{"id":"581f6b31-57cd-4131-9e3f-5655f89f33ce","arxiv_id":"2505.21980","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Tracer dye experiments in a Milky Way-mass galaxy simulation show that circumgalactic gas mixing is best predicted by velocity dispersion and shear tensors, with nearly linear diffusion scaling.","lead":"This paper inserts tracer dyes into a simulated galaxy's surrounding gas to measure how quickly gas mixes in different environments. It finds that local velocity dispersion and shear are the best predictors of mixing, and uses this to calibrate diffusion constants used in galaxy simulations.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central correlation ranking may be an artifact of numerical diffusion that scales with velocity dispersion; only a 1 kpc vs 2 kpc comparison is offered, so the robustness claim is not actually tested.","rationale":"The paper is careful and its tracer-dye methodology is well matched to the question; the PDF comparison in Fig. 3 supports sample diversity, and the qualitative agreement between single-cell and eight-cell injection in Fig. B2 and Table 4 is genuine evidence. The reader's conditional verdict is appropriate. My stress-test identifies the same load-bearing point: the robustness of the correlation ranking to numerical diffusion is asserted but not demonstrated. A fixed 1 kpc cell size does not imply a fixed, environment-independent numerical diffusivity in Arepo; numerical diffusion depends on local flow structure, so it can correlate with sigma_vel and shear. The 1 kpc vs 2 kpc comparison cannot distinguish physical mixing from numerical mixing; it only shows what happens when numerical diffusion is stronger. The power-law exponent n ≈ 1.1 and the calibration constants C and C_s in Section 3.2 are quantitatively affected by this ambiguity, although the results would still be useful as a calibration of the numerical scheme. Because the paper itself flags numerical mixing as significant and the requested higher-resolution test is expensive but well-defined, I do not recommend rejection; the appropriate condition is to provide a higher-resolution check or to soften the physical interpretation of the correlations.","tokens_in":36841,"tokens_out":3974,"duration_ms":50457,"concrete_test":"Re-run a subset of roughly 20 dye injections covering all five flow types with the same halo and refinement strategy but at 500 pc (or 250 pc if feasible) minimum cell size, track the dye for 200 Myr, and recompute the Spearman coefficients between |Delta_dye| and sigma_vel, |S*_ij|, rho_over, KE, and B at the same 5, 10, 20 kpc box sizes, plus the power-law exponent n in Section 3.2. If the velocity-derived predictors drop materially (e.g., below about 0.5) or the ranking of predictors changes, the fixed-resolution robustness claim fails; if r_s remains around 0.7-0.9 with the same ordering and n remains near 1.1, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.5 and Fig. 12 show that going from 1 kpc to 2 kpc resolution doubles the dye spread, and the authors explicitly state that gas mixing at this resolution is significantly affected by numerical diffusion. The robustness argument is that a fixed spatial resolution keeps the numerical contribution comparable across environments. However, in a moving-mesh code such as Arepo, numerical diffusivity is not a function of cell size alone; it depends on local velocity jumps, mesh motion, and concentration gradients, and in practice it increases with the same resolved velocity dispersion and shear that the paper identifies as the physical predictors. Thus the Spearman correlations in Section 3.1 and Section 3.3, as well as the diffusion scaling n ≈ 1.1 in Section 3.2, could be produced by numerical diffusion tracing sigma_vel, rather than by resolved turbulent mixing. The only resolution test compares two diffusion-dominated regimes (1 kpc and 2 kpc); it shows that correlations survive when numerical diffusion is stronger, not that they survive when it is weaker. The central claim therefore rests on an untested assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses Arepo zoom-in simulations of a Milky Way-mass halo at 1 kpc and 2 kpc resolution, injects passive dye scalars at 95 CGM locations, and correlates the dye spread extent and shape at 200 Myr with local gas properties. It reports that velocity dispersion and traceless symmetric shear magnitude in small regions (≤ 5 kpc) best predict the extent, while the dispersion of stretching and plane-of-rotation eigenvectors in larger regions (≥ 10 kpc) best predicts the shape. It derives diffusion calibration constants C ≈ 0.09–0.7, Cs ≈ 0.2–0.4, a nearly linear power-law exponent n ≈ 1.1 for the diffusion coefficient versus velocity dispersion, and a superdiffusive temporal scaling. The paper acknowledges that numerical diffusion significantly affects the absolute mixing and that the resolution is insufficient to resolve the turbulent cascade, but it claims the correlation results are robust because the spatial resolution is fixed throughout the CGM.","tokens_in":37016,"tokens_out":4015,"duration_ms":44920,"significance":"If the correlation result is trustworthy, the paper provides practical calibration constants for subgrid-scale mixing models in SPH and moving-mesh codes, with C and Cs values consistent with literature ranges. The dye-injection methodology