Progress · independent proof diagram
Divisors, zero-cycles and one-cycles
The local obstruction vanishes on homologically trivial divisors, zero-cycles and one-cycles. Consequently the divisor–zero-cycle pairing has both full factors on every variety; in dimension three this also gives the complete middle pairing.
Conditional consequences · Source edition v0.5.1
Domain of the result
For , one also obtains .
Read the diagrams from top to bottom. Select a numbered connection for its exact hypotheses and deduction, or follow a node to its source. Analytic, model and comparison assumptions retain their stated scope.
How to read the logic
Read from top to bottom. Parallel premises meet at a numbered connection, then lead to a conclusion. Select a connection to read its hypotheses and deduction; select a node to read the linked statement.
The full scope of each occurrence is recorded below its diagram. A proved conditional implication and an open geometric premise keep their separate statuses.
LOGIC 01
Endpoint vanishing must cover both complementary degrees
Divisors and zero-cycles supply a full pair in every dimension; one-cycles supply the complete middle pair on a threefold.
Scope. Work under the standing actual-form and covariance hypotheses. Vʳ denotes geometrically homologically trivial rational Chow classes; Hʳ imposes every finite-place obstruction condition.
Scroll inside the frame to follow a large branch. “Fit width” shows the whole layout; “Readable size” restores the text to 100%.
Read the 2 connections
Each numbered card gives the exact inputs, the deduction and its sources.
01AND · ALL endpoint inputs and the kernel pairing
Premises and stated hypotheses
- HypothesisHomologically trivial divisors have zero obstruction
This remains an input to the domain calculation. The expanded proof below uses the divisor numerical-to-algebraic comparison and the independent compact/Poincaré construction under (A1), (A2) and (R).
Scope & sources - HypothesisHomologically trivial zero-cycles have zero obstruction
This remains an input to the domain calculation. The public R12 top-degree proof reduces componentwise degree-zero cycles to geometric algebraic equivalence; its actual-form hypotheses remain explicit below.
Scope & sources - ConditionalHomologically trivial one-cycles have zero obstruction
The same-dimensional semistable alteration, original-model curve correction and dominant obstruction injection are expanded below.
Scope & sources - ConditionalLocalize global classes and apply the kernel pairing
Geometric cycle classes commute with completion. Vanishing at every finite place gives Hᵃ=Vᵃ in each covered degree a; both complementary equalities are needed for a full pair.
Scope & sources
How the deduction works
The three local endpoint results imply H¹=V¹, Hⁿ=Vⁿ and Hⁿ⁻¹=Vⁿ⁻¹. Substitute the appropriate equalities in Hʳ×Hⁿ⁺¹⁻ʳ.
Conclusion / target
The one-cycle result covers degree n−1. In dimensions at least four, it does not by itself determine the codimension-two factor.
Scope & sourcesEvidence and further reading
Back to connection 01 in the diagram ↑All 9 nodes: scope, status and sources
ConditionalFull V¹ × Vⁿ, and Vⁿ⁻¹ × H² for n≥2
The one-cycle result covers degree n−1. In dimensions at least four, it does not by itself determine the codimension-two factor.
Back to this node in the diagram ↑HypothesisHomologically trivial divisors have zero obstruction
This remains an input to the domain calculation. The expanded proof below uses the divisor numerical-to-algebraic comparison and the independent compact/Poincaré construction under (A1), (A2) and (R).
Back to this node in the diagram ↑HypothesisHomologically trivial zero-cycles have zero obstruction
This remains an input to the domain calculation. The public R12 top-degree proof reduces componentwise degree-zero cycles to geometric algebraic equivalence; its actual-form hypotheses remain explicit below.
Back to this node in the diagram ↑ConditionalHomologically trivial one-cycles have zero obstruction
The same-dimensional semistable alteration, original-model curve correction and dominant obstruction injection are expanded below.
Back to this node in the diagram ↑ConditionalLocalize global classes and apply the kernel pairing
Geometric cycle classes commute with completion. Vanishing at every finite place gives Hᵃ=Vᵃ in each covered degree a; both complementary equalities are needed for a full pair.
Back to this node in the diagram ↑ConditionalEvery complementary pair is covered when dim X≤3
For a threefold, the middle group V² is the one-cycle group, so h:V²×V²→ℝ is included.
Back to this node in the diagram ↑HypothesisThe dimension satisfies 1≤n≤3
Only codimensions 1 through n occur. In dimension zero the listed positive complementary range is empty.
Back to this node in the diagram ↑ConditionalThe three local endpoint inputs above
Use the expanded conditional divisor and zero-cycle proofs and the conditional one-cycle proof. All three retain their actual-form premises.
Back to this node in the diagram ↑ConditionalH16 on both complementary kernels
The same analytic and reciprocity hypotheses remain required.
Back to this node in the diagram ↑LOGIC 02
The divisor endpoint follows from the independent compact-parameter proof
Geometric homological triviality gives numerical triviality; a nonzero tensor power is algebraically trivial, and rational coefficients remove that power.
