Documentation

TauCeti.RingTheory.PowerSeries.GaussNorm

The Gauss norm of restricted power series #

A restricted power series has finite Gauss norm. At a positive radius, a nonzero restricted series has a last coefficient attaining that norm; the degree of that coefficient is the distinguished degree of the series, and IsDistinguished names the property. Over a nonarchimedean normed ring with multiplicative norm, a pair of distinguished degrees produces a dominant coefficient in a product, and the Gauss norm is therefore multiplicative on restricted series.

The distinguished degree is the datum Weierstrass division and preparation for Tate algebras are organised around. No completeness hypothesis is needed for the norm identities here. The radius is any positive real number, including the unit radius of the usual Tate algebra.

Completeness enters only in the summation section, where a family of restricted series with summable Gauss norms is summed coefficientwise. This is the convergence statement that successive-approximation arguments over a complete nonarchimedean ring run on, and it takes the place of completeness of the Tate algebra for the Gauss norm.

Multiplicativity and the ultrametric inequality together make the Gauss norm a valuation with values in ℝ≥0 on the ring of restricted series, gaussValuation, whose support is trivial. Pulled back to the Tate algebra at radii at most one, these valuations give the Gauss points of the closed unit disc.

Main definitions #

Main results #

References #

The dominant-coefficient argument follows Mathlib's proof of Polynomial.gaussNorm_mul; restrictedness replaces the finite-support argument for attaining the maximum. We use Mathlib's PowerSeries.IsRestricted and PowerSeries.gaussNorm throughout.

A restricted power series has bounded weighted coefficient norms.

theorem TauCeti.PowerSeries.gaussNorm_eq_of_forall_le {c : ℝ} {s : ℕ} {S : Type u_2} [Semiring S] {v : S → ℝ} {a : PowerSeries S} (h : ∀ (m : ℕ), v ((PowerSeries.coeff m) a) * c ^ m ≤ v ((PowerSeries.coeff s) a) * c ^ s) :

If the weighted coefficient in degree s dominates every other one, the Gauss norm is the value it takes there.

structure TauCeti.PowerSeries.IsDistinguished {R : Type u_1} [NormedRing R] (c : ℝ) (s : ℕ) (f : PowerSeries R) :

f is distinguished of degree s at the radius c when its Gauss norm at c is attained in degree s and every later coefficient is strictly smaller.

At the unit radius this is a norm-theoretic analogue of the classical condition that the leading coefficient of f dominates, in the sense of Bosch–Güntzer–Remmert §5.2. At a positive radius, a nonzero restricted series is distinguished of exactly one degree (TauCeti.PowerSeries.exists_isDistinguished and TauCeti.PowerSeries.IsDistinguished.unique), so this is a genuine invariant of f and c rather than extra data.

The first field is the univariate reading of Mathlib's MvPowerSeries.AchievesGaussNorm; the second is what makes the degree unique and pins down the dominant coefficient of a product.

This is unrelated to Polynomial.IsDistinguishedAt, which asks a polynomial to be monic with its remaining coefficients in an ideal.

Instances For

    A distinguished series has positive Gauss norm, without any sign assumption on the radius.

    theorem TauCeti.PowerSeries.IsDistinguished.ne_zero {R : Type u_1} [NormedRing R] {c : ℝ} {s : ℕ} {f : PowerSeries R} (hf : IsDistinguished c s f) :
    f ≠ 0

    A distinguished series is nonzero.

    The coefficient of a distinguished series in its distinguished degree is nonzero.

    The weighted coefficient norms of a distinguished series are bounded above.

    theorem TauCeti.PowerSeries.IsDistinguished.unique {R : Type u_1} [NormedRing R] {c : ℝ} {s t : ℕ} {f : PowerSeries R} (hf : IsDistinguished c s f) (hf' : IsDistinguished c t f) :
    s = t

    The distinguished degree is unique: a series cannot be distinguished of two degrees at the same radius.

    theorem TauCeti.PowerSeries.exists_isDistinguished {R : Type u_1} [NormedRing R] {c : ℝ} {f : PowerSeries R} (hc : 0 < c) (hf : PowerSeries.IsRestricted c f) (hf0 : f ≠ 0) :
    ∃ (s : ℕ), IsDistinguished c s f

    Every nonzero restricted series is distinguished of some degree: its last coefficient attaining the Gauss norm supplies that degree.

