{"id":"b3279a6c-dffe-40c0-a479-271bf7cc611a","arxiv_id":"2411.12809","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A supernova remnant in a cloudy interstellar medium expands at the same rate as in a uniform medium but loses much more energy to cooling at cloud interfaces, reducing its hot gas mass and momentum.","lead":"This paper uses high-resolution 3D simulations to show that supernova remnant expansion in a cloudy, two-phase interstellar medium loses significant energy as shocks hit cold clouds, so the remnant ends up with less hot gas and less final momentum than a uniform medium would give. The finding challenges simple one-dimensional models that only add mass from clouds and ignore this energy drain.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The flat interface-cooling sink and the fitted 1D constants rest on an unresolved-cooling convergence claim that the paper's own resolution tests do not establish.","rationale":"In good faith, the paper is a strong simulation study: it uses a modern GPU hydro code, achieves 1/64 pc resolution in a global SNR, tests ten initial seeds, varies resolution and parameters, and honestly discloses that formal convergence is not claimed and that magnetic fields and other physics are omitted. The qualitative central claim—cloudy SNRs sweep more mass yet lose energy at cloud interfaces, suppressing hot-gas mass and terminal momentum—is robust across seeds and resolution tests and matches the reader's conditional acceptance. The load-bearing concern is narrower: the quantitative energy-sink law (spatially flat, ˙e ≈ −0.2 ˙e_S, t^−11/5) is the part that would generalize previous 1D mass-loading models, and it depends on unresolved interface cooling. The cited convergence precedents concern turbulent mixing layers, not the transient, shock-dominated geometry here. The paper's own Figure 9 shows non-negligible resolution trends, especially with conduction. My proposed check is a feasible 2048^3 reduced-domain run at twice the fiducial resolution; it directly tests whether the sink profile and normalization are stable. Since the reader already recommended CONDITIONAL explicitly because of this convergence question, my read does not change the verdict.","tokens_in":26535,"tokens_out":9535,"duration_ms":110987,"concrete_test":"Run a new reduced-domain simulation with the same seed and physics as TN-2048-X / TY-2048-X but at Δx = 1/128 pc in a 16 pc box (2048^3 cells) for 10 kyr, and compare the angle-averaged ˙e(r,t)/˙e_S(t) profile and the integrated ˙E/˙E_S history against the existing Δx = 1/64 pc runs at t = 5 and 10 kyr without smoothing. If the profile is not flat within ~20% or the integrated sink changes by more than ~20% between the two resolutions, the t^−11/5 flat-sink model is not converged; if it is unchanged, the resolution concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's novel quantitative result is not just that clouds remove energy, which the global time histories support, but that the sink is spatially flat, ˙e ≈ −0.2 ˙e_S, and follows ˙e ∝ t^−11/5, justifying the 1D model in Eqs. (30)-(31). This measurement is made at cloud interfaces where l_cool ≈ 7×10^−5 pc and l_F ≈ 3×10^−5 pc (Sec. 4.3), while Δx = 1/64 pc, so the cooling and Field lengths are unresolved by roughly two orders of magnitude. The convergence argument leans on Fielding et al. (2020) and Tan et al. (2021) for sustained radiative turbulent mixing layers; it is not demonstrated for transient SNR shock-cloud interactions, where cooling is concentrated in dense shocked CNM shells and NTSI corrugations. Section 4.3 itself disclaims formal convergence, reports ~20% changes in hot-gas mass per resolution doubling, and shows conduction runs with systematically higher densities and cooling rates at higher resolution (Figs. 8-9). Thus the measured flat cooling profile and the calibrated constants A = 0.5, C = −0.2 are not securely converged; a resolution-dependent surface cooling rate would change the inferred sink shape and the 1D model. The qualitative claim that mass-loading-only 1D models are incomplete would likely survive, but the specific form of the energy sink is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"Guo, Kim, and Stone present 3D AthenaK simulations of a single supernova remnant expanding into a two-phase, thermally unstable cloudy medium (mean density 10 cm^-3), with radiative cooling and saturated