{"id":"48fb0014-d821-4f56-a690-39468d1e48ad","arxiv_id":"2507.11758","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Closed-form neutron wall-load formulas for spin-polarized fusion ring sources in convex axisymmetric tori, enabling fast gradient-based wall-shape optimization.","lead":"This paper derives exact, fast formulas for how many fusion neutrons hit the walls of a doughnut-shaped reactor when the fuel is spin-polarized. The formulas can be differentiated, so designers can let optimization software shape the walls to spread the neutron load evenly.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Convex-cross-section extension rests on an unproved single-interval visibility claim (Sec. 4.1); a brute-force occlusion test would settle whether Eq. 24 actually gives the correct integration limits.","rationale":"Read in good faith: the paper's core physics is standard, the lengthy algebra was generated by Rubi and is implemented in a released Python package, which gives some independent support, and the rectangular and angled-field derivations appear internally consistent. The only point where the stated scope ('arbitrary convex cross section') exceeds the proven material is the Sec. 4.1 visibility and integration-bounds algorithm. The reader's weakest_assumption identifies exactly this, and I agree. If a brute-force occlusion test closes the gap, the conditional verdict should become ACCEPT (or at least remain CONDITIONAL pending verification rather than being rejected). The concern is not that the authors are wrong, but that the load-bearing geometric claim is asserted rather than demonstrated, and no independent NWL comparison is presented. This is a falsifiable, localized check, not a wholesale attack.","tokens_in":23433,"tokens_out":24350,"duration_ms":314314,"concrete_test":"Take a convex five- or six-sided poloidal polygon that includes sloped inboard segments, a vertical center-stack segment, and an outboard segment. For each target point on each wall segment (including vertices) and for several interior source rings, compute phi_m by the Sec. 4.1 algorithm as implemented in anarrima. Independently, for phi from 0 to pi, construct the segment T-S(phi) and test exact intersection with each conical or cylindrical wall facet using a line-frustum solver (or ray-marching fine enough to resolve tangencies), recording the largest phi with no intersection in (0,1). If independent phi_m differs from the algorithm by more than 1e-6 rad for any tested pair, or if occlusion occurs for phi < phi_m, the visibility claim fails. As a secondary check, evaluate one g integral with the algorithm's bounds against direct quadrature of the underlying ring integral.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.1 is the only step that turns the exactly integrable rectangular/cylindrical cases into formulas for an arbitrary axisymmetric torus with convex poloidal cross section. The enabling assertion is that for such a cross section every source ring is visible from any wall point and the visible part of the ring is always a single symmetric interval [-phi_m, phi_m], with endpoints determined by tangency to an outward-facing wall segment. This is not trivial, because the solid torus is not convex in 3D: a straight chord between a boundary point and an interior ring point can leave the plasma through the inner wall or central hole even though the poloidal cross section is convex. The paper states the claim and gives the per-segment tangency formula (Eq. 24) plus the segment-selection logic (Eq. 25, z_c check), but it supplies no proof that the most restrictive tangent gives the exact first-occlusion angle, that no other wall feature (vertex or inward-facing segment) can block earlier, or that re-entry cannot create gaps. The 1.5% uniformity residuals in Sec. 6 demonstrate only internal consistency of the model, not agreement with true geometry. If the visibility claim fails for some convex shape, all general-convex NWL outputs in Secs. 4 and 6 are wrong; the rectangular formulas would still be correct if used with the right bounds. This is the load-bearing assumption of the paper's main generalization.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper derives closed-form analytic expressions for the neutron wall load (NWL) from filamentary ring sources in axisymmetric toroidal geometries, for rectangular and general convex poloidal cross-sections, with arbitrary spin-polarization emission patterns (A, B/C modes and mixtures) and arbitrary local magnetic-field directions. The toroidal integrals are reduced to algebraic expressions and incomplete elliptic integrals using the Rubi symbolic integration package, and the results are implemented in a Python/JAX library called anarrima. As demonstrations, the paper presents first-wall shapes optimized for uniform NWL for the A, isotropic, and B/C polarization modes. The main generalization to arbitrary convex cross-sections