Scope. K is the original finite extension of ℚp with its original absolute value. Let X/K be smooth projective geometrically integral of dimension n≥1 and D a prescribed rational divisor with zero geometric ℓ-adic class for one ℓ≠p. The actual-form conclusion assumes (A1), (A2) and (R).
Scroll inside the frame to follow a large branch. “Fit width” shows the whole layout; “Readable size” restores the text to 100%.
Read the 4 connections
Each numbered card gives the exact inputs, the deduction and its sources.
01AND · Use cycle-class compatibility with intersection and degree
Premises and stated hypotheses
- HypothesisThe prescribed divisor has zero geometric ℓ-adic class
This is the input for one ℓ≠p. It concerns the class of D on X over the algebraic closure, not only a numerical condition on one model.
Scope & sources
How the deduction works
Multiplying the prescribed divisor by a preserves its zero geometric class. The trace calculation is on the geometric generic fibre; no special-fibre correction or divisor Tate assertion is used.
Conclusion / target
For each complete integral curve C, cycle classes commute with intersection and trace: deg(c₁(L)·[C])=Tr(c₁(L)∪cl(C))=0. Thus L is geometrically numerically trivial.
Scope & sourcesEvidence and further reading
Back to connection 01 in the diagram ↑02AND · Apply the projective Picard comparison
Premises and stated hypotheses
- ConditionalL has degree zero on every geometric integral curve
For each complete integral curve C, cycle classes commute with intersection and trace: deg(c₁(L)·[C])=Tr(c₁(L)∪cl(C))=0. Thus L is geometrically numerically trivial.
Scope & sources
How the deduction works
X over the algebraic closure is projective and L is invertible, so the stated theorem applies to its numerically trivial class.
Conclusion / target
Kleiman, The Picard scheme, Definition 6.1 and Theorem 6.3(b)⇒(a), over an algebraically closed field: numerical triviality is equivalent to a nonzero tensor power being algebraically trivial. The R12 source cites this exact comparison.
Scope & sourcesEvidence and further reading
Back to connection 02 in the diagram ↑03AND · Pass from the tensor power to the rational divisor class
Premises and stated hypotheses
- ConditionalSome nonzero tensor power of L is algebraically trivial
Kleiman, The Picard scheme, Definition 6.1 and Theorem 6.3(b)⇒(a), over an algebraically closed field: numerical triviality is equivalent to a nonzero tensor power being algebraically trivial. The R12 source cites this exact comparison.
Scope & sources
How the deduction works
This implication uses rational coefficients; it does not assert that every numerically trivial integral line bundle already lies in Pic⁰.
Conclusion / target
If L⊗m is algebraically trivial for m≠0, then am[D] is algebraically trivial. Divide by am in the rational Chow group.
Scope & sourcesEvidence and further reading
Back to connection 03 in the diagram ↑04AND · BOTH the algebraic reduction and the conditional actual construction
Premises and stated hypotheses
- Conditional[D] is geometrically algebraically trivial in CH¹ℚ
If L⊗m is algebraically trivial for m≠0, then am[D] is algebraically trivial. Divide by am in the rational Chow group.
Scope & sources - ConditionalThe compact/Poincaré theorem under (A1), (A2), (R)
Use the construction expanded below with r=1. It starts from generic Chow-zero model relations, rational-fibre rigidity and compactness; it does not use the divisor endpoint or the curve-transfer proof.
Scope & sources
Supporting results used in this deduction
- ConditionalAlgebraic equivalence and compact parameters
BB-ALG-TRANSFERThis remains a conditional application of Theorem A; the classical divisor comparison does not certify its draft actual-form premises.
Scope & sources
How the deduction works
Apply the independent compact/Poincaré theorem to the rational divisor class. Its global primitive corrects the original curved pair on X, preserving the exact support and every compact actual complementary test.
Conclusion / target
The R12 Picard corollary supplies Df=c₁(L̄) for a geometrically algebraically trivial rational line-bundle class. A generic Chow-zero comparison handles the particular model character in the prescribed curved pair.
Scope & sourcesEvidence and further reading
Back to connection 04 in the diagram ↑All 8 nodes: scope, status and sources
definitionClear rational coefficients in the divisor class
Choose a positive integer a so that aD is a Cartier divisor on smooth X, and put L=O(aD). The comparison below is made on X over an algebraic closure; rational Chow coefficients permit division by every nonzero integer.
Back to this node in the diagram ↑ConditionalOb¹X([D]) = 0 and exact-support strong G for D
The R12 Picard corollary supplies Df=c₁(L̄) for a geometrically algebraically trivial rational line-bundle class. A generic Chow-zero comparison handles the particular model character in the prescribed curved pair.
Back to this node in the diagram ↑ConditionalAlgebraic equivalence and compact parameters
This remains a conditional application of Theorem A; the classical divisor comparison does not certify its draft actual-form premises.
Back to this node in the diagram ↑Conditional[D] is geometrically algebraically trivial in CH¹ℚ
If L⊗m is algebraically trivial for m≠0, then am[D] is algebraically trivial. Divide by am in the rational Chow group.
