math / data /lean /OAI /Geometry /ComplexTorus /RiemannProjectivity /CoefficientRoot.lean
introvoyz045's picture
Migrated from GitHub
4a56cd6 verified
Raw History Blame Contribute Delete
8.68 kB
import OAI.Geometry.ComplexTorus.RiemannProjectivity.BranchCoefficientsAt
import Mathlib.Analysis.Polynomial.MahlerMeasure
/-! # Coefficient Root -/
namespace OAI
noncomputable section
namespace Geometry.ComplexTorus.RiemannProjectivity.CoefficientRoot.Constructions.CoefficientGrowth
/-!
# RiemannProjectivity: growth bounds for projection coefficients
The proper-cone norm estimate bounds every coordinate on a fiber. The
product polynomial of a finite fiber therefore has polynomially bounded
coefficients, with an explicit constant and degree. Repeated coordinate
values remain separate factors, as required for symmetric coefficients.
-/
open Set _root_.Polynomial
open scoped BigOperators
theorem norm_coeff_prod_X_sub_C_le {ΞΉ : Type*} [Fintype ΞΉ]
(a : ΞΉ β†’ β„‚) (R : ℝ) (ha : βˆ€ i, β€–a iβ€– ≀ R) (j : β„•) :
β€–(∏ i, (Polynomial.X - Polynomial.C (a i))).coeff jβ€– ≀
(Fintype.card ΞΉ).choose j * (max 1 R) ^ Fintype.card ΞΉ := by
classical
have hm : βˆ€ s : Finset ΞΉ,
(∏ i ∈ s, (Polynomial.X - Polynomial.C (a i))).mahlerMeasure =
∏ i ∈ s, max 1 β€–a iβ€– := by
intro s
induction s using Finset.induction_on with
| empty => simp
| @insert i s hi ih => simp [Finset.prod_insert hi, Polynomial.mahlerMeasure_mul, ih,
Polynomial.mahlerMeasure_X_sub_C]
have hprod : (∏ i, max 1 β€–a iβ€–) ≀ (max 1 R) ^ Fintype.card ΞΉ := by
calc
_ ≀ ∏ _ : ΞΉ, max 1 R := Finset.prod_le_prodβ‚€
(fun _ _ => le_trans zero_le_one (le_max_left _ _))
(fun i _ => max_le_max_left 1 (ha i))
_ = _ := by simp
calc
_ ≀ (∏ i, (Polynomial.X - Polynomial.C (a i))).natDegree.choose j *
(∏ i, (Polynomial.X - Polynomial.C (a i))).mahlerMeasure :=
Polynomial.norm_coeff_le_choose_mul_mahlerMeasure _ _
_ = (Fintype.card ΞΉ).choose j * ∏ i, max 1 β€–a iβ€– := by simp only [hm,
Polynomial.natDegree_finsetProd_X_sub_C_eq_card, Finset.card_univ]
_ ≀ _ := mul_le_mul_of_nonneg_left hprod (Nat.cast_nonneg _)
variable {V W : Type*} [NormedAddCommGroup V] [NormedSpace β„‚ V]
[NormedAddCommGroup W]
/-- The linear bound on a fiber gives the polynomial-growth hypothesis
needed for extension of each symmetric coefficient. -/
theorem coefficient_growth_from_fiber_bound {m : β„•} (b : Fin m β†’ W β†’ V)
(l : V β†’L[β„‚] β„‚) (C : ℝ) {U : Set W}
(hb : βˆ€ z ∈ U, βˆ€ i, β€–b i zβ€– ≀ C * β€–zβ€–) (j : β„•) :
βˆ€ z ∈ U,
β€–(∏ i, (Polynomial.X - Polynomial.C (l (b i z)))).coeff jβ€– ≀
((m.choose j : ℝ) * (max 1 (β€–lβ€– * C)) ^ m) * (max 1 β€–zβ€–) ^ m := by
intro z hz
have ha : βˆ€ i, β€–l (b i z)β€– ≀ β€–lβ€– * C * β€–zβ€– := fun i =>
(l.le_opNorm _).trans (by
simpa only [mul_assoc] using mul_le_mul_of_nonneg_left (hb z hz i) (norm_nonneg l))
have hmax : max 1 (β€–lβ€– * C * β€–zβ€–) ≀ max 1 (β€–lβ€– * C) * max 1 β€–zβ€– := by
apply max_le
Β· exact one_le_mul_of_one_le_of_one_le (le_max_left _ _) (le_max_left _ _)
Β· exact mul_le_mul (le_max_right _ _) (le_max_right _ _) (norm_nonneg z)
(le_trans zero_le_one (le_max_left _ _))
calc
_ ≀ (m.choose j : ℝ) * (max 1 (β€–lβ€– * C * β€–zβ€–)) ^ m := by
simpa only [Fintype.card_fin] using norm_coeff_prod_X_sub_C_le
(fun i => l (b i z)) (β€–lβ€– * C * β€–zβ€–) ha j
_ ≀ (m.choose j : ℝ) * (max 1 (β€–lβ€– * C) * max 1 β€–zβ€–) ^ m := by gcongr
_ = _ := by rw [mul_pow, mul_assoc]
variable [NormedSpace β„‚ W] {X : Type*} [TopologicalSpace X] [T2Space X]
/-- Combine genuine local inverse branches with a uniform fiber bound.
