{"id":"5b7f3d23-b5f6-42b7-9943-558e012b5a49","arxiv_id":"hep-th/9907189","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The Type II worldsheet path integral over a D-brane is anomalous unless the brane's normal bundle has vanishing third Stiefel-Whitney class, which is precisely the condition for D-brane charge to be a K-theory class.","lead":"Freed and Witten compute the sign ambiguity, a global anomaly, that the Type II string path integral picks up when a D-brane is present, and show it is governed by the second Stiefel-Whitney class of the brane's normal bundle. The result forces D-brane worldvolumes to carry a Spin-c structure, which is exactly what is needed for D-brane charge to live in K-theory.","discovery_kind":"unclear","skeptic_critique":{"model":"selfhosted/deepseek-v4.1-flash","headline":"Theorem 5.6 is rigorously established for the connection the authors define on the doubled surface; the load-bearing soft spot is the unproven identification of that connection with the physical Quillen connection of the local-boundary-condition problem, which §6 outlines but does not complete.","rationale":"The reader's weakest_assumption is well-targeted: the doubling step is where the geometric content of Theorem 5.6 is concentrated, and §6 is explicit that the geometric index theory for local boundary conditions is only outlined. I mark agreement as partial rather than agree because the reader's framing — 'if the doubling map does not preserve the metric and connection, the holonomy computed in Theorem 5.6 belongs to a different bundle' — slightly misstates the structure: the authors define the connection on the double, so the holonomy is by construction that of the double's connection. The residual question is whether that definition coincides with the Quillen connection of the boundary-value problem, which is what the §6 gluing argument is meant to establish. That is a narrower and more tractable concern than the reader's phrasing suggests, and it does not undermine the proven topological statement (Theorem 5.5, Prop. 5.8), which has three independent proofs. The paper's other acknowledged gap — footnotes 2 and 3, that the cohomological condition is sufficient but not shown necessary, with higher-genus detection deferred to factorization/unitarity — limits the converse direction of the physical claim but not the theorem. The cited machinery (Atiyah-Patodi-Singer, Bismut-Freed, Booss-Bavnbek-Wojciechowski, Nicolaescu) is published and standard, and the self-citations to [W3] and [W4] are consistency checks, not premises. The verdict should remain ACCEPT: the central theorem is proven, the acknowledged gaps are transparently flagged, and neither moves the assessment of the paper's core contribution.","tokens_in":30772,"tokens_out":11621,"duration_ms":213047,"concrete_test":"Complete the §6 gluing computation for the cylinder factor: verify that the determinant line bundle of the Dirac operator on C3 with the antiperiodic condition χ(1) = −χ(−1) is canonically trivial, with trivial metric and connection, and that the gluing map U is an isometry intertwining the boundary-value Dirac operator with D+ on the double up to that factor. If the C3 determinant line carries nontrivial holonomy, the identification of the physical connection with the double's connection fails and Theorem 5.6 does not transfer to the worldsheet measure.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof chain is: L = L' ⊗ (ratio), where L' is identified with L+γ on the double (Prop. 5.8, second proof), the ratio's holonomy is computed by the flat index theorem to be the w2(ν) pairing (Prop. 5.9), and L+γ is trivializable with its canonical connection (Prop. 5.10). The weak link is the identification of L' with L+γ as bundles with metric and connection, which rests on the gluing map U of §6 (Figures 1–2). The authors state plainly: 'We only indicate a rough outline of the arguments; we have not carried through the details.' Their fallback — 'we adapted the practical point of view of defining the metric and connection directly from the Dirac operator on the double' — makes Theorem 5.6 true by construction, but shifts the burden to showing that this is the connection the worldsheet path integral actually produces. If the determinant line of the cylinder factor C3 with the antiperiodic condition χ(1) = −χ(−1) is not canonically trivial with trivial connection, the physical and double-induced connections differ by a flat line bundle whose holonomy need not be ±1 in the same way, and the anomaly formula would not transfer. A secondary, acknowledged limitation is footnotes 2 and 3: the exclusion of Q with w2(ν) ≠ 0 is proven only when the anomaly is detectable by a torus family; higher-genus detection is deferred to unproven factorization/unitarity arguments.","agreement_with_reader":"partial"},"referee_report":{"model":"selfhosted/deepseek-v4.1-flash","summary":"The paper computes the global worldsheet anomaly for Type II superstrings ending on a D-brane Q inside a spin ten-manifold Y. With the B-field topologically trivial, the pfaffian of the open-string Dirac operator with the local boundary conditions (5.3) is not a function but a section of a line bundle over the space of bosons. Theorem 5.5 identifies the topology of that line bundle: it is the complexification of a real line bundle whose class in H^1(Z;Z/2) is the pushforward of w2(ν), the second Stiefel-Whitney class of the normal bundle, equivalently first Chern class the pushforward of W3(ν). Theorem 5.6 computes the holonomy of the canonical connection as ±1, with the sign the pairing of the boundary family with w2(ν). Cancellation against the A-field then forces A to be a Spin^c connection rather than an honest U(1) connection, so that Q must be Spin^c (W3(ν)=0) when B is trivial, with the twisted generalization (1.12)/(6.9) when B is not. Supporting results include closed-string anomaly cancellation (Theorem 4.5, Corollary 4.6), the R^10 conformal statement (Theorem 4.7), and a Čech description of the A- and B-fields in §6.","tokens_in":31060,"tokens_out":5923,"duration_ms":106581,"significance":"This is the perturbative derivation of the Spin^c/K-theory condition on D-branes that had previously been known only from the nonperturbative analysis of [MM, W4, W3], and it is the paper that made the w2(ν) anomaly a standard tool. The argument is unusually well supported for a result of this kind: Proposition 5.8 is given three independent proofs (the Agranovich-Dynin boundary-symbol computation, KO-theoretic excision, and a characteristic-class computation), and the topological Theorem 5.5 does not depend on the geometric subtleties discussed below. The paper also supplies concrete, checkable consequences: CP^2 is allowed but must carry half-integral flux, the non-Spin^c CP^2 bundle over S^1 is excluded in R^10, and the D0-brane charge formula (1.14) becomes integral via Atiyah-Singer once A is read as a Spin^c connection. The attribution of the K-theory interpretation to prior work is handled carefully, and the abstract's claim [abstract:s2] is presented as a consistency statement rather than a new derivation.","major_comments":[{"comment":"The geometric index theory for Dirac operators with local boundary conditions is explicitly left incomplete: the section states 'We only indicate a rough outline of the arguments; we have not carried through the details,' and the fallback is that 'we adapted the practical point of view of defining the metric and connection directly from the Dirac operator on the double.' Read literally, Theorem 5.6 is then a theorem about the double-induced connection, and the identification of that connection with the physical Quillen connection of the boundary-value problem rests on two unproved ingredients: that the gluing map U of Figures 1–2 is an isometry of the relevant spinor bundles intertwining the Dirac operators up to the stated cutoffs, and that the determinant line of the cylinder factor C3 with the antiperiodic condition χ(1) = −χ(−1) is canonically trivial as a bundle with metric and connection. A concrete test is to verify the second point directly and to check that U commutes with the Dirac operators to the order needed for the determinant lines to agree. If either fails, the holonomy computed in Theorem 5.6 belongs to a bundle other than the physical one. The topological Theorem 5.5 is unaffected, so the Spin^c condition itself is not at risk; what is at risk is the stronger physical claim that A cannot be a conventional U(1) connection. I would ask the authors to surface this caveat in §5 rather than only in §6, and to state the C3 triviality as an explicit lemma with its proof or a precise reference.","section":"§6, Dirac Operators with Local Boundary Conditions","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The exclusion of branes with w2(ν) ≠ 0 is established only when the anomaly is detectable by a map from the torus ∂Σ × C; footnote 2 defers detection by higher-genus surfaces to factorization and unitarity arguments that the paper does not carry out. Similarly, footnote 3 concedes that (1.12) is only a sufficient condition and that the full requirement is expected to follow from factorization. Since the physical conclusion is that such branes are 'ruled out,' the argument as written establishes the exclusion only for the torus-detectable case. The authors acknowledge this, so my request is not for a new computation but for precision: either make the factorization step explicit, or state the exclusion conditionally on the torus-detectability of the class, so that a reader can see exactly what has been proved.","section":"Footnotes 2 and 3","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The cross-references to the closed-string results are inconsistent. The introduction cites 'Corollary 4.7' for closed-string cancellation and 'Theorem 4.8' for the R^10 statement, but in §4 these are Corollary 4.6 and Theorem 4.7; §1 cites 'Theorem 4.6' for the same cancellation statement. Please reconcile the numbering.","section":"Introduction and §1","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The symbol L_{α,β,θ} is used both for the full determinant line defined earlier in §5 and for the ratio (Pfaff D_{α,β,θ}) ⊗ (Pfaff D'_{α,β,θ})^{-1} introduced in (5.11). One of