The valuation of x ^ n - 1 in a local field #
Let K be a nonarchimedean local field and write v_K for its normalized additive valuation,
read on all of K as WithZero.log ∘ normalizedValuationWithZero K (so that the junk value at
0 is 0). This file computes v_K(x ^ n - 1) for deep principal units and deduces the
one-factor estimate for v_K(α ^ b - 1) in a local field of characteristic zero, that is, in a
finite extension of ℚ_[p].
For x ∈ U(K,i) at a depth i with v_K(ℓ) < (ℓ - 1) * i for every prime ℓ ∣ n, the n-th
power map carries U(K,j) onto U(K, j + v_K(n)) for every j ≥ i and is injective on
U(K,i), so it moves the exact depth of x by v_K(n):
v_K(x ^ n - 1) = v_K(x - 1) + v_K(n) for x ≠ 1.
This plays the role of the isometry of the p-adic logarithm on deep units, without any
logarithm in the statement.
In characteristic zero, with residue characteristic p and e = v_K(p), every unit α has a
power γ = α ^ M in U(K, e + 1), where the formula applies at every exponent. If γ ≠ 1 this
gives v_K(α ^ b - 1) ≤ v_K(γ - 1) + e v_p(b). Separating nonunits (where
v_K(α ^ b - 1) ≤ 0) and roots of unity (where α ^ b takes only finitely many values) gives,
for every fixed α, a constant C with
v_K(α ^ b - 1) ≤ C (1 + log max(1, |b|)) whenever α ^ b ≠ 1.
Main results #
TauCeti.log_normalizedValuationWithZero_pow_sub_one: the exact formulav_K(x ^ n - 1) = v_K(x - 1) + v_K(n)for a deep unitx ≠ 1.TauCeti.log_normalizedValuationWithZero_zpow_sub_one_le: ifx ^ M ≠ 1lies in a deep step, thenv_K(x ^ b - 1) ≤ v_K(x ^ M - 1) + v_K(|b|).TauCeti.exists_log_normalizedValuationWithZero_zpow_sub_one_le: in characteristic zero, the one-factor estimatev_K(α ^ b - 1) ≤ C (1 + log max(1, |b|))for every fixedα ∈ K.
References #
- Kunrui Yu, Linear forms in p-adic logarithms II, Compositio Math. 74 (1990), 15–113, §1.
- Kunrui Yu, Linear forms in p-adic logarithms III, Compositio Math. 91 (1994), 241–276, §1.
- J.-P. Serre, Corps Locaux, Chapter XIV, §4, Proposition 9.
The valuation of x ^ n - 1 for a deep unit. For (n : K) ≠ 0, a depth i with
v_K(p) < (p - 1) * i for every prime p ∣ n, and x ∈ U(K,i) with x ≠ 1,
v_K(x ^ n - 1) = v_K(x - 1) + v_K(n).
The one-factor estimate, explicit form. Let x ∈ Kˣ have a power x ^ M ≠ 1 in a
positive-depth step U(K,i), and let b be an integer with (|b| : K) ≠ 0 such that
v_K(ℓ) < (ℓ - 1) * i for every prime ℓ ∣ b. Then
v_K(x ^ b - 1) ≤ v_K(x ^ M - 1) + v_K(|b|).
Nonunits. If x is not a unit of 𝒪[K] and b ≠ 0, then v_K(x ^ b - 1) ≤ 0: a positive
valuation would make x ^ b, hence x, a unit.
The one-factor estimate. Let K be a nonarchimedean local field of characteristic zero,
that is, a finite extension of ℚ_[p]. For every fixed α ∈ K there is a constant C,
depending on α and K, such that
v_K(α ^ b - 1) ≤ C (1 + log max(1, |b|)) for every integer b with α ^ b ≠ 1.