and the explicit treatment of different CGM flow environments are strengths, as is the transparent acknowledgment of numerical diffusion and the Reynolds-number analysis. However, the central quantitative claims—especially the n ≈ 1.1 scaling and the robustness statement—rest on an untested assumption about numerical diffusivity, so the significance is conditional on a stronger resolution or numerical-diffusion test.","major_comments":[{"comment":"The claim that fixed spatial resolution makes the correlation results robust is not tested by the 1 kpc versus 2 kpc comparison. Both runs are in the numerically diffusive regime: the paper states that gas mixing at these resolutions is significantly affected by numerical diffusion, and Fig. 12 shows that the dye spread doubles when going from 1 kpc to 2 kpc. In a moving-mesh code, numerical diffusivity is not a function of cell size alone; it depends on local velocity jumps, mesh motion, and concentration gradients, and it can plausibly increase with the same velocity dispersion and shear that the paper identifies as physical predictors. The offered test shows only that the correlations survive when numerical diffusion is stronger; it does not show they survive when numerical diffusion is weaker. A higher-resolution run (for example, a few hundred pc in the outer CGM, which Section 3.6 identifies as the required resolution) or a quantitative estimate of the numerical diffusivity contribution is needed to support the robustness claim.","section":"Section 3.5, Fig. 12"},{"comment":"The diffusion coefficient is defined from the same dye spread Δ_dye that is used to fit the power law and to compute C and Cs. The paper correctly frames this as calibration rather than prediction, but the fitted exponent n ≈ 1.1 is therefore not an independent test of physical turbulent diffusion; it is a fit to the simulation's effective diffusivity, which includes a dominant numerical component at 1 kpc resolution. Without separating numerical from physical diffusion, the n ≈ 1.1 scaling may describe how numerical diffusivity scales with velocity dispersion rather than how CGM turbulent mixing scales. The statement in Section 4, point (iii), that the exponent is 'robust to changes in resolution' is supported only by the 1 kpc versus 2 kpc comparison, which does not reduce numerical diffusivity.","section":"Section 3.2, Eqs. (6)–(8), Fig. 8"},{"comment":"The superdiffusion/hyperballistic conclusion is based on power-law exponents β = 1.06, 1.16, and 1.63, but the simulations are in a regime where numerical diffusion is significant, and the paper itself lists numerical diffusion as a possible cause of the superdiffusive behaviour. The spread in β across resolutions and injection sizes is large, and the average is skewed by outliers (β > 4 for a few dyes). Without a demonstration that β converges as numerical diffusion is reduced, the abstract's statement that the linear temporal dependence 'suggests superdiffusion in the CGM' is not established. The caution later in the text that the behaviour may not apply to the general CGM is welcome but is in tension with the abstract's unqualified claim.","section":"Section 3.5 and Table 4"}],"minor_comments":[{"comment":"The word 'resollution' should be 'resolution' in the sentence about stellar mass and H I column densities changing with resolution.","section":"Introduction, paragraph 4"},{"comment":"The phrase 'the the dye both advects and diffuses' contains a duplicated article and should read 'the dye both advects and diffuses'.","section":"Section 2.2"},{"comment":"The text 'standard + 2kpcsspatial refinement' contains a typo ('kpcss') and should be 'standard + 2 kpc spatial refinement'.","section":"Section 3.5, first paragraph"},{"comment":"The notation Δ_dye is used elsewhere in the paper for the 1st–99th percentile spread, but in Section 3.2 it is redefined as the standard deviation for the diffusion calculation; this change of definition should be flagged more prominently to avoid confusion.","section":"Section 3.2, Eq. (6)"},{"comment":"The acknowledgements thank the referee by name ('our referee, Douglas Rennehan'); this is unconventional in a submitted manuscript and should be removed or anonymized before publication.","section":"Acknowledgements"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about numerical diffusion is valid and load-bearing. The paper's own Fig. 12 and Section 3.5 show that absolute mixing is not converged and that numerical diffusion is significant. The robustness claim requires either a higher-resolution test or a quantitative treatment of numerical diffusivity; without that, the correlation rankings and the fitted diffusion exponent could be artifacts of numerical diffusion tracing velocity dispersion. The manuscript is otherwise well executed and transparent, and the conclusions would be valuable if the robustness concern is addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does something genuinely useful: it injects passive tracer dyes into a cosmological zoom-in of a Milky Way-mass CGM with magnetic fields and asks which local gas properties predict the extent and shape of mixing. That is a natural question, and this is the first systematic answer in a realistic CGM context. The analysis is thorough, with multiple flow types, bootstrap uncertainties, and robustness checks. The main result, that velocity dispersion and shear tensor magnitudes predict dye spread better than density, temperature, or metallicity, is plausible and consistent with expectations. The derived scalings (D ∝ sigma_vel^1.1, Smagorinsky constant ~0.2–0.4) will be handy for people calibrating subgrid mixing models in SPH or mesh codes, and the paper is honest that these are calibration constants from a fitting procedure, not predictions.