§3, p. 6 — an optimization problem (1) always has the same optimal value as its inference dual (2)
ProvedLogicBenders.Generic.strong_inference_dualityLet be a domain, a feasible set and a real-valued objective. Consider problem (1), subject to , , and its inference dual (2),
which asks for the largest bound such that holds at every satisfying . With optimal values taken in (an infeasible minimization has value , an unbounded one , and vice versa for maximization), the two problems have the same optimal value:
This is the strong duality property of the inference dual. It holds with no assumption on or , and it is the step of the proof of Theorem 1 that identifies the value of the subproblem dual with the optimal value of the subproblem.
Formalization Note. The dual variable ranges over the extended reals, so the convention of §3 is built in: when is empty every is feasible and both sides are ; when is unbounded below on only is feasible and both sides are .
import Mathlib import Definitions.Def_LogicBenders_Generic_Setting
namespace LogicBenders.Generic
theorem strong_inference_duality {D : Type*} (S : Set D) (f : D → ℝ) :
optVal1 S f = infDualVal1 S f := by sorry
end LogicBenders.Generic
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.