{"id":"dc09c15b-e42f-4547-ad03-a492b3f61aed","arxiv_id":"2507.00655","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A resolution of a G2-orbifold carries a smooth 1-parameter family of irreducible, non-flat G2-instantons, and the curve they trace in moduli space is injective.","lead":"This paper constructs a smooth 1-parameter family of G2-instantons, special solutions of the Yang-Mills equations, on a compact 7-manifold obtained by resolving a symmetric orbifold. If correct, it is the first example of a family of such instantons on a compact G2-manifold, showing that symmetries can allow instantons to vary continuously.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 5.11's f-independent Schauder estimate is only sketched; since it underpins the linear estimate in Proposition 5.16 and hence the fixed-point argument in Theorem 3.13, the proof should be completed before the result is accepted as stated.","rationale":"The paper is carefully structured and gives a largely self-contained analytic framework: the algebraic deformation theory, the gluing construction, and the example in Section 6 are detailed, and the claimed family of G2-instantons is genuinely new. The reader's weakest-assumption analysis correctly identifies the t-uniform linear estimate as the most fragile input, and my independent reading confirms that the proof of Proposition 5.11 is the specific point where the necessary f-independence is asserted but not fully demonstrated. This is not a discovered error, but an unfinished proof of a load-bearing estimate. The resolution-metric caveat in Remark 6.21 is a real clarification issue but does not undermine the stated theorems. Since the reader already issued a CONDITIONAL verdict requiring precisely this gap to be closed, my stress-test does not move the verdict: the paper should be accepted only after Proposition 5.11 is supplied with a complete proof establishing the claimed uniformity in f.","tokens_in":58545,"tokens_out":14327,"duration_ms":166723,"concrete_test":"Provide a complete proof of Proposition 5.11, tracking all constants: verify that the coefficients of L_{\\hat A+t\\hat a}L^* are uniformly bounded in C^{k,\\alpha} with respect to the rescaled metric on R\\times \\tau^{-1}(B_{t^{-1}\\kappa}(0)), independent of f\\in F and t\\in(0,T), and that the weighted Schauder estimates of [Bar86] or [LM85] apply with a constant depending only on those uniform bounds. If such a uniform bound cannot be shown, construct a sequence f_n\\in F for which the Schauder constant blows up; this would locate a concrete failure of the linear estimate and thus of Theorem A.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central existence theorem (Theorem A, via Theorem 3.13) depends on Point 2: the t-uniform a priori estimate for the H-invariant cokernel of L_{\\tilde A+\\tilde a}L^*. This is established in Proposition 5.16, whose proof in turn relies on Proposition 5.17 and, crucially, on Proposition 5.11: a Schauder estimate for the model operator L_{\\hat A+t\\hat a}L^* on R\\times \\tau^{-1}(B_{t^{-1}\\kappa}(0)) with constant c_{SI} claimed to be independent of t and of the family parameter f. The proof of Proposition 5.11 is only a sketch: on the non-compact part it rescales to compare the operator with L_{A_\\infty+tR(\\tau)^*\\hat a}L^*, where tR is uniformly bounded but not small, and the f-independence of the constant is asserted without a displayed argument. If the constant fails to be uniform in f, or if the lower-order t\\hat a terms accumulate in the relevant weighted C^{k,\\alpha} norms, then the contraction in Theorem 3.13 has no f-independent contraction constant and the entire 1-parameter family construction collapses. This is the single most load-bearing unproved step in the paper. The resolution-metric caveat in Remark 6.21, by contrast, is explicitly acknowledged and does not invalidate Theorem B as stated, since the theorem only claims existence of a G2-manifold arising from the orbifold, not that it is isometric to Joyce's original resolved example.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops an equivariant gluing construction for G2-instantons over generalized Kummer resolutions and applies it to Joyce's orbifold from [Joy96b, Example 18]. The main abstract result (Theorem A, comprising Theorem 3.13 and Propositions 4.5, 4.7, 5.16) gives conditions under which an H-equivariant family of almost-instantons on a Kummer resolution can be perturbed to a genuine family of G2-instantons, including estimates on derivatives with respect to the family parameter. The concrete result (Theorem B, comprising Propositions 6.31, 6.33, 6.34) produces, for sufficiently small scale t, an SO(14)-bundle over a compact G2-manifold arising from the orbifold and a smooth family of G2-instantons parametrized by a compact interval F in R minus