Deep units are squares, at the sharp depth #
In a nonarchimedean local field of characteristic different from two, every unit in
U(K, 2 v_K(2) + 1) is a square. The resulting openness of the square subgroup, including
in residue characteristic two, is proved in PowerSubgroup.Open and supplies neighborhoods
in which a nonzero value stays in one square class.
The exponent is natCastValuation K 2 h2, so no choice of a dyadic base field is needed.
The inclusion is the case p = 2, i = v_K(2) + 1 of
TauCeti.map_powMonoidHom_unitFiltration_of_prime: squaring carries U(K, v_K(2) + 1) onto
U(K, 2 v_K(2) + 1).
The depth is sharp: U(K, 2 v_K(2)) is not contained in the squares. Since
𝓂[K]^(2 v_K(2)) = 4 𝒪[K], an element 1 + 4c of that depth is a square exactly when
c = t² + t for some integral t, so any c whose residue lies outside the image of the
Artin–Schreier map t ↦ t² + t of 𝓀[K] gives a non-square. That map identifies 0 and -1,
so on the finite residue field it is not surjective. The argument is uniform in the residue
characteristic: when it is odd, v_K(2) = 0 and the statement is that some unit is not a square.
Main results #
TauCeti.unitFiltration_le_range_powMonoidHom_two:U(K, 2 v_K(2) + 1) ⊆ (Kˣ)².TauCeti.isSquare_of_eq_one_add_four_mul: a residue Artin–Schreier condition makes1 + 4ma square.TauCeti.valuation_one_add_four_mul: in residue characteristic two,1 + 4mis a unit.TauCeti.not_mem_range_of_eq_sq_mul_one_add_four_mul: a nonsquare of the formζ² (1 + 4m)has residue ofmoutside the Artin–Schreier range.TauCeti.not_unitFiltration_le_range_powMonoidHom_two:U(K, 2 v_K(2)) ⊄ (Kˣ)².TauCeti.exists_mem_unitFiltration_not_isSquare: a nonsquare witness inU(K, 2 v_K(2)).TauCeti.unitFiltration_le_range_powMonoidHom_two_iff:U(K, n) ⊆ (Kˣ)²exactly when2 v_K(2) + 1 ≤ n.TauCeti.dyadicLevel: the valuationv_K(2)used in the square interface.TauCeti.unitFiltration_le_squareandTauCeti.not_unitFiltration_le_square: the same sharp square theorem expressed usingSubgroup.square Kˣ.TauCeti.valuation_sq_sub_one_ne_pow_oddandTauCeti.not_isSquare_one_add_pow_odd: below depth2 v_K(2), a square cannot differ from one to exact odd order, so1 + π^(2k+1)is not a square whenk < v_K(2).TauCeti.exists_integerUnit_residue_not_isSquare: away from residue characteristic two there is a unit of𝒪[K]whose residue is a nonsquare.TauCeti.isSquare_zpow_mul_iff: forwof even valuation,π ^ m * wis a square exactly whenmis even andwis a square.TauCeti.not_isSquare_mul_of_isUniformizer_of_even_toAdd_normalizedValuation: a uniformizer times an element of even valuation is not a square.TauCeti.not_isSquare_of_isUniformizer: a uniformizer is not a square.
References #
- O. T. O'Meara, Introduction to Quadratic Forms, §63A, local square theorem.
- J. Neukirch, Algebraic Number Theory, Chapter II, §5.
- J.-P. Serre, A Course in Arithmetic, Chapter II, §3, for the case
K = ℚ₂.
The dyadic level v_K(2), expressed using the supplied natural-valued valuation.
Instances For
In characteristic different from two, the ideal (4) of 𝒪[K] is 𝓂[K] ^ (2 v_K(2)).
In characteristic different from two, the valuation of 4 is that of π ^ (2 v_K(2)), for
an irreducible element π of 𝒪[K].
Every unit of depth 2 v_K(2) + 1 is a square. This includes dyadic local fields;
only characteristic two itself is excluded.
A unit w = 1 + 4m is a square if the residue of m lies in the range of
t ↦ t² + t.
In residue characteristic two, 4 ∈ 𝓂[K], so every 1 + 4n with n ∈ 𝒪[K] is a unit.
In residue characteristic two, if a nonsquare is ζ² (1 + 4n) with n ∈ 𝒪[K], the residue of
n lies outside the range of t ↦ t² + t.
The local square theorem is sharp: not every unit of depth 2 v_K(2) is a square. In
residue characteristic two the witness is 1 + 4c for any c whose residue is not of the form
t² + t; for K = ℚ₂ this says that 5 ∈ 1 + 4ℤ₂ is not a square. In odd residue
characteristic the depth is zero and the statement is that some unit is not a square.
There is a nonsquare in the last unit-filtration step not entirely contained in the squares.
Every unit of depth 2 * dyadicLevel K + 1 is a square.
The sharp depth cannot be decreased: units at 2 * dyadicLevel K are not all squares.
The exact depth of the local square theorem: U(K, n) consists of squares if and only if
n ≥ 2 v_K(2) + 1.
Below twice the valuation of two, a square cannot differ from one to exact odd order.
For k < v_K(2), the unit represented by 1 + π^(2k+1) is not a square.
Squares in a nonarchimedean local field. Written against a uniformizer π, an element
π ^ m * w with w of even valuation is a square exactly when m is even and w is a square.
No hypothesis on the residue characteristic is needed.
A uniformizer times an element of even valuation has odd valuation, so it is not a square.
A uniformizer is not a square.
Away from residue characteristic two, there is a unit of 𝒪[K] whose residue is a
nonsquare.