{"id":"07bb8117-a925-447b-ac48-7389bd8ec247","arxiv_id":"2501.12448","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Radiative two-temperature MAD simulations of M87* with Kawazura et al. (2019) turbulent electron heating match EHT total intensity but over-predict beam-scale linear polarization (about 30% versus the observed under 10%), because electrons stay too hot (Ti/Te about 5).","lead":"A single researcher ran 11 computer simulations of the black hole M87*, tracking electrons and ions separately, and predicted what the Event Horizon Telescope should see. The simulated images match the observed shape and brightness but show about three times more polarized light than detected, suggesting the standard turbulent model of electron heating makes electrons too hot for M87*.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The over-polarization and R≈5 result hinges on the K19 heating fraction δ_e in the low-β near-horizon MAD flow; if K19 over-heats electrons there, the claimed tension with EHT polarization disappears.","rationale":"The reader's weakest-assumption analysis correctly identifies the K19 sub-grid heating prescription as the load-bearing input: the paper's headline over-polarization result is a direct consequence of the heating split δ_e(β_i, Te/Ti) from Eq. 13. My stress-test agrees with that identification. The concern is not an internal inconsistency—the simulations do implement K19 faithfully, and the reported R≈5 and <|m|>≈30% follow from that implementation. The issue is external validity: the conclusion 'turbulent heating cannot reproduce EHT polarization' is only as strong as the assumption that K19 describes the actual electron heating throughout the near-horizon MAD flow. In particular, K19's δ_e→1 at low β_i is precisely the property that makes electrons hot in the jet and suppresses Faraday depolarization; alternative physically motivated models (R19 reconnection, K20/Satapathy compressive driving) can yield lower δ_e in the same regions, cooler electrons, and sufficient depolarization. The paper is transparent about this, explicitly deferring R19 simulations and compressive-driving corrections to future work, so the paper is not misleading. But the interpretive payload—that EHT polarization requires R_high≈80–160 and challenges turbulent heating—remains conditional on K19. A concrete R19 re-run of a representative spin (e.g., a*=0.9, where the over-polarization is strongest) would settle whether the tension is robust or a K19 artifact. Because the paper's own framing already presents the result as a survey under one heating model and flags the uncertainty, the appropriate verdict remains CONDITIONAL, unchanged from the reader's assessment.","tokens_in":30341,"tokens_out":5463,"duration_ms":61288,"concrete_test":"Re-run the a* = 0.9 and a* = 0 radiative simulations with the Rowan et al. (2019) reconnection heating prescription substituted for K19, keeping resolution, initial conditions, density rescaling, and image pipeline identical; then compute R in the inner 25 r_g and the 20 μas-blurred <|m|>. If R rises above ~10 and <|m|> falls below ~10% while total-intensity statistics remain EHT-consistent, the paper's over-polarization conclusion is K19-specific rather than a robust challenge to turbulent heating in MAD flows.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central conclusion—that turbulent heating cannot reproduce EHT polarization because the 230 GHz emitting region has R=Ti/Te≈5 and <|m|>≈30%—depends entirely on the Kawazura et al. (2019) heating prescription, Eq. 13–14. In that prescription, Q_i/Q_e = 35 (β_i/15)^{-1.4} e^{-0.1 Te/Ti}, so in the strongly magnetized near-horizon MAD flow (β_i ≪ 1) the electron heating fraction δ_e approaches unity: almost all dissipated energy goes to electrons. This is what drives the moderate R≈5 and the weak Faraday depolarization. The paper does not establish that K19 is valid in this regime; the flow contains reconnection current sheets, compressive fluctuations, and relativistic electrons, all outside the gyrokinetic Alfvénic-turbulence regime from which K19 was calibrated. If the true δ_e is lower at low β—as suggested by Rowan et al. (2019) reconnection heating or by Satapathy et al. (2023, 2024) compressive-driving corrections—electrons would be cooler, R higher, Faraday depth larger, and <|m|> could fall within the EHT range. The paper itself acknowledges this in Section 5, listing R19 simulations and compressive-driving models as necessary future work, so the concern is explicit but unresolved. A secondary, related issue is the numerical heating rate q_v: if the grid-scale dissipation is artificially high, even the K19-specific R≈5 would be an artifact. The paper cites resolution-independence tests but calls for a different numerical implementation to verify this. Thus the strongest claim is not yet robust to the choice of sub-grid electron heating physics.