isotropic thermal conduction, at grid resolutions up to 1/64 pc, and compare with a uniform-medium run and analytic Sedov-Taylor theory. The central claim is that in the cloudy medium the remnant still expands roughly as r ∝ t^(2/5) but sweeps up more mass and continuously loses energy at shock-cloud interfaces, yielding less hot gas mass, thermal energy, and radial momentum by the WNM shell-formation time than in a uniform medium of the same mean density. The authors measure a hot-gas mass-loading rate ~0.5 dM_S/dt and a spatially flat energy sink ~−0.2 de_S/dt, and use these measured rates to construct a minimal 1D spherical model with mass source and energy sink terms that they compare with the ST, White-Long, and Pittard solutions.","tokens_in":26803,"tokens_out":7865,"duration_ms":77821,"significance":"If the quantitative sink measurement is correct, this is an important and timely result: it identifies a physical process omitted from standard mass-loading treatments of SNR evolution and offers a simple parameterization for 1D models. The paper's strengths are the uniform-medium control, the comparison to the analytic ST solution, the multiple initial-condition seeds, the passive-scalar phase diagnostics, and the honest discussion of resolution limitations. The qualitative conclusion that interface cooling is energetically important is well supported. However, the specific claims of a flat cooling profile and the fitted constants A=0.5, C=−0.2 are not yet on secure footing because the cooling and Field lengths at the cloud interfaces are unresolved by two orders of magnitude, and the 1D model is calibrated to the very simulations it is then used to describe. These issues make the quantitative part of the paper provisional but do not invalidate the main qualitative message.","major_comments":[{"comment":"The central quantitative claim, that the interface energy sink is spatially flat at d(e)/dt ≈ −0.2 d(e_S)/dt and follows t^(−11/5), is not established by the resolution study as presented. The manuscript states in §4.3 that the cooling length and Field length at the cloud interface are l_cool ≈ 7×10^(−5) pc and l_F ≈ 3×10^(−5) pc, while the finest cells are Δx = 1/64 pc, so these scales are unresolved by roughly two orders of magnitude. The same section explicitly disclaims formal numerical convergence, and Figs. 8 and 9 show ~20% changes in hot-gas mass per resolution doubling and systematically higher density, pressure, and cooling rates in the conduction runs at higher resolution. Since the flat cooling profile is the direct input to the fitted constants in Eqs. (30)–(31), a resolution-dependent surface cooling rate would change the inferred sink shape and normalization. I ask for a quantitative resolution study of the radial cooling profile itself (not only the global mass and energy), and for a test of whether the Fielding et al. (2020) / Tan et al. (2021) convergence argument for sustained mixing layers applies to the transient, shock-dominated interaction studied here. The qualitative conclusion that cooling is important is likely to survive, but the specific shape and amplitude of the sink are not yet secure.","section":"§4.3, Figs. 8–9"},{"comment":"The 1D model is explicitly calibrated to the simulations: the constants A=0.5 and C=−0.2, the radial shapes k=φ=1, and the activation times for the source and sink terms are all chosen to match the measured mass-loading and cooling rates in Fig. 12. As a result, the statement that this model “better describes the structure of the simulated SNR” is a consistency check rather than an independent validation. To make the model scientifically load-bearing, the authors should demonstrate at least one out-of-sample prediction—for example, using the same A and C for the different heating-rate or larger-scale perturbation runs (T...-h10, T...-w16) and comparing the predicted radial structure with the simulations. If that is not possible, the calibrated and illustrative nature of the model should be stated more prominently in the abstract and conclusions.","section":"§5, Eqs. (30)–(31)"}],"minor_comments":[{"comment":"In the sentence reporting the seed-to-seed scatter, “~ %30 in energy” appears to be a typo for “~30% in energy”; please fix.","section":"§4.4"},{"comment":"The caption contains the fragment “The is a change by a factor of ~20% when doubling the resolution”; this should read “There is a change...”