Coordinate-free Riemann–Roch #
This file exposes the Riemann–Roch stack over divisors indexed by intrinsic valuation-subring
places. The kernel-checked chart proofs are transported across chartToPlace.
Main definitions and results #
FunctionField.RRspace,FunctionField.ell, andFunctionField.genus.FunctionField.AdeleSpaceandFunctionField.adeleFilt.FunctionField.IsCanonicalandFunctionField.duality.FunctionField.riemann_rochand coordinate-free corollaries C1–C6.
A function belongs to the intrinsic Riemann–Roch space of D when its normalized valuation
at every place is bounded by the coefficient of D there.
Equations
- FunctionField.memRRspace k K D f = ∀ (v : FunctionField.Place k K), v.valuation f ≤ WithZero.exp (D v)
Instances For
The Riemann–Roch space of an intrinsic divisor.
Equations
- FunctionField.RRspace k K D = { carrier := {f : K | FunctionField.memRRspace k K D f}, add_mem' := ⋯, zero_mem' := ⋯, smul_mem' := ⋯ }
Instances For
The Riemann–Roch dimension of an intrinsic divisor.
Equations
- FunctionField.ell k K D = Module.finrank k ↥(FunctionField.RRspace k K D)
Instances For
The defect of an intrinsic divisor.
Equations
- FunctionField.defect k K D = FunctionField.Divisor.deg k K D + 1 - ↑(FunctionField.ell k K D)
Instances For
The intrinsic genus is the supremum of the defects of all intrinsic divisors. The Riemann–Roch argument below proves that this supremum is attained and agrees with every chart calculation.
Equations
- FunctionField.genus k K = (sSup {d : ℤ | ∃ (D : FunctionField.Divisor k K), FunctionField.defect k K D = d}).toNat
Instances For
The intrinsic membership condition agrees with the coordinate construction.
The intrinsic Riemann–Roch space is the chart space after reindexing divisors.
Riemann–Roch dimension is preserved by the chart equivalence.
The index of specialty of an intrinsic divisor.
Equations
- FunctionField.indexOfSpecialty k K D = (↑(FunctionField.ell k K D) - (FunctionField.Divisor.deg k K D + 1 - ↑(FunctionField.genus k K))).toNat
Instances For
The dimension of the intrinsic adele quotient by A(D) plus the diagonal.
Equations
- FunctionField.finrankAdeleQuotient k K D = Module.finrank k (↥(FunctionField.AdeleSpace k K) ⧸ FunctionField.adeleFilt k K D + FunctionField.diagonalSubmodule k K)
Instances For
Pulling an intrinsic divisor back to the chart preserves degree.
Intrinsic defect is preserved by the chart equivalence.
The intrinsic supremum definition of genus agrees with the chart construction.
The numerical characterization of a canonical coordinate divisor.
A divisor is canonical exactly when it has degree 2g - 2 and Riemann–Roch dimension g.
The intrinsic specialty index agrees with the coordinate construction.
The intrinsic adele quotient is linearly equivalent to its chart presentation.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The intrinsic and chart adele quotients have the same dimension.
The intrinsic adele quotient dimension is the index of specialty.
The rank change in the intrinsic filtration-and-diagonal sandwich.
Equations
- FunctionField.sandwichRank k K D E = FunctionField.Chart.sandwichRank k K ((FunctionField.divisorEquivChart k K).symm D) ((FunctionField.divisorEquivChart k K).symm E)
Instances For
The intrinsic sandwich rank is the change in deg D - ell(D).
The dimension of intrinsic Omega(D) is the index of specialty of D.
Every function field admits an intrinsic canonical divisor.
Coordinate-free duality.
Riemann's inequality for intrinsic divisors.
The coordinate-free Riemann–Roch theorem.
C1: a canonical divisor has Riemann–Roch dimension equal to the genus.
C2: a canonical divisor has degree 2g - 2.
C3: divisors of degree at least 2g - 1 are nonspecial.
C4: the specialty index is ℓ(W-D).
The core Clifford inequality for intrinsic divisors: whenever ℓ(D) and ℓ(W − D) are
both positive for a canonical W, 2(ℓ(D) − 1) ≤ deg D. No degree bounds are needed.
C5: Clifford's inequality for intrinsic divisors in its textbook form. The degree
bounds 0 ≤ deg D ≤ 2g − 2 are kept for fidelity to the standard statement but are not needed;
see clifford_of_ell_pos.
C6: there exists an intrinsic nonspecial divisor.