{"id":"3fadac7b-8f1e-428a-a1c9-8a8e9188832b","arxiv_id":"2607.21277","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"An axisymmetric general synthetic iterative scheme (AxiGSIS) solves the Shakhov kinetic equation with prescribed electrostatic fields, reproducing reference kinetic solutions with much faster steady-state convergence.","lead":"This paper builds an axisymmetric solver that couples kinetic and macroscopic equations to speed up simulations of rarefied gas and charged-particle flows. The solver matches slower reference simulations while using far fewer cells and iterations, especially at low Knudsen numbers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"GSIS consistency for the new axisymmetric/electrostatic terms is asserted, not shown: at a GSIS fixed point the three-step CIS prediction is not proven to have zero kinetic residual, so the accelerated solver may converge to a state that is not the kinetic solution.","rationale":"The paper's strongest claim—that AxiGSIS reproduces reference kinetic solutions and accelerates convergence—depends on the assumption that the synthetic-iteration fixed point coincides with the steady state of the underlying kinetic model. The reader's weakest assumption identifies precisely this: the consistency argument in §3.2 is inherited from Cartesian GSIS and asserted rather than proven for the new axisymmetric and electrostatic terms. I find this concern load-bearing. The equations of the algorithm do not automatically provide the claimed consistency: the HoT sources used in the synthetic system are frozen before the macroscopic update, the VDF correction (Eq. 51) alters only the equilibrium part, and after the three CIS prediction steps the kinetic residual is not guaranteed to vanish. In the axisymmetric case with T_geo and E-field source terms, any mismatch between the frozen HoTs and those computed from the final corrected VDF means the synthetic equations are not the exact moment system. The absence of a residual-based convergence check or grid-convergence study makes this difficult to rule out empirically. At the same time, the external Taylor–Couette validation and the favorable comparisons with 3D GSIS for the neutral nozzle provide partial support for the axisymmetric kinetic solver itself; the weakness is specifically in the GSIS fixed-point argument. The proposed residual-based test would directly settle whether the concern lands. Since the reader already returned a CONDITIONAL verdict, my stress test does not move the verdict; it sharpens the condition that should be met before the advertised speed and accuracy are taken at face value.","tokens_in":21570,"tokens_out":6468,"duration_ms":68752,"concrete_test":"After AxiGSIS converges for the Kn=0.01 charged-particle sphere case (or the neutral nozzle case at Kn=0.026), freeze the converged f and W and compute the normalized kinetic residual R_{i,α} = r_i(g_S−f)/τ − T_{x,r}^{(2)} − T_E^{(2)} − T_geo (Eq. 32), then perform one additional CIS sweep. If the residual norm is not reduced to the same order as when CIS itself converges (e.g., relative residual ≤1e-5), the GSIS fixed point is not a kinetic steady state. As a second leg, recompute HoTs from the converged f and W and re-solve the synthetic equations once; if the macroscopic residual changes by more than the convergence tolerance, the frozen-HoT assumption in Eq. (42) is violated. These two checks distinguish GSIS convergence from kinetic-equation convergence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that the converged AxiGSIS fixed point solves the axisymmetric kinetic equation (3). Section 3.2 attempts to justify this with the statement: 'At convergence, Δf→0 and the HoT-corrected synthetic equations are consistent with the original kinetic moment equations' (p. 15). That inference is not established for the axisymmetric/charged-particle setting. The HoT source in Eq. (42) is frozen from f^{k+1/2}, the VDF after only three CIS steps (Eqs. 33–34), before the macroscopic synthetic solve updates W to W^{k+1}. The VDF is then corrected by replacing only its Maxwellian part (Eq. 51). At a fixed point of this map, the CIS increments Δf^{k,s} need not each vanish; the converged f^{k+1} is not shown to make the kinetic residual R_{i,α} in Eq. (32) zero. Hence consistency requires that at convergence the frozen HoTs equal Π(f^{k+1})−Π_NSF(W^{k+1}) (and analogously for q), and that the synthetic solve plus Eq. (51) reproduces the kinetic moments; neither is proved. The axisymmetric geometric term T_geo (Eq. 28) and electrostatic source terms (Eq. 40) only make the gap larger, since the synthetic equations in Eq. (37) must match those terms exactly. Without such a check, agreement with CIS in Figs. 5 and 7 cannot fully distinguish a shared discretization bias from true convergence to the kinetic equation; no grid-convergence or residual-based verification is reported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes AxiGSIS, an axisymmetric finite-volume discrete velocity solver for steady rarefied gas flows and charged-particle transport under prescribed electrostatic fields. The physical space is reduced to the two-dimensional meridian plane while the molecular velocity space remains three-dimensional. The solver follows the GSIS paradigm: three conventional iterative kinetic steps produce a predicted VDF