Back to this node in the diagram ↑ConditionalSome nonzero tensor power of L is algebraically trivial
Kleiman, The Picard scheme, Definition 6.1 and Theorem 6.3(b)⇒(a), over an algebraically closed field: numerical triviality is equivalent to a nonzero tensor power being algebraically trivial. The R12 source cites this exact comparison.
Back to this node in the diagram ↑ConditionalL has degree zero on every geometric integral curve
For each complete integral curve C, cycle classes commute with intersection and trace: deg(c₁(L)·[C])=Tr(c₁(L)∪cl(C))=0. Thus L is geometrically numerically trivial.
Back to this node in the diagram ↑HypothesisThe prescribed divisor has zero geometric ℓ-adic class
This is the input for one ℓ≠p. It concerns the class of D on X over the algebraic closure, not only a numerical condition on one model.
Back to this node in the diagram ↑ConditionalThe compact/Poincaré theorem under (A1), (A2), (R)
Use the construction expanded below with r=1. It starts from generic Chow-zero model relations, rational-fibre rigidity and compactness; it does not use the divisor endpoint or the curve-transfer proof.
Back to this node in the diagram ↑LOGIC 03
The top-degree endpoint has an explicit public R12 proof
A zero geometric top-degree class has degree zero on each geometric component. Point differences are algebraically trivial and enter the compact-parameter construction.
Scope. Let X/K be smooth projective geometrically integral of dimension n≥1 over the original valued finite extension K/ℚp, and let Z be a prescribed rational zero-cycle. Assume zero geometric ℓ-adic class for one ℓ≠p and retain (A1), (A2), (R).
Scroll inside the frame to follow a large branch. “Fit width” shows the whole layout; “Readable size” restores the text to 100%.
Read the 4 connections
Each numbered card gives the exact inputs, the deduction and its sources.
01AND · Trace the geometric top-degree cycle class
Premises and stated hypotheses
- HypothesisThe prescribed zero-cycle has zero geometric ℓ-adic class
Use the actual geometric class in H²ⁿ(X over the algebraic closure, ℚℓ(n)), for one ℓ≠p.
Scope & sources
How the deduction works
The trace is normalized so that a geometric point has class of degree one. This gives zero degree separately on every component.
Conclusion / target
The trace of a zero-cycle class on a smooth projective connected geometric component is its degree. Restrict the zero cohomology class to each component and take that trace.
Scope & sourcesEvidence and further reading
Back to connection 01 in the diagram ↑02AND · Use the componentwise degree-zero equality
Premises and stated hypotheses
- ConditionalThe degree on every geometric component is zero
The trace of a zero-cycle class on a smooth projective connected geometric component is its degree. Restrict the zero cohomology class to each component and take that trace.
Scope & sources
How the deduction works
The cycle has finite support and rational coefficients, so this is a finite identity of cycles on each component.
Conclusion / target
Over the algebraic closure all closed points have degree one. Choose x₀ on each component; the equality follows from Σᵢaᵢ=0 on that component.
Scope & sourcesEvidence and further reading
Back to connection 02 in the diagram ↑03AND · BOTH the point-difference expression and connected families
Premises and stated hypotheses
- ConditionalWrite each component as Σᵢaᵢ([xᵢ]−[x₀])
Over the algebraic closure all closed points have degree one. Choose x₀ on each component; the equality follows from Σᵢaᵢ=0 on that component.
Scope & sources - ConditionalPoint differences are fibres of connected cycle families
On a smooth connected projective geometric component V, the diagonal in V×V is the family of points. Its fibres at x and x₀ give [x]−[x₀]. The curve-correspondence formulation used by R12 gives the same algebraic equivalence; its finite witnesses descend in the common construction below.
Scope & sources
How the deduction works
Algebraic equivalence is additive, so the finite rational sum of the displayed point differences is algebraically trivial.
Conclusion / target
Every degree-zero class on a smooth connected projective geometric component is a rational sum of point differences. Those differences are algebraically trivial.
Scope & sourcesEvidence and further reading
- R12: r12p:top-degree, geometric point-difference argument
- Curve-family representation of algebraic equivalence
04AND · BOTH geometric algebraic equivalence and the actual construction
Premises and stated hypotheses
- Conditional[Z] is geometrically algebraically trivial
Every degree-zero class on a smooth connected projective geometric component is a rational sum of point differences. Those differences are algebraically trivial.
Scope & sources - ConditionalThe compact/Poincaré theorem under (A1), (A2), (R)
Use the construction below with r=n. The prescribed curvature is an actual degree-n model character. R12 first combines Chern polynomials by Adams projection and tensor products; it makes no claim for arbitrary atomic actual forms.
Scope & sources
Supporting results used in this deduction
- ConditionalAlgebraic equivalence and compact parameters
BB-ALG-TRANSFERThe top-degree corollary applies the conditional algebraic-equivalence theorem. Recovering its public proof does not upgrade the analytic foundations.
Scope & sources
How the deduction works
The R12 top-degree proof reduces the generic zero-cycle class to geometric algebraic equivalence and invokes the compact/Poincaré theorem. Its primitive is on all of X, so the correction retains exactly Xᵃⁿ∖|Z|ᵃⁿ.