pi Z, all infinitesimally irreducible and non-flat, with injective map into the moduli space. The proof combines Joyce's Kummer theorem, rigid Hermitian-Yang-Mills connections on ALE crepant resolutions due to Degeratu-Walpuski, a family fixed-point argument in weighted Holder spaces, and a detailed linear analysis using a Z2 symmetry to kill the obstruction space.","tokens_in":58830,"tokens_out":15789,"duration_ms":191479,"significance":"If established, this is the first smooth one-parameter family of G2-instantons over a compact G2-manifold, and it extends the existing gluing technology to codimension-six singular strata and to families whose linearized instanton operator has nontrivial cokernel. The paper is carefully structured and contains detailed proofs of most analytic steps, including the pregluing construction, the quadratic estimates, the family fixed-point argument, and the moduli-space injectivity argument; the explicit construction of the flat families in Section 6 and the nontrivial non-tangency result in Proposition 6.18 are valuable contributions. The paper is also transparent about caveats, notably the possible non-isometry of the resolution metric with Joyce's original example in Remark 6.21. These strengths make the overall approach credible. However, as stated, the central claims rest on two points that need further work: the uniform-in-f Schauder estimate is only sketched, and the injectivity statement needs a length restriction on the parameter interval or a quotient by the 2pi-periodicity.","major_comments":[{"comment":"Proposition 5.11, which supplies the f-independent Schauder estimate for the model operator L L^* on R times the inverse image of the ball of radius t^{-1} kappa, is presented only as a proof sketch. The claimed t-independence and f-independence of the constant c_SI are asserted rather than proved: in the exterior region the estimate is reduced to equation (5.2) for the operator built from A_infty + tR(τ)^*a, where tR is only uniformly bounded and not small, and the lower-order terms tR(τ)^*a are said to be uniformly bounded without a displayed weighted-norm argument. This estimate is used in Proposition 5.17 and then in Proposition 5.16, which supplies assumption (2) of Theorem 3.13; the f-independence of the contraction constant in the fixed-point argument and the derivative estimates in Proposition 3.18 all inherit this uniformity. Since this is a load-bearing analytic input, the proof should be completed, or a precise reference covering this non-asymptotically-dilation-invariant family with a uniform constant should be supplied.","section":"Section 5.1, Proposition 5.11"},{"comment":"The injectivity statement in Proposition 6.34 is false for arbitrary compact intervals F in R minus pi Z of length at least 2pi. Remark 6.30 states that the preglued connections for parameters f_1 and f_2 are gauge equivalent whenever f_1 - f_2 is an integer multiple of 2pi, and by the uniqueness in the fixed-point construction the same periodicity passes to the perturbed instantons. The proof of Proposition 6.35(1) rules out distinct limiting theta-values via Proposition 6.3, but it does not exclude f'_infty = f_infty + 2pi, which gives the same value of theta in S^1. Consequently Theorem B as stated needs the additional hypothesis that F has length strictly less than 2pi, or the injectivity claim must be formulated for the induced curve on R/2piZ. This is a load-bearing correction to the main example.","section":"Section 6.5, Propositions 6.34 and 6.35; Remark 6.30"}],"minor_comments":[{"comment":"Proposition 5.16 and the corresponding bullet in Proposition 6.29 state the linear estimate for b in Omega^1 direct sum Omega^7, but the operator L L^* acts on Omega^6 direct sum Omega^0, as used in Theorem 3.13(2) and in Propositions 5.17 and 5.18. The domain in these statements should be corrected.","section":"Section 5.2, Proposition 5.16, and Section 6.4, Proposition 6.29"},{"comment":"The caveat that the Calabi-Yau metric used for the resolution may not be isometric to the one in Joyce's original example is explicitly acknowledged, but Theorem B should state in its formulation that the G2-manifold is one arising from the orbifold via the Kummer construction and is not claimed to be isometric to Joyce's original resolved example.","section":"Remark 6.21"},{"comment":"The passage from the family of flat bundles to the fixed bundle E_0 uses isomorphisms F'_f and F_f; a brief sentence explaining why the modified F_f is still orientation preserving and well-defined on the quotient after the cut-off modification would improve readability.","section":"Section 6.2, Proposition 6.13 and Definition 6.16"}],"recommendation":"major_revision","confidential_remarks":"The paper is a serious and substantial extension of Walpuski's gluing framework and of the