Appendix B, p. 40 — from x₀ = 3.3, |x_j − x⋆_j| < ε̄ for j = 4, …, 11
ProvedIQCAlg.HeavyBall.iterates_4_to_11counterexampleexact-computationheavy-ballp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Run the Heavy-ball method with the gradient (4.11), the tuning of Proposition 1 at , (i.e. , ), the starting point and the initialization . Then
where is the cycle of (B.3).
This supplies the eight consecutive iterates required by the tail argument.
Preamble
import Mathlib import Definitions.Def_IQCAlg_HeavyBall_Setting
Formal statement
namespace IQCAlg.HeavyBall
/-- p. 40: for the Heavy-ball run from `x_0 = 3.3` (with `x_{−1} = x_0`) with Proposition 1's
tuning at `L = 25`, `m = 1`, on the gradient (4.11), the iterates `x_4, …, x_11` are each
within `ε̄` of their limit points `x⋆_k`. -/
theorem iterates_4_to_11 :
∀ j : ℕ, 4 ≤ j → j ≤ 11 →
|hbRun gradF (hbAlpha 25 1) (hbBeta 25 1) (33 / 10) j - cyc j| < epsBar := by sorry
end IQCAlg.HeavyBall
Source
Lessard, Recht & Packard, Analysis and Design of Optimization Algorithms via Integral Quadratic Constraints, arXiv:1408.3595v7, p. 40, "if x₀ = 3.3, then the iterates x₄, x₅, …, x₁₁ are each within a distance ε̄ of their respective limit points"
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.