Conclusion / target
The complete companion source proves r12p:top-degree. Apply it to the actual model character of the prescribed pair, then subtract its global actual primitive. The test space is all compact actual scalar tests on Xᵃⁿ.
Scope & sourcesEvidence and further reading
Back to connection 04 in the diagram ↑All 9 nodes: scope, status and sources
definitionThe degree condition is geometric and componentwise
For the stated geometrically integral X there is one geometric component. If that assumption is dropped, require degree zero on every geometric connected component; total degree zero across different components is insufficient.
Back to this node in the diagram ↑ConditionalObⁿX([Z]) = 0 and exact-support strong G for Z
The complete companion source proves r12p:top-degree. Apply it to the actual model character of the prescribed pair, then subtract its global actual primitive. The test space is all compact actual scalar tests on Xᵃⁿ.
Back to this node in the diagram ↑ConditionalAlgebraic equivalence and compact parameters
The top-degree corollary applies the conditional algebraic-equivalence theorem. Recovering its public proof does not upgrade the analytic foundations.
Back to this node in the diagram ↑Conditional[Z] is geometrically algebraically trivial
Every degree-zero class on a smooth connected projective geometric component is a rational sum of point differences. Those differences are algebraically trivial.
Back to this node in the diagram ↑ConditionalWrite each component as Σᵢaᵢ([xᵢ]−[x₀])
Over the algebraic closure all closed points have degree one. Choose x₀ on each component; the equality follows from Σᵢaᵢ=0 on that component.
Back to this node in the diagram ↑ConditionalThe degree on every geometric component is zero
The trace of a zero-cycle class on a smooth projective connected geometric component is its degree. Restrict the zero cohomology class to each component and take that trace.
Back to this node in the diagram ↑HypothesisThe prescribed zero-cycle has zero geometric ℓ-adic class
Use the actual geometric class in H²ⁿ(X over the algebraic closure, ℚℓ(n)), for one ℓ≠p.
Back to this node in the diagram ↑ConditionalPoint differences are fibres of connected cycle families
On a smooth connected projective geometric component V, the diagonal in V×V is the family of points. Its fibres at x and x₀ give [x]−[x₀]. The curve-correspondence formulation used by R12 gives the same algebraic equivalence; its finite witnesses descend in the common construction below.
Back to this node in the diagram ↑ConditionalThe compact/Poincaré theorem under (A1), (A2), (R)
Use the construction below with r=n. The prescribed curvature is an actual degree-n model character. R12 first combines Chern polynomials by Adams projection and tensor products; it makes no claim for arbitrary atomic actual forms.
Back to this node in the diagram ↑LOGIC 04
Compact Poincaré families supply the global primitive for both endpoints
Descend finite algebraic data, form an additive norm family over K, use compactness and rational-fibre rigidity, and correct the original model character.
Scope. Let K/ℚp be finite with its original absolute value, X/K smooth projective geometrically integral with n=dim X, and z=[Z]∈CHʳ(X)ℚ geometrically algebraically trivial, 1≤r≤n. All actual-form assertions are under (A1), (A2), (R) for every smooth projective variety used. K and its absolute value remain fixed; a finite Galois extension is used only for algebraic family data.
Scroll inside the frame to follow a large branch. “Fit width” shows the whole layout; “Readable size” restores the text to 100%.
Read the 3 connections
Each numbered card gives the exact inputs, the deduction and its sources.
01AND · ALL algebraic data for the native norm family
Premises and stated hypotheses
- HypothesisFinite curve and Chow identities over one Galois L/K
Start with geometric algebraic triviality. The curves, correspondences, marked points and rational-equivalence witnesses use finitely many coefficients. Enlarge once to finite Galois L/K, with rational basepoints on all curves, and obtain the finite expression for zL.
Scope & sources - ConditionalA Poincaré family is additive and has fibre zL
Define Θ=pXJ*(pXC*Γ·pCJ*c₁(P)). Refined Gysin base change gives Θt=Γ*c₁(Pt), and the rigidified Picard group law gives Θt+t′=Θt+Θt′ and Θ₀=0 in Chow. Products and sums combine the finitely many families.
Scope & sources - ConditionalEvaluation and algebraic norm give Θᴺa = d z
Set A=ResL/K J and Θᴺ=(hA)*(id×evaluation)*Θ. The finite-flat/refined-Gysin identity gives Θᴺa=h*Θt=h*zL=d z. Evaluation is a group map, so rational fibres remain additive and the zero fibre is zero.
Scope & sources
How the deduction works
Poincaré families on the finitely many Jacobians combine on their product J. Weil restriction and algebraic finite-flat pushforward then give the required additive family over K.
Conclusion / target
Here d=[L:K] is the entire finite field degree. The norm operates on algebraic Chow families, not on an already constructed actual primitive.
Scope & sourcesEvidence and further reading
Back to connection 01 in the diagram ↑02AND · ALL actual comparison, rigidity and compactness premises
Premises and stated hypotheses
- Draft · audit pendingA Chow-invariant actual Bott–Chern obstruction
BB-ACTUAL-OBSTRUCTION(A1): strict actual model characters, pullback compatibility and a global actual primitive for a generic degree-r Chow-zero difference. Its registered analytic status remains draft.
