The Galois coset-norm kit + the relative ramification index #
Two axiom-free kits feeding the c2c4 assembly of the analytic hunram (docs/orchestration/p15f2c2c-handoff.md).
Part 1 — the coset norm. For H ≤ K closed subgroups of Gal(ℚ̄₂/ℚ₂) with finite index
[K : H], the product cosetNorm H K x = ∏_{c ∈ K ⧸ H} (out c) • x of the K-cosets of H
applied to x:
- is
K-invariant, hence lands infixedField K, whenx ∈ fixedField H(cosetNorm_mem); - has spectral norm
‖cosetNorm H K x‖ = ‖x‖ ^ [K : H](norm_cosetNorm, fromnorm_galois).
So for a tower k ≤ L (H = L.fixingSubgroup ≤ K = k.fixingSubgroup) it realizes
‖x‖ ^ [L : k] ∈ ‖k^×‖ for every x ∈ L^× — the input to Part 2's divisibility.
Part 2 — the relative ramification index. Over B13 DyadicUnitFiltration data (kept as
hypotheses, half-A idiom, so statements stay axiom-free), for a tower F ≤ L: relE with
‖π_F‖ = ‖π_L‖ ^ relE, the tower formula e_L = relE * e_F, and the divisibility
relE ∣ n whenever ‖x‖ ^ n ∈ ‖F^×‖ for all x ∈ L^×. c2c4 instantiates twice
([L : k] ⟹ relE ∣ 2, ⟨t⟩-orbit ⟹ relE ∣ r).
Axiom-free throughout (Ax = ∅, std-3).
Part 1 — the coset norm #
The coset norm: the product over the K-cosets of H of the coset representatives
applied to x. Uses Quotient.out representatives; the value is representative-independent on
fixedField H (smul_eq_of_quot_eq).
Equations
- GQ2.GaloisCosetNorm.cosetNorm H K x = ∏ c : ↥K ⧸ H.subgroupOf K, ↑(Quotient.out c) • x
Instances For
Representative independence (the well-definedness (i)): on fixedField H, two elements of
K in the same H-coset act identically.
Each factor is nonzero, so the coset norm of a nonzero element is nonzero.
(iii) the norm formula: ‖cosetNorm‖ = ‖x‖ ^ [K : H] (each factor is ‖x‖ by
norm_galois).
(ii) K-invariance: on fixedField H the coset norm lands in fixedField K
(left multiplication by g ∈ K permutes the cosets, and the finite product is reordered).
c2c4-facing package: for a nonzero x ∈ fixedField H, the power ‖x‖ ^ [K : H] is the
spectral norm of a nonzero element of fixedField K (the coset norm). This is the
‖x‖ ^ n ∈ ‖(fixedField K)^×‖ input consumed by relE_dvd.
Part 2 — the relative ramification index of a tower F ≤ L #
The relative ramification index e(L/F) of a tower F ≤ L, in norm vocabulary: the
integer m with ‖π_F‖ = ‖π_L‖ ^ m (it exists by norm_eq_zpow, as π_F ∈ F ≤ L).
Equations
- GQ2.GaloisCosetNorm.relE FF FL hFL = ⋯.choose
Instances For
Defining property of relE: ‖π_F‖ = ‖π_L‖ ^ e(L/F).
Tower multiplicativity of the absolute ramification index: e_L = e(L/F) · e_F (both
uniformizers meet ‖2‖ = ‖π‖^e, and ‖π_F‖ = ‖π_L‖^{e(L/F)}; zpow-injectivity on
‖π_L‖ ∈ (0,1)).
e(L/F) ≥ 1 (a genuine relative index): from tower multiplicativity and e_L, e_F ≥ 1.
The divisibility feeding c2c4: if ‖x‖ ^ n lies in ‖(fixedField F)^×‖ for every
x ∈ L^×, then e(L/F) ∣ n. (Test at x = π_L: ‖π_L‖^n = ‖π_F‖^a = ‖π_L‖^{e(L/F)·a}.)