Documentation

Challenge.DivisorLineBundle

Challenge: the divisor-line-bundle dictionary on a smooth proper curve #

This contract captures the remaining A3 existence and equivalence deliverable: a scheme-level dictionary relating chosen invertible-sheaf representatives, the Picard group, and Weil divisor classes. The affine tilde restriction isomorphism and global tilde invertibility are now proved in checked code. A checked equivalence reduces existence of the dictionary to the full Picard comparison together with a divisor-class/Picard equivalence; chosen line bundles and their compatibilities are constructed from those data. The contract does not prescribe a canonical affine-chart normalization. On affine schemes, the localized tensor maps have now been assembled into an objectwise tilde tensor-product isomorphism. This gives a canonical injective module-Picard to scheme-Picard map. For affine Dedekind domains, the divisor map has exactly the principal divisors as kernel, divisor classes are equivalent to its range, and the chosen tilde line bundles are tensor-additive. The chosen Picard representative is now identified with the explicit inverse fractional ideal attached to the divisor. Relative to these fixed comparisons, multiplication of explicit inverse ideals now determines named tensor/addition equivalences for the chosen affine modules, with a checked pure-tensor formula in the fraction field. Tilde transports this to AffineDivisorLocalization.ExplicitIdeal.lineBundleAddIso, and the named chartwise consumer CurveDivisorDescent.localLineBundleAddIso identifies O(D + E) with O(D) ⊗ O(E) on every compatible affine chart. At the raw common-extension level, CommonExtension.extendedInverseIdealTensorAddEquiv and its _tmul theorem identify tensor addition with multiplication in the common fraction field. The _baseChange_tmul and _baseChange theorems prove that multiplication-direction equivalence commutes with a further flat epimorphic scalar extension through AlgebraTensorModule.distribBaseChange. The fixed chosen-module comparison is now exposed by CommonExtension.lineBundleModuleBaseChangeEquivExtendedInverseIdeal and its _tmul theorem, and lineBundleModuleTensorAddEquiv_baseChange proves the corresponding tensor-to-sum square. The genuine curve consumer CurveDivisorDescent.localLineBundleModuleTensorAddEquiv_baseChangeOnCommonAffineOpen derives the restriction algebra, flat-epimorphism, fraction-field, and scalar-tower data for one canonical affine restriction leg. Thus the fixed chosen-module tensor-to-sum map is natural on that raw module-level leg. The generic Scheme.Modules.tildePullbackIsoExtendScalars comparison identifies affine pullback of a tilde module with tilde of its extension of scalars. At the common-extension level, extendedInverseIdealAddIso lifts raw inverse-ideal addition through tilde, while lineBundlePullbackIsoExtendedInverseIdealViaExtendScalars and lineBundleTensorPullbackIsoExtendedInverseIdealsViaExtendScalars give the direct comparisons from the pulled-back fixed bundle and pulled-back fixed tensor to the corresponding raw extended-ideal tilde sheaves. The theorem lineBundleAddIso_pullback_viaExtendScalars proves the resulting deterministic sum-to-tensor square, and CurveDivisorDescent.localLineBundleAddIso_pullbackViaExtendScalarsOnCommonAffineOpen specializes it to one genuine affine restriction leg. CommonExtension.restrictionIsoExtendedInverseIdealOfIsOpenImmersion_eq_viaExtendScalars identifies the legacy single-factor restriction mate with the restriction-to-pullback comparison followed by ViaExtendScalars, and CurveDivisorDescent.localRestrictionIsoExtendedInverseIdeal_eq_viaExtendScalarsOnCommonAffineOpen is its genuine affine-open consumer. The project-original Scheme.Modules.tildePullbackGenerator API exposes the top-section pullback generator, identifies it with the transported pullback-unit section, and evaluates the affine tilde/base-change comparison on it. Naturality of the affine tilde tensor comparison is checked in both hom and inverse orientations, and its top-section evaluator computes pure tensor sections. Together with ModularCurves.pullback_δ_unit_tensorSection, these formulas prove the generic AffineTilde.tildePullbackIsoExtendScalars_tensor_hom base-change square. The comparison CommonExtension.lineBundleTensorPullbackIsoExtendedInverseIdealsViaCanonicalTensorator factors through that canonical cotensorator and the two factorwise scalar-extension comparisons. The direct comparison has the same hom, hence is the same isomorphism, and the genuine curve specialization proves this equality on one fixed common-affine restriction leg. Transporting the inverse pullback tensorator through the legacy restriction-to-pullback comparison gives CommonExtension.lineBundleTensorRestrictionIsoViaPullback. The