cyclic_orbit_poly_subspace_fractional_linear_scale
Given fields B and M with an algebra structure of M over B, a natural number r with 2 ≤ r, elements x, y ∈ M such that (minpoly B x).natDegree ≥ 2 * r - 1, elements a, b, c, d ∈ B with a * d - b * c ≠ 0, and y = (algebraMap B M a * x + algebraMap B M b) / (algebraMap B M c * x + algebraMap B M d), the pointwise scalar multiple (algebraMap B M c * x + algebraMap B M d) ^ (r - 1) • cyclic_orbit_poly_subspace B r y equals cyclic_orbit_poly_subspace B r x.
Statement
open Pointwise in theorem cyclic_orbit_poly_subspace_fractional_linear_scale {B M : Type*} [Field B] [Field M] [Algebra B M] {r : ℕ} (hr : 2 ≤ r) {x y : M} (hx : 2 * r - 1 ≤ (minpoly B x).natDegree) (a b c d : B) (hdet : a * d - b * c ≠ 0) (hy : y = (algebraMap B M a * x + algebraMap B M b) / (algebraMap B M c * x + algebraMap B M d)) : (algebraMap B M c * x + algebraMap B M d) ^ (r - 1) • cyclic_orbit_poly_subspace B r y = cyclic_orbit_poly_subspace B r x
- Source
- arXiv:2611.04947 Lemma 3.2, eq (5)
- Verified
- 05 Nov 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 3
- cyclic_orbit_poly_subspaceGiven a field B, a field M which is an algebra over B, a natural number r, and an element x ∈ M, cyclic_orbit_poly_subspace B r x is the…
- cyclic_orbit_denom_ne_zeroIf x ∈ M has minimal polynomial degree at least 2 over B and a, b, c, d ∈ B satisfy ad - bc ≠ 0, then cx + d ≠ 0 in M.
- cyclic_orbit_poly_subspace_pow_memGiven a field B, a field M which is an algebra over B, a natural number r, an element x ∈ M, and an index i : Fin r, x^(i : ℕ) is in…
Read back from the Lean
Let B and M be fields such that M is an algebra over B. Let r ≥ 2 be a natural number, and let x, y ∈ M. Assume that the degree of the minimal polynomial of x over B is at least 2 * r - 1. Let a, b, c, d ∈ B satisfy a * d - b * c ≠ 0, and assume y = (a * x + b) / (c * x + d) (where a, b, c, d are mapped into M via algebraMap). Then the pointwise scalar multiplication of the B-submodule cyclic_orbit_poly_subspace B r y by (c * x + d)^(r - 1) equals cyclic_orbit_poly_subspace B r x.
Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.
The proof, explained
To show that the fractional linear scaling of the cyclic orbit subspace of y equals that of x, we establish both inclusions. The forward inclusion follows by clearing denominators on each basis element y^i and observing that the resulting element is the evaluation of a polynomial in x of degree at most r - 1. The reverse inclusion is obtained symmetrically using the inverse fractional linear transformation, and the two inclusions are combined using the determinant relation to yield equality.
bound the degree of the numerator and denominator factors
The degree of a linear polynomial is at most 1, so the product has degree at most i + (r - 1 - i) = r - 1 < r.
uses
natDegree_mul_lehave h1 : natDegree (C a * X + C b : B[X]) ≤ 1 := by refine le_trans (natDegree_add_le _ _) ?_ simp only [natDegree_C, max_le_iff] refine ⟨?_, ?_⟩ · refine le_trans (natDegree_mul_le) ?_ simp [natDegree_C, natDegree_X] · omega have h2 : natDegree (C c * X + C d : B[X]) ≤ 1 := by refine le_trans (natDegree_add_le _ _) ?_ simp only [natDegree_C, max_le_iff] refine ⟨?_, ?_⟩ · refine le_trans (natDegree_mul_le) ?_ simp [natDegree_C, natDegree_X] · omega refine lt_of_le_of_lt (natDegree_mul_le) ?_ refine lt_of_le_of_lt (add_le_add (natDegree_pow_le.trans (Nat.mul_le_mul_left i h1)) (natDegree_pow_le.trans (Nat.mul_le_mul_left (r - 1 - i) h2))) ?_ simp only [mul_one] omega
evaluation of any polynomial of degree less than r lies in the subspace
Expanding the polynomial in the monomial basis shows its evaluation is a linear combination of powers x^i for i < r.