author's thesis. The two major comments are fixable within the scope of the manuscript: the Schauder estimate in Proposition 5.11 needs a completed proof, and the injectivity statement needs a length restriction or a quotient by the 2pi-periodicity. I would not reject the paper, but I would not accept it in the present form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline: this is a serious construction paper that probably delivers the first smooth 1-parameter family of G2-instantons on a compact G2-manifold. The novelty is real: it extends Walpuski's gluing to codimension-6 singular strata, to families of connections, and to connections with non-trivial cokernel, using a Z2-invariance trick. The paper is largely self-contained, the analytic setup is careful, and the algebraic example (Theorem B) is concrete and convincing.\n\nWhat is actually new: the Z2-invariant perturbation scheme, and the proof that the family is injective in moduli space via a flat-bundle monodromy argument. The non-tangency result (Prop 6.18) and the contradiction argument in Prop 6.35 are nice pieces of work. The paper also proves infinitesimal rigidity and non-flatness of the constructed instantons.\n\nSoft spots: the load-bearing estimate is Proposition 5.11, the f-independent Schauder estimate for the model operator. It is only a proof sketch, and the constant's independence from f is asserted. This estimate underpins Proposition 5.16, which gives the t-uniform linear estimate needed for the fixed-point theorem. If the constant is not uniform in f, the contraction argument in Theorem 3.13 has no f-independent contraction constant and the family construction collapses. This is a real gap, but I think it is likely fixable: the operator is a lower-order perturbation of a translation-invariant operator, and the sketch outlines the rescaling argument. It needs to be written out, not a fatal obstruction.\n\nThe other caveat, Remark 6.21, is explicitly acknowledged: the resolved metric may not be isometric to the one in Joyce's original example. Since Theorem B only claims existence of a G2-manifold from the orbifold, this is minor, though it should be clearly stated.\n\nOverall: a solid, important paper with one sketched analytic step that must be completed before the result is accepted as stated. I would send it to a serious referee.\n\nRecommendation: engage with it, but the referee should demand a full proof of Prop 5.11.","headline":"A likely first smooth family of G2-instantons, but one sketched estimate needs to be completed before the result is fully established.","tokens_in":59395,"tokens_out":2125,"would_cite":true,"duration_ms":24935,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C07","53C25","53C29"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper constructs the first smooth one-parameter family of G₂-instantons over a compact G₂-manifold, extending the gluing method of [Wal13a] and using a ℤ₂ symmetry of the orbifold to cancel the obstruction that would force instantons…","keywords":["G₂-instantons","generalised Kummer construction","G₂-orbifolds","gluing construction","Hermitian Yang–Mills connections","ALE Calabi–Yau manifolds","moduli space of instantons","infinitesimal rigidity"],"falsifier":"Fix the model geometry $\\mathbb{R} \\times \\hat Z_\\zeta$ with the connection $\\hat A + t\\hat a_f$, where $\\hat a_f = \\frac{f}{\\sqrt{7}}(\\mathrm{id}\\otimes_{\\mathbb{R}}i)\\,ds$, and search for $\\mathbb{Z}_2$-invariant solutions of $L_{\\hat A + t\\hat a_f}L^*_{\\hat A + t\\hat a_f} b = 0$ with $\\|b\\|_{C^0_\\beta} < \\infty$ for some $\\beta < 0$, or check directly whether the constant in Proposition 5.11 stays bounded as $f$ ranges over a compact interval $F \\subset \\mathbb{R}\\setminus\\pi\\mathbb{Z}$. Finding a nonzero solution of that decay, or an estimate constant that blows up with $f$, would contradict Assumption 5.15 and collapse the family construction; finding none, as the paper expects, supports it.","tokens_in":58272,"feed_emoji":"∞","tokens_out":18133,"duration_ms":174810,"temperature":0.7,"pith_summary":"$G_2$-instantons — connections whose curvature satisfies $F_A \\wedge \\psi = 0$ for the coassociative 4-form $\\psi$ — are the Yang–Mills minima in seven dimensions, and a standing question is what their moduli spaces look like. This paper claims that on a compact $G_2$-manifold produced by Joyce's generalised Kummer construction, instantons need not be isolated: for each sufficiently small gluing scale it produces a whole smooth interval of distinct instantons on a single $\\operatorname{SO}(14)$-bundle, all non-flat and infinitesimally irreducible, mapping injectively into the moduli space modulo gauge. The construction extends the gluing method of [Wal13a] to families of connections whose linearised instanton operator has a non-trivial cokernel, and