Proof of Theorem 1, display (16), p. 43 — the integral (1) splits into tail integrals over the distinct utility levels a_1 < … < a_k = u_i
ProvedHetExponomial.ChoiceProb.eq_16In the HEC model with and rates , let be the distinct values of the ideal utilities . Fix a choice with , and write . For a level and a lower limit put
Then the integral in (1) decomposes along the levels :
The -th summand is the line of display (16) that carries the set . For its first term is , since , and the lowest level has no negative term.
This is display (16) in the proof of Theorem 1. It splits the integral (1) into pieces on which the positive parts disappear, so that each piece can be integrated explicitly.
Formalization Note The levels are level u, the increasing enumeration of the finite set of values of , indexed from . The paper's is level u ⟨w-1, _⟩, and the hypothesis level u k = u i pins (shifted by one) as the level of choice . The summand for a Lean index subtracts the tail integral from only when , that is, when the level is not the lowest. Each tail integral is a Bochner integral over . Its integrand is integrable because for .
import Mathlib import Definitions.Def_HetExponomial_ChoiceProb_Model
namespace HetExponomial.ChoiceProb
theorem eq_16 {n : ℕ} (hn : 2 ≤ n) (u lam : Fin n → ℝ) (hlam : ∀ j, 0 < lam j) (i : Fin n)
(k : Fin (levels u).card) (hk : level u k = u i) :
∫ z in Set.Ioi 0, lam i * Real.exp (-(∑ j, lam j * max (u j - u i + z) 0)) =
lam i * ∑ w ∈ Finset.univ.filter (fun w => w ≤ k),
(tailInt u lam i (level u w) (u i - level u w) -
if 0 < w.val then
tailInt u lam i (level u w)
(u i - level u ⟨w.val - 1, Nat.lt_of_le_of_lt (Nat.sub_le _ _) w.isLt⟩)
else 0) := by sorry
end HetExponomial.ChoiceProb
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.