.","section":"Fig. 8 caption"},{"comment":"The legend entry “Yed cond” should read “Yes cond”.","section":"Fig. 13 legend"},{"comment":"The notation Ω_i in Eq. (29) is not defined; please specify that it is the solid angle of the spherical shell used for the angle average and how the finite-difference time derivative is computed.","section":"Eq. (29)"},{"comment":"The bullet list in §3.1 uses “T able 1” with a line break; also, the “Thermal Cond.” column in Table 2 could be labeled more clearly, and the notation “T...-2048” used in §4.3 and captions should be defined once (it appears to mean both the TN and TY series).","section":"§3.1 and Table 2"},{"comment":"The phrase “we suggest the constants A = 0.5 and C = −0.2” understates the fact that these values are fitted to the simulations; “we adopt” or “we calibrate” would be more accurate.","section":"§5, after Eq. (30)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the simulations appear carefully done. My recommendation is driven by the gap between the quantitative claims in the abstract (flat energy sink, fitted model constants) and the convergence support actually provided. A revision that either adds a targeted resolved shock-cloud test or explicitly reframes the flat-sink and 1D-model claims as resolution-dependent and illustrative would be acceptable. I would not support rejection, because the qualitative message is well anchored by the uniform control and multiple realizations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a well-executed simulation paper with a believable headline result: supernova remnants in a cloudy multiphase medium lose significant energy at shock-cloud interfaces, so late-time momentum is lower than uniform-medium estimates and mass-loading-only 1D models are incomplete. The paper earns that conclusion with multiple seeds, resolution tests, a uniform-medium control, and an honest discussion of what is and is not converged.\n\nThe genuinely new piece is the direct measurement of the energy sink: it is spatially flat, about -0.2 times the Sedov reference rate, and follows the t^-11/5 scaling that self-similarity requires. The 1D model that includes both mass source and energy sink reproduces the simulated structure qualitatively, which the earlier mass-loading models cannot. The resolution (up to 1/64 pc) with thermal conduction is a step up from Kim & Ostriker (2015), and the use of passive scalars to separate contributions from CNM, WNM, and ejecta is a useful diagnostic. The paper is also refreshingly explicit about its own limits: no formal convergence claim, the 1D model is an illustration, and the constants A=0.5 and C=-0.2 are calibrated to the simulations.\n\nThe soft spot is exactly where the stress-test places it. The flat cooling profile is measured at cloud interfaces where the cooling length and Field length are unresolved by about two orders of magnitude. The convergence argument leans on Fielding et al. (2020) and Tan et al. (2021) for sustained radiative turbulent mixing layers, and the paper's own Section 4.3 disclaims formal convergence, reports ~20% changes per resolution doubling, and shows conduction runs with systematically higher densities and cooling rates at higher resolution. So the specific form of the sink—flat profile, amplitude, and scaling—is not securely established. That matters because this form is the paper's main quantitative contribution. This is not a disqualifying shortcoming: the global energy and momentum time histories support the qualitative conclusion, and the authors are appropriately cautious. But the calibrated 1D constants should not be used as physical inputs without further evidence.\n\nThis paper is for ISM and galaxy-formation modelers who implement SN feedback. It deserves a serious referee. The right request would be a clearer resolution study targeting interface cooling, ideally with a subgrid model or a convergence test at higher resolution, and an explicit statement of which features of the sink are robust. I would accept it conditionally and would cite it for the qualitative result.","headline":"Solid simulation paper with a credible central result, but the novel flat energy sink and its scaling are calibrated rather than converged; the qualitative conclusion that mass-loading-only 1D models are incomplete is robust.","tokens_in":27365,"tokens_out":4628,"would_cite":true,"duration_ms":40674,"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 