Skip to content

Architecture Generalization Backlog

Chris Dare edited this page Nov 15, 2026 · 1 revision

Generalization backlog

Append-only. One row per altitude finding that could not be acted on inside the chunk that found it.

This file exists because a reviewer who notices that a theorem holds in greater generality usually notices it while reviewing a frozen chunk that may not touch the ancestor module. Before this file existed, that observation had nowhere to go: the reviewer could either suppress it or make it a blocking needs_changes the chunk could not legally satisfy. Both outcomes lost the finding.

Rules

  • Append only. Never delete a row; change its state in place.
  • A row is cheap. Record a lift you are unsure of rather than not recording it.
  • A falsified lift is a result, not a retraction. abstraction-tree.md is explicit that a failed unification recorded with its counterexample is a successful architecture outcome. Keep the row and mark it FALSIFIED.
  • Read this file in Phase 0, before choosing an implementation approach. A row here may already answer where the work belongs.

States

State Meaning
UNVERIFIED The lift is not yet confirmed in merged canonical code. A proof may have been compiled in a scratch probe or an unmerged candidate; record that evidence in the source note.
CONFIRMED <PR> The lift was implemented and merged.
FALSIFIED <counterexample> The general statement is false. Leaves stay separate.

Row schema

### <date> — <leaf declaration>
- chunk:              <chunk id>
- reviewing commit:   <sha>
- found by:           <reviewer or advisor name>
- proposed ancestor:  <module or namespace>
- weaker hypotheses:  <the binder/typeclass set the proof actually needs>
- state:              UNVERIFIED | CONFIRMED <PR> | FALSIFIED <counterexample>

Rows

2026-09-27 — generic Fourier--Mukai ShiftCompatibility (issue #1369)

  • chunk: 1369-shiftcompat-unification
  • reviewing commit: 00196ff8
  • found by: altitude-scout
  • proposed ancestor: CategoryTheory.NatTrans.CommShift, defined in module Mathlib.CategoryTheory.Shift.CommShift; the Fourier--Mukai record packages a selected CategoryTheory.Functor.CommShift structure and compatibility evidence for the supplied comparison isomorphism
  • weaker hypotheses: any natural transformation between functors carrying selected shift structures for an additive monoid; no isomorphism, Fourier--Mukai transform, triangulatedness, dg presentation, or cotwist shift is required
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Shift/CommShift.lean:360 pinned Mathlib source
  • source note: The pinned CategoryTheory.NatTrans.CommShift class asserts generic compatibility of a natural transformation with its source and target shift structures, leaving both structures as inputs. The CategoryTheory.NatTrans namespace is nested directly under CategoryTheory; Shift belongs to the defining module path, not the declaration namespace. The issue's record packages the independently selected Fourier--Mukai target structure alongside this proof. CategoryTheory.Functor.CommShift.ofIso can instead transport the source structure across the comparison, but that produces the transported choice rather than proving compatibility with an independently selected target. CategoryTheory.Functor.isTriangulated_of_iso consumes the selected structures and this compatibility evidence to transfer exactness. Stacks Project §13.3 (tag 05QK) describes the analogous compatibility for 2-morphisms between triangulated functors, in the additional setting where the functors are exact: Stacks Project §13.3, tag 05QK.
  • state: UNVERIFIED

2026-09-23 — K-flat resolutions and derived pullback for ringed topoi (planned)

  • chunk: sf8-5-task12-kflat-pullback
  • reviewing commit: e2332372922d884eea345f3ad79c536e66c22d84
  • found by: altitude-scout
  • proposed ancestor: ringed-site/topos module complexes and their derived pullback interface
  • weaker hypotheses: a morphism of ringed topoi and an arbitrary unbounded complex of modules; no scheme, affine, flatness, boundedness, quasicoherence, or Noetherian hypothesis
  • pin status: UPSTREAM-ONLY (absent from pinned Mathlib)
  • source note: Stacks Project §21.17, Lemma 21.17.11, gives an objectwise K-flat replacement with flat terms; §21.18, Lemmas 21.18.2–.3, constructs derived pullback and its composition law for ringed-topoi morphisms. The cited lemmas do not supply a functorial replacement object with natural comparison, and the derived-functor existence argument is not a Lean producer. This is a broader mathematical analogue, not a dependency or theorem of the scheme chunk. Sources: flat resolutions, derived pullback.
  • state: UNVERIFIED

2026-09-22 — compare the explicit affine H^{-1} witness with CategoryTheory.Tor (planned)

  • chunk: sf8-5-tor-witness-comparison

  • reviewing commit: b8306d6b

  • found by: altitude-scout

  • proposed ancestor: CategoryTheory.Tor, with a later comparison from the abstract left-derived tensor calculation to the displayed affine cochain representative

  • weaker hypotheses: an abelian monoidal preadditive category with projective resolutions and a chosen projective resolution; no distinguished ℤ → ZMod 2, no particular mapping-cone presentation, and no scheme-level pullback claim

  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Monoidal/Tor.lean:44

  • source note: The explicit nonzero H^{-1} witness is now proved in merged PR #1457. Mathlib's Tor derives the second tensor factor; the concrete ModuleCat specialization, selected projective-resolution comparison, chain/cochain indexing bridge, and restriction-of-scalars comparison remain unverified. Do not duplicate this row for the same lift.

  • state: UNVERIFIED

  • progress note (2026-09-22, after PR #1464): The concrete ℤ → ZMod 2 specialization, selected projective-resolution comparison, chain/cochain bridge, and restriction-of-scalars Tor comparison are merged. Only the proposed abstract left-derived tensor comparison under the weaker hypotheses above remains UNVERIFIED.

  • clarification (2026-09-22): The source note above records the status at the original #1457 review; its statement that the concrete comparison remained unverified is superseded by the post-#1464 progress note. The row's UNVERIFIED state applies only to the abstract lift.

2026-09-21 — ZModTwoNonflatDerived.baseChangedConeIso (planned)

  • chunk: sf8-5-nonflat-derived-effect
  • reviewing commit: 49c33eda
  • found by: altitude-scout
  • proposed ancestor: CochainComplex.mappingCone.mapHomologicalComplexIso
  • weaker hypotheses: an additive functor between preadditive categories and an arbitrary cochain morphism; no scalar-extension ring map, ℤ, or ZMod 2 is needed for the cone-transport comparison
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/Algebra/Homology/HomotopyCategory/MappingCone.lean:636
  • source note: This is the generic mapped-cone transport already suited to the representative-level comparison; whether a repository-facing wrapper is useful beyond the witness is unverified.
  • state: UNVERIFIED
  • progress note (2026-09-22, after PR #1457): The private witness comparison uses Mathlib's generic mapped-cone transport directly; no repository-facing wrapper was added, and whether it has independent consumers remains unverified. This is not an unimplemented prerequisite for the concrete witness.

2026-09-21 — ZModTwoNonflatDerived.affineKProjectivePullbackObject_homology_negOne_not_isZero (planned)

  • chunk: sf8-5-nonflat-derived-effect
  • reviewing commit: 49c33eda
  • found by: altitude-scout
  • proposed ancestor: ProjectiveResolution.isoLeftDerivedObj, potentially followed by a comparison with the repository's bounded-above-projective affine pullback lane
  • weaker hypotheses: an additive functor out of an abelian category with projective resolutions and a selected projective resolution; this omits the specific affine ring map and does not assert a general scheme-level derived pullback theorem
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Abelian/LeftDerived.lean:112
  • source note: The pinned theorem computes a left-derived functor from a projective resolution, but its chain indexing and its relation to the repository's K-projective representative have not been compared here.
  • state: UNVERIFIED
  • progress note (2026-09-22, after PR #1464): The concrete ℤ → ZMod 2 specialization and its chain/cochain degree bridge are proved; the proposed abstract comparison to ProjectiveResolution.isoLeftDerivedObj under the weaker hypotheses above remains unverified.

2026-09-21 — planned K-projective representative bridge (drop K.IsStrictlyLE d)

  • chunk: sf8-5-nonflat-derived-effect
  • reviewing commit: 49c33eda
  • found by: hypothesis-elimination-scout
  • proposed ancestor: KProjectiveHomotopyCategory.ofBoundedAboveProjectives and AlgebraicGeometry.DerivedCategory.Dqc.affineKProjectivePullbackObjIso
  • weaker hypotheses: keep [∀ n : ℤ, Projective (K.X n)], deliberately omit [K.IsStrictlyLE d] from the generic representative bridge
  • pin status: PIN-CONFIRMED DerivedAlgGeo/Algebra/Homology/DerivedCategory/KProjective.lean:89-94 and DerivedAlgGeo/AlgebraicGeometry/DerivedCategory/Dqc/AffineKProjectivePullback.lean:65-73
  • source note: A 3.2-second bounded ~/.elan/bin/lake env lean /dev/stdin probe instantiated both the object expression and affineKProjectivePullbackObjIso f K d without this binder. Both sites failed with failed to synthesize instance of type class K.IsStrictlyLE d.
  • state: FALSIFIED compiler witness: the current generic bridge cannot admit a complex without a strict upper bound.

2026-09-21 — planned K-projective representative bridge (drop degreewise projectivity)

  • chunk: sf8-5-nonflat-derived-effect
  • reviewing commit: 49c33eda
  • found by: hypothesis-elimination-scout
  • proposed ancestor: KProjectiveHomotopyCategory.ofBoundedAboveProjectives and AlgebraicGeometry.DerivedCategory.Dqc.affineKProjectivePullbackObjIso
  • weaker hypotheses: keep [K.IsStrictlyLE d], deliberately omit [∀ n : ℤ, Projective (K.X n)] from the generic representative bridge
  • pin status: PIN-CONFIRMED DerivedAlgGeo/Algebra/Homology/DerivedCategory/KProjective.lean:89-94 and DerivedAlgGeo/AlgebraicGeometry/DerivedCategory/Dqc/AffineKProjectivePullback.lean:65-73
  • source note: A 6.8-second bounded ~/.elan/bin/lake env lean /dev/stdin probe instantiated both the object expression and affineKProjectivePullbackObjIso f K d without this binder. Both sites failed with failed to synthesize ∀ (n : ℤ), Projective (K.X n) (after the pinned typeclass-heartbeat diagnostic).
  • state: FALSIFIED compiler witness: the current generic bridge cannot admit a merely bounded-above complex without degreewise projectivity.