theorem CommonExtension.lineBundleAddIso_restrict_viaPullback proves the deterministic tensor/addition square for the legacy restriction comparisons, and CurveDivisorDescent.localLineBundleAddIso_restrictViaPullbackOnCommonAffineOpen is its genuine curve specialization on that same fixed leg. Equality-induced multiplication of raw extended inverse-ideal tilde sheaves is now natural across two chart rings: Boundary.overlapInverseIdealExtensionEq_add derives the comparison for a divisor sum, CommonExtension.extendedInverseIdealAddIso_naturality proves the resulting sum-to-tensor square, and CurveDivisorDescent.localExtendedInverseIdealAddIso_naturalityViaPairwiseIntersection is the proper-smooth curve consumer whose factorwise equalities come through the pairwise intersection. The chosen coordinate-divisor restriction comparisons now satisfy the corresponding cross-chart tensor/addition square: CommonExtension.lineBundleAddIso_restrict_viaPullback_naturality proves the generic square, and CurveDivisorDescent.localLineBundleAddIso_restrictViaPullback_naturalityViaPairwiseIntersection is its proper-smooth pairwise-intersection consumer. The named transports CurveDivisorDescent.localLineBundleCoordinateIso and CurveDivisorDescent.localLineBundleAddSourceCoordinateIso expose the proof-defined global-divisor bundles in those coordinate presentations, and localLineBundleAddIso_restrictViaPullback_naturalityViaPairwiseIntersection_global aligns the square with the actual proof-defined local addition maps and pairwise-intersection transitions. The definition CurveDivisorDescent.localLineBundleTensorChosenOverlapIsoOfPairwiseModelOnProperSmoothCurve transports the whole pairwise-intersection tensor model to Mathlib's chosen pullback, and CurveDivisorDescent.localLineBundleAddIso_chosenOverlap_commViaTransportedTensorModel proves that the proof-defined local addition maps commute there with this whole transported tensor transition and the raw sum transition. Conditional on an already supplied divisor cocycle CDE for D + E, CurveDivisorDescent.localLineBundleTransportedTensorModelDescentData constructs the full changeObjects target on the chartwise tensor objects, and CurveDivisorDescent.localLineBundleAddDescentDataIsoViaTransportedTensorModel gives its full DescentData isomorphism with the source sum cocycle. The theorem CurveDivisorDescent.localLineBundleTransportedTensorModelEffective_of_sumCocycleEffective transfers an already supplied effectivity witness, while CurveDivisorDescent.localLineBundleTransportedTensorModelDescentData_hom_on_chosenOverlap identifies the chosen transition with the whole transported tensor model under exact per-overlap alignment of CDE's Homs with the raw sum family. Separately, strong monoidality of pullback composition is now checked first for presheaves of modules and then for scheme modules. LineBundleDescent.pullbackOverlapIsoOfModel_tensor identifies transport of a tensor transition through the exact-pseudofunctor comparison with the cotensorator-conjugated tensor of the two factor transports. The curve-level theorem localLineBundleTensorChosenOverlapIsoOfPairwiseModel_eq_tensorRawOverlapIsoFamily specializes this to the two raw divisor transitions, and CurveDivisorDescent.localLineBundleAddIso_chosenOverlap_commViaFactorwiseRawTensor proves the resulting local-add Hom square. Conditional on supplied divisor cocycles for D, E, and D + E whose chosen-overlap Homs equal the raw geometric transitions, localLineBundleAddDescentDataIsoViaTransportedTensorModel_hom_commViaFactorwiseCocycleOverlaps proves that the local components of the existing full addition DescentData isomorphism satisfy this factorwise chosen-overlap square. This is a chosen-overlap presentation of the existing full isomorphism, not a standalone factorwise equality for the target Hom, an arbitrary-refinement factorwise formula, or an independently assembled tensor descent datum or morphism. The conditional consumers still do not construct the required raw arbitrary-divisor cocycles or prove unconditional effectivity. Constructing an independent factorwise tensor descent datum and its descent-data morphism, and constructing a CurveDivisorDescent.DescendedTensorAdditive witness, remain open. On a same-chart principal open, equality of divisor coefficients proves equality of the localized inverse ideals, equivalently equality after common-field extension, and constructs an isomorphism of the chosen restricted tilde bundles. Meanwhile, AffineDivisorLocalization.Boundary.OverlapInverseIdealExtensionEq isolates cross-chart ideal-extension equality and AffineDivisorLocalization.CommonExtension.RestrictionIdentifiesExtendedInverseIdeal isolates the restriction/base-change identification. Affine tilde base change and flat