Scope & sources - Draft · audit pendingUniform constancy along nearby rational fibres
BB-RATIONAL-FIBRE-RIGIDITY(R): one ordinary neighbourhood of 0∈A(K) gives Ωb=Ω₀ on all of X. The uniform actual-form assertion remains draft.
Scope & sources - ConditionalA positive multiple ma enters that neighbourhood
A(K) is compact because A is projective over the local field. Choose V with V−V inside the neighbourhood and cover A(K) by finitely many translates; two multiples of a in one translate give the required positive multiple.
Scope & sources - ConditionalThe difference Δ has generic Chow character zero
The additive native family has Θma−mΘa=0 and Θ₀=0. This is a relation of Chow classes, so (A1) applies to Δ. Merely having an abelian parameter variety would not suffice.
Scope & sources
Supporting results used in this deduction
- Internal proof availableCompact additive maps into torsion-free groups
BB-COMPACT-RIGIDITYCompactness is the proved group-theoretic step; the actual local constancy and model-character comparison remain separately displayed premises.
Scope & sources
How the deduction works
Choose m>0 with ma near zero. Rigidity gives Ωma=Ω₀, while Chow additivity makes Δ=Ωma−Ω₀−m(Ωa−Ω₀) generically Chow-zero. By (A1), D H=Δ; thus βa=−H/m has the required derivative.
Conclusion / target
Use a model character Ω of the additive native family on X×A. The primitive retains the restricted finite-flat traces of original scalar–curvature generators; no trace of an arbitrary primitive is required.
Scope & sourcesEvidence and further reading
Back to connection 02 in the diagram ↑03AND · ALL norm-family, primitive and exact-support inputs
Premises and stated hypotheses
- ConditionalAn additive family over K has a rational fibre d z
Here d=[L:K] is the entire finite field degree. The norm operates on algebraic Chow families, not on an already constructed actual primitive.
Scope & sources - ConditionalCompactness produces Dβa = Ωa−Ω₀
Use a model character Ω of the additive native family on X×A. The primitive retains the restricted finite-flat traces of original scalar–curvature generators; no trace of an arbitrary primitive is required.
Scope & sources - Draft · audit pendingGreen pairs on the prescribed support complement
BB-SUPPORTED-GREEN-PAIR(A2): choose (GZ,ΩZ) for the original prescribed cycle, with ΩZ representing its actual obstruction. The pair and its all-test identity retain their draft analytic source status.
Scope & sources - ConditionalCorrect the prescribed curvature by ΨZ=(βa−Hz)/d
The difference Ωa−Ω₀−dΩZ has generic Chow class Θᴺa−d z=0. By (A1), choose a GLOBAL actual Hz with DHz equal to that difference. Then D((βa−Hz)/d)=ΩZ; both the minus sign and the full d=[L:K] are retained.
Scope & sources - ConditionalγZ=GZ−ΨZ on the original support complement
ΨZ is global on Xᵃⁿ, so restriction introduces no extra removed set. Actual Stokes cancels ΩZ in the curved identity for every compact actual test of bidegree (n−r,n−r), preserving the original cycle coefficients.
Scope & sources
Supporting results used in this deduction
- ConditionalAlgebraic equivalence and compact parameters
BB-ALG-TRANSFERThis is the independent conditional compact/Poincaré route, with its actual-form inputs visible.
Scope & sources - ConditionalRemoving an exact global curvature
BB-GREEN-CORRECTIONThe formal global correction preserves the exact complement once the curved pair and global actual primitive are supplied.
Scope & sources
How the deduction works
The native norm fibre is d z. Compactness gives Dβa=Ωa−Ω₀. Generic Chow-zero comparison gives DHz=Ωa−Ω₀−dΩZ; hence ΨZ=(βa−Hz)/d has DΨZ=ΩZ. Subtract ΨZ from the prescribed pair and apply actual Stokes.
Conclusion / target
The construction uses no divisor endpoint and no correspondence-kernel transfer theorem. All final forms and tests are native over K, on exactly Xᵃⁿ∖|Z|ᵃⁿ.
Scope & sourcesEvidence and further reading
Back to connection 03 in the diagram ↑All 16 nodes: scope, status and sources
ConditionalObʳX(z) = 0 and strong G for the original Z
The construction uses no divisor endpoint and no correspondence-kernel transfer theorem. All final forms and tests are native over K, on exactly Xᵃⁿ∖|Z|ᵃⁿ.
Back to this node in the diagram ↑ConditionalAlgebraic equivalence and compact parameters
This is the independent conditional compact/Poincaré route, with its actual-form inputs visible.
Back to this node in the diagram ↑ConditionalRemoving an exact global curvature
The formal global correction preserves the exact complement once the curved pair and global actual primitive are supplied.
Back to this node in the diagram ↑ConditionalAn additive family over K has a rational fibre d z
Here d=[L:K] is the entire finite field degree. The norm operates on algebraic Chow families, not on an already constructed actual primitive.
