§4.1, p. 26 — the prox of the indicator of ℝⁿ₊ is (v)₊
ProvedBoydADMM.Prox.prox_nonnegOrthantadmmp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-bookp2o-v1projectionproximal-operator
Let , , and let be the indicator function of the nonnegative orthant . Then the -update
is unique and equals the nonnegative part of :
That is, a vector minimizes over if and only if .
This is the simplest example of a projection-type ADMM update; it appears in nonnegative least squares and in every problem with sign constraints.
Formalization Note The indicator of is encoded as the domain with the zero function; is the vector of componentwise maxima with .
Preamble
import Mathlib import Definitions.Def_BoydADMM_Prox_Basic open Matrix
Formal statement
namespace BoydADMM.Prox
theorem prox_nonnegOrthant {n : ℕ} (ρ : ℝ) (hρ : 0 < ρ) (v : EuclideanSpace ℝ (Fin n))
(x : EuclideanSpace ℝ (Fin n)) :
IsProx (nonnegOrthant n) (fun _ => 0) ρ v x ↔ x = posPartVec v := by sorry
end BoydADMM.Prox
Source
Boyd, Parikh, Chu, Peleato, Eckstein, Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers, Found. Trends Mach. Learn. 3(1) (2011), p. 26, §4.1
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.