rw [p.as_sum_range_C_mul_X_pow' hp] rw [map_sum] refine Submodule.sum_mem _ ?_ intro i hi rw [Finset.mem_range] at hi rw [map_mul, aeval_C, aeval_X_pow] rw [← Algebra.smul_def] exact Submodule.smul_mem _ _ (cyclic_orbit_poly_subspace_pow_mem r x ⟨i, hi⟩)
algebraic identity clearing denominators
Multiplying g^(r-1) by (f/g)^i cancels g^i to leave f^i * g^(r-1-i).
have : g ^ (r - 1) = g ^ (r - 1 - i) * g ^ i := by rw [← pow_add, Nat.sub_add_cancel (by omega)] rw [div_pow, this] have hgi : g ^ i ≠ 0 := pow_ne_zero i hg rw [mul_assoc, mul_div_cancel₀ _ hgi] ring
forward inclusion of the scaled subspace
Every basis element scaled by the denominator factor is the evaluation of a polynomial of degree < r, so it lies in the subspace of x.
dsimp [cyclic_orbit_poly_subspace, Submodule.pointwise_smul_def] rw [Submodule.map_span] rw [Submodule.span_le] rintro _ ⟨_, ⟨i, rfl⟩, rfl⟩ dsimp [DistribSMul.toLinearMap] rw [hy] rw [cyclic_orbit_poly_subspace_fractional_linear_scale_pow_sub_mul_div _ _ hcx i.1 r i.2] have hp : aeval x ((C a * X + C b) ^ (i : ℕ) * (C c * X + C d) ^ (r - 1 - i) : B[X]) = (algebraMap B M a * x + algebraMap B M b) ^ (i : ℕ) * (algebraMap B M c * x + algebraMap B M d) ^ (r - 1 - i) := by simp only [map_mul, map_pow, map_add, aeval_C, aeval_X] rw [← hp] exact cyclic_orbit_poly_subspace_fractional_linear_scale_aeval_mem r x _ (cyclic_orbit_poly_subspace_fractional_linear_scale_poly_deg a b c d i.1 r i.2)
properties of the inverse fractional linear transformation
The denominator of the inverse transformation is non-zero, x is recovered by the inverse matrix, and the product of the two denominators equals the determinant.
have hy' : y * (algebraMap B M c * x + algebraMap B M d) = algebraMap B M a * x + algebraMap B M b := by rw [hy, div_mul_cancel₀ _ hcx] have h_denom : (algebraMap B M (-c) * y + algebraMap B M a) * (algebraMap B M c * x + algebraMap B M d) = algebraMap B M (a * d - b * c) := by simp only [map_neg, map_sub, map_mul] calc (-algebraMap B M c * y + algebraMap B M a) * (algebraMap B M c * x + algebraMap B M d) _ = -algebraMap B M c * (y * (algebraMap B M c * x + algebraMap B M d)) + algebraMap B M a * (algebraMap B M c * x + algebraMap B M d) := by ring _ = -algebraMap B M c * (algebraMap B M a * x + algebraMap B M b) + algebraMap B M a * (algebraMap B M c * x + algebraMap B M d) := by rw [hy'] _ = algebraMap B M a * algebraMap B M d - algebraMap B M b * algebraMap B M c := by ring have hdet_M : algebraMap B M (a * d - b * c) ≠ 0 := by intro h0 have : a * d - b * c = 0 := (injective_iff_map_eq_zero (algebraMap B M)).mp (RingHom.injective _) _ h0 exact hdet this have hdenom_ne : algebraMap B M (-c) * y + algebraMap B M a ≠ 0 := by intro h0 have : algebraMap B M (a * d - b * c) = 0 := by rw [← h_denom, h0, zero_mul] exact hdet_M this refine ⟨hdenom_ne, ?