Projection onto a product acts blockwise
ProvedBoydADMM.Constrained.product_projectionadmmconvex-optimizationp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-bookp2o-v1projections
Let and let be nonempty closed convex sets. For block vectors and , projection onto their Cartesian product is exactly componentwise projection:
This is the first geometric identity needed to turn product-space ADMM into parallel updates.
Formalization Note The product projection minimizes the sum , which is the Euclidean metric on the stacked vector. Each component projection uses the published nearest-point predicate.
Preamble
import Mathlib import Definitions.Def_BoydADMM_Constrained_Geometry
Formal statement
namespace BoydADMM.Constrained
/-- §5.1.2, p. 35: projection onto a Cartesian product acts blockwise. -/
theorem product_projection {N n : ℕ} (hN : 0 < N)
(A : Fin N → Set (Vec n))
(hAne : ∀ i, (A i).Nonempty) (hAclosed : ∀ i, IsClosed (A i))
(hAconvex : ∀ i, Convex ℝ (A i))
(v p : Blocks N n) :
IsBlockProjection (productSet A) v p ↔
∀ i, RandomGradFree.Nonsmooth.IsMetricProjection (A i) (v i) (p i) := by sorry
end BoydADMM.Constrained
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. 35, §5.1.2, product projection display; https://doi.org/10.1561/2200000016
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.