epimorphic tensor/span comparison now prove the latter for every affine open immersion, without a classical localization presentation. Hence the packaged cross-chart extension condition, which also records the scalar towers, constructs the actual chosen restriction isomorphism. For two distinct chart rings, extension equality is proved for the principal divisors defined by the same rational function, giving the actual restriction isomorphism on every common affine overlap satisfying the standing Dedekind and common-fraction-field assumptions. For arbitrary divisors, open-immersion prime extension proves extended-ideal equality from contracted coefficient equality on any common Dedekind affine overlap; the classical-localization theorem is now only a specialization. For restrictions of one ambient divisor, equality of the two coordinate maps to the curve makes the contracted ambient points, coefficients, extended ideals, and actual chosen restrictions agree automatically. The resulting checked consumer is CurveDivisorDescent.localLineBundleRestrictionIsoOfCommonMap. More concretely, localLineBundleRestrictionIsoOnCommonAffineOpen derives every algebraic and geometric input canonically from a chosen common nonempty Dedekind affine subopen. Under separatedness, localLineBundleRestrictionIsoOnIntersection specializes this to the full intersection of two nonempty affine charts, conditional only on its coordinate ring being Dedekind. For the intended proper curve over a field, localLineBundleRestrictionIsoOnProperCurveIntersection derives the separatedness input from properness. With smooth relative dimension one, localLineBundleRestrictionIsoOnProperSmoothCurveIntersectionOfNormality further replaces the intersection Dedekind hypothesis by exactly the existing mapwise normality contract. Canonical Picard surjectivity is proved equivalent to the reverse tensor-unit/local-rank-one comparison, while the forward affine gap is isolated as the checked localization predicate AffineTilde.TildeReflectsInvertibility. The chart API identifies height-one primes with ambient codimension-one points and proves order compatibility automatically from a Dedekind-domain instance. Over a field, locally standard-smooth relative dimension one now proves both dimension at most one and integral closedness; AffineChart.dedekindOrderCompatibilityOfSmoothRelativeCurve is the unconditional consumer for the intended charts. The historical AffineChart.SmoothRelativeCurveRingConditions and AffineChart.SmoothRelativeCurveNormality predicates remain compatibility wrappers, not open field-base hypotheses. The local divisor-to-Picard map has exactly the principal divisors as kernel, descends injectively to chart divisor classes, and identifies them with its scheme-Picard range. AffineTilde.tildeReflectsInvertibility now discharges the former affine localization predicate and supplies the forward tensor-inverse comparison. The remaining A3 boundary begins with packaging the checked raw arbitrary-divisor overlap family as the required DivisorCocycle, selecting or supplying a universe-zero nonempty affine coordinate cover, and applying ordinary module effectivity to that concrete datum. The remaining boundary for descending the fixed chartwise addition maps now has the full transported tensor-object descent datum, its DescentData isomorphism with an already supplied sum cocycle, conditional transfer of effectivity, and its chosen transition under raw-Hom alignment. Strong monoidality of pullback composition, factorwise chosen-overlap tensor coherence, and the resulting local-add Hom square are checked. Under supplied cocycles for both factors and their sum with exact raw-Hom alignment, the local components of the existing full addition DescentData isomorphism also satisfy that chosen-overlap square. Constructing the required raw cocycles, an independent factorwise tensor descent datum and its descent-data morphism, an arbitrary-refinement factorwise formula, proving effectivity without a supplied sum-cocycle witness, and constructing a CurveDivisorDescent.DescendedTensorAdditive witness remain open. Principal coherence, object separation, rational normalization and principal detection, exactness, Picard surjectivity, and the stronger global comparison for arbitrary invertible sheaves also remain. The residue-degree product formula remains the separately registered A2 prerequisite in the roadmap; the affine and local A3 modules here do not import it. The companion module CurveAffineChart isolates the upstream-compatible sufficient input PicardIso.PullbackMonoidalData: given strong monoidality of sheaf pullback, checked code transports the divisor and class maps to Pic(U) while preserving the exact principal kernel and range equivalence. The option-free selective AINTLIB