2026-09-22 — existence of a minimizing object for rouquierDim (planned)

  • chunk: rou1-919-rouquier-dimension
  • reviewing commit: e18e20f9c3768269eedb622da04b9ab9f26ce127
  • found by: altitude-scout
  • proposed ancestor: CategoryTheory.Triangulated.rouquierDim, via ENat.exists_eq_iInf
  • weaker hypotheses: a nonempty index type and any function into ℕ∞; for rouquierDim, HasZeroObject supplies a local Nonempty C witness. No finite-dimension assumption or [IsTriangulated C] is needed.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/Data/ENat/Lattice.lean:115-116
  • source note: The pinned lemma gives an index attaining the infimum even when its value is ⊤. The definition still returns only a scalar and does not package that object. “Non-attainment” therefore describes the lack of a selected witness in the definition, not the absence of an existential minimizer. Whether to expose the stronger existence theorem is a separate API decision.
  • state: UNVERIFIED

2026-09-22 — classical generators are strong when one strong generator exists (planned)

  • chunk: rou1-919-rouquier-dimension
  • reviewing commit: e18e20f9c3768269eedb622da04b9ab9f26ce127
  • found by: altitude-scout
  • proposed ancestor: CategoryTheory.Triangulated.Generators, with the iterated-envelope composition theorem as the reusable result
  • weaker hypotheses: a triangulated category with some strong generator and a chosen classical generator; no scheme, stability, or geometric hypotheses.
  • pin status: UPSTREAM-ONLY (absent at the pinned revision)
  • source note: Stacks Project tag 0FXA proves that every classical generator is strong when the category has a strong generator. Pinned Mathlib records this as a TODO at .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Generators.lean:40-41; its current strong-to-classical theorem is at lines 196-200. This sharpens the planned “finite dimension gives a classical generator” consequence, but requires the composition-of-envelopes work and is outside this chunk.
  • state: UNVERIFIED

2026-09-22 — finite Rouquier witnesses avoid nonempty-index minimization

  • chunk: rou1-919-rouquier-dimension
  • reviewing commit: e18e20f9c3768269eedb622da04b9ab9f26ce127
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Triangulated.rouquierDim finite-bound and finite/strong-generator characterizations
  • weaker hypotheses: for any ι : Sort* and f : ι → ℕ∞, (⨅ i, f i) ≤ n ↔ ∃ i, f i ≤ n and (⨅ i, f i) ≠ ⊤ ↔ ∃ i, f i ≠ ⊤ compile without [Nonempty ι]; specializing to Rouquier dimension requires no local [Nonempty C] or [IsTriangulated C] proof assumption.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/Order/CompleteLattice/Defs.lean:313-315, .lake/packages/mathlib/Mathlib/Data/ENat/Basic.lean:325-326, and .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Generators.lean:86-90
  • source note: /tmp/rou1-919-no-nonempty-probe.lean compiled the arbitrary-index lemmas and the finite-bound, finite/strong-generator, classical-generator, and zero-stage statements for the proposed rouquierDim infimum formula (named candidateDim in the probe) against this exact commit. Finite witnesses follow from iInf_lt_iff at n + 1 and ⊤, followed by ENat.lt_coe_add_one_iff; the proof need not use ENat.exists_eq_iInf. The issue body suggests iInf_le_iff, but its pinned signature is a universal lower-bound condition, not an existential witness. triangEnvelopeIter_succ has only the existing pretriangulated context, whereas triangEnvelopeIter_add requires [IsTriangulated C]. This lift does not remove HasZeroObject from the public API, which the reused generation-time/envelope API requires.
  • state: L (proof-witness verified)

2026-09-22 — zero Rouquier dimension needs no inhabited infimum index

  • chunk: rou1-919-rouquier-dimension
  • reviewing commit: e18e20f9c3768269eedb622da04b9ab9f26ce127
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Triangulated.rouquierDim zero-stage characterization through ENat.iInf_eq_zero
  • weaker hypotheses: for any ι : Sort* and f : ι → ℕ∞, (⨅ i, f i) = 0 ↔ ∃ i, f i = 0 compiles without [Nonempty ι]; the Rouquier specialization needs no local [Nonempty C] or [IsTriangulated C] proof assumption.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/Data/ENat/Lattice.lean:71 and .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Generators.lean:66-68
  • source note: /tmp/rou1-919-no-nonempty-probe.lean compiled the arbitrary-index lemma and the zero-stage Rouquier equivalence using ENat.iInf_eq_zero and generationTime_eq_zero_iff'; this proof route does not need an inhabited-index instance.
  • state: L (proof-witness verified)

2026-09-23 — affineKProjectiveSchemeModulePullbackComparison

  • chunk: sf8-5-affine-kprojective-scheme-comparison
  • reviewing commit: b3d0ad4aeb971972e47e7b4be28a06e801cdc4e1
  • found by: abstraction-adversary
  • proposed ancestor: DerivedAlgGeo/Algebra/Homology/DerivedCategory/KProjective.lean, beside CategoryTheory.kProjectiveLocusDerivedFunctor
  • weaker hypotheses: abelian C, D, E; additive F : C ⥤ E and G : C ⥤ D; exact T : D ⥤ E; and e : F ≅ G ⋙ T. The comparison needs no ring map, scheme, or tilde functor beyond data supplying e.
  • source note: The reviewer searched this repository's KProjective.lean, BoundedAboveProjective.lean, and ExactFunctor.lean, plus pinned Mathlib's homotopy-category, exact-derived-functor, K-projective, and localization APIs. Existing composition/comparison results do not provide this general K-projective-locus localization transport. In the reviewed affine proof, homotopy-category composition and exact-functor factorization through localization use only the hypotheses above (implementation lines 95–130). The generic API is outside this frozen chunk and is deferred; do not move or duplicate the geometric declaration to implement it.
  • state: UNVERIFIED

2026-09-23 — SerreFunctorData.fullyFaithful without Hom-finiteness

  • chunk: srf1-897-full-faithfulness
  • reviewing commit: 74a05e716502cd3034958530c8134021fd2688c1
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Linear.SerreFunctor.Equivalence
  • weaker hypotheses: retain [Field k] [Category C] [Preadditive C] [Linear k C] and the Serre duality data, but omit [HomFinite k C] and all replacement reflexivity hypotheses.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/LinearAlgebra/Dual/Defs.lean:229 and .lake/packages/mathlib/Mathlib/LinearAlgebra/Dual/Lemmas.lean:288-306
  • source note: /tmp/srf1-hypothesis-probes.lean attempted the double-dual Hom equivalence without HomFinite; Mathlib failed to synthesize Module.IsReflexive k (A ⟶ B) for Module.evalEquiv. The full-faithfulness chain needs reflexivity of each Hom module. A probe replacing Hom-finiteness by [∀ A B, Module.IsReflexive k (A ⟶ B)] compiled, but over the existing [Field k] root Mathlib infers FiniteDimensional from reflexivity, so that is not a genuine weakening. The adjunction from both duality structures does not itself use Hom-finiteness; this row concerns the double-dual step.
  • state: FALSIFIED compiler witness: without Hom-finiteness or reflexivity, Module.evalEquiv k (A ⟶ B) has no Module.IsReflexive instance.

2026-09-23 — SerreFunctorData.fullyFaithful over a commutative ring

  • chunk: srf1-897-full-faithfulness
  • reviewing commit: 74a05e716502cd3034958530c8134021fd2688c1
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Linear.SerreFunctor, if its existing scalar binder is ever generalized
  • weaker hypotheses: replace [Field k] by [CommRing k] while retaining finite generation [Module.Finite k (A ⟶ B)] for each Hom module.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/LinearAlgebra/Dual/Defs.lean:229 and .lake/packages/mathlib/Mathlib/LinearAlgebra/Dual/Lemmas.lean:258-288
  • source note: /tmp/srf1-hypothesis-probes.lean attempted Module.evalEquiv R M with [CommRing R] [Module.Finite R M]; pinned Mathlib could not synthesize Module.IsReflexive R M. Finite generation over a commutative ring does not supply the finite-projective/free reflexivity hypotheses used by the available instances. This falsifies the weakening with finite generation alone; an explicit reflexivity or finite-projective condition is a different hypothesis set. The existing Serre root is field-based, and this finding is outside the frozen #897 chunk.
  • state: FALSIFIED compiler witness: Module.evalEquiv R M fails with failed to synthesize Module.IsReflexive R M.

2026-09-24 — stabilization of a lifted localization-heart filtration (planned)

  • chunk: sf11-4a-1496-noetherian-cover
  • reviewing commit: ce58cc9e44340c07c89836ba2a81fdd07aaee6b4
  • found by: altitude-scout
  • proposed ancestor: Mathlib.CategoryTheory.Subobject.NoetherianObject
  • weaker hypotheses: any category C, object E : C, and a filtration F : ℕ ⥤ MonoOver E; no scheme or t-structure hypotheses.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Subobject/NoetherianObject.lean:88
  • source note: isNoetherianObject_iff_isEventuallyConstant supplies ACC once a localization filtration is lifted into subobjects of a fixed global heart object. The current FiltrationLiftHypotheses lifts to a chain of objects but records no common ambient object or embeddings, so this application remains unverified.
  • state: UNVERIFIED
  • progress note (2026-09-24): The categorical transfer from an explicit fixed-ambient subobject-chain lift is proved in this SF11.4 continuation. The geometric lift for a section-open or affine localization, the bounded D_T heart restriction, and their comparison remain UNVERIFIED.
  • progress note (2026-09-24): Successive binary joins now turn pointwise subobject lifts into a fixed-ambient monotone chain. A triangulated t-exact restriction induces an exact heart functor and therefore preserves those joins, so the heart-level transfer needs only chainwise pointwise lifts. The bounded-component geometric pointwise lift is still not supplied by the existing SLocal owner-data fields.
  • progress note (2026-09-24): A mono/epi-preserving functor between the abelian hearts maps image subobjects to image subobjects. Fixed-target mono extension therefore implies pointwise subobject lifting, provided the relevant target-heart object and its chosen source ambient lift are paired. All-arrow extension at every source-heart target is a stronger criterion: applying it to a zero arrow supplies the target-object lift as well. Constructing even the required mono extensions for the bounded coherent base-change component remains UNVERIFIED.