rests on a visibility algorithm in Section 4.1 that determines the integration limits by tangency to outward-facing wall segments.","tokens_in":23659,"tokens_out":17414,"duration_ms":209945,"significance":"If the formulas are correct, they provide a fast, differentiable surrogate for the previously numerical part of NWL scoping calculations, which is genuinely useful for systems codes and for building intuition about spin-polarization effects. Concrete strengths are the use of machine-checked symbolic integration, the public availability of the code and example scripts, the explicit statement of physical limitations (no scattering, no beam-target center-of-mass corrections), and the honest treatment of the polarization physics as taken from prior literature rather than fit to the new formulas. The optimized uniform-NWL shells with sub-1.5% residuals are a useful proof-of-concept, though they test only internal consistency. The main weaknesses are the unproved and potentially fragile visibility claim that underlies the general-convex formulas and the absence of any independent numerical validation of the long analytic expressions.","major_comments":[{"comment":"The claim that for any convex poloidal cross section every source ring is visible from any wall point and that the limiting angles ±φm are always set by tangency to an outward-facing wall segment is stated without proof. This is the only step that converts the exactly integrable rectangular/cylindrical formulas into the general-convex formulas used in Sections 4 and 6, so it is load-bearing. The projected chord from target to source in the poloidal plane is not a straight line: its cylindrical radius satisfies R(λ) = sqrt((1−λ)²r² + λ²p² + 2λ(1−λ)rp cosφ), so a tangency condition derived for the straight segment between the target and a source at radius p, as in Eq. (24), is not evidently equivalent to a 3D visibility test against the solid of revolution. The algorithm also tests only conical wall segments and never checks the vertex circles where adjacent segments meet, although a limiting sightline can pass through a vertex without being tangent to either segment. I request either a rigorous proof of the visibility statement with the exact hypotheses on the cross section, or a brute-force verification by ray-surface intersection on a battery of convex shapes, including sloped segments and vertex circles. Without this, the correctness of the general-convex NWL formulas is not established; the rectangular formulas would remain exact if supplied with the correct bounds.","section":"Sec. 4.1 (Eqs. 24-25)"},{"comment":"The final expressions are presented without any independent numerical check. The defining integrands are simple enough to be evaluated by adaptive quadrature or Monte Carlo, and such a comparison would also catch transcription errors in the typesetting of these very long formulas. The uniform-shell examples in Section 6 only demonstrate that the optimization routine converges on the model's own formulas; they cannot detect a wrong constant, a missing term, or a misapplied integration limit. Please add a validation section comparing the analytic g-expressions to direct numerical integration of the defining integrals (e.g., Eqs. (4), (8), (10), (15), (19), (21)) over a grid of p, r, z, α, and β, including both r < p and r > p cases, and for at least one nontrivial convex cross section with brute-force occlusion of the ring.","section":"Secs. 3-4 (Eqs. 16, 17, 22, 23, 27-30)"},{"comment":"The paper asserts that the F and E terms for arbitrary integration limits follow from the fixed-limit formulas by replacing u or v with r and setting φ = φm/2, but it does not give the general-φm F/E coefficients. This is not a trivial claim, because the fixed-limit formulas such as Eq. (20) have been simplified using the specific values of the limits, and it is not shown that the same coefficient forms survive for arbitrary φm. Since the general-normal formulas in Section 4.2 (Eqs. 26-30) rely on these F/E terms, the authors should either print the general-φm coefficients explicitly or provide a derivation showing that the adapted fixed-limit coefficients are limit-independent. As written, a reader implementing a tilted wall element cannot verify this essential piece without redoing the symbolic integration.","section":"Sec. 4.2 (Eqs. 6, 20, 26)"}],"minor_comments":[{"comment":"The section outline lists Section 5 before Section 4 in the paper's own ordering; please reorder the sentences so the numerical order matches the actual section sequence.","section":"Sec. 1.2"},{"comment":"The text says the geometry is \"shown in Figure 1,\" but the square-cross-section torus is Figure 2; Figure 1 shows the polarization emission triangles.","section":"Sec. 2, first paragraph"},{"comment":"The naming scheme for the h and g terms is introduced piecemeal across Sections 3 and 4; a table collecting the geometry subscripts (Ri, Rf, Ro, H, V, I, A, c, B, a) and the meaning of the trailing L, F, E letters would make the paper much easier to use.","section":"Secs. 3-4 notation"},{"comment":"The statement that F and E are \"as for sin²θ (see Eq. 16)\" should explicitly say that the amplitude and parameter are the same as in Eq. (16); otherwise a reader may carry over the wrong amplitude when reusing the component integrals.","section":"Appendix C, Eq. (41)"},{"comment":"The abstract claims the formulas \"can be evaluated very quickly,\" but no timing benchmark is reported; a short runtime comparison against quadrature would make this concrete and would also serve as a partial validation.","section":"Abstract and Sec. 5"}],"recommendation":"major_revision","confidential_remarks":"The paper is a formula-derivation contribution with released code, and it fits the journal reasonably. My main concern is the unproved visibility claim in Section 4.1: it is the only bridge to the paper's headline generalization, and it may fail at vertex circles or for sloped convex cross-sections. If the authors cannot supply a proof, I would prefer the paper to restrict its general claims to rectangular/cylindrical geometries and to present the convex case explicitly as an approximation validated by brute-force ray tracing, rather than as an exact result. I also find the complete absence of independent numerical validation unusual for a paper whose contribution is a library of extremely dense closed forms; a quadrature comparison should be a routine part of the revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a useful, honest paper. Schwartz has done what prior multifilament treatments left numeric: evaluated the toroidal integral for filament ring sources in closed form for isotropic, sin-squared (A-mode), and B/C-mode emission, including angled magnetic fields and arbitrary target-normal components. The formulas are long but built from algebra, trig, and incomplete elliptic integrals, and he ships a jax-compatible package with Carlson elliptic integral implementations, so they are fast and differentiable. That is a real contribution for scoping studies and gradient-based wall-shape optimization. The optimized uniform-NWL examples are reasonable demonstrations; the sub-1.5% residuals show internal consistency, not validation, and the paper does not overclaim.\n\nThe biggest soft spot is Section 4.1. The rectangular and floor formulas follow from exact geometry, and I believe those. But the generalization to arbitrary convex poloidal cross sections relies on the claim that every source ring is visible from any wall point and that the visible interval is always a single symmetric [-phi_m, phi_m] bounded by tangents to outward-facing segments. The 3D solid torus is not convex; a sightline can leave through the central hole, so this is not automatic. The paper states the claim and gives tangency formulas but no proof. The stress-test note lands: a brute-force occlusion test for a few non-trivial convex shapes would settle it. If the claim fails, the Section 4 formulas are wrong, though the rectangular and floor formulas still stand.\n\nSecond, there is no independent numerical verification of the dense formulas, no comparison to adaptive quadrature, ray tracing, or an existing NWL code. Given the length of Equations 16, 22, and 27-30, transcription errors are plausible and would be hard for a referee to catch by inspection. This is solvable; a few convergence tables in an appendix would do.\n\nThe notation is heavy, but the naming conventions are explained and the Appendix B component integrals help. Limitations are stated plainly: no scattering, no beam-target or knock-on sources, convex-only cross sections apart from the dipole exterior.\n\nWho this is for: fusion systems modelers who need fast, differentiable wall loads for optimization. It is not a transport paper and does not claim to be. I would send it out. The referee should ask for numerical checks of the general-convex integration bounds and at least one independent quadrature or Monte Carlo comparison.","headline":"Closed-form, differentiable NWL formulas are a genuinely useful scoping tool, but the general-convex extension rests on an unproved visibility claim and the dense algebra has not been checked against numerics.","tokens_in":24179,"tokens_out":2097,"would_cite":true,"duration_ms":28633,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper derives exact, differentiable formulas for the neutron wall load from spin-polarized fusion rings, turning wall-shape optimization into a fast gradient problem.","keywords":["neutron wall load","spin-polarized fusion","axisymmetric torus","analytic integration","elliptic integrals","wall shape optimization","polarization modes","tokamak"],"falsifier":"Take a convex poloidal cross section with an extended straight inboard segment and a target near one end; for each source ring, compare the visible intervals computed by the