f^{k+1/2}; high-order stress and heat-flux corrections are frozen from this prediction; a macroscopic synthetic system with axisymmetric geometric and electrostatic source terms is then iterated; and the VDF is corrected by replacing only its Maxwellian part via Eq. (51). The scheme is validated on Taylor–Couette flow, neutral nozzle expansion, charged-particle flow past an electrostatic sphere, and charged-particle nozzle flow, with reported savings in iteration counts and core-hours relative to CIS and to a 3D GSIS solver.","tokens_in":21989,"tokens_out":3534,"duration_ms":40477,"significance":"If the central consistency claim holds, the contribution is valuable: it extends GSIS to axisymmetric geometries and to charged-particle kinetic transport with imposed electrostatic fields, a combination not previously available in this framework. The paper contains a genuinely useful algorithmic construction: detailed finite-volume discretizations, the T-UCE treatment of the axisymmetric geometric term, velocity-space integration compensation, a consistent half-range boundary treatment, and an MPI implementation. The Taylor–Couette comparison with the external reference of Tibbs et al. is a genuine independent validation, and the documented speedups in near-continuum regimes are substantial. The main caveat is that the load-bearing consistency statement in §3.2 is asserted rather than proved, and the numerical validation does not include a grid-convergence or residual-based verification that would rule out a converged solution different from the kinetic steady state.","major_comments":[{"comment":"The central claim that AxiGSIS converges to the solution of the axisymmetric kinetic equation (3) is not established. After three CIS prediction steps, the HoTs are frozen from f^{k+1/2}=f^{k,3}; the synthetic equations are solved; and the VDF is corrected by replacing only the Maxwellian part, Eq. (51). A fixed point of this outer iteration does not by itself imply that the kinetic residual R_{i,α} in Eq. (32) vanishes, nor that the frozen HoTs equal Π(f^{k+1})−Π_NSF(W^{k+1}) and the analogous heat-flux relation. This gap is larger in the present setting because the axisymmetric geometric term T_geo (Eq. 28) and the electrostatic source terms (Eq. 40) must be reproduced exactly by the synthetic system. Please provide either a proof of consistency specific to these terms or a numerical verification: for example, report the kinetic residual norm at the converged AxiGSIS state and demonstr","section":"§3.2, p.15, Eq. (51)"},{"comment":"The validation lacks grid-convergence studies and quantitative error metrics. In the neutral nozzle case, AxiGSIS is compared with a 3D GSIS solver from the same group (Ref. [41]); in the charged-particle cases, AxiGSIS is compared only with CIS on the same mesh and velocity grid. Agreement between two solvers that share the same geometric and electrostatic discretizations cannot fully separate discretization bias from true convergence to the kinetic equation. Please add at least one systematic refinement study — in both physical space and velocity space — for, say, Taylor–Couette flow and one charged-particle case, and report quantitative errors (e.g., L1 or L2 differences against a fine-grid reference) rather than only line/symbol agreement.","section":"§4.2–4.4, Figs. 3, 5, 7"},{"comment":"The charged-particle validation is entirely against the same CIS implementation and uses only prescribed electrostatic fields. This is acceptable for a solver demonstration, but it makes the electrostatic acceleration terms the least independently tested part of the scheme. A comparison against an independent reference (a different kinetic solver, or a manufactured solution with a known analytic force field) would significantly strengthen the claim that the new velocity-space acceleration discretization and synthetic source terms are correct.","section":"§4.3, Table 2"}],"minor_comments":[{"comment":"The table formatting is corrupted: the columns 'core-hours' and 'Iter' are not separated, so entries such as '617' and '7525671' cannot be parsed. Please reformat the table with clear column separators.","section":"Table 3"},{"comment":"The under-relaxation factor η_φ is introduced but its value is never specified. State the value used in the charged-particle cases.","section":"Appendix A, Eq. (A.6)"},{"comment":"The notation E_θ in Eq. (15) is used before the statement that E_θ=0 for the considered axisymmetric fields; define the azimuthal component consistently in the coordinate system introduction to avoid confusion.","section":"§2.1"},{"comment":"Contour comparisons between CIS and AxiGSIS are shown as filled contours and black contour lines; at the reproduced scale these are difficult to distinguish. Consider using overlaid line plots with quantitative differences for the contour-level comparisons.","section":"Fig. 7"}],"recommendation":"major_revision","confidential_remarks":"To the editor: the paper is technically substantial and likely publishable after revision. The main issue is not the algorithmic idea but the absence of a consistency proof and the reliance for validation on same-group or same-solver comparisons. I would ask the authors for either a specific consistency argument for the axisymmetric/electrostatic extension or a residual-based convergence verification. The citation pattern is partially self-referential (the neutral nozzle reference is from the same group), but the external Taylor–Couette benchmark and the documented speedups provide sufficient independent evidence for the method's practical value."