2026-09-24 — anchored_chain_of_pointwise_lifts with local join structure

  • chunk: sf11-4d-pointwise-anchored
  • reviewing commit: 46d220ce422f8ea2aa1c712de1c08634522e7b56
  • found by: abstraction-adversary
  • proposed ancestor: DerivedAlgGeo/CategoryTheory/Subobject/NoetherianObject.lean
  • weaker hypotheses: binary joins only in Subobject X and in the target ambient subobject poset, rather than category-wide images and binary coproducts
  • pin status: UNCOMPILED
  • source note: The current proof uses Mathlib's category-wide instances to synthesize the two joins. A locally quantified semilattice formulation may suffice mathematically, but its Lean instance binding and theorem type have not been tested; do not present it as an available API.
  • state: UNVERIFIED

2026-09-24 — SLocalSlicingHypotheses.restriction_eq_of_phase_iff

  • chunk: sf11-4a-1496-noetherian-cover
  • reviewing commit: ce58cc9e44340c07c89836ba2a81fdd07aaee6b4
  • found by: hypothesis-elimination-scout
  • proposed ancestor: DerivedAlgGeo/AlgebraicGeometry/DerivedCategory/Stability/SLocal.lean
  • weaker hypotheses: drop _L : SLocalSlicingHypotheses R s; retain the open, compactness, both restrictions, and the phase comparison.
  • pin status: Focused copy of the existing proof compiled from /dev/stdin at this base.
  • state: UNVERIFIED

2026-09-24 — Coh.restrict_jointlyReflectsIsomorphisms

  • chunk: sf11-4a-1496-noetherian-cover
  • reviewing commit: ce58cc9e44340c07c89836ba2a81fdd07aaee6b4
  • found by: hypothesis-elimination-scout
  • proposed ancestor: DerivedAlgGeo/AlgebraicGeometry/Modules/Coherent/Noetherian.lean
  • weaker hypotheses: drop [IsLocallyNoetherian X]; retain the open cover and the premise that every restricted map is an isomorphism.
  • pin status: Focused copy of the existing proof compiled from /dev/stdin at this base.
  • state: UNVERIFIED

2026-09-24 — isNoetherianObject_of_finite_jointlyReflectsIsomorphisms

  • chunk: sf11-4a-1496-noetherian-cover
  • reviewing commit: 1187331a9df2c7be6907d2385eb3329675facd6b
  • found by: abstraction-adversary
  • proposed ancestor: DerivedAlgGeo/CategoryTheory/Subobject/NoetherianObject.lean
  • weaker hypotheses: jointly reflect isomorphisms only for monomorphisms, rather than all morphisms; no scheme or t-structure hypothesis
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Subobject/Basic.lean:298
  • source note: The generic detector applies joint reflection to Subobject.ofLE, which is monic at the pinned Mathlib API. A theorem with reflection restricted to monomorphisms may therefore suffice, but the weaker statement has not been compiled. This generic API change is outside the SF11.4a geometry chunk.
  • state: UNVERIFIED

2026-09-24 — compare the open free-Yoneda sheaf with extension by zero

  • chunk: sf8-flat-generators
  • reviewing commit: 300a828ce99b9b7b02e7f77320bd1aacde3fc84
  • found by: altitude-scout
  • proposed ancestor: the extension-by-zero functor j_! for an open immersion
  • weaker hypotheses: an open immersion of ringed spaces and a module sheaf on its source; no scheme-specific free-Yoneda construction is needed for the upstream stalk formula
  • pin status: UPSTREAM-ONLY (Stacks Project Lemma 6.31.8, tag 00A7; no pinned Lean comparison established)
  • source note: Stacks gives zero stalks off the open and the original stalk on it for j_!. This does not prove that the repository's canonical SheafOfModules.freeYonedaSheaf is j_! of the restricted structure sheaf. The current scheme-level result proves the stalk formula directly and makes no such identification.
  • state: UNVERIFIED

2026-09-21 — CommShift₂Int output transport of ObjectProperty.lift₂

  • chunk: dt1-928-exact-bifunctor-restriction
  • reviewing commit: 858ca700829706455ddc7ea3f4ca9fff4ffa951b
  • found by: altitude-scout
  • proposed ancestor: CategoryTheory.Functor.CommShift₂ via a general fully faithful output-composition transport
  • weaker hypotheses: coherence-only statement for arbitrary C₁, C₂, D, D', additive-commutative shift monoid M, shifts, explicitly compatible CommShift₂Setup data, a fully faithful shift-compatible H : D ⥤ D', a bifunctor postcomposition isomorphism, and a CommShift₂ witness after postcomposition; no object property, closure witness, Pretriangulated, or ExactBifunctor fields
  • state: UNVERIFIED

2026-09-23 — liftNatTrans_commShift

  • chunk: dt1-928-exact-bifunctor-restriction
  • reviewing commit: 8a61f0e94cd9c414f172d97535f62c1684a91c21
  • found by: abstraction-adversary
  • proposed ancestor: DerivedAlgGeo/CategoryTheory/Shift/CommShift.lean, as a general faithful-postcomposition reflection lemma for NatTrans.CommShift
  • weaker hypotheses: arbitrary categories with shifts by an additive monoid; F, G : E ⥤ D and H : D ⥤ E' commute with shifts; H is faithful; and τ : F ⟶ G commutes with shifts after whiskering by H. No object property, closure, or triangulated hypotheses.
  • evidence: the private commShift_of_whiskerRight proof uses only H.map_injective, H's shift naturality, and the assumed shift compatibility of the whiskered transformation; liftNatTrans_commShift applies it to the restricted natural transformation.
  • state: UNVERIFIED

2026-09-23 — ExactBifunctor.lift₂ with distinct full subcategories

  • chunk: dt1-928-exact-bifunctor-restriction
  • reviewing commit: ffb17b7b8cac3feecda2288845167f9470f79dd1
  • found by: altitude-scout
  • proposed ancestor: CategoryTheory.Functor.ExactBifunctor.lift₂, generalized to distinct domain, second-input, and output ObjectProperty full subcategories
  • weaker hypotheses: an ambient bifunctor on three (possibly distinct) triangulated categories with a supplied ExactBifunctor, three object properties whose full subcategories inherit triangulated structures, and an explicit closure witness from the first two properties into the output property; no single-category or same-property identification among the three positions
  • source note: This is a plausible API generalization inferred from the current restriction's use of one category and property for all three positions. The distinct-category coherence transport was not proved or compiled during this review.
  • state: UNVERIFIED

2026-09-26 — generation-time +1 submultiplicativity without P.Nonempty

  • chunk: rou1-920-envelope-composition
  • reviewing commit: b674a436dd002da2487d744531f804ddbb795628
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.ObjectProperty.generationTime and the planned envelope-composition API
  • weaker hypotheses: the scalar inequality P.generationTime R + 1 ≤ (P.generationTime Q + 1) * (Q.generationTime R + 1) needs no separate [P.Nonempty] binder. The iterated-envelope composition law and zero-stage fixed-point statement also hold without [P.Nonempty]; the fixed-point statement additionally needs no [IsTriangulated C].
  • pin status: PIN-CONFIRMED for the corollary proof route
  • source note: CategoryTheory.ObjectProperty.generationTime_add_one_submultiplicative proves the scalar inequality by splitting the empty-generator case from finite generation-time witnesses. triangEnvelopeIter_compose handles an empty property by showing all its iterates are bottom; the nonempty branch uses the fixed-point lemma. The fixed-point proof uses triangEnvelopeIter_succ, which needs only the shared pretriangulated context, while the composition proof uses the triangulated stage-addition API.
  • state: L (proof-witness verified)

2026-09-26 — products of retract-closed object properties

  • chunk: rou1-920-envelope-composition
  • reviewing commit: 12b1f252d33ca17406fc85197fa060c33794fc78
  • found by: abstraction-adversary
  • proposed ancestor: Mathlib.CategoryTheory.ObjectProperty.Retract
  • weaker hypotheses: a category with binary products and an object property Q closed under binary products; no additive, shift, or triangulated structure
  • pin status: UPSTREAM-ONLY (the pinned ObjectProperty/Retract.lean and ObjectProperty/FiniteProducts.lean provide no such closure instance)
  • source note: The private retractClosure_isClosedUnderBinaryProducts proof constructs products of two retracts componentwise and transfers the product property across an isomorphism. Whether this belongs as a general closure API in Mathlib's retract module remains to be confirmed.
  • state: UNVERIFIED

2026-09-26 — products of extension-closed object properties

  • chunk: rou1-920-envelope-composition
  • reviewing commit: 12b1f252d33ca17406fc85197fa060c33794fc78
  • found by: abstraction-adversary
  • proposed ancestor: Mathlib.CategoryTheory.Triangulated.Subcategory
  • weaker hypotheses: the existing pretriangulated context, binary products, and binary-product closure of both input properties; no IsTriangulated or nonemptiness assumption
  • pin status: UPSTREAM-ONLY (the pinned Subcategory.lean has only the more restrictive closure instance requiring a triangulated object property)
  • source note: The private extensionProduct_prop_prod and extensionProduct_isClosedUnderBinaryProducts proofs form products of the input distinguished triangles using Mathlib's productTriangle_distinguished. Whether to expose this weaker closure result as general Subcategory API remains to be confirmed.
  • state: UNVERIFIED