tangency formula with direct three-dimensional ray tracing. If any visible-interval endpoint is set by a line of sight that crosses the wall rather than being tangent to an outward-facing segment, the general-convex formulas fail; likewise, any convex cross section where a source ring's near arc is occluded would refute the claim.","tokens_in":23202,"feed_emoji":"⚛️","tokens_out":5495,"duration_ms":61529,"temperature":0.7,"pith_summary":"This paper derives closed-form analytic formulas for the neutron wall load (NWL) on the walls of an axisymmetric torus from thin ring-shaped fusion sources, including the directional neutron emission produced by spin-polarized fuel. The formulas work for arbitrary polarization mixtures (the A, B, and C modes and combinations) and for magnetic fields that are angled out of the purely toroidal direction. Because the expressions are fast to evaluate and automatically differentiable, they can replace numerical ray tracing in scoping studies and can be embedded directly in gradient-based optimization of first-wall shape. As demonstrations, the paper constructs toroidal shells with nearly uniform NWL for each main polarization mode, with residual nonuniformity below 1.5 percent. The price of this speed is scope: scattering is omitted, and the general convex-wall formulas rely on a geometric visibility assumption about which parts of a source ring a wall point can see.","feed_headline":"Spin-polarized fusion wall loads now computable in closed form","feed_subtitle":"Exact, differentiable expressions let designers shape walls to spread neutron load evenly, to under 1.5 percent error.","key_machinery":"The carrying object is the reduced geometric intensity g for a filamentary ring source: an integral over the visible toroidal angle of p cos(nu) times the angular emission factor (1 for isotropic emission, $sin^{2}$ $\\theta$ for the A mode, 1/4 + 3/4 $cos^{2}$ $\\theta$ for the B/C modes) divided by |$\\Delta$|^2. The paper evaluates all such integrals by decomposing the numerator into monomials over denominators (h - cos phi)^{3/2} or (h - cos phi)^{5/2}, each of which integrates to algebraic leading terms plus incomplete elliptic integrals F and E. The second piece of machinery is the visibility algorithm: for convex cross sections the integration limits are fixed by tangency of the line of sight to an outward-facing wall segment, computed from a stated formula, and arbitrary wall normals are assembled from horizontal and vertical component formulas.","core_discovery":"The central claim is that the neutron wall load from any axisymmetric filamentary ring source with the spin-polarized DT emission pattern can be reduced to algebraic functions of the source, target, and wall parameters plus incomplete elliptic integrals F and E. For rectangular-cross-section tori the paper writes exact intensities on inboard, floor, and outboard walls for isotropic emission, $sin^{2}$ $\\theta$ (A-mode), and 1/4 + 3/4 $cos^{2}$ $\\theta$ (B/C-mode) angular distributions, with the magnetic field either purely toroidal or at arbitrary axisymmetric angles. For arbitrary convex poloidal cross sections, the same kernels apply after replacing the fixed integration limits with bounds computed from sightlines tangent to outward-facing wall segments, and a wall element with arbitrary orientation is handled by a weighted sum of horizontal and vertical normal components. The author demonstrates the resulting formulas through gradient-optimized first-wall shapes with uniform NWL and provides an open-source implementation.","pith_inferences":["The visibility algorithm is stated for convex poloidal cross sections; extending the same kernels to non-convex geometries, such as diverted tokamaks or mirror end cells, would require up to two visible intervals per ring and is left for future work.","The absence of scattering means these formulas bound the direct component of wall load; adding a scattering correction on top of the analytic direct term could give a hybrid tool that is faster than full Monte Carlo while retaining accuracy in blankets.","The B/C modes' reduction of center-stack loading suggests a systems-level trade: sacrificing some fusion power (C mode) or keeping rate unchanged (B mode) to shrink the center stack and lower aspect ratio, a consequence the paper mentions but does not optimize."],"forward_implications":["NWL estimates that previously required numerical toroidal integration or Monte Carlo transport can be evaluated analytically, making wall-load scoping studies nearly instantaneous.","Because the formulas are automatically differentiable, first-wall shape, plasma shape, and polarization mix can be optimized jointly with constraints such as maximum wall load or peaking factor.","In the square-torus example, B/C polarization reduces center-stack neutron load by 43 percent at the midplane, which could relax shielding requirements on the center stack of a spherical