OAI/GroupTheory/PolycyclicRecognition/Algebra/RescaledWordSums.lean
2 declarations, 0 contain sorry, 0 declared as axioms
Exponential bounds for actual finite word expansions
The maximizing-basis exchange estimate controls every word in the finite list. A common surplus or deficit in the coefficients passes to the norm of the whole sum, uniformly over the selected maximizing basis.
- theorem
OAI.PolycyclicRecognition.exists_uniform_rescaled_word_boundline 23Source documentation
All words, rather than just the original letters, have uniformly bounded coordinates after normalization by their own weight.
theorem exists_uniform_rescaled_word_bound (X : I → L) (S : Finset (FreeMagma I)) (b₀ : Basis (Fin (finrank ℝ L)) ℝ L) : ∃ C : ℝ, 0 ≤ C ∧ ∀ (cost : S → ℝ) (β : WordBasisSelections X S), IsMaximizingColumnBasis (wordListCoordinates X S b₀) cost β.val → ∀ r : ℝ, 0 ≤ r → ∀ w : S, ‖(exponentialRescaleBasis (WordBasisSelections.basis X S β) (fun j => cost (β.val j)) r).equivFun (Real.exp (r * cost w) • evalBracketWord X w)‖ ≤ C := by …Mentions: OAI.PolycyclicRecognition.WordBasisSelections, OAI.PolycyclicRecognition.IsMaximizingColumnBasis, OAI.PolycyclicRecognition.wordListCoordinates, OAI.PolycyclicRecognition.exponentialRescaleBasis, OAI.PolycyclicRecognition.WordBasisSelections.basis, OAI.PolycyclicRecognition.evalBracketWord
- theorem
OAI.PolycyclicRecognition.norm_rescaled_word_sum_leline 48Source documentation
A finite expansion with a common exponential margin retains that margin in the actual normalized coordinate norm. The index may repeat words.
theorem norm_rescaled_word_sum_le {J : Type*} [Fintype J] (X : I → L) (S : Finset (FreeMagma I)) (b : Basis (Fin (finrank ℝ L)) ℝ L) (cost : S → ℝ) (r : ℝ) {C R θ : ℝ} (hR : 0 ≤ R) (hword : ∀ w : S, ‖b.equivFun (Real.exp (r * cost w) • evalBracketWord X w)‖ ≤ C) (w : J → S) (a : J → ℝ) (ha : ∀ j, |a j| ≤ R * Real.exp (r * (cost (w j) + θ))) : ‖b.equivFun (∑ j, a j • evalBracketWord X (w j))‖ ≤ (Fintype.card J : ℝ) * C * R * Real.exp (r * θ) := by …