2026-09-26 — exact-functor transport of finite triangulated-generation stages

  • chunk: rou1-921-generation-functor-transport
  • reviewing commit: e83b614078d491b7f1d8dab085bf494fddb32ebc
  • found by: altitude-scout
  • proposed ancestor: CategoryTheory.Triangulated.Generators, for pointwise transport of ObjectProperty.triangEnvelopeIter; the generation-time and Rouquier-dimension consequences remain downstream
  • weaker hypotheses: an exact functor between triangulated categories and an object property with a finite generation-stage bound; essential surjectivity is needed only for the category-level Rouquier-dimension inequality, and full faithfulness is not needed
  • pin status: UPSTREAM-ONLY (the transport and dimension statements are absent at the pinned revision)
  • source note: Olander, Ample line bundles and generation time, §4, uses that finite-stage generated subcategories are preserved by exact functors (published PDF p. 304, lines 324-328). Its Lemma 7 proves that an essentially-surjective exact functor does not increase countable Rouquier dimension (PDF p. 303, lines 224-230); specializing its stage argument to a singleton gives the planned ordinary Rouquier-dimension inequality. This is a literature precedent, not a pinned Lean API. At the pin, triangEnvelopeIter and its stage recurrence are defined at .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Generators.lean:62-80, while Functor.IsTriangulated and its additive/product-preservation instances are at .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Functor.lean:181-235; the searched Mathlib tree has no functor transport of these stages or Rouquier-dimension API. The quasi-inverse triangulated structure is already supplied for Equivalence.IsTriangulated at .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Adjunction.lean:186-198. Source: Olander, published PDF. Merged implementation in PR #1590: triangEnvelopeIter_map_obj and triangEnvelopeIter_map_le are in DerivedAlgGeo/CategoryTheory/Triangulated/Generators/Functor.lean; generation-time and Rouquier-dimension consequences are in DerivedAlgGeo/CategoryTheory/Triangulated/Dimension/Functor.lean.
  • state: CONFIRMED #1590

2026-09-26 — generation transport needs only pretriangulated categories (planned)

  • chunk: rou1-921-generation-functor-transport
  • reviewing commit: e83b614078d491b7f1d8dab085bf494fddb32ebc
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.ObjectProperty envelope transport and CategoryTheory.Triangulated.Dimension.Functor numerical consequences
  • weaker hypotheses: independently universe-polymorphic C and D, each with Category, HasZeroObject, HasShift _ ℤ, Preadditive, additive shifts, and Pretriangulated; F : C ⥤ D with F.CommShift ℤ and F.IsTriangulated. Neither category needs IsTriangulated (octahedral). Pointwise transport additionally binds P : ObjectProperty C, n : ℕ, X : C, and the source stage-membership proof; the map inequality binds only P,n; generation-time monotonicity binds arbitrary P,Q. None needs property nonemptiness, ContainsZero, closure under isomorphisms, F.Full, F.Faithful, or F.EssSurj. Additivity and binary-product preservation of F are inferred, not additional assumptions. Dimension monotonicity adds F.EssSurj; strong-generator transport adds G : C and (singleton G).IsStrongTriangulatedGenerator.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Functor.lean:181-235 and Triangulated/Generators.lean:62-74
  • source note: /tmp/rou1-921-hypothesis-probes.lean compiled complete proofs of pointwise tower transport, the ObjectProperty.map inequality, generation-time monotonicity, Rouquier-dimension monotonicity and equivalence invariance, generation-time equivalence invariance, and strong-generator transport with the above category hypotheses. The tower proof follows the left-associated triangEnvelopeIter_succ, so never uses the octahedral recurrence or the composition law. Verified using LEAN_NUM_THREADS=2 ~/.elan/bin/lake env lean; only unused-section-variable warnings in three scratch helper lemmas remain. No implementation module was changed by this scout. Merged implementation in PR #1590 uses the recorded Pretriangulated C and Pretriangulated D context, without an IsTriangulated (octahedral) assumption; see the stage and generation-time results in DerivedAlgGeo/CategoryTheory/Triangulated/Generators/Functor.lean and DerivedAlgGeo/CategoryTheory/Triangulated/Dimension/Functor.lean.
  • state: CONFIRMED #1590

2026-09-26 — equivalence transport can construct the inverse's exact data (planned)

  • chunk: rou1-921-generation-functor-transport
  • reviewing commit: e83b614078d491b7f1d8dab085bf494fddb32ebc
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Triangulated.Dimension.Functor
  • weaker hypotheses: the pretriangulated category context recorded above, E : C ≌ D, [E.functor.CommShift ℤ], and [E.functor.IsTriangulated]; omit separate inverse-shift, inverse-exactness, and E.CommShift assumptions. Generation-time invariance additionally binds arbitrary P,Q : ObjectProperty C, without either being replete.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Shift/Adjunction.lean:617-624 and Triangulated/Adjunction.lean:61
  • source note: /tmp/rou1-921-hypothesis-probes.lean compiled dim_equivalence and time_equivalence using letI := E.commShiftInverse ℤ, letI := E.commShift_of_functor ℤ, and letI := E.toAdjunction.isTriangulated_rightAdjoint. The proof constructs the compatible inverse shift, rather than claiming an independently chosen inverse shift is automatically compatible. The generation-time proof also compiled P.isoClosure.triangEnvelopeIter n = P.triangEnvelopeIter n, accounting for the isomorphism closure of ObjectProperty.map. Merged implementation in PR #1590 constructs the inverse compatibility locally in generationTime_map_eq_of_equiv and rouquierDim_eq_of_equiv; the strong-generator equivalence result is exists_isStrongTriangulatedGenerator_iff_of_equiv.
  • state: CONFIRMED #1590

2026-09-26 — Rouquier dimension and strong generators under retract-dense functors (planned)

  • chunk: rou1-921-generation-functor-transport
  • reviewing commit: e83b614078d491b7f1d8dab085bf494fddb32ebc
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Triangulated.Dimension.Functor
  • weaker hypotheses: the pretriangulated category and exact-functor context recorded above, replacing [F.EssSurj] by ∀ Y : D, ∃ X : C, Nonempty (Retract Y (F.obj X)); strong-generator transport also binds G : C and its singleton strong-generation proof.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/ObjectProperty/Retract.lean:34
  • source note: /tmp/rou1-921-hypothesis-probes.lean compiled dim_retract_dense and strong_retract_dense with complete proofs. Apply pointwise singleton-stage transport and then prop_of_retract to the target envelope stage. Essential surjectivity supplies this premise using (F.objObjPreimageIso Y).symm.retract; that specialization also compiled. This proposes an explicit hypothesis, not a new carrier or typeclass; whether to expose the stronger API in #921 remains an implementation scope decision. Merged implementation in PR #1590 proves rouquierDim_le_of_retract_coverage in DerivedAlgGeo/CategoryTheory/Triangulated/Dimension/Functor.lean and isStrongTriangulatedGenerator_map_of_retract_coverage in DerivedAlgGeo/CategoryTheory/Triangulated/Generators/Functor.lean.
  • state: CONFIRMED #1590

2026-09-26 — replace exactness by additivity and shift compatibility (planned)

  • chunk: rou1-921-generation-functor-transport
  • reviewing commit: e83b614078d491b7f1d8dab085bf494fddb32ebc
  • found by: hypothesis-elimination-scout
  • proposed ancestor: distinguished-triangle step of finite-stage transport
  • weaker hypotheses: retain the pretriangulated category context, [F.CommShift ℤ], and [F.Additive]; omit [F.IsTriangulated].
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Functor.lean:185
  • source note: /tmp/rou1-921-failed-hypothesis-probes.lean attempted F.map_distinguished T hT with exactly these assumptions. The compiler reports failed to synthesize instance of type class F.IsTriangulated. This falsifies deleting exactness from this proof route; it is not a compiler-produced mathematical counterexample to every alternate statement.
  • state: UNVERIFIED

2026-09-26 — omit compatibility for a separately chosen inverse shift (planned)

  • chunk: rou1-921-generation-functor-transport
  • reviewing commit: e83b614078d491b7f1d8dab085bf494fddb32ebc
  • found by: hypothesis-elimination-scout
  • proposed ancestor: Equivalence inverse triangulated-functor derivation
  • weaker hypotheses: retain [E.functor.CommShift ℤ], [E.functor.IsTriangulated], and an independently supplied [E.inverse.CommShift ℤ]; omit [E.CommShift ℤ].
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Adjunction.lean:53-61
  • source note: /tmp/rou1-921-failed-hypothesis-probes.lean attempted E.toAdjunction.isTriangulated_rightAdjoint for that inverse shift. The compiler reports failed to synthesize instance of type class E.toAdjunction.CommShift ℤ. The verified construction above avoids this failure by choosing the compatible inverse shift locally; it does not erase compatibility from the adjunction theorem.
  • state: UNVERIFIED

2026-09-26 — transport of shift-closed properties for arbitrary additive shifts (planned)

  • chunk: rou1-921-triangulated-functor-transport
  • reviewing commit: 65b2dff1ef8aaa5a986c8695f9e59dcc3b236c56
  • found by: abstraction-adversary (review round 1)
  • proposed ancestor: DerivedAlgGeo/CategoryTheory/ObjectProperty/Shift.lean
  • weaker hypotheses: categories with shifts by an additive monoid A, [F.CommShift A], and pointwise transport of the generating property; no preadditivity, zero object, or triangulated structure
  • pin status: PIN-CONFIRMED
  • source note: Replacing ℤ by A in the complete shift-closure transport proof compiles unchanged; only F.mapIso and F.commShiftIso are used. The proposed owner is outside the frozen #921 file list, so this lift is deferred.
  • state: UNVERIFIED

2026-09-26 — transport of arbitrary-shape limit closures (planned)