_, h_denom⟩ have h_num : (algebraMap B M d * y + algebraMap B M (-b)) * (algebraMap B M c * x + algebraMap B M d) = algebraMap B M (a * d - b * c) * x := by simp only [map_neg, map_sub, map_mul] calc (algebraMap B M d * y + -algebraMap B M b) * (algebraMap B M c * x + algebraMap B M d) _ = algebraMap B M d * (y * (algebraMap B M c * x + algebraMap B M d)) - algebraMap B M b * (algebraMap B M c * x + algebraMap B M d) := by ring _ = algebraMap B M d * (algebraMap B M a * x + algebraMap B M b) - algebraMap B M b * (algebraMap B M c * x + algebraMap B M d) := by rw [hy'] _ = (algebraMap B M a * algebraMap B M d - algebraMap B M b * algebraMap B M c) * x := by ring have : (algebraMap B M d * y + algebraMap B M (-b)) / (algebraMap B M (-c) * y + algebraMap B M a) = ((algebraMap B M d * y + algebraMap B M (-b)) * (algebraMap B M c * x + algebraMap B M d)) / ((algebraMap B M (-c) * y + algebraMap B M a) * (algebraMap B M c * x + algebraMap B M d)) := by rw [mul_div_mul_right _ _ hcx] rw [this, h_denom, h_num, mul_div_cancel_left₀ _ hdet_M]
scaling by a non-zero scalar in B preserves the subspace
Since the determinant is non-zero in B, multiplying by its power is an invertible B-linear map on the subspace, preserving it.
have hdeg : 2 ≤ (minpoly B x).natDegree := by omega have hcx : algebraMap B M c * x + algebraMap B M d ≠ 0 := cyclic_orbit_denom_ne_zero hdeg a b c d hdet have hle1 := cyclic_orbit_poly_subspace_fractional_linear_scale_le r x y a b c d hcx hy obtain ⟨hdenom_ne, hy_inv, hgh⟩ := cyclic_orbit_poly_subspace_fractional_linear_scale_inv a b c d x y hdet hcx hy have hle2 := cyclic_orbit_poly_subspace_fractional_linear_scale_le r y x d (-b) (-c) a hdenom_ne hy_inv have hu : (a * d - b * c) ^ (r - 1) ≠ 0 := pow_ne_zero (r - 1) hdet have h_scalar : (algebraMap B M ((a * d - b * c) ^ (r - 1))) • cyclic_orbit_poly_subspace B r x = cyclic_orbit_poly_subspace B r x := by ext z constructor · rintro ⟨w, hw, rfl⟩ change (algebraMap B M ((a * d - b * c) ^ (r - 1))) * w ∈ cyclic_orbit_poly_subspace B r x rw [← Algebra.smul_def] exact Submodule.smul_mem _ _ hw · intro hz refine ⟨(algebraMap B M ((a * d - b * c) ^ (r - 1))⁻¹) * z, ?_, ?_⟩ · rw [← Algebra.smul_def] exact Submodule.smul_mem _ _ hz · change (algebraMap B M ((a * d - b * c) ^ (r - 1))) * ((algebraMap B M ((a * d - b * c) ^ (r - 1))⁻¹) * z) = z rw [← mul_assoc, ← map_mul, mul_inv_cancel₀ hu, map_one, one_mul]
subspace equality by two-sided inclusion
The product of denominators equals the scalar, so combining both one-sided inclusions and antisymmetry yields equality.
have h_comm : (algebraMap B M c * x + algebraMap B M d) * (algebraMap B M (-c) * y + algebraMap B M a) = algebraMap B M (a * d - b * c) := by rw [mul_comm, hgh] have h_prod : ((algebraMap B M c * x + algebraMap B M d) * (algebraMap B M (-c) * y + algebraMap B M a)) ^ (r - 1) = algebraMap B M ((a * d - b * c) ^ (r - 1)) := by rw [h_comm, map_pow] rw [mul_pow] at h_prod have h_assoc : ((algebraMap B M c * x + algebraMap B M d) ^ (r - 1) * (algebraMap B M (-c) * y + algebraMap B M a) ^ (r - 1)) • cyclic_orbit_poly_subspace B r x = (algebraMap B M c * x + algebraMap B M d) ^ (r - 1) • ((algebraMap B M (-c) * y + algebraMap B M a) ^ (r - 1) • cyclic_orbit_poly_subspace B r x) := by rw [mul_smul] have hle : cyclic_orbit_poly_subspace B r x ≤ (algebraMap B M c * x + algebraMap B M d) ^ (r - 1) • cyclic_orbit_poly_subspace B r y := by rw [← h_scalar, ← h_prod, h_assoc] exact smul_le_smul_left _ hle2 exact le_antisymm hle1 hle
Polynomial.natDegree_mul_le- degree of a product is at most sum of degrees
Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.