the two should be renamed.","section":"Eq. (5.11)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The letter L is heavily overloaded: it denotes the 'line bundle' on Q on which A is a connection, the line bundles L^±_α over the space of bosons, and the line bundle L_B over loop space. Distinct notation would make the Spin^c discussion in §1 and the determinant-line discussion in §5 easier to follow.","section":"§1 and §5","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"[DZ] (Dai–Zhang) appears in the reference list but I could not locate a citation to it in the text; either cite it or remove it.","section":"References","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The [CY] entry has a doubled comma ('Yin,,') and the arXiv number given, hep-th/9803931, does not appear to correspond to the Cheung–Yin paper; please verify the bibliographic data.","section":"References","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The odd-rank case is handled by 'replace F' with F' plus a trivial line bundle and apply the previous,' which changes the pfaffian by a factor. A sentence explaining why the reduction is legitimate in the pfaffian line would help the reader.","section":"Third proof of Proposition 5.8","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Please state explicitly what the pairing (Σ, w2(Q)) denotes and why the formula is well defined for a flat B-field, since this is the example that makes the twisted case concrete.","section":"Eq. (1.13)","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"accept","confidential_remarks":"This is a landmark paper and the central topological result (Theorem 5.5) is independent of the geometric subtleties flagged above, so I do not regard the acknowledged gaps in §6 and footnotes 2–3 as blocking. My major comments are requests for precision about what has been proved, not challenges to the Spin^c conclusion. The paper is well within the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"selfhosted/deepseek-v4.1-flash","letter":"The thing to know: this paper computes the open-string global anomaly directly on the worldsheet and gets w2(nu) as the sign, which forces the A-field to be a Spin-c connection rather than a U(1) one. That condition was already known from K-theory in Witten's earlier work, and the authors say so plainly in section 1 and again in section 6. What is new is the derivation, and it is a real derivation, not a restatement. The reader's take is right on this and the circularity worry is correctly low: [W3] and [W4] are used as a consistency check on a condition obtained independently here, not as premises. The abstract's framing at [abstract:s2] is a bit grander than the body, which is really about the anomaly itself, but that is a small thing. What the paper does well: Proposition 5.8 gets three genuinely different proofs (Agranovich-Dynin boundary symbol, KO excision, characteristic classes), and having all three is not padding, because each checks a different link. The flat index theorem argument in 5.9 is the right tool and is used correctly. The CP^2 and CP^2-bundle-over-S^1 examples are concrete and checkable, and the D0-brane charge integrality payoff at (1.14) is the kind of thing that makes the abstract claim at [abstract:s2] more than rhetoric. Where it is soft: section 6 says outright that the geometric index theory for local boundary conditions is only outlined, and the metric and connection on the physical pfaffian bundle are defined by fiat from the double. The stress-test is right that this is the weak link, but I think it overstates the damage. The ratio in 5.11 is flat by 5.9, and its class is computed topologically by excision, so the holonomy is a topological invariant of the ratio; a mismatch in the doubling map would have to produce a flat discrepancy with nontrivial holonomy, and nothing in the argument suggests one. Footnote 3 is the other acknowledged limit: (1.12) is sufficient, not necessary, with the converse deferred to factorization and unitarity. That is honest and minor. This is for anyone working on D-branes, anomalies, or the worldsheet origin of K-theory classifications; the first two sections are readable without the index theory. It deserves referees, and I would cite it.","headline":"A genuine perturbative derivation of the Spin-c condition on D-branes, with one acknowledged geometric gap that matters less than it looks because the relevant connection is flat.","tokens_in":31780,"tokens_out":1188,"would_cite":true,"duration_ms":21173,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":["81T30","58J28","81T50"],"pacs":["11.25.