reports that supernova remnants in a cloudy interstellar medium still expand roughly like Sedov-Taylor blasts (radius ∝ t^(2/5)), but they lose energy continuously where their shocks hit cold clouds, ending with less hot gas…","keywords":["supernova remnant","multiphase interstellar medium","shock-cloud interaction","radiative cooling","turbulent mixing layer","thermal conduction","Sedov-Taylor stage","mass loading"],"falsifier":"Run the same cloudy-medium SNR at Δx = 1/128 pc and directly measure the angle-averaged energy sink ė(r). If the flat, ≈ -0.2 ė_S profile changes by more than the ~20% spread already seen between 1/8 pc and 1/64 pc runs, or if its time scaling departs from $t^{{-11/5}}$, then the claimed energy sink and the calibrated 1D model constants are numerical artifacts rather than converged physics.","tokens_in":26312,"feed_emoji":"💥","tokens_out":6090,"duration_ms":62758,"temperature":0.7,"pith_summary":"This paper reports high-resolution 3D hydrodynamic simulations of a supernova remnant expanding into a two-phase cloudy interstellar medium. Its central claim is that the remnant still expands roughly like a Sedov-Taylor blast (radius ∝ t^(2/5)) but behaves very differently in energy and mass: it sweeps up more mass from cold clouds while continuously losing thermal energy as shocks hit the clouds and the turbulent interfaces radiatively cool. As a result, at the time the outer shock becomes radiative the remnant contains roughly half the hot gas mass and about 80-90% of the radial momentum expected for a uniform medium of the same mean density, and the pressure-driven snowplow stage is suppressed. The paper further claims that most previous 1D models adding only a mass-loading term miss the dominant energy sink, and that a minimal 1D model with both a mass source and a spatially flat energy sink reproduces the simulated structure.","feed_headline":"Cold clouds bleed supernova remnants of energy and momentum","feed_subtitle":"High-resolution 3D runs reveal a flat interface-cooling rate that 1D mass-loading models miss, cutting final momentum.","key_machinery":"The load-bearing machinery is a set of direct 3D hydrodynamic simulations with radiative cooling and saturating thermal conduction, run at up to Δx = 1/64 pc so that shock-cloud interactions for most clouds are resolved. The key diagnostic is the direct measurement of the mass and energy source/sink terms in the angle-averaged evolution equations: a mass loading rate and an energy sink rate whose time scalings ($t^{{-1}}$ and $t^{{-11/5}}$, respectively) match the conditions for a self-similar solution with impurity terms. These measured rates are then used to construct a minimal 1D spherical model with both a mass source and an energy sink.","core_discovery":"On the paper's own terms, the discovery is that the energetics of a supernova remnant in a cloudy medium are controlled by radiative cooling in shock-cloud interfaces, not just by mass loading. Direct measurements from the simulations show a mass loading rate ṗ ≈ 0.5 ṗ_S that scales with the Sedov reference rate (and hence as $t^{{-1}}$), and a spatially flat energy loss rate ė ≈ -0.2 ė_S scaling as $t^{{-11/5}}$, where ṗ_S and ė_S are the standard reference rates for a uniform medium. This energy sink persists throughout the nominal Sedov-Taylor stage, so the hot gas mass, thermal energy, and terminal momentum are all reduced compared with uniform-medium predictions even though the expansion law stays close to r ∝ $t^{{2/5}}$. Thermal conduction is a secondary effect: it smooths the morphology and raises the hot-gas density by a factor of 3-5, but does not change the global dynamics.","pith_inferences":["If the t^{-11/5} flat energy sink is a robust feature of turbulent mixing layers, galaxy-scale simulations that do not resolve cloud interfaces are likely missing a distributed cooling channel; a subgrid prescription assigning a fixed ~20% of the Sedov energy flux to interface cooling would encode this cheaply.","The measured mass-loading time dependence (t^{-1}) matches the earlier evaporation-based self-similar ansatz, suggesting that framework can be rehabilitated simply by adding the measured energy sink rather than by abandoning impurity models.","The suppression of the momentum boost implies feedback implementations that tie star-formation regulation to SNR momentum may