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents eleven 3D two-temperature, radiative GRMHD (2TGRRMHD) simulations of M87* in the magnetically arrested state, covering spins a* = -0.9 to +0.9, together with eleven matched single-fluid runs in the KORAL code. The radiative runs adopt the Kawazura et al. (2019) turbulent electron-heating prescription and are calibrated to M87*'s mass, distance, and 230 GHz flux density, with small post hoc density rescaling factors tabulated (Table 2). The main results are: (i) radiation and two-temperature physics leave the horizon magnetic flux, jet efficiency, and spindown parameter nearly unchanged relative to single-fluid runs (Section 4.1); (ii) the radiative runs self-consistently produce R = Ti/Te ≈ 5 in the 230 GHz emitting region, well described by fitted Mościbrodzka et al. (2016) parameters R_low ≈ 2 and R_high ≈ 5, with an effective adiabatic index Γgas ≈ 1.55; (iii) the simulated images match EHT total-intensity statistics but are over-polarized (⟨|m|⟩ ≈ 30% against the EHT range of 5.7-10.7%); and (iv) the polarization mode β2 follows the Palumbo et al. (2020) trend with spin, parameterized by a two-parameter BZ-motivated fit. The paper interprets the over-polarization as a challenge to the turbulent-heating interpretation of M87*'s polarization and candidly lists the heating-prescription, numerical-dissipation, and magnetization-cutoff caveats.","tokens_in":30569,"tokens_out":42871,"duration_ms":367427,"significance":"The suite is the first systematic 2TGRRMHD spin survey of M87* and a practical resource for the EHT interpretation community: horizon fluxes, fitted R-beta parameters (Tables 1 and 3), per-simulation rescaling factors (Table 2), the β2(a*) fitting function (Eq. 37), and the public KORAL code are all provided. The manuscript is unusually candid about its own limitations; Section 5 explicitly flags the restricted regime of validity of the K19 prescription, the uncertain numerical heating rate, and the σ_cut choice. If the over-polarization result survives correction of Eq. (13) and alternative heating prescriptions, it is a substantive falsifiable challenge to the standard turbulent-heating interpretation of M87*'s polarization, implying much cooler electrons (R_high ≈ 80-160) or modifications to the aligned-MAD paradigm, and the follow-up R19 and compressive-driving tests are already specified. The secondary result that two-temperature and radiative effects leave φ_BH, η, and s unchanged strengthens the basis for spin inference from single-fluid MAD libraries.","major_comments":[{"comment":"Equation (13) as printed is inconsistent with the rest of the paper, and since the heating partition drives the headline R ≈ 5 result, this is a load-bearing issue. With Q_i/Q_e = 35 (β_i/15)^(-1.4) e^(-0.1 Te/Ti), at β_i = 0.01 one gets Q_i/Q_e ≈ 10^6 and δ_e ≈ 10^-6; at β_i = 1, δ_e ≈ 7×10^-4; and δ_e exceeds 1/2 only for β_i ≳ 190. This is the opposite of the claim in Section 4.2 that the K19 prescription 'delivers most of the heat to electrons in the most highly magnetized regions (δ_e > 0.5 when β_i ≪ 1)'. It is also opposite to what the code must be doing: the moderate R ≈ 5 and weak Faraday depolarization reported in Sections 4.3-4.4 require substantial electron heating at low β_i, whereas the printed formula would leave electrons nearly unheated, producing R ≫ 5, strong depolarization, and ⟨|m|⟩ far below 30%. I suspect a sign error in the exponent or an inverted ratio in Eq. (13); the authors should correct the printed formula, verify it against the form actually implemented in KORAL, and check that Sections 4.2, 4.4, and 5 consistently describe the same δ_e(β_i, Te/Ti) behavior.","section":"Eq. (13)-(14); Section 4.2"},{"comment":"The stress-test concern about the K19 prescription's regime of validity lands. Even with a corrected Eq. (13), the model is calibrated from gyrokinetic simulations of Alfvénic turbulence, and the near-horizon MAD flow contains reconnection current sheets, compressive fluctuations, and relativistic electrons that are outside that calibration