Lemma 2, p. 572 — additive invariance of the choice distribution
ProvedBoundedNV.Pooling.lemma2_additive_invariancebounded-rationalitylogit-choicep2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1
Let be a measurable decision domain, and suppose the logit normalizer of a utility at parameter is finite and positive. Adding a constant to every utility value does not change the choice distribution:
This result permits the cost-based logit law used here to represent the paper's profit-based behavioral choice.
Formalization Note The normalizer assumptions make the density a probability density rather than an artifact of Lean's default value for a nonintegrable integral.
Preamble
import Mathlib import Definitions.Def_BoundedNV_Pooling_Logit open MeasureTheory
Formal statement
namespace BoundedNV.Pooling
/-- Lemma 2, p. 572: a common additive shift of utility preserves the choice CDF. -/
theorem lemma2_additive_invariance (Y : Set ℝ) (u : ℝ → ℝ) (β a : ℝ)
(hY : MeasurableSet Y)
(hu : IntegrableOn (fun v => Real.exp (u v / β)) Y)
(hZ : 0 < ∫ v in Y, Real.exp (u v / β))
(hβ : 0 < β) (y : ℝ) :
logitCDF Y (fun v => u v + a) β y = logitCDF Y u β y := by sorry
end BoundedNV.Pooling
Source
Su, Bounded Rationality in Newsvendor Models, Manufacturing & Service Operations Management 10(4), 2008, p. 572 (PDF 7), Lemma 2; proof p. 586 (PDF 21), eq. (35)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.