\n\nThe soft spot is exactly where the stress-test lands: the claim that the correlation ranking is robust because spatial resolution is fixed. In Arepo, numerical diffusion is not simply a function of cell size; it depends on local velocity jumps and gradients, so it can scale with the same sigma_vel and shear that the paper identifies as physical drivers. The 1 kpc vs 2 kpc comparison shows the ranking survives when numerical diffusion is stronger, but it does not show what happens when numerical diffusion is reduced. That would need a several-hundred-pc run, which is expensive but not impossible for a single halo. So the abstract's phrasing that the correlation results are \"robust thanks to fixed spatial resolution\" is a bit too strong; a more accurate statement would be that they are robust to a doubling of numerical diffusion, but a physical origin is not independently confirmed.\n\nThat caveat is real but not fatal. The paper never oversells the absolute mixing rates, it flags the superdiffusion exponent as potentially numerical, and the direction alignment with shear eigenvectors is harder to explain away. If the referee asks for a higher-resolution test or a more careful statement about numerical diffusion, the paper can be revised accordingly.\n\nI'd send this to a serious referee. It deserves time even if, on a second reading, the numerical-diffusion confound turns out to be severe. The community needs this kind of calibration data, and the paper is transparent about what it can and cannot constrain.","headline":"A careful tracer-dye study of CGM mixing; the correlations are plausible and useful for calibrating subgrid models, but the robustness claim rests on a resolution comparison that does not fully rule out numerical diffusion driving the ranking.","tokens_in":37547,"tokens_out":1683,"would_cite":true,"duration_ms":20378,"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 a simulated Milky Way-mass halo, local velocity dispersion and pure-shear deformation determine how far and in which direction circumgalactic gas mixes.","keywords":["circumgalactic medium","gas mixing","turbulent diffusion","velocity dispersion","shear tensor","Smagorinsky model","cosmological zoom-in simulations","passive tracer dyes"],"falsifier":"A higher-resolution simulation (e.g., ~300 pc or better in the same halo) that reduces numerical diffusion while keeping the same physics: if the Spearman correlations between velocity dispersion/shear and dye spread drop significantly or change ranking compared to the 1 kpc run, the claim that fixed resolution preserves relative mixing rates would be falsified.","tokens_in":36621,"feed_emoji":"🌌","tokens_out":4809,"duration_ms":44171,"temperature":0.7,"pith_summary":"This paper aims to identify which local gas properties control how much and in which direction gas mixes in the circumgalactic medium (CGM) of a Milky Way-mass galaxy. Using cosmological magnetohydrodynamic zoom-in simulations with tracer dyes injected in 95 diverse CGM environments, the authors show that after 200 Myr the extent of dye spread is best predicted by velocity dispersion and the magnitude of the traceless symmetric shear tensor measured in small (≤5 kpc) regions, while the shape of the mixed dye aligns best with the dispersion of stretching and plane-of-rotation eigenvectors in larger (≥10 kpc) regions. The paper uses these correlations to calibrate diffusion constants for subgrid mixing models, finding a nearly linear (n≈1.1) power-law relation between diffusion coefficient and velocity dispersion, consistent with superdiffusive mixing. If correct, these results give simulation codes a direct, physically motivated way to model metal, heat, and magnetic-field diffusion in the CGM.","feed_headline":"Shear and velocity spread forecast CGM gas mixing","feed_subtitle":"Tracer dyes in a Milky Way-mass halo show pure shear predicts the extent and shape of mixed gas.","key_machinery":"The central objects are the tracer dyes, passive scalars injected into single cells and advected with the flow, and the velocity-gradient decomposition into symmetric (strain), traceless symmetric (pure shear), and antisymmetric (vorticity) tensors derived from the velocity field's partial derivatives. Eigenvectors of the traceless symmetric tensor define stretching/compression directions; the antisymmetric tensor's eigenvector gives the plane of rotation. The paper correlates dye spread magnitude and direction with these tensor statistics and velocity dispersion at multiple box sizes and times, and fits power-law and scaling-constant relations to convert the correlation into diffusion coefficients usable in subgrid models.","core_discovery":"The central claim is that the local velocity structure, specifically the velocity dispersion and the pure-shear (traceless symmetric) part of the velocity gradient tensor, determines both how far and in which direction gas from a point source mixes. Extent after 200 