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper extends the general synthetic iterative scheme to axisymmetric geometries and to charged-particle transport under prescribed electrostatic fields. That is genuinely new, and it is a non-trivial extension: the cylindrical geometric source terms and the velocity-space acceleration terms have to be treated consistently in both the kinetic and synthetic equations, and the authors do that carefully. The Taylor-Couette validation against Tibbs et al. is genuinely external, and the neutral-nozzle comparison against the same group's 3D GSIS is reassuring. The speedups reported are large and plausible, and the authors are honest that the gain vanishes at high Knudsen number.\n\nWhat the paper does well is the discretization detail: the T-UCE treatment of the geometric term, the velocity-space integration compensation, and the consistent boundary treatment for the synthetic equations are all described concretely, not hand-waved.\n\nThe soft spots are the ones the reader flagged, and they are real but not fatal. There is no grid-convergence study and no quantitative error metric, so 'excellent agreement' is visual. The charged-particle cases compare AxiGSIS only against CIS from the same code; that checks consistency of the acceleration, not accuracy of the physics. The efficiency comparison against 3D GSIS is not controlled (different mesh sizes and core counts), though as an end-to-end cost comparison it is still meaningful. The stress-test concern about the consistency proof is fair as far as it goes: the paper asserts 'at convergence Δf→0' without proving it or checking the kinetic residual numerically. I think the claim is very likely true for a converged fixed point—if the three CIS substeps from the fixed VDF leave it unchanged, the frozen HoTs are exact—but the authors should demonstrate it, e.g., with a residual plot or a grid-refinement study. That is a revision request, not a desk-reject reason.\n\nWho is this for? Anyone doing deterministic kinetic simulations of axisymmetric rarefied or charged-particle flows, especially in the near-continuum regime. It deserves a serious referee; I would send it out and ask for grid convergence and a residual-based consistency check, plus at least one independent charged-particle validation if available.","headline":"A worthwhile engineering extension of GSIS to axisymmetric and charged-particle flows; sounds right, but the missing grid-convergence study and the asserted consistency proof keep it from being more than conditionally solid.","tokens_in":22426,"tokens_out":2559,"would_cite":true,"duration_ms":25345,"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":"A solver that alternates kinetic and macroscopic updates reaches axisymmetric rarefied gas and charged-particle steady states in a fraction of the iterations.","keywords":["rarefied gas dynamics","general synthetic iterative scheme","axisymmetric kinetic equation","discrete velocity method","charged-particle transport","electrostatic field","Knudsen number","finite-volume method"],"falsifier":"On a refined sequence of physical and velocity grids for the Kn=0.01 charged-particle sphere or the Pin=1000 Pa nozzle, run AxiGSIS and the conventional iterative scheme to the same residual tolerance; if the converged surface pressure, heat flux, or centerline profiles diverge by more than discretization error between schemes, the consistency claim fails.","tokens_in":21453,"feed_emoji":"⚡","tokens_out":6299,"duration_ms":66040,"temperature":0.7,"pith_summary":"This paper is trying to establish that the general synthetic iterative scheme, previously confined to Cartesian neutral-gas settings, can be built for axisymmetric geometries and for charged particles moving through prescribed electrostatic fields. The new solver works on a two-dimensional meridian plane while keeping a full three-dimensional molecular velocity space, so it retains the circumferential-velocity effects that matter in cylindrical flows. It alternates kinetic iterations with fast macroscopic synthetic iterations: the kinetic solver feeds nonequilibrium stress and heat flux into the macroscopic equations, and the corrected macroscopic state is fed back into the distribution function. On four benchmark flows the converged results match reference kinetic solutions, while iteration counts drop from thousands to dozens in near-continuum regimes, cutting computational cost accordingly. If the claim holds, practical simulations of axisymmetric rarefied and charged-particle flows, such as nozzles, thrusters, and beam devices, become far more affordable.","feed_headline":"Axisymmetric gas solver: dozens of iterations, not thousands","feed_subtitle":"Extends fast synthetic coupling to charged particles; matches kinetic solutions with far fewer iterations.","key_machinery":"The load-bearing mechanism is the GSIS coupling itself: the kinetic solver supplies nonequilibrium stress and heat flux, which are converted into high-order corrective terms by subtracting their Navier–Stokes–Fourier estimates; these terms are frozen while the macroscopic conservative