Eq. (5.4) — the first approximation from the direct-route policy does not exceed it
ProvedBellmanRouting.PolicySpace.approx_one_le_approx_zerodynamic-programmingp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1shortest-pathsuccessive-approximations
Let and for . Let be the direct-route policy (5.2), for and , and let be given by (5.3): for , . Then
This is the first step of the monotone decrease (5.5).
Formalization Note The paper prints (5.2) for , which gives . Read literally, (5.4) is then false whenever . For : . The paper's own justification (" represents the minimum time for a path with at most one stop") requires , which is what is used here. This is equivalent to the printed (5.2) under .
Preamble
import Mathlib import Definitions.Def_BellmanRouting_PolicySpace_Routing
Formal statement
namespace BellmanRouting.PolicySpace
theorem approx_one_le_approx_zero {n : ℕ} (hn : 1 ≤ n)
(t : Fin (n + 1) → Fin (n + 1) → ℝ) (ht : ∀ i j, i ≠ j → 0 < t i j) :
∀ i, approx t 1 i ≤ approx t 0 i := by sorry
end BellmanRouting.PolicySpace
Source
Bellman, On a routing problem, Quart. Appl. Math. 16 (1958), p. 89, Section 5, Eqs. (5.2)–(5.4)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.