-w","11.25.Uv"],"model":"selfhosted/deepseek-v4.1-flash","headline":"A hidden minus sign forces D-branes to be Spin-c","keywords":["D-branes","global anomalies","pfaffian line bundle","Spin-c structure","K-theory classification of D-brane charge","index theory on manifolds with boundary","B-field","Type II superstring"],"falsifier":"Take the explicit family the paper constructs: a disc times a circle mapped into $2$ so that the boundary torus wraps a generator of $H_2(2)$ on which $w_2(2)$ is nonzero. If the pfaffian returned with the same sign after transport around that loop, the anomaly formula would fail. Conversely, a consistent worldsheet theory on the non-Spin$^c$ five-manifold built as a $2$ bundle over $S^1$ with conjugation monodromy would refute the Spin$^c$ restriction.","tokens_in":32014,"feed_emoji":"🧵","tokens_out":4720,"duration_ms":87009,"temperature":0.7,"pith_summary":"Freed and Witten ask whether the Type II string worldsheet path integral is globally well defined when the worldsheet has a boundary ending on a D-brane. For closed worldsheets the answer is yes, given that spacetime is spin and ten-dimensional; once a boundary is present, the fermion path integral — the pfaffian of the worldsheet Dirac operator — is not a number but a section of a line bundle, and the question becomes whether that bundle admits a flat section of unit norm. The paper computes the obstruction: the bundle's natural connection is flat, yet transporting the pfaffian around a loop of worldsheets multiplies it by $(-1)^{(C, w_2(Q))}$, the pairing of the boundary torus with the second Stiefel-Whitney class of the brane. The consequence is that the brane's apparent $U(1)$ gauge field cannot be an ordinary gauge field, since it must supply exactly the same sign; instead it is a Spin$^c$ connection, and the brane worldvolume must be Spin$^c$. That restriction is precisely the one already required for D-brane charge to be a class in K-theory, so a perturbative worldsheet argument lands on the same condition as the nonperturbative classification.","feed_headline":"A hidden minus sign forces D-branes to be Spin-c","feed_subtitle":"The open-string path integral flips sign around a loop unless the brane's gauge field is a Spin-c connection.","key_machinery":"The load-bearing device is a doubling trick. The Dirac operator on a surface with boundary and the local boundary conditions (5.3) — which mix right- and left-handed spinors and flip the sign on the normal-bundle part — is identified with a chiral Dirac operator on the closed double surface, with the sign encoded in a real line bundle $2$ supported in a collar. That identification turns an index problem with boundary into a closed-manifold problem, where the determinant-line-bundle machinery applies: the curvature formula, the holonomy as an adiabatic limit of the exponentiated $2$-invariant, and the Atiyah-Patodi-Singer flat index theorem. The difference between the two boundary conditions is then computed by the Agranovič-Dynin formula as the index of a family of Dirac operators on the boundary circle, which reduces to the standard statement that the pfaffian line bundle of a family of spin circles coupled to a real bundle $F'$ is the pullback of $w_2(F')$.","core_discovery":"The central result, Theorem 5.6, is that the canonical connection on the pfaffian line bundle of the open-string Dirac operator is flat, and that its holonomy around a family of worldsheets fibered over a circle is $(-1)^{(C,w_2(Q))}$, where $Q$ is the D-brane worldvolume and the pairing is over $W = (1.2)$'s $W = (1.2)$'s boundary torus $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s boundary torus $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s boundary torus $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s boundary torus $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s $W = (1.2)$'s boundary torus $W = (1.2)$'s boundary torus, that is, the integral of $w_2(Q)$ over $W = (1.2)$'s boundary torus. Equivalently (Theorem 5.5), the line bundle is the complexification of a real line bundle whose class in $H^1(Z;Z/2)$ is the pushforward of $w_2(2)$, and the topological class of the complex line bundle is the pushforward of $W_3(2)$.","pith_inferences":["The argument is a clean template for other boundary conditions: any brane whose worldvolume fermions see a sign ambiguity in a real bundle should be re-read as a twisted (Spin$^c$-like) object, and the same doubling computation should produce the relevant characteristic class without any input from the closed-string sector.","Because the anomaly is a flat $±1$ holonomy rather than a curvature, the condition is invisible to any local differential-geometric check of the low-energy effective action; one should expect the same to be true for the analogous conditions in M-theory and for orientifold and Type I settings the paper explicitly sets aside.","The paper's heuristic section suggests that the result could also be reached by imagining a 9-$ar 9$ pair and quantizing 9-$p$ strings; that route depends on assuming brane-antibrane pairs can be created, so the anomaly computation is the conservative version of the same statement, and the two should agree wherever both apply.","The demand that the pfaffian bundle be flat with holonomy in $±1$ is a stronger statement than the topological condition $W_3(\\nu)=0$; a brane could pass the K-theory test and still fail the geometric one, so the paper's footnote that the topological class may vanish while the holonomy does not is worth pursuing as a separate constraint."],"forward_implications":["A D-brane can wrap $Q$ only if $W_3(2)=0$, so $Q$ must be Spin$^c$; the paper's own example $Q=\\mathbb{CP}^2$ is allowed but requires half-integral flux, while the non-Spin$^c$ five-manifold built as a $\\mathbb{CP}^2$ bundle over $S^1$ with conjugation monodromy is excluded even though it embeds in $\\mathbb{R}^{10}$.","With a topologically nontrivial $B$-field the condition becomes $\\zeta|_Q = W_3(\\nu)$, where $\\zeta\\in H^3(Y;\\mathbb{Z})$ is the characteristic class of $B$, so the allowed branes are exactly those whose normal bundles have Thom classes in the appropriate twisted K-group.","The D0-brane charge of a brane wrapping $Q_0$, naively $\\int_{Q_0}\\hat A(Q_0)\\exp(c_1(L))$, is an integer automatically, because with $L$ read as a Spin$^c$ line bundle the integral is the index of a Dirac operator; for $\\mathbb{CP}^2$ with trivial $L$ the naive value $\\pm 1/8$ is avoided without postulating new conservation laws.","For closed worldsheets the anomaly cancels whenever $Y$ is spin of dimension ten, and for $Y=\\mathbb{R}^{10}$ it cancels separately for left- and right-movers, so the whole obstruction is localized at the boundary of the worldsheet.","The product of the Levi-Civita holonomy in the spin representation with the $A$-field holonomy, $\\mathrm{Tr}\\,P\\exp(\\oint\\omega)\\cdot\\exp(i\\oint A)$, is globally well defined even though neither factor is, which is exactly the statement that the brane carries a Spin$^c$ structure built from the metric and $A$."],"supporting_citations":[{"why":"Supplies the canonical metric and connection on the determinant line bundle of a family of Dirac operators, on which the curvature formula (3.1) and holonomy formula (3.2) rest.","marker":"[BF]"},{"why":"Supplies the global anomaly formula identifying the holonomy of the determinant line bundle with the adiabatic limit of the exponentiated $2$-invariant.","marker":"[W1]"},{"why":"Supplies the flat index theorem that turns the ratio of $2$-invariants for a flat bundle into a K-theory pushforward, the step that reduces Lemma 4.8 and Proposition 5.9 to topology.","marker":"[APS2]"},{"why":"Supplies the pfaffian square root of the determinant line bundle and the cohomological formula $c_1(2)=2$ for spin bundles, used in the third proof of Proposition 5.8.","marker":"[F1]"},{"why":"Supplies the Agranovič-Dynin formula for elliptic boundary-value problems, which computes the difference of indices between the two boundary conditions as a boundary family index.","marker":"[BW]"},{"why":"Supplies the gluing law and the sign $(-1)^{k}$ for $2$ under orientation reversal, used to evaluate the invariants on the doubled surface.","marker":"[DF]"},{"why":"Supplies the nonperturbative statement that D-brane charge takes values in K-theory, the classification the anomaly computation is claimed to reproduce.","marker":"[MM]"},{"why":"Supplies the requirement that the normal bundle carry a Spin$^c$ structure for the brane to define a K-theory class, and the twisted K-theory statement for nonzero $B$.","marker":"[W4]"}],"fun_headline_variants":["Anomaly cancellation pins D-brane topology to Spin-c","Anomaly cancellation lets D-brane charge sit in K-theory","Pfaffian holonomy turns the open-string sign into a Spin-c test","D-brane worldvolume must be Spin-c for global anomaly cancellation","Verify before promotion: core claim repeats malformed torus text and w2(2) typo"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The anomaly is computed for a Dirac operator on the doubled surface, and the metric and connection on the physical boundary-value problem are defined by declaring them to come from that double; the paper states in section 6 that the geometric index theory for local boundary conditions is only outlined and the details are not carried through.","fun_headline_variants_meta":{"raw":{"variants":["Anomaly cancellation pins D-brane topology to Spin-c","Anomaly cancellation lets D-brane charge sit in K-theory","Pfaffian holonomy turns the open-string sign into a Spin-c test","D-brane worldvolume must be Spin-c for global anomaly cancellation","Verify before promotion: core claim repeats malformed torus text and w2(2) typo"]},"model":"selfhosted/deepseek-v4.1-flash","effort":"low","attempts":1,"cost_usd":0.0,"raw_usage":{"total_tokens":3519,"prompt_tokens":1532,"completion_tokens":1987,"prompt_tokens_details":null,"completion_tokens_details":{"reasoning_tokens":1900}},"tokens_in":1532,"tokens_out":1987,"duration_ms":34038,"image_pages":[],"temperature":1.0,"reasoning_budget":50,"reasoning_tokens":1900,"transport_provider":"selfhosted","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-10-29T18:26:55.617098+00:00","model_set":{"reader":"selfhosted/deepseek-v4.1-flash"},"falsifier":"Take the explicit family the paper constructs: a disc times a circle mapped into $2$ so that the boundary torus wraps a generator of $H_2(2)$ on which $w_2(2)$ is nonzero. If the pfaffian returned with the same sign after transport around that loop, the anomaly formula would fail. Conversely, a consistent worldsheet theory on the non-Spin$^c$ five-manifold built as a $2$ bundle over $S^1$ with conjugation monodromy would refute the Spin$^c$ restriction.","supporting_citations":[],"review_version":1}