Lean source view module on GitHub
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
open Pointwise Polynomiallemma cyclic_orbit_poly_subspace_fractional_linear_scale_poly_deg {B : Type*} [Field B] (a b c d : B) (i : ℕ) (r : ℕ) (hi : i < r) : natDegree ((C a * X + C b) ^ i * (C c * X + C d) ^ (r - 1 - i) : B[X]) < r := by have h1 : natDegree (C a * X + C b : B[X]) ≤ 1 := by refine le_trans (natDegree_add_le _ _) ?_ simp only [natDegree_C, max_le_iff] refine ⟨?_, ?_⟩ · refine le_trans (natDegree_mul_le) ?_ simp [natDegree_C, natDegree_X] · omega have h2 : natDegree (C c * X + C d : B[X]) ≤ 1 := by refine le_trans (natDegree_add_le _ _) ?_ simp only [natDegree_C, max_le_iff] refine ⟨?_, ?_⟩ · refine le_trans (natDegree_mul_le) ?_ simp [natDegree_C, natDegree_X] · omega refine lt_of_le_of_lt (natDegree_mul_le) ?_ refine lt_of_le_of_lt (add_le_add (natDegree_pow_le.trans (Nat.mul_le_mul_left i h1)) (natDegree_pow_le.trans (Nat.mul_le_mul_left (r - 1 - i) h2))) ?_ simp only [mul_one] omegalemma cyclic_orbit_poly_subspace_fractional_linear_scale_aeval_mem {B M : Type*} [Field B] [Field M] [Algebra B M] (r : ℕ) (x : M) (p : B[X]) (hp : p.natDegree < r) : aeval x p ∈ cyclic_orbit_poly_subspace B r x := by rw [p.as_sum_range_C_mul_X_pow' hp] rw [map_sum] refine Submodule.sum_mem _ ?_ intro i hi rw [Finset.mem_range] at hi rw [map_mul, aeval_C, aeval_X_pow] rw [← Algebra.smul_def] exact Submodule.smul_mem _ _ (cyclic_orbit_poly_subspace_pow_mem r x ⟨i, hi⟩)lemma cyclic_orbit_poly_subspace_fractional_linear_scale_pow_sub_mul_div {M : Type*} [Field M] (g f : M) (hg : g ≠ 0) (i r : ℕ) (hi : i < r) : g ^ (r - 1) * (f / g) ^ i = f ^ i * g ^ (r - 1 - i) := by have : g ^ (r - 1) = g ^ (r - 1 - i) * g ^ i := by rw [← pow_add, Nat.sub_add_cancel (by omega)] rw [div_pow, this] have hgi : g ^ i ≠ 0 := pow_ne_zero i hg rw [mul_assoc, mul_div_cancel₀ _ hgi] ringlemma cyclic_orbit_poly_subspace_fractional_linear_scale_le {B M : Type*} [Field B] [Field M] [Algebra B M] (r : ℕ) (x y : M) (a b c d : B) (hcx : algebraMap B M c * x + algebraMap B M d ≠ 0) (hy : y = (algebraMap B M a * x + algebraMap B M b) / (algebraMap B M c * x + algebraMap B M d)) : (algebraMap B M c * x + algebraMap B M d) ^ (r - 1) • cyclic_orbit_poly_subspace B r y ≤ cyclic_orbit_poly_subspace B r x := by dsimp [cyclic_orbit_poly_subspace, Submodule.pointwise_smul_def] rw [Submodule.map_span] rw [Submodule.span_le] rintro _ ⟨_, ⟨i, rfl⟩, rfl⟩ dsimp [DistribSMul.toLinearMap] rw [hy] rw [cyclic_orbit_poly_subspace_fractional_linear_scale_pow_sub_mul_div _ _ hcx i.1 r i.2] have hp : aeval x ((C a * X + C b) ^ (i : ℕ) * (C c * X + C d) ^ (r - 1 - i) : B[X]) = (algebraMap B M a * x + algebraMap B M b) ^ (i : ℕ) * (algebraMap B M c * x + algebraMap B M d) ^ (r - 1 - i) := by simp only [map_mul, map_pow, map_add, aeval_C, aeval_X] rw [← hp] exact cyclic_orbit_poly_subspace_fractional_linear_scale_aeval_mem r x _ (cyclic_orbit_poly_subspace_fractional_linear_scale_poly_deg a b c d i.1 r i.2)lemma cyclic_orbit_poly_subspace_fractional_linear_scale_inv {B M : Type*} [Field B] [Field M] [Algebra B M] (a b c d : B) (x y : M) (hdet : a * d - b * c ≠ 0) (hcx : algebraMap B M c * x + algebraMap B M d ≠ 0) (hy : y = (algebraMap B M a * x + algebraMap B M b) / (algebraMap B M c * x + algebraMap B M d)) : algebraMap B M (-c) * y + algebraMap B M a ≠ 0 ∧ x = (algebraMap B M d * y + algebraMap B M (-b)) / (algebraMap B M (-c) * y + algebraMap B M a) ∧ (algebraMap B M (-c) * y + algebraMap B M a) * (algebraMap B M c * x + algebraMap B M d) = algebraMap B M (a * d - b * c) := by have hy' : y * (algebraMap B M c * x + algebraMap B M d) = algebraMap B M a * x + algebraMap B M b := by rw [hy, div_mul_cancel₀ _ hcx] have h_denom : (algebraMap B M (-c) * y + algebraMap B M a) * (algebraMap B M c * x + algebraMap B M d) = algebraMap B M (a * d - b * c) := by simp only [map_neg, map_sub, map_mul] calc (-algebraMap B M c * y + algebraMap B M a) * (algebraMap B M c * x + algebraMap B M d) _ = -algebraMap B M c * (y * (algebraMap B M c * x + algebraMap B M d)) + algebraMap B M a * (algebraMap B M c * x + algebraMap B M d) := by ring _ = -algebraMap B M c * (algebraMap B M a * x + algebraMap B M b) + algebraMap B M a * (algebraMap B M c * x + algebraMap B M d) := by rw [hy'] _ = algebraMap B M a * algebraMap B M d - algebraMap B M b * algebraMap B M c := by ring have hdet_M : algebraMap B M (a * d - b * c) ≠ 0 := by intro h0 have : a * d - b * c = 0 := (injective_iff_map_eq_zero (algebraMap B M)).mp (RingHom.injective _) _ h0 exact hdet this have hdenom_ne : algebraMap B M (-c) * y + algebraMap B M a ≠ 0 := by intro h0 have : algebraMap B M (a * d - b * c) = 0 := by rw [← h_denom, h0, zero_mul] exact hdet_M this refine ⟨hdenom_ne, ?_, h_denom⟩ have h_num : (algebraMap B M d * y + algebraMap B M (-b)) * (algebraMap B M c * x + algebraMap B M d) = algebraMap B M (a * d - b * c) * x := by simp only [map_neg, map_sub, map_mul] calc (algebraMap B M d * y + -algebraMap B M b) * (algebraMap B M c * x + algebraMap B M d) _ = algebraMap B M d * (y * (algebraMap B M c * x + algebraMap B M d)) - algebraMap B M b * (algebraMap B M c * x + algebraMap B M d) := by ring _ = algebraMap B M d * (algebraMap B M a * x + algebraMap B M b) - algebraMap B M b * (algebraMap B M c * x + algebraMap B M d) := by rw [hy'] _ = (algebraMap B M a * algebraMap B M d - algebraMap B M b * algebraMap B M c) * x := by ring have : (algebraMap B M d * y + algebraMap B M (-b)) / (algebraMap B M (-c) * y + algebraMap B M a) = ((algebraMap B M d * y + algebraMap B M (-b)) * (algebraMap B M c * x + algebraMap B M d)) / ((algebraMap