  • chunk: rou1-921-triangulated-functor-transport
  • reviewing commit: 65b2dff1ef8aaa5a986c8695f9e59dcc3b236c56
  • found by: abstraction-adversary (review round 1)
  • proposed ancestor: DerivedAlgGeo/CategoryTheory/ObjectProperty/LimitsClosure.lean
  • weaker hypotheses: an arbitrary family J : α → Type, categories on each J a, preservation of each J a-shaped limit by F, and pointwise transport of the generating property; no WalkingPair specialization
  • pin status: PIN-CONFIRMED
  • source note: The complete limitsClosure_le proof compiles for P.limitsClosure J; no proof step uses the binary-product shape. The proposed owner is outside the frozen #921 file list, so this lift is deferred.
  • state: UNVERIFIED

2026-09-26 — shift closure commutes with isomorphism closure for arbitrary additive shifts (planned)

  • chunk: rou1-921-triangulated-functor-transport
  • reviewing commit: 65b2dff1ef8aaa5a986c8695f9e59dcc3b236c56
  • found by: abstraction-adversary (review round 1)
  • proposed ancestor: DerivedAlgGeo/CategoryTheory/ObjectProperty/Shift.lean
  • weaker hypotheses: one category with shifts by any additive monoid A and an arbitrary object property P; no preadditivity, zero object, or triangulation
  • pin status: PIN-CONFIRMED
  • source note: The existing three-line argument for P.isoClosure.shiftClosure A = P.shiftClosure A compiles with only the additive-shift assumptions. The proposed owner is outside the frozen #921 file list, so this lift is deferred.
  • state: UNVERIFIED

2026-09-27 — extension of the resolution-degree-zero augmentation target (planned)

  • chunk: sf8-554-total-zero-comparison
  • reviewing commit: 1d29d6696bb048fc2a3e648fab4b39931d836488
  • found by: altitude-scout
  • proposed ancestor: HomologicalComplex.extendSingleIso in Mathlib.Algebra.Homology.Embedding.Extend
  • weaker hypotheses: an arbitrary category with zero morphisms and a zero object, decidable equality on both complex index types, an embedding of arbitrary complex shapes, and an equality identifying the image of the single supported degree; no schemes, abelianness, or flatness
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/Algebra/Homology/Embedding/Extend.lean:293
  • source note: At Mathlib pin 520045ab14e26149ee970e2e617ca04b09bde5d6, extendSingleIso e X i i' h identifies the extension of the single complex at i with the single complex at i'. Specializing to embeddingDownNat, i = 0, and i' = 0 supplies the objectwise identification of the inner complexes in the target of freeYonedaSheafCoproductTotalAugmentationToSingleZero. It does not identify a total complex with the input, and its declaration is an objectwise isomorphism rather than the required natural isomorphism of totalization functors. The functorial assembly and total comparison are still obligations; no proof was compiled by this research scout.
  • state: UNVERIFIED

2026-09-27 — exchanging the supported axis before a single-zero total comparison (planned)

  • chunk: sf8-554-total-zero-comparison
  • reviewing commit: 1d29d6696bb048fc2a3e648fab4b39931d836488
  • found by: altitude-scout
  • proposed ancestor: HomologicalComplex₂.totalFlipIso in Mathlib.Algebra.Homology.TotalComplexSymmetry
  • weaker hypotheses: a preadditive category, arbitrary input and output complex shapes with total-shape structures in both orders and compatible symmetry signs, existence of the chosen bicomplex total, and decidable equality on the output indices; no schemes, flatness, or abelianness
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/Algebra/Homology/TotalComplexSymmetry.lean:114
  • source note: This is the canonical isomorphism K.flip.total c ≅ K.total c. The repository already proves its generic naturality in Algebra/Homology/SpectralSequence/TotalFlipNaturality.lean:47. It permits a second-axis support comparison to reuse a first-axis comparison, but does not itself collapse a single supported axis. The existing singleZeroTotalIso in Algebra/Homology/SpectralSequence/FilteredTotalComplexAdjacent.lean:410 and its naturality in TotalQuasiIso.lean:248 are explicitly restricted to AddCommGrpCat, so they do not instantiate directly for X.Modules. CategoryTheory/Sites/SheafCohomology/Cech/TotalComparison.lean:154 demonstrates composition through this flip and the existing single-zero comparison for abelian groups. A generic extraction and the required bicomplex adapter remain to be checked; none was compiled by this scout.
  • state: UNVERIFIED

2026-09-27 — natural recovery of the supported term of a single complex (planned)

  • chunk: sf8-554-total-zero-comparison
  • reviewing commit: 1d29d6696bb048fc2a3e648fab4b39931d836488
  • found by: altitude-scout
  • proposed ancestor: HomologicalComplex.singleCompEvalIsoSelf in Mathlib.Algebra.Homology.Single
  • weaker hypotheses: an arbitrary category with zero morphisms and a zero object, an arbitrary complex shape, decidable equality on its indices, and a chosen supported degree; no preadditivity or geometric assumptions
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/Algebra/Homology/Single.lean:103
  • source note: The natural isomorphism single V c j ⋙ eval V c j ≅ 𝟭 V recovers the surviving degree, while the adjacent isZero_single_comp_eval treats all other degrees. This supplies canonical component data for the single-axis comparison; it is an evaluation theorem and does not subsume the direct-sum total comparison or its differential compatibility. The related pinned singleMapHomologicalComplex in Algebra/Homology/Additive.lean:266 already makes single-complex formation commute with any functor preserving zero morphisms, but likewise contains no totalization result. Source inspection did not find a general single-complex total theorem at the pin. The three-candidate cap was reached on disk, so no external sources were fetched. No proof was compiled by this scout.
  • state: UNVERIFIED

2026-09-27 — single-zero total comparison outside abelian groups (planned)

  • chunk: sf8-554-total-zero-comparison
  • reviewing commit: 1d29d6696bb048fc2a3e648fab4b39931d836488
  • found by: hypothesis-elimination-scout
  • proposed ancestor: the canonical HomologicalComplex₂.singleZeroTotalIso and singleZeroTotalIso_naturality beside Mathlib's total-complex API, extracted from the current spectral-sequence consumers
  • weaker hypotheses: {C : Type u} [Category.{v} C] [Preadditive C] [HasZeroObject C], an arbitrary A : CochainComplex C ℤ, and only [(singleZeroBicomplex A).HasTotal (ComplexShape.up ℤ)]; naturality also takes an arbitrary B, its selected-total instance, and f : A ⟶ B. No Abelian C, all-coproducts, boundedness, flatness, or scheme assumption.
  • pin status: PIN-CONFIRMED Mathlib 520045ab14e26149ee970e2e617ca04b09bde5d6
  • source note: .lake/sf8_hypothesis_minimal.lean copied the existing comparison and naturality proof into a scratch namespace, replacing AddCommGrpCat with C. The only proof adjustment was to expose generic componentwise cancellation for an isomorphism of complexes, proved by HomologicalComplex.comp_f and the iso identities. The complete proof, naturality, and the composite totalFlipIso ≪≫ singleZeroTotalIso compiled with LEAN_NUM_THREADS=2 /home/chris-dare/.elan/bin/lake env lean .lake/sf8_hypothesis_minimal.lean. The printed axiom sets for the comparison and naturality contain only propext, Classical.choice, and Quot.sound. The subsequent support witness below removes even the selected-total assumptions. These are scratch witnesses, not merged public declarations.
  • state: L (proof-witness verified)

2026-09-27 — single support supplies the required total coproducts (planned)

  • chunk: sf8-554-total-zero-comparison
  • reviewing commit: 1d29d6696bb048fc2a3e648fab4b39931d836488
  • found by: hypothesis-elimination-scout
  • proposed ancestor: the same canonical single-zero bicomplex comparison beside Mathlib's total-complex API
  • weaker hypotheses: {C : Type u} [Category.{v} C] [Preadditive C] [HasZeroObject C] and an arbitrary A : CochainComplex C ℤ; no supplied HasTotal, countable coproducts, finite biproducts, or arbitrary coproducts. Naturality adds only B : CochainComplex C ℤ and f : A ⟶ B.
  • pin status: PIN-CONFIRMED Mathlib 520045ab14e26149ee970e2e617ca04b09bde5d6
  • source note: .lake/sf8_hypothesis_support.lean reorganizes the existing surviving-summand/inverse calculation into a cofan with point A.X n. Its universal property selects the (0,n) summand; every other summand is zero. GradedObject.CofanMapObjFun.hasMap then supplies singleZeroHasTotal. The comparison, its naturality, and the second-axis comparison through totalFlipIso compile without any colimit assumption. .lake/sf8_hypothesis_narrow.lean additionally compiles the same witness with only Mathlib.Algebra.Homology.Single and Mathlib.Algebra.Homology.TotalComplexSymmetry imports, using the underlying HomologicalComplex.single instead of the definitionally equal CochainComplex.singleFunctor spelling. Command: LEAN_NUM_THREADS=2 /home/chris-dare/.elan/bin/lake env lean .lake/sf8_hypothesis_narrow.lean. Exit code 0; comparison and naturality depend only on propext, Classical.choice, and Quot.sound. No stable Lean source was edited. This records the support argument for the planned extraction, not a general coproduct-existence theorem for arbitrary totals.
  • state: L (proof-witness verified)

2026-09-27 — drop the zero object from the single-zero comparison (attempted)

  • chunk: sf8-554-total-zero-comparison
  • reviewing commit: 1d29d6696bb048fc2a3e648fab4b39931d836488
  • found by: hypothesis-elimination-scout
  • proposed ancestor: the canonical single-zero bicomplex constructor
  • weaker hypotheses: {C : Type u} [Category.{v} C] [Preadditive C] and A : CochainComplex C ℤ, deliberately omitting HasZeroObject C
  • pin status: PIN-CONFIRMED Mathlib 520045ab14e26149ee970e2e617ca04b09bde5d6
  • source note: .lake/sf8_hypothesis_nozero.lean, compiled with LEAN_NUM_THREADS=1 /home/chris-dare/.elan/bin/lake env lean, fails at (CochainComplex.singleFunctor (CochainComplex C ℤ) 0).obj A with the exact diagnostic failed to synthesize instance of type class HasZeroObject (CochainComplex C ℤ). The canonical single complex needs an actual zero object to populate every unsupported degree. This is a boundary of the current construction, not a claim about all possible alternate APIs.
  • state: FALSIFIED compiler witness: the current single-complex constructor does not accept this hypothesis deletion.