Back to this node in the diagram ↑HypothesisFinite curve and Chow identities over one Galois L/K
Start with geometric algebraic triviality. The curves, correspondences, marked points and rational-equivalence witnesses use finitely many coefficients. Enlarge once to finite Galois L/K, with rational basepoints on all curves, and obtain the finite expression for zL.
Back to this node in the diagram ↑ConditionalA Poincaré family is additive and has fibre zL
Define Θ=pXJ*(pXC*Γ·pCJ*c₁(P)). Refined Gysin base change gives Θt=Γ*c₁(Pt), and the rigidified Picard group law gives Θt+t′=Θt+Θt′ and Θ₀=0 in Chow. Products and sums combine the finitely many families.
Back to this node in the diagram ↑ConditionalEvaluation and algebraic norm give Θᴺa = d z
Set A=ResL/K J and Θᴺ=(hA)*(id×evaluation)*Θ. The finite-flat/refined-Gysin identity gives Θᴺa=h*Θt=h*zL=d z. Evaluation is a group map, so rational fibres remain additive and the zero fibre is zero.
Back to this node in the diagram ↑ConditionalCompactness produces Dβa = Ωa−Ω₀
Use a model character Ω of the additive native family on X×A. The primitive retains the restricted finite-flat traces of original scalar–curvature generators; no trace of an arbitrary primitive is required.
Back to this node in the diagram ↑Internal proof availableCompact additive maps into torsion-free groups
Compactness is the proved group-theoretic step; the actual local constancy and model-character comparison remain separately displayed premises.
Back to this node in the diagram ↑Draft · audit pendingA Chow-invariant actual Bott–Chern obstruction
(A1): strict actual model characters, pullback compatibility and a global actual primitive for a generic degree-r Chow-zero difference. Its registered analytic status remains draft.
Back to this node in the diagram ↑Draft · audit pendingUniform constancy along nearby rational fibres
(R): one ordinary neighbourhood of 0∈A(K) gives Ωb=Ω₀ on all of X. The uniform actual-form assertion remains draft.
Back to this node in the diagram ↑ConditionalA positive multiple ma enters that neighbourhood
A(K) is compact because A is projective over the local field. Choose V with V−V inside the neighbourhood and cover A(K) by finitely many translates; two multiples of a in one translate give the required positive multiple.
Back to this node in the diagram ↑ConditionalThe difference Δ has generic Chow character zero
The additive native family has Θma−mΘa=0 and Θ₀=0. This is a relation of Chow classes, so (A1) applies to Δ. Merely having an abelian parameter variety would not suffice.
Back to this node in the diagram ↑Draft · audit pendingGreen pairs on the prescribed support complement
(A2): choose (GZ,ΩZ) for the original prescribed cycle, with ΩZ representing its actual obstruction. The pair and its all-test identity retain their draft analytic source status.
Back to this node in the diagram ↑ConditionalCorrect the prescribed curvature by ΨZ=(βa−Hz)/d
The difference Ωa−Ω₀−dΩZ has generic Chow class Θᴺa−d z=0. By (A1), choose a GLOBAL actual Hz with DHz equal to that difference. Then D((βa−Hz)/d)=ΩZ; both the minus sign and the full d=[L:K] are retained.
Back to this node in the diagram ↑ConditionalγZ=GZ−ΨZ on the original support complement
ΨZ is global on Xᵃⁿ, so restriction introduces no extra removed set. Actual Stokes cancels ΩZ in the curved identity for every compact actual test of bidegree (n−r,n−r), preserving the original cycle coefficients.
Back to this node in the diagram ↑LOGIC 05
One-cycles on arbitrary smooth projective varieties
A same-dimensional alteration supplies the regular model; descend the zero obstruction and correct the original prescribed cycle on X.
Scope. For n≥3 let X/K be smooth projective geometrically integral and Z a rational one-cycle with zero geometric ℓ-adic class for one ℓ≠p. For n=2, the source uses the divisor endpoint input.
Scroll inside the frame to follow a large branch. “Fit width” shows the whole layout; “Readable size” restores the text to 100%.
Read the 2 connections
Each numbered card gives the exact inputs, the deduction and its sources.
01AND · BOTH analytic detection and injective pullback
Premises and stated hypotheses
- HypothesisThe native model-Chern detector and generic Chow comparison
Test every vertical cycle, including full multiplicities and native residue degrees. The companion analytic detector is an explicit hypothesis.
Scope & sources - HypothesisAll-degree dominant obstruction pullback is injective
Retain the actual covariance theorem, all-degree injection, and full degrees. An alteration is not treated as finite away from the given Z.
Scope & sources
How the deduction works
ObV(f*z)=f*ObX(z)=0, so injectivity gives ObX(z)=0. No analytic pushforward of an upstairs Green form is asserted.
Conclusion / target
Regard the F-model over K. Native special-fibre degrees acquire the positive residue-field factor; the numerical detector kills curvature. Dominant injection retains all field and function-field degrees.