pullback-monoidal cone and its tensor-section evaluator now enter the A3 route through a separate companion module; the Challenge itself does not import that larger cone. For gluing, the separately importable modules AffineDivisorLocalization, CurveLineBundleDescent, CurveLineBundleLocality, CurveDivisorDescent, CurveDivisorPicardDescent, CurveLineBundleCompatibleFamilies, and CurveLineBundleCocycleForcesNormalization construct the actual affine O(D) on a coordinate cover and prove that every principal divisor gives a trivial line bundle on each chart. Same-chart coefficient agreement on a principal open constructs the actual restriction isomorphism between the corresponding chartwise divisor bundles. The raw arbitrary-divisor overlap family satisfies the pointwise all-index cocycle, including repeated indices. The generic normalization_of_iso_cocycle theorem derives its diagonal identity, and the derived LineBundleCocycle.normalization theorem is its concrete structure-level consumer. LineBundleCocycle stores exact-pseudofunctor-map overlap isomorphisms and the pointwise all-index cocycle; it has no independent normalization field. Packaging the raw divisor family as this structure and then as CurveDivisorDescent.DivisorCocycle remains open. Locality of Tau Ceti invertibility is proved for arbitrary scheme open covers, and ordinary module effectivity is proved for every OpenCover.{0}, but neither theorem has yet been applied to a concrete arbitrary-divisor cocycle system. A prestack instance for the scheme-module pseudofunctor now has a checked consumer producing fully faithful descent and hence object separation, but no such instance is currently available. For a divisor-indexed cocycle family, objectwise effectivity, tensor-additivity, and triviality of the zero-divisor bundle construct an actual divisor-to-Picard homomorphism: the bundle for -D is an explicit inverse to the bundle for D. Coherent triviality of the principal cocycles supplies zero triviality and descends this map to divisor classes under object separation. Exact principal kernel is equivalent, under coherent principal triviality and object separation, to ExplicitInverse.TrivialLineBundleDetectsPrincipal: only principal divisors have globally trivial descended line bundle. A global trivialization is proved to make the divisor principal on every chart. The witness-level condition TrivialDescendedLineBundleHasGlobalPrincipalWitness says that one rational function gives all chart restrictions; under agreement of the chosen global and scheme orders, checked code proves this condition equivalent to principal detection in both directions. It is therefore not an additional gluing hypothesis. Constructing compatible rational data from the explicit line-bundle cocycle remains absent. Exactness gives an equivalence with the actual Picard range, and adding surjectivity gives the full divisor-class/Picard equivalence without a global tensor-inverse comparison. Such a comparison is still required only by the stronger DivisorPicard.Dictionary, which records the class of every invertible sheaf. Chosen-pullback transport and the raw all-index overlap equation are checked. Conditional on an already supplied sum cocycle, the full transported tensor-object descent datum, its change-of-objects DescentData isomorphism, transfer of a supplied effectivity witness, and chosen-transition identification under raw-Hom alignment are also checked. Strong monoidality of pullback composition now identifies the whole transported transition with the cotensorator-conjugated tensor of the two separately transported raw factor transitions, and the corresponding local-add Hom square is checked. Under supplied factor and sum cocycles with exact raw-Hom alignment, the local components of the existing full addition DescentData isomorphism satisfy the same factorwise chosen-overlap square. No arbitrary-divisor DivisorCocycle or divisor cocycle system, concrete Type-0 effectivity application, global divisor-line-bundle family, unconditional effectivity, standalone factorwise target-Hom equality, arbitrary-refinement factorwise formula, independent factorwise tensor descent datum or descent-data morphism, CurveDivisorDescent.DescendedTensorAdditive witness, coherent principal triviality, prestack/object-separation, rationally normalized cocycle data, geometric principal detection, exact kernel, Picard surjectivity, or global tensor-inverse comparison is asserted. No additional localization Challenge is needed; the weighted product formula remains the separate A2 prerequisite. These remaining packaging, global compatibility, and equivalence results are open, so the contract earns no Mazur completion credit. The Challenge's published import list remains unchanged; solvers use these checked route modules explicitly.