VectorSpaceOpt.perturbationValue
DefinitionVectorSpaceOpt_perturbationValuedefinitionextended-realperturbationvalue-function
For a right-hand side , define the perturbed feasible set by and . The perturbation value is the extended-real infimum
The EReal codomain records for an infeasible perturbation and for an objective unbounded below, instead of inheriting an arbitrary real sInf value. This is the value function used in both convexity and monotonicity milestones and is the sensitivity object underlying Luenberger's multiplier interpretation. It also provides reusable infrastructure for later perturbation and stability results. The indexing subtype records feasibility without selecting an optimizer.
Definition code
import Definitions.Def_VectorSpaceOpt_coneLE
namespace VectorSpaceOpt
/-- The possibly infinite optimal value under the perturbed constraint `G x ≤_P z`. -/
noncomputable def perturbationValue
{X Z : Type*}
[AddCommGroup Z] [Module ℝ Z]
(P : ConvexCone ℝ Z) (Ω : Set X) (f : X → ℝ) (G : X → Z) (z : Z) : EReal :=
⨅ x : {x : X // x ∈ Ω ∧ coneLE P (G x) z}, ((f x : ℝ) : EReal)
end VectorSpaceOptSource
David G. Luenberger, Optimization by Vector Space Methods, Wiley, 1969, public scan https://sites.science.oregonstate.edu/~show/old/142_Luenberger.pdf, Chapter 8, §8.3, definition of ω(z), printed p. 216; physical PDF p. 236.