Scope & sourcesEvidence and further reading
Back to connection 01 in the diagram ↑02AND · ALL geometric, descent and support inputs
Premises and stated hypotheses
- HypothesisA regular strictly semistable alteration of the SAME dimension
de Jong gives finite F/K and V/F of dimension n with a projective generically finite dominant K-map V→X and a regular projective strictly semistable model. Operational Chow pullback between the smooth generic varieties preserves the geometric cycle class and codimension n−1.
Scope & sources - ConditionalA pure virtual-bundle curve certificate on the model of V
Its generic character is f*[Z], and its degree is zero against every vertical (n−1)-dimensional integral subvariety.
Scope & sources - ConditionalNumerical detection and dominant Bott–Chern injection
Regard the F-model over K. Native special-fibre degrees acquire the positive residue-field factor; the numerical detector kills curvature. Dominant injection retains all field and function-field degrees.
Scope & sources - HypothesisCorrect the prescribed pair on X by a global actual primitive
Take (GZ,ΩZ) for the original Z. Zero obstruction gives DΨ=ΩZ globally; γ=GZ−Ψ on the exact original complement. No support from the alteration is deleted.
Scope & sources
How the deduction works
The numerical certificate kills the upstairs obstruction. Covariance and dominant injectivity kill the downstairs class; subtract a global primitive from the original pair on X.
Conclusion / target
The result holds over the original K, on Xᵃⁿ∖|Z|ᵃⁿ, against every compact actual (1,1) test.
Scope & sourcesEvidence and further reading
Back to connection 02 in the diagram ↑All 7 nodes: scope, status and sources
ConditionalObⁿ⁻¹X([Z]) = 0 and strong G on the exact original complement
The result holds over the original K, on Xᵃⁿ∖|Z|ᵃⁿ, against every compact actual (1,1) test.
Back to this node in the diagram ↑HypothesisA regular strictly semistable alteration of the SAME dimension
de Jong gives finite F/K and V/F of dimension n with a projective generically finite dominant K-map V→X and a regular projective strictly semistable model. Operational Chow pullback between the smooth generic varieties preserves the geometric cycle class and codimension n−1.
Back to this node in the diagram ↑ConditionalA pure virtual-bundle curve certificate on the model of V
Its generic character is f*[Z], and its degree is zero against every vertical (n−1)-dimensional integral subvariety.
Back to this node in the diagram ↑ConditionalNumerical detection and dominant Bott–Chern injection
Regard the F-model over K. Native special-fibre degrees acquire the positive residue-field factor; the numerical detector kills curvature. Dominant injection retains all field and function-field degrees.
Back to this node in the diagram ↑HypothesisThe native model-Chern detector and generic Chow comparison
Test every vertical cycle, including full multiplicities and native residue degrees. The companion analytic detector is an explicit hypothesis.
Back to this node in the diagram ↑HypothesisAll-degree dominant obstruction pullback is injective
Retain the actual covariance theorem, all-degree injection, and full degrees. An alteration is not treated as finite away from the given Z.
Back to this node in the diagram ↑HypothesisCorrect the prescribed pair on X by a global actual primitive
Take (GZ,ΩZ) for the original Z. Zero obstruction gives DΨ=ΩZ globally; γ=GZ−Ψ on the exact original complement. No support from the alteration is deleted.
Back to this node in the diagram ↑LOGIC 06
The weighted numerical correction on a regular SNC model
A divisor-kernel calculation and geometric homology annihilate the obstruction functional; numerical duality then solves the vertical correction equation.
Scope. A supplied integral regular flat projective model has special fibre ΣᵢmᵢDᵢ with strict normal-crossing reduced support and smooth closed strata; n≥3. The prescribed one-cycle has zero geometric ℓ-adic class for one ℓ≠p. All mᵢ are positive and all groups rational.
Scroll inside the frame to follow a large branch. “Fit width” shows the whole layout; “Readable size” restores the text to 100%.
Read the 3 connections
Each numbered card gives the exact inputs, the deduction and its sources.
01AND · ALL weighted-kernel and vanishing steps
Premises and stated hypotheses
- Conditionalker A₁ = M ker δ + im γ₀
The principal-fibre relation gives the mᵢmⱼ weighted incidence matrix. Finite-dimensional edge correction and two divisor Hodge-index energy identities force compatibility. No middle-dimensional standard-conjecture sign is used.
Scope & sources - ConditionaltΘ vanishes on im γ₀
Each double-stratum cycle occurs twice with opposite signs under Σjᵢ*. Projection therefore kills the edge contribution.
Scope & sources - ConditionaltΘ vanishes on M ker δ by geometric homology
Numerically compatible divisors have compatible geometric ℓ-adic classes. Purity kills outgoing Čech differentials, proper base change lifts the class, and ΣmᵢTrDi equals generic trace. The generic z factor is zero. No algebraic or Frobenius-fixed lift is required.
Scope & sources
How the deduction works
Signed edge cycles cancel as actual supported cycles. For a compatible divisor tuple, the geometric cycle class of z and the multiplicity-weighted trace make the pairing zero.
Conclusion / target
ker A₁ = M ker δ + im γ₀. The two summands are killed for different reasons.
