if it drives electron temperatures lower while still mapping onto an $R(\\beta)$-like model, it could resolve the too-variable and too-optically-thin discrepancies without abandoning the parametric approach.","The weak effect of nonthermal electrons found at millimeter wavelengths may not persist at higher frequencies or for brighter sources; applying the kappa-distribution models to such cases could expose larger circular-polarization differences.","The appendix's observation that raising $R_{\\rm low}$ makes $R(\\beta)$ images resemble those of a much larger $R_{\\rm high}$ suggests that polarimetric scoring with $R_{\\rm low}=1$ fixed may have biased inferred temperature ratios upward; verifying this could remove part of the apparent tension between turbulent-heating models and the best-fit $R(\\beta)$ model."],"forward_implications":["If the convergence result holds, existing model libraries built at these grid resolutions do not need to be regenerated at higher resolution to draw conclusions about millimeter emission statistics.","If the turbulent-heating models are indeed practically indistinguishable from the R10 parametric models, parameter inference based on $R(\\beta)$ electron temperatures remains a serviceable shortcut for interpreting current Sgr A* observations.","Radiative cooling can be neglected when modeling Sgr A* millimeter emission from MADs with turbulent heating at these accretion rates.","Circular polarization is the most discriminating observable: it distinguishes thermal from nonthermal electron distributions and favors an intermediate black hole spin ($a_*\\approx 0.5$) for Sgr A*.","None of the modeled MADs matches the observed variability or optical depth of Sgr A*, which points to physical ingredients beyond the thermal turbulent-heating picture."],"supporting_citations":[{"why":"It introduces the electron-entropy tracking scheme and argues that grid-scale dissipation in turbulent torus simulations stands in for the turbulent heating divided between protons and electrons.","marker":"Ressler et al. 2015"},{"why":"It provides the proton-to-electron heating ratio used to set electron temperatures in all turbulent-heating runs.","marker":"Kawazura et al. 2019"},{"why":"It defines the $R(\\beta)$ prescription whose R1 through R160 models serve as the parametric comparator.","marker":"Moscibrodzka et al. 2016"},{"why":"It supplies the GRMHD solver and the radiative-transfer infrastructure used for all simulations in the paper.","marker":"Ryan et al. 2015"},{"why":"It supplies the Sgr A* millimeter light curves and the modulation-index definition used in the observational comparison.","marker":"Wielgus et al. 2022a"},{"why":"It provides earlier two-temperature Sgr A* MAD simulations with turbulent heating that the present work extends to longer durations and adds polarimetric convergence tests.","marker":"Dexter et al. 2020a"}],"fun_headline_variants":["Turbulent electron heating fails to beat simpler black hole recipe","Black hole simulations converge but still miss Sgr A* polarization","Two-temperature black hole models match simpler recipes in practice","R(beta) recipe matches turbulent heating for black hole emission"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole comparison treats the numerical dissipation that appears at the grid scale in the ideal simulation as a stand-in for the physical turbulent cascade that heats electrons, and the paper does not independently validate that equivalence.","fun_headline_variants_meta":{"raw":{"variants":["Turbulent electron heating fails to beat simpler black hole recipe","Black hole simulations converge but still miss Sgr A* polarization","Two-temperature black hole models match simpler recipes in practice","R(beta) recipe matches turbulent heating for black hole emission"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000792,"raw_usage":{"total_tokens":3514,"prompt_tokens":993,"completion_tokens":2521,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":609,"completion_tokens_details":{"reasoning_tokens":2452}},"tokens_in":609,"tokens_out":2521,"duration_ms":18621,"temperature":1.0,"reasoning_tokens":2452,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T19:35:55.049823+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run one of the zero-spin MAD models with roughly an order of magnitude more grid points in each direction and compare the distributions of the modulation index, rotation measure, and polarization fractions over the same time interval; if they move outside the run-to-run scatter among the four current resolutions, the claimed convergence plateau is false.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It provides the proton-to-electron heating ratio used to set electron temperatures in all turbulent-heating runs."},{"cited_title":"R., Dolence , J","cited_arxiv_id":null,"evidence_quote":"It supplies the GRMHD solver and the radiative-transfer infrastructure used for all simulations in the paper."}],"review_version":1}