need to lower the momentum per supernova in clumpy, metal-rich environments, while mixed-morphology remnants should appear brighter and denser toward the center.","A testable extension is to vary the cloud volume filling factor or metallicity (cooling strength) and check whether the coefficients 0.5 and -0.2 change systematically, which would reveal whether the energy sink is set by interface surface area or by the cooling curve."],"forward_implications":["Radius-based estimates of SNR age and energy using the mean ambient density remain approximately valid, since the expansion stays near r ∝ t^{2/5}.","Mass, thermal energy, and momentum at shell formation are reduced by roughly factors of 2, 2-3, and 1.1-1.25 relative to uniform mean-density expectations, and the pressure-driven snowplow stage essentially disappears.","Thermal conduction mainly changes observable morphology (denser, smoother hot gas and enhanced central X-ray emission) without altering the global momentum budget.","One-dimensional impurity models need both a mass source and an energy sink, with rates ≈0.5 ṗ_S and ≈-0.2 ė_S, to describe the hot gas structure.","The suppression of the late momentum boost implies that the momentum delivered to the interstellar medium may be lower than classical uniform-medium predictions."],"supporting_citations":[{"why":"Establishes that interface mass and energy fluxes converge even when cooling lengths are unresolved, supporting the resolution claim in Sec 4.3.","marker":"Fielding et al. 2020"},{"why":"Same convergence result for turbulent mixing layers, used to justify not resolving the Field length.","marker":"Tan et al. 2021"},{"why":"The constant mass-loading 1D model that the simulations are inconsistent with; serves as the comparison baseline.","marker":"Pittard 2019"},{"why":"The self-similar conductive-evaporation solution whose mass-loading scaling (t^{-1}) matches the measured rate.","marker":"White & Long 1991"},{"why":"The classical evaporation theory predicting a modified expansion law that the simulations do not find.","marker":"McKee & Ostriker 1977"},{"why":"Provides the earlier simulation setup and the mean-density predictions used as references for mass and momentum.","marker":"Kim & Ostriker 2015"}],"fun_headline_variants":["Cloudy gas saps supernova remnants of momentum, not just mass","Supernova remnants lose energy to cloud interfaces, cutting momentum","Shock-cloud cooling drains supernova energy, slashing terminal momentum","Mass loading alone can't explain supernova remnant slowdown in clouds","Clouds steal supernova energy via flat cooling rate, limiting momentum"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central assumption is that unresolved structure at the cold-cloud/hot-gas interface—cooling lengths of order $10^{{-5}}$ pc and the associated Field length—does not change the integrated mass and energy exchange rates, so the measured flat energy sink is real physics and not a resolution artifact.","fun_headline_variants_meta":{"raw":{"variants":["Cloudy gas saps supernova remnants of momentum, not just mass","Supernova remnants lose energy to cloud interfaces, cutting momentum","Shock-cloud cooling drains supernova energy, slashing terminal momentum","Mass loading alone can't explain supernova remnant slowdown in clouds","Clouds steal supernova energy via flat cooling rate, limiting momentum"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001085,"raw_usage":{"total_tokens":4583,"prompt_tokens":1042,"completion_tokens":3541,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":658,"completion_tokens_details":{"reasoning_tokens":3452}},"tokens_in":658,"tokens_out":3541,"duration_ms":25608,"temperature":1.0,"reasoning_tokens":3452,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:10:26.395490+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same cloudy-medium SNR at Δx = 1/128 pc and directly measure the angle-averaged energy sink ė(r). If the flat, ≈ -0.2 ė_S profile changes by more than the ~20% spread already seen between 1/8 pc and 1/64 pc runs, or if its time scaling departs from $t^{{-11/5}}$, then the claimed energy sink and the calibrated 1D model constants are numerical artifacts rather than converged physics.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The constant mass-loading 1D model that the simulations are inconsistent with; serves as the comparison baseline."}],"review_version":1}