The same local polynomial has both holomorphic coefficients and the
quantitative growth estimate. -/
theorem local_bounded_fiber_polynomial (a : X β†’ V) {f : X β†’ W} (hf : IsProperMap f)
(C : ℝ) (hbound : βˆ€ x, β€–a xβ€– ≀ C * β€–f xβ€–)
{y : W} (hy : y ∈ BranchCoefficientsAt.Constructions.BranchCoefficients.holomorphicRegularValues a f)
(l : V β†’L[β„‚] β„‚) :
βˆƒ (m : β„•) (U : Set W) (P : W β†’ Polynomial β„‚), IsOpen U ∧ y ∈ U ∧
(βˆ€ j, DifferentiableOn β„‚ (fun z => (P z).coeff j) U) ∧
(βˆ€ z ∈ U, βˆ€ j, β€–(P z).coeff jβ€– ≀
((m.choose j : ℝ) * (max 1 (β€–lβ€– * C)) ^ m) * (max 1 β€–zβ€–) ^ m) ∧
βˆ€ z ∈ U, (P z).Monic ∧
βˆ€ c, (P z).eval c = 0 ↔ βˆƒ x, f x = z ∧ l (a x) = c := by
obtain ⟨m, U, b, hU, hyU, hb, hbij⟩ := BranchCoefficientsAt.Constructions.BranchCoefficients.finite_holomorphic_branches a hf hy
have hp := BranchCoefficientsAt.Constructions.BranchCoefficients.polynomial_of_holomorphic_branches a f b hb
(fun z hz => (hbij z hz).2) l
refine ⟨m, U, _, hU, hyU, hp.1, ?_, hp.2⟩
intro z hz j
apply coefficient_growth_from_fiber_bound (fun i w => a (b i w)) l C ?_ j z hz
intro w hw i
have hfw : f (b i w) = w := ((hbij w hw).2 _).mpr ⟨i, rfl⟩
simpa only [hfw] using hbound (b i w)
end Geometry.ComplexTorus.RiemannProjectivity.CoefficientRoot.Constructions.CoefficientGrowth
namespace Geometry.ComplexTorus.RiemannProjectivity.CoefficientRoot.Constructions.RootCoefficientBound
/-! # RiemannProjectivity: coefficient bounds from bounded roots, including multiplicities -/
open _root_.Polynomial
open scoped BigOperators
theorem prod_max_norm_le (s : Multiset β„‚) (R : ℝ) (h : βˆ€ z ∈ s, β€–zβ€– ≀ R) :
(s.map (fun z => max 1 β€–zβ€–)).prod ≀ (max 1 R) ^ s.card := by
induction s using Multiset.induction_on with
| empty => simp
| @cons z s ih =>
have hz : β€–zβ€– ≀ R := h z (Multiset.mem_cons_self _ _)
have hs : βˆ€ w ∈ s, β€–wβ€– ≀ R := fun w hw => h w (Multiset.mem_cons_of_mem hw)
have hn : 0 ≀ (s.map (fun w => max 1 β€–wβ€–)).prod := by
apply Multiset.prod_nonneg
intro a ha
obtain ⟨w, _, rfl⟩ := Multiset.mem_map.mp ha
exact zero_le_one.trans (le_max_left _ _)
simp only [Multiset.map_cons, Multiset.prod_cons, Multiset.card_cons, pow_succ]
calc
_ ≀ max 1 R * (max 1 R) ^ s.card :=
mul_le_mul (max_le_max_left 1 hz) (ih hs) hn (zero_le_one.trans (le_max_left _ _))
_ = _ := mul_comm _ _
