Documentation

TauCeti.RingTheory.Affinoid.SupSeminorm

Point norms and the supremum seminorm of an affinoid algebra #

Let A be a K-algebra over a complete nonarchimedean field K. For a maximal ideal x of A and f : A, the value |f(x)| is the spectral norm over K of the residue of f in A ⧸ x. When A is affinoid this residue field is finite over K, and the spectral norm is the unique extension of the norm of K to it; no embedding into a chosen algebraic closure enters. The supremum seminorm |f|sup is the supremum of |f(x)| over all classical points x, with value zero for the zero algebra.

For an affinoid algebra the point norms are multiplicative, ultrametric and compatible with scalars, and they are bounded by every complete ultrametric algebra norm on A. In particular |f|sup is a finite, power-multiplicative ring seminorm, its kernel is the nilradical, algebra homomorphisms of affinoid algebras do not increase it, and integral injective ones preserve it. It is unchanged by passing to the reduction A ⧸ nilradical A, it is the maximum of its values on the quotients by the finitely many minimal primes of A, and a monic equation f ^ d + ∑_{i < d} aᵢ f ^ i = 0 bounds it by max_{i < d} |aᵢ|sup ^ (1 / (d - i)).

The bound by a complete algebra norm uses automatic continuity, TauCeti.IsAffinoidAlgebra.continuous_algHom, of the residue map into A ⧸ x with its spectral norm, followed by the power-multiplicativity of the spectral norm. The kernel computation uses that affinoid algebras are Jacobson rings.

Main definitions #

Main results #

References #

theorem AlgHom.spectralNorm_map {K : Type u_1} [NontriviallyNormedField K] {E : Type u_2} {L : Type u_3} [Field E] [Algebra K E] [Field L] [Algebra K L] (ψ : E →ₐ[K] L) (y : E) :
spectralNorm K L (ψ y) = spectralNorm K E y

The spectral norm over K is invariant under K-algebra maps of fields.

noncomputable def TauCeti.pointNorm (K : Type u_1) [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] (x : MaximalSpectrum A) (f : A) :

The absolute value |f(x)| of f : A at a maximal ideal x: the spectral norm over K of the residue of f in the field A ⧸ x. For an affinoid algebra this residue field is finite over K, and the spectral norm is the unique extension of the norm of K to it.

Instances For
    noncomputable def TauCeti.supSeminorm (K : Type u_1) [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] (f : A) :

    The supremum seminorm |f|sup of f : A: the supremum of its absolute values |f(x)| at the maximal ideals x of A. It is zero on the zero algebra, which has no maximal ideals.

    Instances For
      theorem TauCeti.pointNorm_nonneg {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {x : MaximalSpectrum A} (f : A) :
      0 ≤ pointNorm K x f
      @[simp]
      theorem TauCeti.pointNorm_algebraMap {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {x : MaximalSpectrum A} (c : K) :
      pointNorm K x ((algebraMap K A) c) = ‖c‖
      @[simp]
      theorem TauCeti.pointNorm_zero {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {x : MaximalSpectrum A} :
      pointNorm K x 0 = 0
      @[simp]
      theorem TauCeti.pointNorm_one {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {x : MaximalSpectrum A} :
      pointNorm K x 1 = 1
      theorem TauCeti.pointNorm_eq_spectralNorm {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {L : Type u_4} [Field L] [Algebra K L] {x : MaximalSpectrum A} {ψ : A →ₐ[K] L} (hψ : x.asIdeal ≤ RingHom.ker ψ) (f : A) :
      pointNorm K x f = spectralNorm K L (ψ f)

      The absolute value at a classical point can be computed in any field L over K that receives the residue field: if a K-algebra map ψ : A → L vanishes on x, then |f(x)| is the spectral norm of ψ f.

      theorem TauCeti.pointNorm_eq_of_sub_mem {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {x : MaximalSpectrum A} {f g : A} (h : f - g ∈ x.asIdeal) :
      pointNorm K x f = pointNorm K x g

      Elements congruent modulo a classical point have the same absolute value there.

      theorem TauCeti.pointNorm_comap {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {B : Type u_3} [CommRing B] [Algebra K B] (φ : A →ₐ[K] B) (x : MaximalSpectrum B) (hx : (Ideal.comap φ x.asIdeal).IsMaximal) (a : A) :
      pointNorm K { asIdeal := Ideal.comap φ x.asIdeal, isMaximal := hx } a = pointNorm K x (φ a)

      The absolute value at a classical point is unchanged by pulling the point back along an algebra map, whenever the pullback is again a maximal ideal.

