Solutions of are exactly the zeros of the projected residual
ProvedSolodovSvaiterVI.Alg21.mem_viSol_iff_residual_eq_zerop2o-batch-p200bp2o-gran-per-chapterp2o-plan-paperp2o-v1projection-methodsvariational-inequalities
Let be a nonempty closed convex subset of and any map. Let be the solution set of and the projected residual. Then for every ,
This is what makes a valid stopping test in Algorithm 2.1, and it is how the convergence proof certifies that a limit point is a solution.
Formalization Note The paper assumes nonempty throughout; here only is assumed (it follows from and is what makes well defined). No continuity of is needed.
Preamble
import Mathlib import Definitions.Def_SolodovSvaiterVI_Alg21_viSol import Definitions.Def_SolodovSvaiterVI_Alg21_residual
Formal statement
namespace SolodovSvaiterVI.Alg21
/-- Solodov–Svaiter, Section 2, p. 767: solutions of `VI(F, C)` coincide with the zeros of the
projected residual `r(x) = x − P_C[x − F(x)]`, i.e. `x ∈ S ⟺ r(x) = 0`. -/
theorem mem_viSol_iff_residual_eq_zero {n : ℕ}
(F : EuclideanSpace ℝ (Fin n) → EuclideanSpace ℝ (Fin n))
(C : Set (EuclideanSpace ℝ (Fin n))) (hCne : C.Nonempty) (hCc : IsClosed C)
(hCcv : Convex ℝ C) (x : EuclideanSpace ℝ (Fin n)) :
x ∈ viSol F C ↔ residual F C x = 0 := by sorry
end SolodovSvaiterVI.Alg21
Source
Solodov & Svaiter, A New Projection Method for Variational Inequality Problems, SIAM J. Control Optim. 37 (1999), p. 767, Section 2 (first paragraph)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.