Lemma 14, p. 23 — first arrow q → z: ‖z‖^{λ,K}_F ≤ L(1 + 1/λ)‖q‖^{λ,K}_F
ProvedDIGing.Push.lemma_14distributed-optimizationlipschitz-gradientp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1small-gain
Let each satisfy Assumption 4, and let . Let be any sequence in and any point. Put , for and . Then for every and every ,
This is the first gain of the small gain cycle ; the paper restates Lemma 5 for Push-DIGing.
Formalization Note Only Assumption 4 is used, so the sequence is arbitrary (not necessarily a run of the algorithm) and is any point.
Preamble
import Mathlib import Definitions.Def_DIGing_Undir_Common import Definitions.Def_DIGing_Push_Setting
Formal statement
namespace DIGing.Push
theorem lemma_14 {n p : ℕ} (f : Fin n → EuclideanSpace ℝ (Fin p) → ℝ) (Lc : Fin n → ℝ)
(h4 : DIGing.Undir.Assumption4 f Lc) (x : ℕ → DIGing.Undir.Stack n p) (xstar : EuclideanSpace ℝ (Fin p)) :
∀ K : ℕ, ∀ lam : ℝ, 0 < lam → lam < 1 →
DIGing.Undir.ergK lam K (fun k => DIGing.Undir.frob (DIGing.Undir.zSeq f x k)) ≤
DIGing.Undir.Lmax Lc * (1 + 1 / lam) * DIGing.Undir.ergK lam K (fun k => DIGing.Undir.frob (x k - DIGing.Undir.ones xstar)) := by sorry
end DIGing.Push
Source
arXiv:1607.03218v3, Lemma 14, p. 23 (restating Lemma 5, p. 10)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.