      @[simp]

      The absolute value at a classical point is unchanged by pulling the point back along an algebra map into an affinoid algebra.

      The absolute value at a classical point is homogeneous for scalars of K: |(c • f)(x)| = ‖c‖ * |f(x)|.

      theorem MaximalSpectrum.pointNorm_pow {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] [IsUltrametricDist K] (x : MaximalSpectrum A) (f : A) (n : ℕ) :

      |(f ^ n)(x)| = |f(x)| ^ n at every classical point x.

      The absolute value |f(x)| vanishes exactly when f lies in the maximal ideal x.

      The absolute value at a classical point is multiplicative: |(f * g)(x)| = |f(x)| * |g(x)|.

      The absolute value at a classical point is ultrametric.

      The units of an affinoid algebra are the functions without zeros: f is a unit if and only if |f(x)| ≠ 0 at every classical point x.

      The absolute value of the inverse of a unit is the inverse absolute value: |u⁻¹(x)| = |u(x)|⁻¹.

      Point norms are bounded by complete algebra norms. On an affinoid algebra with a complete ultrametric algebra norm, |f(x)| ≤ ‖f‖ at every classical point x.

      theorem TauCeti.supSeminorm_eq_iSup {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] (f : A) :
      supSeminorm K f = ⨆ (x : MaximalSpectrum A), pointNorm K x f
      theorem TauCeti.supSeminorm_nonneg {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] (f : A) :
      theorem TauCeti.supSeminorm_le {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {f : A} {r : ℝ} (hr : 0 ≤ r) (h : ∀ (x : MaximalSpectrum A), pointNorm K x f ≤ r) :

      A nonnegative bound for all point norms bounds the supremum seminorm.

      If the absolute values of f at the classical points are bounded, then each of them is at most the supremum seminorm: |f(x)| ≤ |f|sup.

      The supremum seminorm vanishes on the zero algebra, which has no maximal ideals.

      @[simp]
      theorem TauCeti.supSeminorm_zero {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] :
      @[simp]
      theorem TauCeti.supSeminorm_algebraMap {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] [Nontrivial A] (c : K) :
      @[simp]
      theorem TauCeti.supSeminorm_one {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] [Nontrivial A] :
      theorem TauCeti.isUnit_one_sub_of_supSeminorm_lt_one {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {a : A} (hb : BddAbove (Set.range fun (x : MaximalSpectrum A) => pointNorm K x a)) (ha : supSeminorm K a < 1) :
      IsUnit (1 - a)

      Perturbations of 1 by elements of supremum seminorm < 1 are units. If the absolute values of a at the classical points are bounded and |a|sup < 1, then 1 - a is a unit.

      @[simp]
      theorem TauCeti.supSeminorm_neg {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] [IsUltrametricDist K] (f : A) :
      theorem TauCeti.supSeminorm_smul {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] [IsUltrametricDist K] (c : K) (f : A) :

      The supremum seminorm is homogeneous for scalars of K: |c • f|sup = ‖c‖ * |f|sup.

      theorem TauCeti.supSeminorm_pow {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] [IsUltrametricDist K] (f : A) {n : ℕ} (hn : n ≠ 0) :
      supSeminorm K (f ^ n) = supSeminorm K f ^ n

      |f ^ n|sup = |f|sup ^ n for n ≠ 0, including on the zero algebra. For n = 0 the statement fails on the zero algebra.

      The supremum seminorm is power-multiplicative: |f ^ n|sup = |f|sup ^ n for n ≥ 1.

      theorem TauCeti.supSeminorm_map_of_ker_le_nilradical {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {B : Type u_3} [CommRing B] [Algebra K B] {φ : A →ₐ[K] B} (hφ : Function.Surjective ⇑φ) (hker : RingHom.ker φ ≤ nilradical A) (a : A) :

      A surjective algebra map whose kernel consists of nilpotent elements preserves supremum seminorms, since every classical point of the source contains the kernel and so comes from a classical point of the target.