2026-09-27 — replace preadditivity by zero morphisms for totalization (attempted)

  • chunk: sf8-554-total-zero-comparison
  • reviewing commit: 1d29d6696bb048fc2a3e648fab4b39931d836488
  • found by: hypothesis-elimination-scout
  • proposed ancestor: HomologicalComplex₂.total
  • weaker hypotheses: {C : Type u} [Category.{v} C] [HasZeroMorphisms C] [HasZeroObject C], deliberately omitting Preadditive C
  • pin status: PIN-CONFIRMED Mathlib 520045ab14e26149ee970e2e617ca04b09bde5d6
  • source note: .lake/sf8_hypothesis_nopreadditive.lean, compiled with LEAN_NUM_THREADS=1 /home/chris-dare/.elan/bin/lake env lean, fails at HomologicalComplex₂.total (C := C) with the exact diagnostic failed to synthesize instance of type class Preadditive C. The existing proof also explicitly uses Preadditive.comp_add, Preadditive.add_comp, and the integer-unit signs in the total differential. Single support does not remove the preadditive parameter from this canonical total API. A different construction would be outside this proof-based scout's scope.
  • state: FALSIFIED compiler witness: zero morphisms alone do not support Mathlib's existing total-complex construction.

2026-09-27 — natural extension of a single complex for the free-Yoneda augmentation target (planned)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: altitude-scout
  • proposed ancestor: HomologicalComplex.singleCompExtendIso, as a direct extension of Mathlib's Algebra/Homology/Embedding/Extend.lean API
  • weaker hypotheses: any category with zero morphisms and a zero object, arbitrary complex shapes with decidable index equality, an embedding e, degrees i and i', and e.f i = i'; no schemes, sheaves, abelian structure, coproducts, preadditivity, or fixed natural/integer grading are needed for the natural isomorphism before totalization
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/Algebra/Homology/Embedding/Extend.lean:293
  • source note: At Mathlib revision 520045ab14e26149ee970e2e617ca04b09bde5d6, the cited HomologicalComplex.extendSingleIso supplies the objectwise canonical comparison. The proposed lift packages those same components as single C c i ⋙ e.extendFunctor C ≅ single C c' i'; it is not a claim that this natural wrapper is already declared in the pin. A unique scratch probe, .lake/sf8_554_target_altitude_single_extend_20260927.lean, compiled this naturalization, its ChainComplex.single₀ / embeddingDownNat specialization, and its cochain lift using pinned CategoryTheory.NatIso.mapHomologicalComplex (Mathlib/Algebra/Homology/Additive.lean:196). The only printed axioms were propext, Classical.choice, and Quot.sound. This finding is distinct from the objectwise ancestor noted in pending PR #1600: it checks arbitrary-shape naturality and the reusable outer-complex transport. The mapped single presentation still needs an explicit comparison with the flipped singleZeroBicomplex presentation before composing with that PR's singleZeroFlipTotalIso; an Iso.refl probe did not elaborate, which is evidence only against that attempted definitional identification. Whiskering with pinned HomologicalComplex₂.totalFunctor (Mathlib/Algebra/Homology/TotalComplex.lean:467) requires that functor's ambient total-existence assumptions. No total augmentation quasi-isomorphism or K-flatness claim follows from these comparisons.
  • state: UNVERIFIED

2026-09-27 — omit explicit decidable index equality from natural single extension (planned)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: HomologicalComplex.singleCompExtendIso, extending Algebra/Homology/Embedding/Extend.lean
  • weaker hypotheses: [Category C] [HasZeroObject C] [HasZeroMorphisms C], arbitrary index types and shapes, e : c.Embedding c', i, i', and h : e.f i = i'; neither [DecidableEq ι] nor [DecidableEq ι'] is an explicit parameter
  • level: L (proof-witness verified)
  • pin status: PIN-CONFIRMED Mathlib 520045ab14e26149ee970e2e617ca04b09bde5d6
  • source note: .lake/sf8_554_target_hypothesis_verified_20260927.lean compiled with Lean v4.32.1, exit 0. Its SF8554TargetHypothesisScout20260927.singleCompExtendIso uses the altitude scout's complete naturality proof unchanged under open scoped Classical. #print confirms both decidability parameters are absent; #print axioms reports only propext, Classical.choice, and Quot.sound. This supplies classical decisions inside the signature; it does not make single independent of its chosen decisions. Preserving ambient instances in the generic public wrapper remains useful for definitional interoperability.
  • state: UNVERIFIED

2026-09-27 — construct the zero morphisms needed by natural single extension (planned)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: HomologicalComplex.singleCompExtendIso, with the existing CategoryTheory.Limits.HasZeroObject.zeroMorphismsOfZeroObject
  • weaker hypotheses: [Category C] [HasZeroObject C], arbitrary index types and shapes, embedding, degrees and e.f i = i'; zero morphisms are chosen locally and index equality is supplied classically
  • level: L (proof-witness verified)
  • pin status: PIN-CONFIRMED Mathlib/CategoryTheory/Limits/Shapes/ZeroMorphisms.lean:258
  • source note: In .lake/sf8_554_target_hypothesis_verified_20260927.lean, SF8554TargetHypothesisDerivedZeros20260927.singleCompExtendIso compiles using letI : HasZeroMorphisms C := HasZeroObject.zeroMorphismsOfZeroObject (C := C) in both its result type and proof. SF8554TargetHypothesisNativeDerivedZeros20260927.mappedSingleZeroExtendDownNatIso separately verifies the same deletion for the native ChainComplex.single₀ / embeddingDownNat specialization and its outer cochain lift. Both print only the standard three axioms. A further compiled example proves agreement with any ambient HasZeroMorphisms C by Subsingleton.elim. This agreement is propositional: Mathlib's comment immediately preceding the constructor explicitly warns that the constructed instance need not be definitionally equal to an additive category's existing instance. This is a verified logical redundancy, not a recommendation to replace the ambient-instance API.
  • state: UNVERIFIED

2026-09-27 — eliminate the output-degree equality by fixing its value (planned)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: HomologicalComplex.singleCompExtendIso
  • weaker hypotheses: category, zero object, zero morphisms, arbitrary shapes, embedding e and source degree i; the target is single C c' (e.f i)
  • level: L (proof-witness verified)
  • pin status: PIN-CONFIRMED
  • source note: singleCompExtendAtImageIso in .lake/sf8_554_target_hypothesis_verified_20260927.lean compiles by applying the generic comparison with i' := e.f i and rfl. This removes a redundant parameter and its equality proof together; it does not justify deleting the equality while keeping an independently selected i'.
  • state: UNVERIFIED

2026-09-27 — natural single extension lifts along any outer complex shape (planned)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: HomologicalComplex.singleCompExtendIso, consumed by pinned CategoryTheory.NatIso.mapHomologicalComplex
  • weaker hypotheses: the category, zero object, zero morphisms and embedding comparison above, plus an arbitrary κ : Type* and d : ComplexShape κ; no outer [DecidableEq κ], fixed ℤ, preadditivity, abelian structure, boundedness, or total-existence assumption
  • level: L (proof-witness verified)
  • pin status: PIN-CONFIRMED Mathlib/Algebra/Homology/Additive.lean:196
  • source note: SF8554TargetHypothesisScout20260927.mappedSingleCompExtendIso in .lake/sf8_554_target_hypothesis_verified_20260927.lean compiles directly as NatIso.mapHomologicalComplex (singleCompExtendIso e i i' h) d. The two required zero-preservation instances are inferred from the single and extension functors. Its printed signature contains none of the omitted assumptions and its axioms are the standard three. The scratch file imports only Mathlib.Algebra.Homology.Embedding.Extend; the altitude witness's TotalComplex import is unnecessary for these comparisons. This is an outer complex comparison and asserts no totalization quasi-isomorphism.
  • state: UNVERIFIED

2026-09-27 — native zero-degree extension needs no caller index data (planned)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: natural single extension specialized to ChainComplex.single₀ and ComplexShape.embeddingDownNat
  • weaker hypotheses: only {C : Type u} [Category.{v} C] [HasZeroObject C] [HasZeroMorphisms C]; no index types, decidability instances, embedding, degrees, or equality supplied by the caller
  • level: L (proof-witness verified)
  • pin status: PIN-CONFIRMED
  • source note: In .lake/sf8_554_target_hypothesis_verified_20260927.lean, the namespace SF8554TargetHypothesisNative20260927 rechecks the altitude witness with the ambient decidability parameters preserved in its generic helper. #print singleZeroExtendDownNatIso and #print mappedSingleZeroExtendDownNatIso show exactly the four parameters above. The equality is discharged by simp, the two concrete index instances are synthesized, and the lift uses NatIso.mapHomologicalComplex. Both the specialization and its lift compile with no additional assumptions.
  • state: UNVERIFIED

2026-09-27 — weaken the category of natural single extension to CategoryStruct (failed)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: the same pinned single/extension API with only [CategoryStruct C]
  • weaker hypotheses: remove category laws while retaining objects, morphisms, identities and composition
  • source note: .lake/sf8_554_target_hypothesis_failures_20260927.lean attempts both HasZeroObject C and HomologicalComplex C (ComplexShape.up ℤ) under [CategoryStruct C]. Lean reports failed to synthesize instance of type class Category ... C at lines 10 and 11. This falsifies literal deletion in the existing API; it is not a counterexample about a different encoding. Functors, complex categories, and the naturality proof all use the category structure.
  • state: FALSIFIED (the proposed signature does not elaborate)