equations, with consistent axisymmetric geometric and electrostatic source terms, are iterated; then the equilibrium part of the distribution function is refreshed with the corrected macroscopic state. Around this core, the method stacks a finite-volume discrete velocity discretization, a trigonometric upwind conservative scheme for the cylindrical velocity-space angular transport, and an analytical equilibrium","core_discovery":"The paper's central claim is that the general synthetic iterative idea extends to axisymmetric geometries and to charged-particle transport under prescribed electrostatic fields. The resulting solver, AxiGSIS, solves the kinetic Shakhov equation on a 2D meridian plane with a full 3D velocity space; every few kinetic steps it extracts nonequilibrium stress and heat flux, subtracts Navier–Stokes–Fourier values to define high-order corrections, and iterates a macroscopic synthetic system until those corrections are the only nonequilibrium input. The corrected macroscopic state is then fed back to the distribution function. For four benchmark flows—Taylor–Couette flow, neutral nozzle expansion,","pith_inferences":["Inference: If the consistency claim survives, the same machinery should extend to self-consistent Poisson coupling, where particle density updates the potential; that would position AxiGSIS as a candidate for steady-state plasma sheath and thruster simulations, though the paper does not demonstrate it.","Inference: A natural stress test is a systematic grid-convergence study at small Knudsen number with strong electrostatic forcing, comparing AxiGSIS and the conventional kinetic scheme on successively refined physical and velocity meshes; the paper compares solutions on a single grid but reports no mesh-convergence study.","Inference: Because axisymmetric reduction removes one physical dimension while keeping the full velocity space, the speedup over a 3D solver should grow roughly with the number of azimuthal cells needed in a full 3D simulation; the neutral-nozzle numbers suggest an order-of-magnitude or more for devices with near-circular symmetry."],"forward_implications":["AxiGSIS reproduces reference kinetic solutions for all four benchmark flows, including the tangential velocity profile in Taylor–Couette flow and centerline profiles in the nozzle and sphere cases.","Near the continuum limit the acceleration is large: for charged-particle sphere flow at Kn=0.01, iterations fall from 3060 to 28 and core-hours from 328 to 10.8; for the charged nozzle at Pin=1000 Pa, iterations fall from 5671 to 17 and core-hours from 752 to 6.","At high Knudsen number, exemplified by Kn=1 for the sphere case, the acceleration is small because conventional kinetic iteration is already efficient.","For the neutral axisymmetric nozzle flow, AxiGSIS matches a full 3D GSIS solution while using roughly 3% of the physical cells and about a twentieth of the core-hours.","Charged-particle surface pressure and heat flux are reduced by a repulsive electrostatic field, and AxiGSIS captures this response in agreement with the conventional kinetic solver."],"fun_headline_variants":["Axisymmetric solver: dozens of iterations for gas and charged-particle flows","General synthetic iterative solver extends to axisymmetric charged-particle flows","Iterative synthetic solver cuts gas and charged-particle simulations to dozens of steps","New solver combines axisymmetric rarefied gas and electrostatic charged-particle flows","AxiGSIS: fast synthetic iterative solver for axisymmetric flows with charged particles"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The argument assumes that after three kinetic prediction steps, freezing the high-order nonequilibrium corrections and iterating the synthetic macroscopic equations, including the new axisymmetric geometric and electrostatic source terms, still converges to the same steady state as the underlying kinetic equation; this consistency is asserted from prior Cartesian development rather than proved for these new terms.","fun_headline_variants_meta":{"raw":{"variants":["Axisymmetric solver: dozens of iterations for gas and charged-particle flows","General synthetic iterative solver extends to axisymmetric charged-particle flows","Iterative synthetic solver cuts gas and charged-particle simulations to dozens of steps","New solver combines axisymmetric rarefied gas and electrostatic charged-particle flows","AxiGSIS: fast synthetic iterative solver for axisymmetric flows with charged particles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001044,"raw_usage":{"total_tokens":4219,"prompt_tokens":733,"completion_tokens":3486,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":477,"completion_tokens_details":{"reasoning_tokens":3392}},"tokens_in":477,"tokens_out":3486,"duration_ms":24221,"temperature":1.0,"reasoning_tokens":3392,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T07:55:35.769164+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a refined sequence of physical and velocity grids for the Kn=0.01 charged-particle sphere or the Pin=1000 Pa nozzle, run AxiGSIS and the conventional iterative scheme to the same residual tolerance; if the converged surface pressure, heat flux, or centerline profiles diverge by more than discretization error between schemes, the consistency claim fails.","supporting_citations":[],"review_version":1}