Extension of the main theorem, p. 871 — with √· replaced by t^k, ½ ≤ k ≤ 1, the minimum still occurs at a vertex of D
ProvedSimpsonInv.Vertex.theorem_I_rpowconcave-minimizationextreme-pointinventoryp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let , let be real constants, , and let be a constant with . Let be the domain of THEOREM I ( and ) and assume . Then the function
attains its minimum over at a vertex of .
This covers correlated demand, where the standard deviation of demand over a time grows like instead of ; is the linear-programming case.
Formalization Note Vertex means extreme point. (equation (8)) and are added as for THEOREM I. The power is Real.rpow; on its base is nonnegative. The range is the page's "between ½ and 1", endpoints included because the page treats as the limiting case where "the result is known".
Preamble
import Mathlib import Definitions.Def_SimpsonInv_Vertex_Setting
Formal statement
namespace SimpsonInv.Vertex
theorem theorem_I_rpow {m : ℕ} (k c S0 Sn : ℝ) (r T : ℕ → ℝ)
(hk : 1 / 2 ≤ k) (hk1 : k ≤ 1)
(hr : ∀ i ∈ Finset.Icc 1 (m + 1), 0 < r i)
(hD : (dom (m := m) T S0 Sn).Nonempty) :
∃ v ∈ Set.extremePoints ℝ (dom (m := m) T S0 Sn),
IsMinOn (costPow k c r T S0 Sn) (dom (m := m) T S0 Sn) v := by sorry
end SimpsonInv.Vertex
Source
Simpson, In-Process Inventories, Operations Research 6 (1958), p. 871, Extension of the main theorem, Correlated demand
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.