{"axiom_check_command":"cd ~/workspace/millennium-lab/lean && lake env lean CheckAxioms_r5.lean","axiom_check_log":"~/workspace/millennium-lab/lean/axioms_r5.log","blocking_piece":"The Dolbeault isomorphism H^q(X, O_X) ≃ H^{0,q}(X) is still blocked, and the Hodge decomposition is still blocked. What changed in round 5: the sheaf long exact sequence is now PROVED for the actual exponential sequence instead of assumed. The remaining bridge to Lefschetz (1,1) is now exactly the two analytic inputs, both located precisely: (1) the Dolbeault identification H^2(X, O_X) ≃ H^{0,2}(X), which cannot be stated because mathlib v4.13.0 has no differential forms on manifolds (verified by grep in round 4: no AlternatingMap under Mathlib/Geometry/Manifold, no exterior derivative, no de Rham or Dolbeault cohomology, no sheaves A^{0,q} of smooth (0,q) forms, no fine sheaf acyclicity); (2) the Hodge decomposition map ι whose (0,2) component is the Dolbeault isomorphism composed with the map induced by the inclusion. The Lefschetz bridge lefschetz_one_one_of_expSeq is proved conditional on exactly those two inputs, with the Chern class map now the one from the exponential sequence itself. Nothing here claims the Hodge conjecture, or any Millennium problem, is solved.","build_command":"cd ~/workspace/millennium-lab/lean && lake build","build_exit_code":0,"build_log":"~/workspace/millennium-lab/lean/build_r5.log","build_status":"compiled","build_targets":"4584/4584","date":"2026-11-21","file":"~/workspace/millennium-lab/lean/Hodge/LongExactSequence.lean","lean_toolchain":"leanprover/lean4:v4.13.0","lean_version":"4.13.0","mathlib_version_or_commit":"v4.13.0","piece_formalized":"Candidate (a), the strongest one: the long exact sequence segment induced by the exponential sequence of sheaves, with the connecting homomorphism set up and c1 identified with it, for the ACTUAL exponential sequence 0 -> Z -> O_X -> O_X^* -> 0 on a complex manifold. (A) The three sheaves as TopCat.Sheaf AddCommGrp on the opens site: locallyConstantIntSheafAdd (Z, locally constant integer-valued functions, built from a LocalPredicate), holomorphicSheafAddCommGrp (O_X, additive group from the round-3 commutative ring sheaf), holomorphicSheafUnitsAddCommGrp (O_X^*, nowhere zero holomorphic functions with pointwise multiplication written additively, via Additive of the round-3 CommGroup). The inclusion expSeqIncl sends n to 2 pi i n; expSeqExp is the exponential map; they form expSeqShortComplex. (B) Exactness at O_X, proved at section level from the two analytic inputs: intOfExpEqOne_isLocallyConstant (a holomorphic section f with exp f = 1 is locally constant with values in 2 pi i Z, from continuity of f and discreteness of Z, via a norm bound |n(y) - n(x0)| < 1/2 and Int.one_le_abs) and exists_local_log (every nowhere zero holomorphic section has a holomorphic logarithm locally, from Complex.differentiableAt_log on the slit plane, with the log built as log (c * g) - log c on a ball of radius ||a||/2). The kernel is proved categorically: expSeqKernelLift builds the lift of any map killed by exp to a map into Z, expSeqKernelLift_comp shows the lift composed with the inclusion is the original map, and expSeq_isLimit_kernelFork (as a def, since IsLimit is data) plus ShortComplex.exact_of_f_is_kernel give expSeqShortComplex_exact. (C) Epi: expSeqExp_isLocallySurjective via TopCat.Presheaf.isLocallySurjective_iff and exists_local_log, then expSeqExp_epi via CategoryTheory.Sheaf.epi_of_isLocallySurjective (applied directly; infer_instance fails to find that instance). (D) Assembly: expSeq_shortExact via ShortExact.mk' from exactness, expSeqIncl_mono (injective on sections since 2 pi i != 0), and the epi. expSeq_chernClass is sheafChernClass of that short exact sequence, i.e. c1 : H^1(X, O_X^*) ->+ H^2(X, Z). Exactness at both joints (expSeq_chernClass_exact_left/right) and surjectivity onto the kernel of H^2(Z) -> H^2(O_X) (expSeq_chernClass_surj_onto_kernel) by instantiating the round-4 abstract lemmas. lefschetz_one_one_of_expSeq is the round-4 bridge with the abstract sequence replaced by the exponential sequence. (E) Byproduct: in any abelian category a mono that is also an epi is an iso (isIso_of_mono_of_epi_of_abelian), hence every abelian category is balanced (balanced_of_abelian, confirmed to synthesize for the sheaf category). All 17 new theorems and the balanced instance depend only on the standard axioms propext, Classical.choice, Quot.sound, with 0 sorries.","round":5,"sorries":[],"sorry_count":0,"technical_notes":"Mathlib v4.13.0 findings from this round, for future rounds. (1) IsLimit is data, not a Prop (structure IsLimit in Mathlib/CategoryTheory/Limits/IsLimit.lean:54, no : Prop annotation; IsLimit (KernelFork.ofι ...) : Type 1). A kernel-fork limit must be a def, not a theorem; use ShortComplex.exact_of_f_is_kernel with the def applied to I M. (2) The general sheaf_hasExt instance does not fire by typeclass search at a concrete site: its universe levels are left as metavariables and the search times out at 20000 heartbeats. Fix: give the instance concretely with the TopCat.Sheaf head and the universe levels pinned (HasExt.{1, 0, 1} (TopCat.Sheaf AddCommGrp.