(4.8) under MFCQ, proof of Theorem 4.2, p. 10 — the multipliers of an approximate KKT sequence are bounded
ProvedAugLagLLC.KKT.mfcq_multipliers_boundedconstraint-qualificationmfcqmultiplier-estimatesp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Consider the nonlinear program "minimize subject to (), ()" on with finite index sets and continuously differentiable , , . Let be feasible and satisfy MFCQ, and let . Suppose there are coefficients and with for all , whenever , and
Then the sequences and are bounded: there is with
Applied to (4.11) along a subsequence of Algorithm 3.1 converging to , this gives the boundedness (4.8) of the multiplier estimates , , , .
Formalization Note. The norms are the sup norms on ι → ℝ and κ → ℝ, matching the of the page. Boundedness for all is equivalent to boundedness from some index on, since finitely many terms are bounded.
Preamble
import Mathlib import Definitions.Def_AugLagLLC_KKT_ConstraintQualifications
Formal statement
namespace AugLagLLC.KKT
open Filter Topology
/-- (4.8) under MFCQ, proof of Theorem 4.2, p. 10, without the algorithm: in the setting of
`cpld_limit_kkt` with MFCQ at `xs` in place of CPLD, the coefficient sequences `aₖ`, `bₖ` are
bounded (in the sup norm). -/
theorem mfcq_multipliers_bounded {n : ℕ} {ι κ : Type*} [Fintype ι] [Fintype κ]
{F : EuclideanSpace ℝ (Fin n) → ℝ} {H : ι → EuclideanSpace ℝ (Fin n) → ℝ}
{G : κ → EuclideanSpace ℝ (Fin n) → ℝ}
(hF : ContDiff ℝ 1 F) (hH : ∀ i, ContDiff ℝ 1 (H i)) (hG : ∀ j, ContDiff ℝ 1 (G j))
{xs : EuclideanSpace ℝ (Fin n)} (hHx : ∀ i, H i xs = 0) (hGx : ∀ j, G j xs ≤ 0)
(hmfcq : MFCQAt H G xs)
{y : ℕ → EuclideanSpace ℝ (Fin n)} (hy : Tendsto y atTop (𝓝 xs))
(a : ℕ → ι → ℝ) (b : ℕ → κ → ℝ) (hb : ∀ k j, 0 ≤ b k j)
(hbG : ∀ k j, G j xs < 0 → b k j = 0)
(hres : Tendsto (fun k => gradient F (y k) + ∑ i, a k i • gradient (H i) (y k)
+ ∑ j, b k j • gradient (G j) (y k)) atTop (𝓝 0)) :
∃ C : ℝ, ∀ k, ‖a k‖ ≤ C ∧ ‖b k‖ ≤ C := by sorry
end AugLagLLC.KKT
Source
Andreani, Birgin, Martínez & Schuverdt, On augmented Lagrangian methods with general lower-level constraints, HAL hal-01295437v1, p. 10, proof of Theorem 4.2, (4.8) under MFCQ
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.