2026-09-27 — omit zero morphisms without installing their construction (failed)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: native single/extension API with automatic zero-morphism synthesis from a zero object
  • weaker hypotheses: [Category C] [HasZeroObject C], with no local choice of HasZeroMorphisms C
  • source note: .lake/sf8_554_target_hypothesis_failures_20260927.lean:17 and line 18 check ChainComplex.single₀ C and ComplexShape.embeddingDownNat.extendFunctor C. Both report failed to synthesize instance of type class HasZeroMorphisms C. The successful explicit local construction is recorded above; the failure is about automatic instance search, not mathematical existence.
  • state: FALSIFIED (the unmodified signature does not elaborate)

2026-09-27 — omit the zero object or replace it by a distinguished object (failed)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: the same pinned single/extension API without [HasZeroObject C]
  • weaker hypotheses: [Category C] [HasZeroMorphisms C], first alone and then additionally [Zero C]
  • source note: .lake/sf8_554_target_hypothesis_failures_20260927.lean checks generic single and extendFunctor, the native single₀ and embeddingDownNat.extendFunctor, and single/single₀ again with [Zero C]. All report failed to synthesize instance of type class HasZeroObject C. The proof uses zero-object uniqueness off the support, and extension uses the zero object outside the embedding's image. A chosen object with a supplied IsZero proof would reintroduce equivalent evidence; a bare distinguished object does not. This records the existing API's requirement, not an impossibility result for every special shape or alternative proof.
  • state: FALSIFIED (the proposed signatures do not elaborate)

2026-09-27 — omit decidable indices without supplying classical decisions (failed)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: the same pinned HomologicalComplex.single API
  • weaker hypotheses: remove [DecidableEq ι] and [DecidableEq ι'] independently, without enabling a classical instance
  • source note: The NoSourceDecidableEq and NoTargetDecidableEq sections of .lake/sf8_554_target_hypothesis_failures_20260927.lean each check the corresponding single. Lean reports respectively failed to synthesize instance of type class DecidableEq ι and failed to synthesize instance of type class DecidableEq ι'. The successful classical wrapper above resolves exactly these failures; this row does not claim that decidability is an essential mathematical hypothesis of a noncomputable comparison.
  • state: FALSIFIED (literal binder deletion without replacement)

2026-09-27 — independently choose a different degree for extended single complexes (failed)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: HomologicalComplex.extendSingleIso or its natural wrapper without e.f i = i'
  • weaker hypotheses: retain independent i : ι and i' : ι', removing their equality constraint
  • source note: The direct reuse attempt in .lake/sf8_554_target_hypothesis_failures_20260927.lean fails with the remaining goal ⊢ e.f i = i'. More decisively, SF8554TargetHypothesisScout20260927.noExtendSingleIsoOffImage in the verified probe proves that e.f i ≠ i' and ¬ IsZero X imply ¬ Nonempty ((((single C c i).obj X).extend e) ≅ (single C c' i').obj X). Evaluation at e.f i would identify X with the target's zero object. The compiler accepts that obstruction with only the standard three axioms. No different proof can supply the unconstrained comparison in this case; fixing i' := e.f i, or imposing a zero-object degeneracy, changes the claim.
  • state: FALSIFIED (proved obstruction for unequal degrees and nonzero input)

2026-09-27 — treat a classical single wrapper as definitionally the native single₀ (failed)

  • chunk: sf8-554-target-comparison
  • reviewing commit: f18bf326077601e1502702b6e6dc154a0a34f057
  • found by: hypothesis-elimination-scout
  • proposed ancestor: direct native specialization of a generic wrapper whose signature fixes classical decidability
  • weaker hypotheses: omit the generic decidability parameters and expect the resulting concrete specialization to identify definitionally with ChainComplex.single₀
  • source note: The retained .lake/sf8_554_target_hypothesis_classical_native_attempt_20260927.lean and its log show the attempted direct specialization failing at line 40: the source uses @HomologicalComplex.single ... ℕ (fun a b => Classical.propDecidable (a = b)) ..., whereas the expected single₀ uses instDecidableEqNat. The attempted simpa only transport did not close this mismatch. This falsifies that attempted definitional reuse, not the existence of a comparison; a different proof could transport between the decisions using their propositional equality. The verified native specialization instead retains the generic helper's ambient decidability parameters and needs no explicit decisions from callers.
  • state: FALSIFIED (attempted definitional identification fails)

2026-09-27 — one-step compatibility suffices for triangulatedness transfer (planned)

  • chunk: 1369-shiftcompat-unification
  • reviewing commit: 00196ff8fa17f1904a68e0bdccd633ae2fe86d64
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Functor.isTriangulated_of_iso and Functor.mapTriangleIso
  • weaker hypotheses: keep the chosen [F₁.CommShift ℤ] and [F₂.CommShift ℤ], but require only NatTrans.CommShiftCore e.hom (1 : ℤ) rather than the full [NatTrans.CommShift e.hom ℤ]; retain [F₁.IsTriangulated] and the usual pretriangulated-category context
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Shift/CommShift.lean:296-318 and .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Functor.lean:258
  • source note: /tmp/issue1369_single_shift_probe.lean copies the triangle-isomorphism proof using only the core compatibility at +1, then transfers distinguishedness by isomorphic_distinguished. The theorem isTriangulated_of_iso_of_core compiled with exit code 0 under ~/.elan/bin/lake env lean /tmp/issue1369_single_shift_probe.lean. The selected target CommShift remains an independent input; this does not manufacture it via Functor.CommShift.ofIso.
  • state: L (proof-witness verified)

2026-09-27 — derive inverse shift compatibility for equivalence transfer (planned)

  • chunk: 1369-shiftcompat-unification
  • reviewing commit: 00196ff8fa17f1904a68e0bdccd633ae2fe86d64
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Equivalence.IsTriangulated, using Equivalence.commShiftInverse and Equivalence.commShift_of_functor
  • weaker hypotheses: an equivalence E, its selected forward [E.functor.CommShift ℤ], and [E.functor.IsTriangulated]; no independently supplied [E.inverse.CommShift ℤ] or E.CommShift ℤ is needed. The inverse shift and adjunction compatibility are derived from the chosen forward shift.
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Shift/Adjunction.lean:617-624 and .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Adjunction.lean:197-221
  • source note: equivalenceIsTriangulated in /tmp/issue1369_hypothesis_probe.lean builds a selected forward shift from the packaged comparison, derives the inverse and equivalence compatibility with Mathlib, and applies Equivalence.IsTriangulated.mk'. The generic wrapper compiled with exit code 0 under ~/.elan/bin/lake env lean /tmp/issue1369_hypothesis_probe.lean.
  • state: L (proof-witness verified)

2026-09-27 — drop all shift compatibility evidence for exactness transfer (attempted)

  • chunk: 1369-shiftcompat-unification
  • reviewing commit: 00196ff8fa17f1904a68e0bdccd633ae2fe86d64
  • found by: hypothesis-elimination-scout
  • proposed ancestor: CategoryTheory.Functor.isTriangulated_of_iso
  • weaker hypotheses: retain both selected functor CommShift ℤ structures and source triangulatedness, but omit compatibility of e.hom with the selected shifts
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Triangulated/Functor.lean:258
  • source note: /tmp/issue1369_drop_comparison_compat.lean attempts Functor.isTriangulated_of_iso α after deleting only [NatTrans.CommShift α.hom ℤ]. Lean reports failed to synthesize instance of type class NatTrans.CommShift α.hom ℤ. The one-step witness above shows that compatibility specifically at +1 is sufficient for a weaker custom wrapper, but some compatibility evidence cannot be dropped from the transported-triangulation proof.
  • state: FALSIFIED (literal compatibility-evidence deletion fails)

2026-09-30 — discretePseudofunctorMap (issue #1635)

  • chunk: 1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: altitude-scout
  • proposed ancestor: a natural-transformation adapter for Functor.toPseudofunctor', owned by CategoryTheory/Bicategory/Functor/LocallyDiscrete.lean; a discrete map specializes it using Functor.whiskerRight φ typeToCat
  • weaker hypotheses: {I B : Type*} [Category* I] [Bicategory B] [Strict B] and {F G : I ⥤ B} (φ : F ⟶ G)`; no presheaf variance, discrete fibers, sheaf condition, Grothendieck topology, groupoid condition, or scheme
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Bicategory/Functor/LocallyDiscrete.lean:146
  • source note: At the pin Functor.toPseudofunctor' already promotes an ordinary functor into any strict bicategory. The precise NatTrans adapter is not declared there; it is a new adapter on that existing root, with app T := φ.app T.as and naturality f := eqToIso (φ.naturality f.as). The three coherence proofs reduce using LocallyDiscrete.eq_of_hom, Strict.leftUnitor_eqToIso, Strict.rightUnitor_eqToIso, and Strict.associator_eqToIso. Its proposed implementation is typechecked in scratch/altitude-probe-1635.lean:29. Functor.whiskerRight is at .lake/packages/mathlib/Mathlib/CategoryTheory/Whiskering.lean:61, with component typeToCat.map (φ.app T). typeToCat is at .lake/packages/mathlib/Mathlib/CategoryTheory/Category/Cat.lean:408, and its map is literally Discrete.functor (Discrete.mk ∘ f) (:410). Scratch examples at :42 and :47 confirm by rfl both Discrete.mk (φ.app T x) and Discrete.mk (g ≫ f) for yoneda.map f. This is a possible foundation for later consolidation, not a request to introduce another pseudofunctor or stack carrier.
  • state: UNVERIFIED

2026-09-30 — StackInGroupoids.representable (issue #1635)