regime. Section 5 concedes that R19 or compressive-driving corrections (Satapathy et al. 2023, 2024) could lower δ_e at low β, raise R, and remove the tension with the EHT ⟨|m|⟩ < 10% constraint. Because this is load-bearing, I ask for a quantitative sensitivity test within the current pipeline: re-render a subset of snapshots with a modified heating partition (or with an R(β) profile spanning the R19/compressive-driving range) and recompute ⟨|m|⟩, m_net, and β2. The citations to Salas et al. (2024) and Liska et al. (2024), which found R ≈ 5-10 under R19-type heating, concern Sgr A* and X-ray binaries and do not substitute for a test in these M87* flows. Alternatively, the abstract and conclusions should state the tension as conditional on K19 rather than as a generic property of turbulent heating.","section":"Section 5; Eqs. (13)-(14); Abstract"},{"comment":"The paper's uncertainty about the numerical dissipation rate q_v is itself load-bearing, because the electron heating rate is δ_e q_v and R ≈ 5 is the quantity that creates the claimed polarization tension. The resolution-independence evidence cited from Mościbrodzka (2024) is an external test and does not directly bound q_v for this suite. The authors should provide, at minimum, the fractional contribution of the computed q_v to the electron internal-energy budget in the r ≲ 5 r_g emitting region, or a resolution study of R and ⟨|m|⟩ for one representative simulation, so the reader can judge how much of R ≈ 5 is set by the heating algorithm rather than physical dissipation. The call in Section 5 for a different numerical implementation is appropriate but is not a substitute for such a bound.","section":"Section 5 (q_v discussion)"},{"comment":"The choice σ_cut = 25 is non-standard (the community default is σ_cut = 1), and the paper reports that 30-50% of the 230 GHz flux originates from the region 1 < σ_i < 25. Since the polarization statistics in Figures 11-12 are computed over this emission, the over-polarization claim is partly defined by this choice. The paper's defense, that a lower σ_cut would require a higher accretion rate that is 'likely not sufficient' to depolarize the images, is plausible but qualitative. A quantitative check (recomputing ⟨|m|⟩ and m_net with σ_cut = 1 and the correspondingly rescaled density, or a short scan of ⟨|m|⟩ versus σ_cut for one simulation) would place this modeling choice on firmer ground and is well within the presented post-processing pipeline.","section":"Section 3.2 (σ_cut = 25)"}],"minor_comments":[{"comment":"The sentence 'the observed large degree of image polarization in M87*' is confusing, since the paper's argument is that the observed ⟨|m|⟩ = 5.7-10.7% is small compared with the simulated ≈30%; please rephrase, for example, 'the detected polarization signal in M87*'.","section":"Section 4.4"},{"comment":"The claim that 'the ratio of the Kerr-Schild poloidal to radial magnetic field strength is proportional to Ω_H' should be 'azimuthal-to-radial' (or 'toroidal-to-poloidal'); as written the phrase is ambiguous given that the BZ monopole field is fundamentally radial. Showing explicitly how the argument of Eq. (37) follows from B_φ/B_r would make the zero-spin and high-spin limits of the fit transparent.","section":"Section 4.4 (Eq. 37)"},{"comment":"The two density-rescaling steps (initial normalization to M_dot = 10^-6 M_dot_Edd at t = 10^4 t_g, the 500 t_g equilibration, and the second rescaling to median F230 = 0.5 Jy near t = 1.05×10^4 t_g, followed by the run to t = 2×10^4 t_g) are described in prose and are easy to misread; a short numbered timeline would improve clarity.","section":"Section 3.1"},{"comment":"The fitted C0 and C1 values for Eq. (37) are given only in the text; adding them to the caption would make the figure self-contained.","section":"Figure 12"},{"comment":"The prograde and retrograde fits to Eq. (37) differ by C1 = 17 deg versus -26 deg; one sentence explaining this offset (e.g., the contribution of the counter-rotating disc to the field pitch angle in the emission region) would help readers use the fit for spin inference.","section":"Section 4.4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a good fit for the journal, and the related-work coverage and code/data availability statements are appropriate; there are no novelty or citation concerns. My recommendation of major revision is driven primarily by the internal inconsistency of Eq. (13) with the text and with the simulation outcomes; this is very likely