theorem norm_coeff_le {p : Polynomial β„‚} (hp : p.Monic) {R : ℝ}
(hroot : βˆ€ z : β„‚, p.eval z = 0 β†’ β€–zβ€– ≀ R) (j : β„•) :
β€–p.coeff jβ€– ≀ (p.natDegree.choose j : ℝ) * (max 1 R) ^ p.natDegree := by
have hroots : βˆ€ z ∈ p.roots, β€–zβ€– ≀ R := fun z hz => hroot z ((mem_roots hp.ne_zero).mp hz)
have hb := prod_max_norm_le p.roots R hroots
rw [← (IsAlgClosed.splits p).natDegree_eq_card_roots] at hb
have hm : p.mahlerMeasure ≀ (max 1 R) ^ p.natDegree := by
rw [mahlerMeasure_eq_leadingCoeff_mul_prod_roots, hp.leadingCoeff, norm_one, one_mul]
exact hb
exact (norm_coeff_le_choose_mul_mahlerMeasure j p).trans
(mul_le_mul_of_nonneg_left hm (Nat.cast_nonneg _))
end Geometry.ComplexTorus.RiemannProjectivity.CoefficientRoot.Constructions.RootCoefficientBound
namespace Geometry.ComplexTorus.RiemannProjectivity.CoefficientRoot.Constructions.MonicFamily
/-! # RiemannProjectivity: reconstructing a monic polynomial from its lower coefficients -/
open _root_.Polynomial
open scoped BigOperators
def ofCoefficients (k : β„•) (a : β„• β†’ β„‚) : β„‚[X] :=
X ^ k + βˆ‘ j ∈ Finset.range k, monomial j (a j)
theorem coeff_ofCoefficients (k j : β„•) (a : β„• β†’ β„‚) :
(ofCoefficients k a).coeff j = (if j = k then 1 else 0) +
(if j < k then a j else 0) := by
classical
simp [ofCoefficients, coeff_monomial, coeff_X_pow, eq_comm]
theorem monic_ofCoefficients (k : β„•) (a : β„• β†’ β„‚) :
(ofCoefficients k a).Monic ∧ (ofCoefficients k a).natDegree = k := by
have hd : (ofCoefficients k a).natDegree ≀ k := by
apply natDegree_le_iff_coeff_eq_zero.mpr
intro j hj
simp [coeff_ofCoefficients, ne_of_gt hj, not_lt_of_ge (Nat.le_of_lt hj)]
have hc : (ofCoefficients k a).coeff k = 1 := by simp [coeff_ofCoefficients]
exact ⟨monic_of_natDegree_le_of_coeff_eq_one k hd hc,
le_antisymm hd (le_natDegree_of_ne_zero (by rw [hc]; exact one_ne_zero))⟩
theorem ofCoefficients_eq {p : β„‚[X]} (hp : p.Monic) {k : β„•} (hd : p.natDegree = k)
{a : β„• β†’ β„‚} (ha : βˆ€ j < k, a j = p.coeff j) : ofCoefficients k a = p := by
ext j
rw [coeff_ofCoefficients]
by_cases hj : j < k
Β· simp [hj, ne_of_lt hj, ha j hj]
Β· by_cases he : j = k
Β· subst j
simp [← hd, hp.coeff_natDegree]
Β· have hk : p.natDegree < j := by omega
simp [he, hj, coeff_eq_zero_of_natDegree_lt hk]
end Geometry.ComplexTorus.RiemannProjectivity.CoefficientRoot.Constructions.MonicFamily
end
end OAI