Myr correlates most strongly (Spearman r_s≈0.86–0.87) with velocity dispersion and |S*_ij| in 5 kpc regions around the injection site, with the correlation peaking ~70–120 Myr after injection, implying a delay in the transfer of local shear to mixing. The direction of spread aligns best (≈70% of dyes within 9.6°) with the dispersion of the stretching eigenvector of the traceless symmetric shear tensor and the plane-of-rotation eigenvector of the antisymmetric (vorticity) tensor evaluated in ≥10 kpc boxes. The derived subgrid calibration constants—scaling constant C≈0.1–0.6 and Smagorinsky constant C_s≈0.2–0.4—match values used in SPH subgrid models, and the dye spread grows as t^β with β≈1.06, indicating superdiffusive/hyperballistic mixing over the 200 Myr window.","pith_inferences":["The correlation framework could be tested observationally by comparing CGM metal distribution with velocity dispersion maps inferred from absorption-line kinematics.","The ~100 Myr delay between shear and maximum correlation suggests that mixing models might need a time-lag or memory term, not just an instantaneous diffusivity.","If numerical diffusion dominates absolute mixing, the derived constants may be numerical rather than physical; higher-resolution runs could reveal whether C and C_s converge to the same values.","The same dye method could be applied to ram-pressure stripping or satellite wakes to test whether the same shear statistics predict mixing in other environments."],"forward_implications":["Subgrid mixing models in SPH and other codes can calibrate diffusivity using velocity dispersion and |S*_ij|, with C≈0.1–0.6 and C_s≈0.2–0.4, rather than assuming universal constants.","The diffusion coefficient scales nearly linearly (n≈1.1) with velocity dispersion, implying a dynamic, time-dependent diffusivity rather than a constant.","Mixing is superdiffusive/hyperballistic over 200 Myr, so constant diffusion coefficients underestimate mixing on CGM timescales.","The radial decrease in velocity dispersion implies less mixing at larger galactocentric radii, affecting metal and magnetic field distribution in the CGM.","Resolving outer CGM turbulence requires a few hundred pc resolution (vs current 1 kpc), providing a target for future simulations."],"supporting_citations":[{"why":"Supplies the shear-based diffusivity formula (Equation 5) that the paper calibrates.","marker":"Smagorinsky 1963"},{"why":"Provides the superdiffusive mixing result and the Mach-number scaling that the paper compares against.","marker":"Colbrook et al. 2017"},{"why":"Gives the range of scaling constants (0.1–1) used in SPH subgrid models that the paper's C values are compared with.","marker":"Williamson et al. 2016"},{"why":"Provides the cosmological MHD simulations of Milky Way-mass halos that this paper's zoom-in runs are based on.","marker":"van de Voort et al. 2021"},{"why":"Defines the Auriga simulation suite and galaxy formation model used for the initial conditions.","marker":"Grand et al. 2017"},{"why":"Introduces the arepo moving-mesh code that the simulations use.","marker":"Springel 2010"},{"why":"Shows how SPH codes use velocity dispersion and shear for subgrid mixing, motivating the calibration approach.","marker":"Wadsley et al. 2008"},{"why":"Demonstrates that the Smagorinsky constant should be space- and time-dependent, motivating the paper's calibration of it.","marker":"Rennehan et al. 2019"}],"fun_headline_variants":["Shear statistics best predict CGM gas mixing","Superdiffusive CGM mixing driven by local shear","Velocity dispersion and shear shape gas mixing","CGM gas mixing forecast by shear and dispersion","Pure shear predicts extent and shape of gas mixing"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the relative ranking of mixing rates across different CGM environments is preserved even though numerical diffusion dominates the total mixing at the 1 kpc resolution used, so a 2 kpc comparison run is treated as sufficient to establish robustness without a higher-resolution run that actually reduces numerical diffusion.","fun_headline_variants_meta":{"raw":{"variants":["Shear statistics best predict CGM gas mixing","Superdiffusive CGM mixing driven by local shear","Velocity dispersion and shear shape gas mixing","CGM gas mixing forecast by shear and dispersion","Pure shear predicts extent and shape of gas mixing"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00083,"raw_usage":{"total_tokens":3693,"prompt_tokens":1081,"completion_tokens":2612,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":697,"completion_tokens_details":{"reasoning_tokens":2542}},"tokens_in":697,"tokens_out":2612,"duration_ms":16077,"temperature":1.0,"reasoning_tokens":2542,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:17:32.484571+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A higher-resolution simulation (e.g., ~300 pc or better in the same halo) that reduces numerical diffusion while keeping the same physics: if the Spearman correlations between velocity dispersion/shear and dye spread drop significantly or change ranking compared to the 1 kpc run, the claim that fixed resolution preserves relative mixing rates would be falsified.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the range of scaling constants (0.1–1) used in SPH subgrid models that the paper's C values are compared with."}],"review_version":1}