      The supremum seminorm is unchanged by reduction: |f|sup is the same in A and in its reduction A ⧸ nilradical A.

      Boundedness of point norms. The absolute values of an element of an affinoid algebra at its classical points are bounded.

      Each absolute value |f(x)| is bounded by the supremum seminorm |f|sup.

      theorem TauCeti.IsAffinoidAlgebra.supSeminorm_le_iff {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] [IsUltrametricDist K] [CompleteSpace K] (hA : IsAffinoidAlgebra K A) {f : A} {r : ℝ} (hr : 0 ≤ r) :
      supSeminorm K f ≤ r ↔ ∀ (x : MaximalSpectrum A), pointNorm K x f ≤ r

      The supremum seminorm is bounded by every complete ultrametric algebra norm.

      The supremum seminorm is submultiplicative: |f * g|sup ≤ |f|sup * |g|sup.

      The supremum seminorm is ultrametric.

      The supremum seminorm of an affinoid algebra vanishes exactly on the elements lying in every maximal ideal.

      The kernel of the supremum seminorm is the nilradical. An element of an affinoid algebra has supremum seminorm zero exactly when it is nilpotent, since an affinoid algebra is a Jacobson ring.

      The supremum seminorm is a norm exactly on reduced affinoid algebras.

      theorem TauCeti.IsAffinoidAlgebra.supSeminorm_map_le {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {B : Type u_3} [CommRing B] [Algebra K B] [IsUltrametricDist K] [CompleteSpace K] (hA : IsAffinoidAlgebra K A) (hB : IsAffinoidAlgebra K B) (φ : A →ₐ[K] B) (a : A) :

      An algebra homomorphism of affinoid algebras does not increase supremum seminorms.

      theorem TauCeti.IsAffinoidAlgebra.supSeminorm_map_of_isIntegral {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {B : Type u_3} [CommRing B] [Algebra K B] [IsUltrametricDist K] [CompleteSpace K] (hB : IsAffinoidAlgebra K B) (φ : A →ₐ[K] B) (hφ : φ.IsIntegral) (hφi : Function.Injective ⇑φ) (a : A) :

      An integral injective algebra homomorphism into an affinoid algebra preserves supremum seminorms, since every classical point of the source lifts to the target.

      theorem TauCeti.IsAffinoidAlgebra.pointNorm_le_supSeminorm_map {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] {B : Type u_3} [CommRing B] [Algebra K B] [IsUltrametricDist K] [CompleteSpace K] (hB : IsAffinoidAlgebra K B) {φ : A →ₐ[K] B} (hφ : Function.Surjective ⇑φ) {x : MaximalSpectrum A} (hx : RingHom.ker φ ≤ x.asIdeal) (a : A) :
      pointNorm K x a ≤ supSeminorm K (φ a)

      At a classical point x of A containing the kernel of a surjective algebra map φ : A → B onto an affinoid algebra, |a(x)| ≤ |φ a|sup: the point x comes from a classical point of B.

      The supremum seminorm is computed on minimal-prime quotients. On an affinoid algebra, |f|sup is the supremum, over the finitely many minimal primes 𝔭 of A, of the supremum seminorm of the residue of f in the affinoid domain A ⧸ 𝔭.

      On a nonzero affinoid algebra the supremum seminorm |f|sup is attained on the quotient by one of its minimal primes.

      The supremum seminorm of an affinoid algebra, bundled as a ring seminorm.

      Instances For
        theorem TauCeti.IsAffinoidAlgebra.supSeminorm_le_iSup_coeff_rpow_of_eval_eq_zero {K : Type u_1} [NontriviallyNormedField K] {A : Type u_2} [CommRing A] [Algebra K A] [IsUltrametricDist K] [CompleteSpace K] (hA : IsAffinoidAlgebra K A) {p : Polynomial A} (hp : p.Monic) {f : A} (hf : Polynomial.eval f p = 0) :
        supSeminorm K f ≤ ⨆ (i : Fin p.natDegree), supSeminorm K (p.coeff ↑i) ^ (1 / (↑p.natDegree - ↑↑i))

        Integral equations bound the supremum seminorm. If f is a root of a monic polynomial p = X ^ d + ∑_{i < d} aᵢ X ^ i over an affinoid algebra, then |f|sup ≤ max_{i < d} |aᵢ|sup ^ (1 / (d - i)).