Scope & sourcesEvidence and further reading
Back to connection 01 in the diagram ↑02AND · ALL inputs of the finite-dimensional correction
Premises and stated hypotheses
- ConditionalThe extension functional tΘ annihilates ker A₁
ker A₁ = M ker δ + im γ₀. The two summands are killed for different reasons.
Scope & sources - ConditionalPerfect numerical duality identifies the transpose image
Projection on double strata and the same diagonal normal term give A₁ᵗ=Aₙ₋₂. Thus (ker A₁)⊥=im Aₙ₋₂ over ℚ.
Scope & sources
How the deduction works
The extension functional annihilates ker A₁. Perfect numerical duality and A₁ᵗ=Aₙ₋₂ then put T in im Aₙ₋₂; choose rational cycle representatives of a preimage.
Conclusion / target
If Tᵢ=jᵢ!Θ and T=Aₙ₋₂u, then Θad=Θ−Σjᵢ*uᵢ has unchanged generic class and componentwise numerical restriction zero.
Scope & sourcesEvidence and further reading
Back to connection 02 in the diagram ↑03AND · BOTH the corrected Chow extension and its operational realization
Premises and stated hypotheses
- ConditionalA corrected extension has zero numerical restriction on every Dᵢ
If Tᵢ=jᵢ!Θ and T=Aₙ₋₂u, then Θad=Θ−Σjᵢ*uᵢ has unchanged generic class and componentwise numerical restriction zero.
Scope & sources - ConditionalThe regular-model Chern character gives a pure operational ξ
The rational K₀-to-Chow isomorphism gives the preimage of the pure class. Uniqueness under the Adams operation forces every other OPERATIONAL component to vanish; cap equality alone is not substituted.
Scope & sources - ConditionalComponentwise numerical zero tests every vertical integral support
Every vertical integral V lies in some Dᵢ. Regular Gysin compatibility and projection identify the supported character degree with (jᵢ!Θad)·[V], which is zero.
Scope & sources
How the deduction works
Apply the pure operational representative lemma to the vertically corrected class. Numerical triviality on each smooth component then tests every vertical subvariety, including singular ones.
Conclusion / target
There is ξ with chₙ₋₁(ξ|X)=z, all other operational character components zero, and zero degree on every integral vertical (n−1)-cycle.
Scope & sourcesEvidence and further reading
Back to connection 03 in the diagram ↑All 10 nodes: scope, status and sources
definitionAₐ is restriction after vertical Gysin; M=diag(mᵢ)
δ compares divisor restrictions on double strata, and γ₀ is signed Gysin of edge constants. Every degree is native over the finite residue field.
Back to this node in the diagram ↑ConditionalA pure K₀ witness with zero degree on every vertical test
There is ξ with chₙ₋₁(ξ|X)=z, all other operational character components zero, and zero degree on every integral vertical (n−1)-cycle.
Back to this node in the diagram ↑ConditionalA corrected extension has zero numerical restriction on every Dᵢ
If Tᵢ=jᵢ!Θ and T=Aₙ₋₂u, then Θad=Θ−Σjᵢ*uᵢ has unchanged generic class and componentwise numerical restriction zero.
Back to this node in the diagram ↑ConditionalThe extension functional tΘ annihilates ker A₁
ker A₁ = M ker δ + im γ₀. The two summands are killed for different reasons.
Back to this node in the diagram ↑Conditionalker A₁ = M ker δ + im γ₀
The principal-fibre relation gives the mᵢmⱼ weighted incidence matrix. Finite-dimensional edge correction and two divisor Hodge-index energy identities force compatibility. No middle-dimensional standard-conjecture sign is used.
Back to this node in the diagram ↑ConditionaltΘ vanishes on im γ₀
Each double-stratum cycle occurs twice with opposite signs under Σjᵢ*. Projection therefore kills the edge contribution.
Back to this node in the diagram ↑ConditionaltΘ vanishes on M ker δ by geometric homology
Numerically compatible divisors have compatible geometric ℓ-adic classes. Purity kills outgoing Čech differentials, proper base change lifts the class, and ΣmᵢTrDi equals generic trace. The generic z factor is zero. No algebraic or Frobenius-fixed lift is required.
Back to this node in the diagram ↑ConditionalPerfect numerical duality identifies the transpose image
Projection on double strata and the same diagonal normal term give A₁ᵗ=Aₙ₋₂. Thus (ker A₁)⊥=im Aₙ₋₂ over ℚ.
Back to this node in the diagram ↑ConditionalThe regular-model Chern character gives a pure operational ξ
The rational K₀-to-Chow isomorphism gives the preimage of the pure class. Uniqueness under the Adams operation forces every other OPERATIONAL component to vanish; cap equality alone is not substituted.
Back to this node in the diagram ↑ConditionalComponentwise numerical zero tests every vertical integral support
Every vertical integral V lies in some Dᵢ. Regular Gysin compatibility and projection identify the supported character degree with (jᵢ!Θad)·[V], which is zero.
Back to this node in the diagram ↑Remaining problem
In higher dimensions, the one-cycle result leaves the codimension-two factor to be determined. The regular-model correction and descent are explicit analytic inputs to this deduction.