Bridging the banana transmission vocabulary to AspPerm #
Bananas/ models a transmission permutation as a raw function ℤ → ℤ
(IsTransmissionPermutation, IsKAffine, kInversions), because that is the
shape the paper's Definition 2.11 writes down and the shape the inversion
counting arguments use. Utilities models the same notion over the
demazure dependency's AspPerm, via Utilities.SatisfiesTransmission
(an inequality against the slipface τ.s).
This file supplies the adapter between those two formulations. Nothing here
is banana-specific: it holds for any connected CFGraph with two marks.
What is proved #
exists_aspPerm_rank_eq_of_isTransmissionPermutation is the substantive half.
Given a raw transmission permutation τ for D, it produces the AspPerm
σ with σ.func = τ and the equality
rank G (D + a • u - b • v) = σ.s (a + 1) b - 1
at every lattice point. TransmissionInequality is the ≥ weakening of this.
The degree normalization deg D = genus G + σ.χ, the other conjunct of
SatisfiesTransmission, looks like an extra side condition but is automatic:
AspPerm.s_eq at the origin gives χ(σ) = σ.s 0 0 - (σ⁻¹).s 0 0, the two
counting lemmas of ThetaNonrecurrence identify those with rank (D - u) + 1
and rank (K - (D - u)) + 1, and Riemann-Roch turns the difference into
deg (D - u) - g + 1 = deg D - g. So
satisfiesTransmission_of_isTransmissionPermutation is unconditional and
constructs a SatisfiesTransmission witness directly.
The AspPerm underlying a raw transmission permutation, together with the
exact marked rank formula it computes.
This is the translation lemma between the two transmission formalisms. The
AspPerm is the one whose slipface is the marked rank surface of D - u, and
σ.s (a + 1) b is exactly rank (D + a·u - b·v) + 1.
Every raw transmission permutation yields the transmission inequalities of
Utilities.Transmission for the corresponding AspPerm, on any test set.
Unlike satisfiesTransmission_of_isTransmissionPermutation this needs no
degree hypothesis, because SatisfiesTransmissionOn carries none.
The degree normalization is automatic.
SatisfiesTransmission requires deg D = g + χ(σ), which looks like an extra
side condition. It is not: for the AspPerm attached to a transmission
permutation of D it follows from Riemann-Roch. Evaluating
AspPerm.s_eq at (0,0) gives
χ(σ) = σ.s 0 0 - (σ⁻¹).s 0 0, and the two counting lemmas identify those
with rank (D - u) + 1 and rank (K - (D - u)) + 1, so
χ(σ) = rank (D - u) - rank (K - (D - u)) = deg (D - u) - g + 1 = deg D - g.
The full bridge, unconditionally: a raw transmission permutation for D
gives an AspPerm satisfying Utilities.SatisfiesTransmission for D.
This gives a direct construction of a SatisfiesTransmission witness.
Specialization to a banana graph: all-submodularity and a torsion witness
already produce a raw transmission permutation for every divisor
(exists_affine_transmission_of_allSubmodular), so they produce a full
AspPerm-level transmission witness for every divisor too.
This is the form in which the Section 4 banana results — for instance
evenlyMarkedTheta_kGeneral, whose KGeneralTransmission conclusion contains
exactly this data — become usable by the AspPerm/wedge machinery.