then uses a $\\mathbb{Z}_2$-symmetry of the underlying orbifold (orientation reversal $y \\mapsto -y$) so that only the invariant part of the cokernel has to vanish. The author notes that this is to his knowledge the first example of a smooth one-parameter family of instantons over a compact $G_2$-manifold, the kind of structure that any future enumerative theory of $G_2$-instantons would have to accommodate.","feed_headline":"First smooth family of G₂-instantons on a compact G₂-manifold","feed_subtitle":"A ℤ₂ symmetry clears the obstructions, producing a curve of distinct non-flat instantons in moduli space.","key_machinery":"The load-bearing object is the linearised instanton operator $L_A = (\\psi_t \\wedge d_A - d^*_A,\\, d^*_A)$ together with its formal adjoint $L^*_A$, acting on weighted Hölder spaces adapted to the Kummer degeneration; the proof runs a fixed-point contraction for the augmented instanton equation $F_A \\wedge \\psi - d^*_A \\xi = 0$ using the right-inverse $L^*_{\\tilde A}(L_{\\tilde A}L^*_{\\tilde A})^{-1}$, and everything depends on a scale-uniform estimate for $L_{\\tilde A}L^*_{\\tilde A}$ on $\\mathbb{Z}_2$-invariant sections, the t-uniform linear estimate of Proposition 5.16. Two named inputs carry the example: the representation family $f_\\theta : \\Gamma \\to \\operatorname{SO}_{\\mathbb{R}}(\\mathbb{C}^7)$ of the crystallographic group $\\Gamma$, giving a one-parameter family of flat bundles whose infinitesimal direction is the kernel of the linearised operator, and the model operator $L_{\\hat A + t\\hat a}L^*_{\\hat A + t\\hat a}$ on $\\mathbb{R} \\times \\hat Z_\\zeta$, the ALE Calabi–Yau resolution of $\\mathbb{C}^3/\\mathbb{Z}_7$, whose vanishing $\\mathbb{Z}_2$-invariant cokernel (Assumption 5.15) is the analytic condition that lets the contraction close.","core_discovery":"The paper's central claim, stated as Theorem A and Theorem B, is that families of $G_2$-instantons can be glued together and then deformed into genuine instantons even when the linearised instanton operator has a non-trivial cokernel. Theorem A is an abstract existence statement: for a degenerating family of $G_2$-manifolds $(\\hat Y_t, \\phi_t)$ arising from the generalised Kummer construction that resolves an orbifold with codimension-6 singular strata, and a finite group $H$ acting by coassociative-form-preserving isometries, any $H$-equivariant family of gluing data whose $H$-invariant cokernel of $L_{\\tilde A} L^*_{\\tilde A}$ vanishes deforms to a smooth $H$-invariant family of $G_2$-instantons with the same regularity as the starting family. Theorem B realises this for Joyce's Example 18 orbifold with $H = \\mathbb{Z}_2$: for every compact interval $F \\subset \\mathbb{R} \\setminus \\pi\\mathbb{Z}$ and all sufficiently small $t$, there is a compact $G_2$-manifold $(\\hat Y, \\phi_t)$ and an $\\operatorname{SO}(14)$-bundle such that $f \\mapsto A_{t,f}$ is a smooth family of $G_2$-instantons, each infinitesimally irreducible and non-flat, and the map $f \\mapsto [A_{t,f}]$ into the moduli space of connections modulo gauge is injective. The mechanism that makes this possible is the $\\mathbb{Z}_2$-action $y \\mapsto -y$ on the orbifold: the one-dimensional obstruction space of the family is anti-invariant under it, so the fixed-point perturbation runs inside the $\\mathbb{Z}_2$-invariant connections, a situation the paper describes as instantons being `$\\mathbb{Z}_2$-unobstructed'.","pith_inferences":["A general recipe suggests itself beyond the paper: whenever an orbifold-resolution admits a finite symmetry under which the obstruction space of a family of gluing data is anti-invariant, a smooth family of instantons should exist with parameter dimension bounded by the invariant kernel; testing this on other Joyce orbifolds or on Spin(7)-instantons would show how much of the mechanism is special ","The paper's own caveat in Remark 6.21 leaves the resolved metric's identity open: the Calabi–Yau metric used here may not be isometric to the one in Joyce's original example, so the theorem may produce instantons on a $G_2$-manifold with a different $G_2$-structure from Joyce's, and whether the family persists on the exact Joyce metric remains an open question.","The injectivity proof yields a reusable criterion: a gauge equivalence between nearby members of a gluing family forces the derivative of the flat family to be an exact form in the limit $t \\to 0$, and checking that derivative is a finite-dimensional representation-theoretic computation that could be repeated for other groups and other representations."],"forward_implications":["Instantons on compact $G_2$-manifolds can form smooth families: the moduli space can contain curves, not only isolated points, even when each member is infinitesimally irreducible.","The gluing method of [Wal13a] now