Fact (13), p. 575, i = 1 — ∂A/∂η₁ at (p/β, c/β) equals E min(D, X♭)
ProvedBoundedNV.ExpFam.fact13_dA_deta1bounded-rationalityexponential-familylog-partitionnewsvendorp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1
Let have density , decision domain , and let , . Let be the behavioral solution, with density of (11), and let be the log-partition function (12). Then is differentiable at and
This is the case of the exponential-family identity (13), , for the newsvendor family, with sufficient statistic , the expected sales at order .
Formalization Note The derivative is asserted with HasDerivAt, which includes differentiability, so the identity cannot hold through Lean's convention that an undefined derivative is . Standing readings , as in the Observation; no finite mean of is assumed. The Fact is stated for this family only, not for abstract exponential families.
Preamble
import Mathlib import Definitions.Def_BoundedNV_ExpFam_Logit import Definitions.Def_BoundedNV_ExpFam_LogPartition
Formal statement
namespace BoundedNV.ExpFam
theorem fact13_dA_deta1 (f : ℝ → ℝ) (hf : IsDemandDensity f)
(p c β : ℝ) (hβ : 0 < β) (hc : 0 < c) (hcp : c < p) :
HasDerivAt (fun t => logPartition f t (c / β))
(BoundedNV.Uniform.logitExp (decisionDomain f) (BoundedNV.Uniform.nvProfit f p c) β (BoundedNV.Uniform.expMin f)) (p / β) := by sorry
end BoundedNV.ExpFam
Source
Su, Bounded Rationality in Newsvendor Models, Manufacturing & Service Operations Management 10(4), 2008, p. 575 (PDF p. 10), Fact, eq. (13), i = 1
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.