{0} (TopCat.of M)) := Hodge.sheaf_hasExt (C := Opens (TopCat.of M)) (J := Opens.grothendieckTopology (TopCat.of M))). The same timeout hits Abelian.Ext.mk₀ and (Abelian.Ext.mk₀ ...).postcomp sheafZ in a bare statement; the concrete HasExt instance unblocks all of them. (3) A section of O_X as an AddCommGrp sheaf does not coerce to a function through a nested ascription ((k.val.app U s : Type-sheaf obj) : U -> C) x; factor the conversion through a def (toHolomorphicSection, and toUnitsSection / ofHolomorphicSection for the other direction) whose result type is exactly the Type-sheaf obj, then coerce. (4) congrFun does not apply to equalities of AddMonoidHoms; use DFunLike.congr_fun for naturality and for extracting (k >> exp).val.app U s = 0 from k >> exp = 0. (5) epi_of_isLocallySurjective is an instance but infer_instance fails to find it; apply it directly (CategoryTheory.Sheaf.epi_of_isLocallySurjective). (6) For an Opens morphism h : U ⟶ V in the opposite category, the underlying inclusion is h.unop, not h.op (h.op lands in the double opposite); use (h.unop y) with parentheses, since f h.unop y parses as (f h.unop) y. (7) twoPiI_ne_zero and expSeq_exact are already declared in Hodge.Exponential (imported via SheafOX); the round-5 names are intOfExpEqOne_eq (using the existing twoPiI_ne_zero) and expSeqShortComplex_exact. (8) Theorem names with section variables (I, M) must be applied to I M when used as proof terms (expSeqShortComplex_exact I M), or Lean reports the Pi type as an application mismatch. (9) theorem foo : _ := proof does not infer the type from the body; state the type explicitly.","theorems_checked":[{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.isIso_of_mono_of_epi_of_abelian","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.balanced_of_abelian","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.mdifferentiable_of_isLocallyConstant","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.intOfExpEqOne_spec","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.intOfExpEqOne_isLocallyConstant","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.exists_local_log","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.intOfExpEqOne_eq","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeqIncl_mono","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeqKernel_exp_eq_one","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeqKernelLift_comp","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeqShortComplex_exact","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeqExp_isLocallySurjective","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeqExp_epi","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeq_shortExact","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeq_chernClass_exact_left","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeq_chernClass_exact_right","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.expSeq_chernClass_surj_onto_kernel","sorry":false},{"axioms":["propext","Classical.choice","Quot.sound"],"name":"Hodge.lefschetz_one_one_of_expSeq","sorry":false}],"theorems_proved":["Hodge.isIso_of_mono_of_epi_of_abelian: in any abelian category, a monomorphism that is also an epimorphism is an isomorphism. Proved from Abelian.monoIsKernelOfCokernel and the cokernel of an epi being zero.","Hodge.balanced_of_abelian: every abelian category is balanced. Mathlib v4.13.0 does not provide this instance; confirmed to synthesize for the sheaf category.","Hodge.mdifferentiable_of_isLocallyConstant: a locally constant complex-valued function on a complex manifold is holomorphic (MDifferentiable), since it is eventually equal to a constant.","Hodge.intOfExpEqOne_spec: the chosen integer n(x) satisfies f(x) = n(x) * 2 pi i, from Complex.exp_eq_one_iff.","Hodge.intOfExpEqOne_isLocallyConstant: the pointwise integer of a holomorphic section with exp f = 1 is locally constant. Proved from continuity of f and discreteness of Z.","Hodge.exists_local_log: every nowhere zero holomorphic section has a holomorphic logarithm on a smaller open set containing a given point, from Complex.differentiableAt_log on the slit plane.","Hodge.intOfExpEqOne_eq: uniqueness of the pointwise integer: if f(x) = n * 2 pi i then the chosen integer at x is n.","Hodge.expSeqIncl_mono: the inclusion Z -> O_X is a monomorphism, injective on sections since 2 pi i != 0.","Hodge.expSeqKernel_exp_eq_one: if k >> exp = 0 then exp of every value of k s is 1. The pointwise kernel fact, factored out for reuse.","Hodge.expSeqKernelLift_comp: the lift to Z composed with the inclusion is the original map k.","Hodge.expSeqShortComplex_exact: the exponential sequence is exact at O_X, via the kernel fork being a limit (expSeq_kernelForkIsLimit, a def since IsLimit is data) and ShortComplex.exact_of_f_is_kernel.","Hodge.expSeqExp_isLocallySurjective: the exponential map of sheaves is locally surjective, via TopCat.Presheaf.isLocallySurjective_iff and exists_local_log.","Hodge.expSeqExp_epi: the exponential map of sheaves is an epimorphism, via CategoryTheory.Sheaf.epi_of_isLocallySurjective.","Hodge.expSeq_shortExact: the exponential sequence 0 -> Z -> O_X -> O_X^* -> 0 is short exact as sheaves of abelian groups, via ShortExact.mk'.","Hodge.expSeq_chernClass_exact_left: exactness at H^1(X, O_X^*): the kernel of c1 is the image of the map induced by exp.","Hodge.expSeq_chernClass_exact_right: exactness at H^2(X, Z): the kernel of the map induced by the inclusion is the image of c1.","Hodge.expSeq_chernClass_surj_onto_kernel: c1 surjects onto the kernel of H^2(X, Z) -> H^2(X, O_X).","Hodge.lefschetz_one_one_of_expSeq: Lefschetz (1,1) for the exponential sequence, conditional only on the Dolbeault isomorphism and the Hodge decomposition, with c1 the map from the exponential sequence itself."]}