B M (-c) * y + algebraMap B M a) * (algebraMap B M c * x + algebraMap B M d)) := by rw [mul_div_mul_right _ _ hcx] rw [this, h_denom, h_num, mul_div_cancel_left₀ _ hdet_M]open Pointwise in/-- Fractional linear scaling: (cx+d)^(r-1) • P_r(y) = P_r(x). -/theorem cyclic_orbit_poly_subspace_fractional_linear_scale {B M : Type*} [Field B] [Field M] [Algebra B M] {r : ℕ} (hr : 2 ≤ r) {x y : M} (hx : 2 * r - 1 ≤ (minpoly B x).natDegree) (a b c d : B) (hdet : a * d - b * c ≠ 0) (hy : y = (algebraMap B M a * x + algebraMap B M b) / (algebraMap B M c * x + algebraMap B M d)) : (algebraMap B M c * x + algebraMap B M d) ^ (r - 1) • cyclic_orbit_poly_subspace B r y = cyclic_orbit_poly_subspace B r x := by have hdeg : 2 ≤ (minpoly B x).natDegree := by omega have hcx : algebraMap B M c * x + algebraMap B M d ≠ 0 := cyclic_orbit_denom_ne_zero hdeg a b c d hdet have hle1 := cyclic_orbit_poly_subspace_fractional_linear_scale_le r x y a b c d hcx hy obtain ⟨hdenom_ne, hy_inv, hgh⟩ := cyclic_orbit_poly_subspace_fractional_linear_scale_inv a b c d x y hdet hcx hy have hle2 := cyclic_orbit_poly_subspace_fractional_linear_scale_le r y x d (-b) (-c) a hdenom_ne hy_inv have hu : (a * d - b * c) ^ (r - 1) ≠ 0 := pow_ne_zero (r - 1) hdet have h_scalar : (algebraMap B M ((a * d - b * c) ^ (r - 1))) • cyclic_orbit_poly_subspace B r x = cyclic_orbit_poly_subspace B r x := by ext z constructor · rintro ⟨w, hw, rfl⟩ change (algebraMap B M ((a * d - b * c) ^ (r - 1))) * w ∈ cyclic_orbit_poly_subspace B r x rw [← Algebra.smul_def] exact Submodule.smul_mem _ _ hw · intro hz refine ⟨(algebraMap B M ((a * d - b * c) ^ (r - 1))⁻¹) * z, ?_, ?_⟩ · rw [← Algebra.smul_def] exact Submodule.smul_mem _ _ hz · change (algebraMap B M ((a * d - b * c) ^ (r - 1))) * ((algebraMap B M ((a * d - b * c) ^ (r - 1))⁻¹) * z) = z rw [← mul_assoc, ← map_mul, mul_inv_cancel₀ hu, map_one, one_mul] have h_comm : (algebraMap B M c * x + algebraMap B M d) * (algebraMap B M (-c) * y + algebraMap B M a) = algebraMap B M (a * d - b * c) := by rw [mul_comm, hgh] have h_prod : ((algebraMap B M c * x + algebraMap B M d) * (algebraMap B M (-c) * y + algebraMap B M a)) ^ (r - 1) = algebraMap B M ((a * d - b * c) ^ (r - 1)) := by rw [h_comm, map_pow] rw [mul_pow] at h_prod have h_assoc : ((algebraMap B M c * x + algebraMap B M d) ^ (r - 1) * (algebraMap B M (-c) * y + algebraMap B M a) ^ (r - 1)) • cyclic_orbit_poly_subspace B r x = (algebraMap B M c * x + algebraMap B M d) ^ (r - 1) • ((algebraMap B M (-c) * y + algebraMap B M a) ^ (r - 1) • cyclic_orbit_poly_subspace B r x) := by rw [mul_smul] have hle : cyclic_orbit_poly_subspace B r x ≤ (algebraMap B M c * x + algebraMap B M d) ^ (r - 1) • cyclic_orbit_poly_subspace B r y := by rw [← h_scalar, ← h_prod, h_assoc] exact smul_le_smul_left _ hle2 exact le_antisymm hle1 hle