covers codimension-6 Kummer resolutions and families of connections with non-trivial cokernel: the obstruction to deforming a family can be cancelled by a finite group symmetry rather than by choosing the gluing data more carefully.","The constructed family injects into the moduli space modulo gauge, so it is a genuine curve of distinct gauge-equivalence classes of non-flat instantons, providing a concrete test object for any Donaldson–Thomas-type count of $G_2$-instantons.","Because the linearised operator has one-dimensional kernel (from the family direction) and the index over the odd-dimensional compact manifold is zero, each member of the family carries a one-dimensional obstruction space; the paper shows such obstructed instantons are nevertheless constructible when the obstructions are anti-invariant under a symmetry."],"supporting_citations":[{"why":"Supplies the gluing construction for G₂-instantons that this paper extends to codimension-6 Kummer resolutions and to families of connections; also the source of the weighted Hölder-space scheme and the pregluing estimates.","marker":"[Wal13a]"},{"why":"Source of Example 18, the flat G₂-orbifold resolved in Theorem B, and of the generalised Kummer construction theorem (Theorem 2.2.1) that produces the manifolds (Ŷ_t, φ_t).","marker":"[Joy96b]"},{"why":"Joyce's Theorem A, the perturbation result that corrects the closed 3-form to a torsion-free G₂-structure; cited as Theorem 2.8 and needed for the metric estimates in the linear analysis.","marker":"[Joy96a]"},{"why":"Provides the ALE Calabi conjecture (Theorem 8.2.3) used to put Ricci-flat ALE Calabi–Yau metrics on the crepant resolutions, plus the book's account of the Kummer construction underlying the setup.","marker":"[Joy00]"},{"why":"Gives the rigid Hermitian Yang–Mills connections on tautological bundles over ALE crepant resolutions (Propositions 2.21 and 2.25) used as gluing data, and Equation (1.8) on intersection forms used to prove the instantons are non-flat.","marker":"[DW16]"},{"why":"Weighted elliptic and Schauder theory on asymptotically flat manifolds; cited in Propositions 5.6 and 5.11 for the model-operator estimates on which the t-uniform linear estimate depends.","marker":"[Bar86]"},{"why":"The Calabi–Yau metric on the crepant resolution appearing in Joyce's original Example 18; Remark 6.21 flags that the paper's resolution metrics may not be isometric to it, leaving a comparison gap with Joyce's manifold.","marker":"[TY91]"}],"fun_headline_variants":["First smooth family of G₂-instantons on a compact manifold","Z₂ symmetry admits curve of rigid G₂-instantons","Z₂-invariant gluing yields distinct non-flat G₂-instantons","Compact G₂-manifold hosts 1-parameter instanton family","Z₂ action clears obstructions for smooth G₂-instanton family"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction rests on a linear estimate for the operator $L_{\\tilde A} L^*_{\\tilde A}$ that must hold with a constant independent of the family parameter $f$ and of the scale $t$; the proof of the underlying Schauder estimate (Proposition 5.11) is only sketched, and if the constant grows with $f$, or a $\\mathbb{Z}_2$-invariant kernel appears, the fixed-point argument that builds the instanton family fails.","fun_headline_variants_meta":{"raw":{"variants":["First smooth family of G₂-instantons on a compact manifold","Z₂ symmetry admits curve of rigid G₂-instantons","Z₂-invariant gluing yields distinct non-flat G₂-instantons","Compact G₂-manifold hosts 1-parameter instanton family","Z₂ action clears obstructions for smooth G₂-instanton family"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000848,"raw_usage":{"total_tokens":3835,"prompt_tokens":1236,"completion_tokens":2599,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":852,"completion_tokens_details":{"reasoning_tokens":2504}},"tokens_in":852,"tokens_out":2599,"duration_ms":24398,"temperature":1.0,"reasoning_tokens":2504,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T21:10:37.748176+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fix the model geometry $\\mathbb{R} \\times \\hat Z_\\zeta$ with the connection $\\hat A + t\\hat a_f$, where $\\hat a_f = \\frac{f}{\\sqrt{7}}(\\mathrm{id}\\otimes_{\\mathbb{R}}i)\\,ds$, and search for $\\mathbb{Z}_2$-invariant solutions of $L_{\\hat A + t\\hat a_f}L^*_{\\hat A + t\\hat a_f} b = 0$ with $\\|b\\|_{C^0_\\beta} < \\infty$ for some $\\beta < 0$, or check directly whether the constant in Proposition 5.11 stays bounded as $f$ ranges over a compact interval $F \\subset \\mathbb{R}\\setminus\\pi\\mathbb{Z}$. Finding a nonzero solution of that decay, or an estimate constant that blows up with $f$, would contradict Assumption 5.15 and collapse the family construction; finding none, as the paper expects, supports it.","supporting_citations":[],"review_version":1}