a transcription error, but it sits in the central equation of the paper and must be fixed and verified against the code. The remaining major comments ask for sensitivity analyses that are feasible with the existing post-processing pipeline. If the author prefers not to add those tests, rephrasing the abstract and conclusions to make the K19-conditionality of the polarization claim explicit would be the minimum acceptable response. The author's candor in Section 5 is a strength and should be preserved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a look. This is the first systematic survey of radiative, two-temperature MAD simulations of M87* across spin, and as a resource it is genuinely useful. The eleven runs are well documented, the time-variability error bars are a nice touch, and the descriptive results — self-consistent Gamma_gas ≈ 1.55, 15% thicker discs, 10% narrower jets, unchanged phi_BH/eta/s versus single-fluid runs — are new and credible. The beta_2(a*) fitting function is a reasonable compact summary, and the trend with spin is consistent with earlier work. The author is also unusually candid about limitations, which earns real credit.\n\nThe headline result is that K19 turbulent heating yields R ≈ 5 in the 230 GHz emitting region and <|m|> ≈ 30%, far above the EHT's <10%. That would be a strong constraint on collisionless plasma heating if it holds. But it depends on the K19 prescription in precisely the regime where that prescription is least secure: low-beta, near-horizon MAD flow with reconnection current sheets and compressive fluctuations. If the true delta_e there is lower, electrons are cooler, R is higher, Faraday depolarization is stronger, and the tension with EHT polarization weakens or disappears. The paper says this itself in Section 5, listing R19 and compressive-driving models as necessary follow-ups, so the concern is explicit but unresolved. A second soft spot is numerical dissipation: q_v may artificially heat electrons. The resolution tests are reassuring but not a different code, and the author asks for one. The sigma_cut = 25 choice is disclosed, and the discussion of shifting to sigma_cut = 1 suggests it would not rescue the polarization. Total-intensity consistency is partly enforced by density rescaling, which is standard and disclosed, so that match is not a free prediction. These are caveats, not fatal flaws; the descriptive survey stands regardless.\n\nI agree with the reader's conditional verdict. The over-polarization claim is a challenge to K19 in particular, not a proof against turbulent heating in general. The paper is honest about this, and the new simulation library deserves to be in the literature. The data being \"available on request\" rather than public is a minor reproducibility annoyance, not a dealbreaker.\n\nA serious editor should send this to peer review. A referee should push on the heating-prescription validity question and the q_v caveat, but the core survey is solid and the interpretation is appropriately hedged.\n\nWould I bring it to reading group? Yes. Would I cite it? Yes, as the simulation resource and for the beta_2 fit.","headline":"The first systematic 2TGRRMHD spin survey of M87* is a solid new resource, and the over-polarization result is a real challenge to K19 heating, though its force depends on a heating prescription whose validity in the near-horizon MAD regime remains unproven.","tokens_in":31335,"tokens_out":1913,"would_cite":true,"duration_ms":22715,"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":"Standard turbulent electron heating in two-temperature radiative simulations of M87* yields $R\\approx 5$ and ~30% beam-scale linear polarization, far above the observed $<10\\%$; matching M87* requires much cooler electrons.","keywords":["M87*","magnetically arrested accretion","two-temperature GRMHD","turbulent electron heating","sub-grid heating prescription","linear polarization","Faraday depolarization","black hole spin-down"],"falsifier":"Run the same eleven-spin, two-temperature, radiative magnetically arrested suite with a reconnection-based electron heating prescription instead of the turbulent one, keeping the same black-hole mass, accretion-rate normalization, and magnetization cut; if any such model yields beam-scale linear polarization below 10% while keeping $R$ near 80–160 in the emitting region, the paper's central claim would be refuted.","tokens_in":29906,"feed_emoji":"🔭","tokens_out":11168,"duration_ms":108811,"temperature":0.7,"pith_summary":"This