tokamak.","Uniform-NWL shells built by gradient optimization for the A, isotropic, and B/C modes show residual nonuniformity below 1.5 percent, confirming that the formulas are usable in practice."],"supporting_citations":[{"why":"Supplies the spin-polarized DT differential cross section (a,b,c modes and angular factors sin^2 theta, 1/4 + 3/4 cos^2 theta) that the wall-load integrands are built from.","marker":"[13]"},{"why":"Defines the neutron wall load quantity and gives the deterministic stellarator method this work extends to analytic axisymmetric form.","marker":"[7]"},{"why":"Earlier ray-tracing NWL calculation whose numerical approach is superseded here by closed-form toroidal integration.","marker":"[10]"},{"why":"Multifilament NWL method with numerically integrated toroidal limits; the paper replaces those limits with analytic bounds.","marker":"[11]"},{"why":"Multifilament method applied to EAST; like [11] it leaves unclear how filament integration limits are determined.","marker":"[12]"},{"why":"Rule-based symbolic integration system used to derive the closed-form component integrals in Appendix B.","marker":"[25]"},{"why":"Carlson's duplication algorithms for symmetric elliptic integrals underlie the differentiable implementations used in the examples.","marker":"[19, 20, 21]"},{"why":"Line-torus intersection reference noting that even circular cross sections require solving a cubic for inward-facing limiting angles, in contrast to the convex outward-facing case.","marker":"[24]"}],"fun_headline_variants":["Spin-polarized fusion wall loads now exact and differentiable","Closed-form neutron wall loads for optimized fusion walls","Analytic spin-polarized wall loads aid fusion design","Exact neutron wall loading formulas for axisymmetric fusion","Uniform wall loading via closed-form spin-polarized fusion"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"For any convex torus cross section, the part of each source ring visible from any wall point is always cut off only by lines of sight that just graze an outward-facing wall segment; the paper states this claim but does not prove it.","fun_headline_variants_meta":{"raw":{"variants":["Spin-polarized fusion wall loads now exact and differentiable","Closed-form neutron wall loads for optimized fusion walls","Analytic spin-polarized wall loads aid fusion design","Exact neutron wall loading formulas for axisymmetric fusion","Uniform wall loading via closed-form spin-polarized fusion"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000228,"raw_usage":{"total_tokens":1438,"prompt_tokens":874,"completion_tokens":564,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":490,"completion_tokens_details":{"reasoning_tokens":489}},"tokens_in":490,"tokens_out":564,"duration_ms":7050,"temperature":1.0,"reasoning_tokens":489,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:02:11.233855+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a convex poloidal cross section with an extended straight inboard segment and a target near one end; for each source ring, compare the visible intervals computed by the tangency formula with direct three-dimensional ray tracing. If any visible-interval endpoint is set by a line of sight that crosses the wall rather than being tangent to an outward-facing segment, the general-convex formulas fail; likewise, any convex cross section where a source ring's near arc is occluded would refute the claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the spin-polarized DT differential cross section (a,b,c modes and angular factors sin^2 theta, 1/4 + 3/4 cos^2 theta) that the wall-load integrands are built from."},{"cited_title":"A deterministic method for the fast evaluation and opti- misation of the 3D neutron wall load for generic stellarator configurations.Nuclear Fusion, 62(7):076040, July 2022","cited_arxiv_id":null,"evidence_quote":"Defines the neutron wall load quantity and gives the deterministic stellarator method this work extends to analytic axisymmetric form."},{"cited_title":"Chapin and William G","cited_arxiv_id":null,"evidence_quote":"Earlier ray-tracing NWL calculation whose numerical approach is superseded here by closed-form toroidal integration."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Multifilament NWL method with numerically integrated toroidal limits; the paper replaces those limits with analytic bounds."},{"cited_title":"Rule-based integration: An extensive system of symbolic integration rules","cited_arxiv_id":null,"evidence_quote":"Rule-based symbolic integration system used to derive the closed-form component integrals in Appendix B."},{"cited_title":"Line-Torus Intersection for Ray Tracing: Alternative Formulations","cited_arxiv_id":null,"evidence_quote":"Line-torus intersection reference noting that even circular cross sections require solving a cubic for inward-facing limiting angles, in contrast to the convex outward-facing case."}],"review_version":1}