  • chunk: 1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: altitude-scout
  • proposed ancestor: the existing stackInGroupoidsOfSheaf constructor, specialized with GrothendieckTopology.Subcanonical.isSheaf_of_isRepresentable; the upstream represented-sheaf API is CategoryTheory/Sites/Canonical.lean
  • weaker hypotheses: arbitrary C, arbitrary subcanonical J, and any {P : Cᵒᵖ ⥤ Type w} [P.IsRepresentable]; the underlying discrete-stack constructor already works for any P supplied with a J-sheaf proof
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/CategoryTheory/Sites/Canonical.lean:151
  • source note: The pinned theorem is universe-general in P and is stronger than a theorem restricted to literal yoneda.obj X. The same file packages the entire sheaf-valued Yoneda functor as GrothendieckTopology.yoneda (:168) and its raised-universe variant uliftYoneda (:176). For relative sites the existing instance .lake/packages/mathlib/Mathlib/CategoryTheory/Sites/SubcanonicalOver.lean:25 proves (J.over X).Subcanonical without geometry. The repository root DerivedAlgGeo/CategoryTheory/Sites/Descent/StackInGroupoids/Discrete.lean:131 already supplies the stack constructor and :28 supplies its pseudofunctor; a new representable carrier, independently assumed descent, or a relative representable carrier is unnecessary. The issue's yoneda.obj X specialization can keep the existing StackInGroupoids owner.
  • state: UNVERIFIED

2026-09-30 — fppfTopology_eq_propQCTopology and etaleTopology_eq_propQCTopology (issue #1635)

  • chunk: 1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: altitude-scout
  • proposed ancestor: AlgebraicGeometry/Sites/QuasiCompact.lean, with precoverage P = propQCPrecoverage P and grothendieckTopology P = propQCTopology P for every open-map property P
  • weaker hypotheses: {P : MorphismProperty Scheme.{u}} (hP : P ≤ fun _ _ f ↦ IsOpenMap f.base) only; no multiplicativity, base-change stability, flatness, local finite presentation, or étaleness is needed for the generic equality
  • pin status: PIN-CONFIRMED .lake/packages/mathlib/Mathlib/AlgebraicGeometry/Sites/QuasiCompact.lean:87
  • source note: The pinned source already proves precoverage P ≤ qcPrecoverage from exactly hP. propQCPrecoverage P is qcPrecoverage ⊓ precoverage P (:112), so le_antisymm (le_inf (precoverage_le_qcPrecoverage_of_isOpenMap hP) le_rfl) inf_le_right proves the equality. congrArg Precoverage.toGrothendieck proves the corresponding topology equality. Both typechecked as scratch examples (:10 and :18) without extra hypotheses. Exact equality names are absent at the pin; this is a pin-supported generic derivation, not a claim that the requested fppf/étale equalities are already named. The fppf specialization repeats Mathlib's own use of the open-map inequality in .lake/packages/mathlib/Mathlib/AlgebraicGeometry/Sites/Fpqc.lean:66. The generic equality is an extension of scheme-site machinery, not neutral category theory: quasi-compact covers and underlying open maps keep its owner under AlgebraicGeometry/Sites/.
  • state: UNVERIFIED

2026-09-30 — individual representables need only objectwise sheafhood

  • chunk: stk2-1-1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: hypothesis-elimination-scout
  • proposed ancestor: existing stackInGroupoidsOfSheaf and planned StackInGroupoids.discreteMap, in CategoryTheory/Sites/Descent/StackInGroupoids/
  • weaker hypotheses: arbitrary category C; Presieve.IsSheaf J (yoneda.obj X) for one constructor, and the two objectwise sheaf proofs for a map; omit [J.Subcanonical] from this proof-witness formulation.
  • pin status: PIN-CONFIRMED
  • source note: private scratch file compiled with the exact bounded command above. PUnit at the top topology verifies that this is strictly weaker than requiring all representables to be sheaves. No new public root is required.
  • state: UNVERIFIED
  • proof classification: L (proof-witness verified; compiled candidate, not merged)

2026-09-30 — restriction needs no explicit lower subcanonical class

  • chunk: stk2-1-1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: hypothesis-elimination-scout
  • proposed ancestor: planned StackInGroupoids.representable_ofLE, with the arbitrary-sheaf restriction equality in Discrete.lean
  • weaker hypotheses: retain [J₂.Subcanonical], h : J₁ ≤ J₂, and X : C; omit [J₁.Subcanonical] and derive it with Mathlib's GrothendieckTopology.Subcanonical.of_le h. More generally retain only a sheaf proof for arbitrary P at J₂, and transport it by Presieve.isSheaf_of_le _ h hP.
  • pin status: PIN-CONFIRMED
  • source note: both resulting equalities compile by rfl in the private scratch file with the exact bounded command above.
  • state: UNVERIFIED
  • proof classification: L (proof-witness verified; compiled candidate, not merged)

2026-09-30 — QC comparison consumes ordinary openness only

  • chunk: stk2-1-1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: hypothesis-elimination-scout
  • proposed ancestor: AlgebraicGeometry.Scheme.precoverage_eq_propQCPrecoverage_of_isOpenMap and induced topology equality, beside Mathlib's Sites/QuasiCompact.lean
  • weaker hypotheses: any P : MorphismProperty Scheme.{u} and P ≤ fun _ _ f ↦ IsOpenMap f.base; omit flatness, local finite presentation, étaleness, stability, multiplicativity, locality, or finiteness assumptions.
  • pin status: PIN-CONFIRMED
  • source note: le_antisymm (le_inf (precoverage_le_qcPrecoverage_of_isOpenMap hP) le_rfl) inf_le_right compiles, and congrArg Precoverage.toGrothendieck gives the topology equality. The parent plans these generic results in DerivedAlgGeo/AlgebraicGeometry/Sites/QuasiCompact.lean.
  • state: UNVERIFIED
  • proof classification: L (proof-witness verified; compiled candidate, not merged)

2026-09-30 — delete subcanonicity without an objectwise sheaf replacement

  • chunk: stk2-1-1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: hypothesis-elimination-scout
  • attempted weaker hypotheses: arbitrary category C, topology J, object X, with no [J.Subcanonical] and no IsSheaf J (yoneda.obj X) proof.
  • pin status: PIN-CONFIRMED
  • source note: compiled bool_yoneda_not_sheaf_top and bool_discrete_not_stack_top for C = Type, J = ⊤, X = Bool. The empty cover of PUnit makes the false and true maps have indistinguishable descent data. Full faithfulness would produce a morphism between their discrete fibre objects, forcing false = true.
  • state: FALSIFIED (compiled counterexample)

2026-09-30 — a morphism and one sheaf endpoint imply the other endpoint is a sheaf

  • chunk: stk2-1-1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: hypothesis-elimination-scout
  • attempted weaker hypotheses: only the source sheaf proof or only the target sheaf proof in discreteMap, retaining the natural transformation.
  • pin status: PIN-CONFIRMED
  • source note: compiled examples give arrows PUnit ⟶ Bool and Bool ⟶ PUnit, while yoneda.obj PUnit is a sheaf at ⊤ and yoneda.obj Bool is not. Thus neither endpoint's proof can generally be reconstructed from the other.
  • state: FALSIFIED (compiled counterexample)

2026-09-30 — reverse topology restriction

  • chunk: stk2-1-1635-representable-stacks
  • reviewing commit: a30793798c7faecef2a83895df639b53934eb3f8
  • found by: hypothesis-elimination-scout
  • attempted weaker hypotheses: use descent for a coarser topology to obtain descent for a finer topology without further proof.
  • pin status: PIN-CONFIRMED
  • source note: compiled conjunction proves yoneda.obj Bool is a sheaf for ⊥ by Presieve.isSheaf_bot and is not a sheaf for ⊤ by the empty-cover counterexample. The associated discrete pseudofunctor fails the ⊤ stack condition by the compiled direct descent proof.
  • state: FALSIFIED (compiled counterexample)

2026-09-30 — dropping the open-map premise from QC comparison (issue #1635)

  • chunk: 1635-representable-stacks
  • reviewing commit: 068871899ba704007712939699dbbe7dc5c00d8f
  • found by: mathematics-adversary
  • proposed ancestor: AlgebraicGeometry/Sites/QuasiCompact.lean
  • weaker hypotheses: arbitrary scheme morphism property, without the open-map premise
  • source note: The review's compiled point-family counterexample takes an infinite affine scheme and its residue-field points. The family is jointly surjective, hence a cover for the unrestricted property, but cannot satisfy the quasi-compact covering condition: finitely many source point opens cover only finitely many target points. The empty-cover and point-family proof witnesses are kept in the private transcript supplement for #1635.
  • state: FALSIFIED (infinite affine residue-field point family)

2026-09-30 — Lattice.pairCharge over a nonsymmetric form (issue #1230, withdrawn)

  • chunk: none (owner decision on #1230)
  • reviewing commit: 3f6453a3
  • found by: second-consumer research agent, spot-checked by hand
  • proposed ancestor: a LinearAlgebra/BilinearForm/ root pairCharge b x y v with b not assumed symmetric, above PeriodDomain.centralCharge (QuadraticForm/ComplexPairing.lean:79) and Mukai.expCharge (CentralCharge/Mukai/Charge.lean:50)
  • weaker hypotheses: a real bilinear form with no symmetry. Neither leaf's API uses symmetry except expCharge_apply, which needs it through polar_realForm.
  • pin status: not applicable; no Mathlib declaration is involved
  • source note: expCharge is centralCharge (realForm b) by definition, so the two were already one root. For nonsymmetric b, polar (realForm b) is the symmetrization of realPairing b, so expCharge = pairCharge (realPairing b) fails. The surviving Lift is add/smul/zero/kernel lemmas of about one line each with no consumer. The one hand-rolled copy of the form is cPair (Exponential/Divisorial.lean:101), which is symmetric (follow-up #1810). Mukai.Graded.pairing has no consumer: n = 2 is realPairing (LinearMap.mul ℝ ℝ) and the odd form is a recorded negative result.
  • state: FALSIFIED (as motivated: nonsymmetry admits no leaf, and expCharge ≠ pairCharge (realPairing b) for nonsymmetric b)

Exported from docs/architecture/generalization-backlog.md at the commit that removed it from the tree.

Home

Architecture

CI and runners

Decision records

Roadmaps

Clone this wiki locally