paper tries to establish that the standard turbulent electron heating prescription adopted in two-temperature, radiative magnetically arrested simulations of M87* makes the electrons in the 230 GHz emitting region only moderately cooler than the ions, with $R = T_{\\rm i}/T_{\\rm e} \\approx 5$. In that temperature state, simulated 20-microarcsecond-scale images carry roughly 30% linear polarization, while the Event Horizon Telescope measures less than 10%. The paper's core claim is therefore that this heating model cannot satisfy the polarization constraints on M87* without invoking much cooler electrons than it predicts. A sympathetic reader would take the main payoff to be a physically motivated prediction of electron temperature that conflicts sharply with post-processing models requiring $R_{\\rm high} \\approx 80$–$160$ for Faraday depolarization.","feed_headline":"Turbulent heating over-polarizes simulated M87* images","feed_subtitle":"Two-temperature models give 30% beam-scale polarization; M87* shows under 10% observed.","key_machinery":"The load-bearing object is the K19 sub-grid electron heating fraction $\\delta_{\\rm e}$, defined by $Q_{\\rm i}/Q_{\\rm e} = 35\\,(\\beta_{\\rm i}/15)^{-1.4}\\,e^{-0.1\\,T_{\\rm e}/T_{\\rm i}}$ and $\\delta_{\\rm e} = 1/(1+Q_{\\rm i}/Q_{\\rm e})$. It partitions viscous dissipation between electrons and ions in each cell from the local ion plasma $\\beta$ $\\beta_{\\rm i}$ and temperature ratio, and it is the sole electron heating input that sets why the simulated electrons stay at $R\\approx 5$ instead of cooling further. The paper also relies on the $R(\\beta_{\\rm gas})$ fitting form of the standard post-processing temperature-ratio model to summarize the simulation data.","core_discovery":"The paper reports eleven 3D magnetically arrested simulations around a $6.5\\times 10^9\\,M_\\odot$ black hole, each evolved with separate electron and ion temperatures, radiative cooling, and the K19 electron heating fraction from gyrokinetic turbulence. It finds the emitting region sits at $R\\approx 5$, with an effective adiabatic index $\\Gamma_{\\rm gas}\\approx 1.55$, and that the images reproduce M87*'s total-intensity ring size, asymmetry, and position angle. The decisive result is polarimetric: the time-averaged, beam-blurred images have $\\langle |m| \\rangle \\approx 30\\%$, several times the observed $<10\\%$, because the moderately warm electrons do not produce enough internal Faraday rotation to depolarize the ring. The paper concludes that producing the observed polarization requires $R_{\\rm high}\\approx 80$–$160$, more than an order of magnitude above what the turbulent heating prescription yields, and that this challenge is specific to the choice of sub-grid electron heating.","pith_inferences":["Inference: If reconnection-based heating also fails to cool electrons enough, the over-polarization may point to non-thermal electron populations, anisotropic electron distributions, or emission-region cuts rather than the heating fraction; the paper notes that several recent Sgr A* two-temperature studies with reconnection heating still find $R\\approx 5$–$10$.","Inference: Because $R$ is not a single-valued function of $\\beta_{\\rm gas}$ in the simulations, with up to an order of magnitude of scatter, image libraries that parameterize $R(\\beta)$ with one curve may mis-rank models; a history-dependent or two-parameter electron-temperature prescription could change which spins and heating models fit the polarization data.","Inference: The 30–50% of 230 GHz flux coming from the moderately magnetized region $1<\\sigma_{\\rm i}<25$ makes the magnetization cut a hidden lever on the polarization fraction; adopting a lower cut with a larger accretion rate could partly compensate the over-polarization, though the paper argues the effect is likely too small to fully resolve the mismatch.","Inference: If no heating prescription yields sufficiently cold electrons, the tension may point toward field configurations between standard turbulent and magnetically arrested states, or toward sub-Maxwellian electron distributions, rather than an error in the temperature ratio alone; the paper sketches these as alternatives."],"forward_implications":["If the K19 prescription is right for M87*, the 230 GHz emission region sits at $R\\approx 5$, so single-fluid libraries that tune $R_{\\rm high}$ to 80–160 to match polarization are effectively invoking electron cooling physics that turbulent heating alone