§5, p. 141 — every feasible deviation has nonnegative split variables
ProvedExecCompLP.LAD.lift_feasibleleast-absolute-deviationslinear-programmingp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let satisfy the nonnegative-weight, ceiling, ranking, and floor constraints (2). For each specified level , set , , and . Then satisfies the transformed constraints (4)–(5), and
This identifies an LP-feasible point with the same objective for every feasible choice of salary weights.
Formalization Note The split variables are functions on all ranked levels, but their signs and defining equations are required only at the specified levels , as in the paper.
Preamble
import Mathlib import Definitions.Def_ExecCompLP_LAD_Setting
Formal statement
namespace ExecCompLP.LAD
theorem lift_feasible {n L : ℕ} [NeZero L]
(x : Fin n → Fin L → ℝ) (sM sm : ℝ) (K : Finset (Fin L))
(s : Fin L → ℝ) (a : Fin n → ℝ) (ha : Feasible x sM sm a) :
LPFeasible x sM sm K s a
(fun k => posPart (salary x a k - s k))
(fun k => negPart (salary x a k - s k)) ∧
lpObj K
(fun k => posPart (salary x a k - s k))
(fun k => negPart (salary x a k - s k)) = obj x K s a := by sorry
end ExecCompLP.LAD
Source
Charnes, Cooper & Ferguson, Optimal estimation of executive compensation by linear programming, Management Science 1(2) (1955), p. 141, §5, equations (4)–(5) and sentence following (5); https://doi.org/10.1287/mnsc.1.2.138
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.