    @[simp]

    Truncating a series just past a degree in which its Gauss norm is attained leaves that norm unchanged.

    theorem TauCeti.PowerSeries.IsDistinguished.trunc {R : Type u_1} [NormedRing R] {c : ℝ} {s : ℕ} {f : PowerSeries R} (hf : IsDistinguished c s f) :

    The truncation of a distinguished series just past its distinguished degree is again distinguished of that degree. It is the polynomial part f⁻ a Weierstrass division divides by.

    The tail f⁺ left by truncating a restricted distinguished series just past its distinguished degree has strictly smaller Gauss norm than the series itself. This is the contraction factor of the Weierstrass division algorithm.

    theorem TauCeti.PowerSeries.isDistinguished_of_norm_coeff_sub_lt {R : Type u_1} [NormedRing R] {c : ℝ} {s : ℕ} {f : PowerSeries R} [IsUltrametricDist R] (hc : 0 < c) {a : R} (ha : ‖a‖ * c ^ s = PowerSeries.gaussNorm norm c f) (hs : ‖(PowerSeries.coeff s) f - a‖ * c ^ s < PowerSeries.gaussNorm norm c f) (hm : ∀ (m : ℕ), s < m → ‖(PowerSeries.coeff m) f‖ * c ^ m < PowerSeries.gaussNorm norm c f) :

    Recognising a distinguished series. At a positive radius, f is distinguished of degree s once its weighted coefficient in degree s is closer than the Gauss norm to an element a of weighted norm equal to the Gauss norm, and every later weighted coefficient norm is smaller than the Gauss norm.

    The sum of two power series with bounded weighted coefficient norms again has bounded weighted coefficient norms at a nonnegative radius.

    The product of two power series with bounded weighted coefficient norms again has bounded weighted coefficient norms at a nonnegative radius.

    theorem TauCeti.PowerSeries.summable_coeff_of_summable_gaussNorm {R : Type u_1} [NormedRing R] {c : ℝ} [CompleteSpace R] {ι : Type u_2} {a : ι → PowerSeries R} (hc : 0 < c) (ha : ∀ (k : ι), PowerSeries.HasGaussNorm norm c (a k)) (hs : Summable fun (k : ι) => PowerSeries.gaussNorm norm c (a k)) (i : ℕ) :
    Summable fun (k : ι) => (PowerSeries.coeff i) (a k)

    Over a complete ring, a family of power series with summable Gauss norms has summable coefficients in every degree.

    theorem TauCeti.PowerSeries.isRestricted_mk_tsum_coeff {R : Type u_1} [NormedRing R] {c : ℝ} [IsUltrametricDist R] [CompleteSpace R] {ι : Type u_2} {a : ι → PowerSeries R} (hc : 0 < c) (ha : ∀ (k : ι), PowerSeries.IsRestricted c (a k)) (hs : Summable fun (k : ι) => PowerSeries.gaussNorm norm c (a k)) :

    Coefficientwise summation of restricted power series. Over a complete nonarchimedean ring, the degreewise sums of a family of restricted power series with summable Gauss norms assemble into a restricted power series.

    This is the convergence statement behind successive-approximation arguments such as Weierstrass division: it plays the role of completeness of the Tate algebra for the Gauss norm.

    The dominant coefficient of a product of distinguished series. If f is distinguished of degree i and g of degree j, then the coefficient of f * g in degree i + j realises the product of the two Gauss norms.

    The Gauss norm is multiplicative on restricted power series at every positive radius.

    theorem TauCeti.PowerSeries.IsDistinguished.mul {R : Type u_1} [NormedRing R] {c : ℝ} {i j : ℕ} {f g : PowerSeries R} [IsUltrametricDist R] [NormMulClass R] (hf : IsDistinguished c i f) (hg : IsDistinguished c j g) (hc : 0 < c) :
    IsDistinguished c (i + j) (f * g)

    The product of series distinguished in degrees i and j is distinguished in degree i + j at a positive radius.

    The Gauss valuation #

    The Gauss valuation at a positive radius c: the Gauss norm f ↦ sup ‖aₙ‖ cⁿ, as a valuation with values in ℝ≥0 on the ring of power series restricted at c.

    Equations
    • One or more equations did not get rendered due to their size.
    Instances For
      @[simp]

      The Gauss valuation is the Gauss norm, read in ℝ.

      @[simp]

      The Gauss valuation of a constant series is the norm of its coefficient.

      @[simp]

      The Gauss valuation of the variable is the radius.

      @[simp]

      The Gauss valuation vanishes only at zero: its support is trivial.