does not provide.","Because radiation feedback leaves $\\phi_{\\rm BH}$, $\\eta$, and the spindown parameter nearly unchanged, conclusions about jet power and black-hole spin-down drawn from non-radiative MAD libraries remain valid even when two-temperature and radiative effects are added.","The larger effective adiabatic index ($\\Gamma_{\\rm gas}\\approx 1.55$) than the usual $13/9$ makes simulated MAD discs about 15% thicker and jets about 10% narrower, so temperature modeling changes global disc structure, not just emissivity.","The polarization-spiral trend with spin persists in radiative two-temperature models; if one ignores the polarization-fraction mismatch, the observed $\\angle\\beta_2$ would select a spin around $a_*\\in[-0.7,0.2]$, and for the weakly spinning retrograde case the image asymmetry is set by the accretion flow rather than the black-hole spin."],"supporting_citations":[{"why":"Supplies the K19 sub-grid electron heating fraction (Eqs. 13–14) that sets the electron temperature state claimed to over-polarize the images.","marker":"Kawazura et al. (2019)"},{"why":"Defines the R(β_gas) two-parameter temperature-ratio prescription that the paper fits to its simulations and uses to frame the comparison with single-fluid image libraries.","marker":"Mościbrodzka et al. (2016)"},{"why":"Provides the observed linear polarization fractions (m_net and ⟨|m|⟩) that the simulated images fail to match.","marker":"Event Horizon Telescope Collaboration et al. (2021a)"},{"why":"Supplies the EHT MAD simulation library and the finding that R_high ≈ 80–160 is needed for sufficient Faraday depolarization.","marker":"Event Horizon Telescope Collaboration et al. (2021b)"},{"why":"Sets the observed 230 GHz flux density (0.5 Jy) and total-intensity ring statistics used to normalize and compare the simulated images.","marker":"Event Horizon Telescope Collaboration et al. (2019a)"},{"why":"Defines the β_2 polarization-spiral statistic and the spin-dependent trend that the paper confirms in radiative two-temperature simulations.","marker":"Palumbo et al. (2020)"},{"why":"Provides the two-temperature radiative GRMHD equations and the numerical heating-rate calculation used by the simulation code.","marker":"Sądowski et al. (2017)"},{"why":"Supplies the magnetically arrested spin-dependent fitting functions for φ_BH, η, and s and the same initial conditions against which the radiative simulations are compared.","marker":"Narayan et al. (2022)"}],"fun_headline_variants":["Simulated M87* rings are too polarized, hinting at missing Faraday rotation","M87* models overpolarize by 3x: turbulent electron heating too weak","Two-temperature M87* sims: 30% polarization vs observed <10%","Electron heating model fails M87* polarization test: 30% vs <10%","M87* polarization mismatch: turbulent heating yields 30%, observed <10%"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result assumes the K19 turbulent-heating formula correctly divides dissipated energy between electrons and ions everywhere in the flow, including the strongly magnetized jet; if the true electron share is smaller there, electrons would be cooler and the over-polarization would disappear.","fun_headline_variants_meta":{"raw":{"variants":["Simulated M87* rings are too polarized, hinting at missing Faraday rotation","M87* models overpolarize by 3x: turbulent electron heating too weak","Two-temperature M87* sims: 30% polarization vs observed <10%","Electron heating model fails M87* polarization test: 30% vs <10%","M87* polarization mismatch: turbulent heating yields 30%, observed <10%"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000338,"raw_usage":{"total_tokens":1981,"prompt_tokens":1169,"completion_tokens":812,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":785,"completion_tokens_details":{"reasoning_tokens":702}},"tokens_in":785,"tokens_out":812,"duration_ms":7512,"temperature":1.0,"reasoning_tokens":702,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T17:13:17.908675+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same eleven-spin, two-temperature, radiative magnetically arrested suite with a reconnection-based electron heating prescription instead of the turbulent one, keeping the same black-hole mass, accretion-rate normalization, and magnetization cut; if any such model yields beam-scale linear polarization below 10% while keeping $R$ near 80–160 in the emitting region, the paper's central claim would be refuted.","supporting_citations":[],"review_version":1}