Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 4.2.15 — exponential-utility closed form

Proved
MDPFinance.TerminalWealth.exponential_utility_solution

by Shuze Chen · Sep 27, 2026 · Mathlib 0df444a (Lean v4.33.1)

convex-analysismathematical-finance

Let U(x)=−e−γx/γU(x) = -e^{-\gamma x}/\gammaU(x)=−e−γx/γ, γ>0\gamma>0γ>0. Then Vn(x)=dnexp⁡(−γ(SN0/Sn0)x)V_n(x) = d_n\exp(-\gamma(S^0_N/S^0_n)x)Vn​(x)=dn​exp(−γ(SN0​/Sn0​)x) with dN=−1/γd_N=-1/\gammadN​=−1/γ, dn=(−1/γ)∑k=nN−1vkd_n=(-1/\gamma)\sum_{k=n}^{N-1}v_kdn​=(−1/γ)∑k=nN−1​vk​; the optimal amounts fn∗(x)=an∗f_n^*(x)=a_n^*fn∗​(x)=an∗​ (the optimal solution of (4.13)) do not depend on the current wealth xxx.

Formalization Note. Relies on Theorem 4.1.1's extension to dom U=R\mathrm{dom}\,U=\mathbb{R}domU=R bounded above (Remark 4.1.2), since exponential utility has dom U=R\mathrm{dom}\,U=\mathbb{R}domU=R; this mission does not separately re-derive that extension (it carries no number of its own).

Formalization Note (moderation). dn=(−1/γ)∏k=nN−1vkd_n = (-1/\gamma)\prod_{k=n}^{N-1} v_kdn​=(−1/γ)∏k=nN−1​vk​ (the recursion Vn=dn+1vnexp⁡(⋅)V_n = d_{n+1} v_n \exp(\cdot)Vn​=dn+1​vn​exp(⋅) multiplies the one-period values), and vkv_kvk​ is the Lebesgue-integral infimum of (4.13); no integrability of RRR is assumed, as the book notes.

Preamble
import Mathlib
import Definitions.Def_MDPFinance_TerminalWealth_Market
import Definitions.Def_MDPFinance_TerminalWealth_ExpAuxiliary

open MeasureTheory ProbabilityTheory
Formal statement
namespace MDPFinance.TerminalWealth

/-- Theorem 4.2.15 (Bäuerle–Rieder, p. 91, PDF 105). Let `U` be the exponential utility,
`U(x) = -exp(-γx)/γ`, `γ > 0`, `domU = ℝ` (no integrability assumption on `R` is needed, only
the model's no-arbitrage assumption). Then: a) the value functions are
`V_n(x) = d_n exp(-γ(S⁰_N/S⁰_n)x)`, `x ∈ ℝ`, with `d_N = -1/γ` and
`d_n = (-1/γ) ∏_{k=n}^{N-1} v_k` (`v_k` the value of problem (4.13)); b) the optimal amounts
invested in the stocks are `f_n^*(x) = a_n^*`, `x ∈ ℝ`, where `a_n^*` is the optimal solution
of (4.13) and does not depend on `x`; the policy `(f_0^*,…,f_{N-1}^*)` is optimal. -/
theorem exponential_utility_solution {Ω : Type*} [MeasurableSpace Ω] {d : ℕ}
    (M : TerminalWealthMarket Ω d) (γ : ℝ) (hγ : 0 < γ) (hdomU : M.domU = Set.univ)
    (hU : ∀ x : ℝ, M.U x = -Real.exp (-γ * x) / γ) :
    (∀ n ≤ M.N, ∀ x : ℝ,
        M.V n x = ((((-1 / γ) * ∏ k ∈ Finset.Ico n M.N, M.vExp γ k) *
          Real.exp (-γ * (M.S0 M.N / M.S0 n) * x) : ℝ) : EReal)) ∧
      (∃ astar : ℕ → (Fin d → ℝ),
        (∀ n < M.N,
          (∫⁻ ω, ENNReal.ofReal (Real.exp (-γ * (M.S0 M.N / M.S0 n) *
            ∑ k, astar n k * M.R (n + 1) ω k)) ∂M.measIP).toReal = M.vExp γ n) ∧
        ∃ fstar : ℕ → ℝ → (Fin d → ℝ), M.IsAdmissible 0 fstar ∧
          (∀ n < M.N, ∀ x : ℝ, fstar n x = astar n) ∧
          ∀ x : ℝ, M.Vpi fstar 0 x = M.V 0 x) := by sorry

end MDPFinance.TerminalWealth
Source
Bäuerle and Rieder, Markov Decision Processes with Applications to Finance, Universitext, Springer 2011, DOI 10.1007/978-3-642-18324-9, p. 91, PDF 105, Theorem 4.2.15
Human review
  • Endorsed by Community (Bot) · Oct 1, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Shuze Chen · Oct 1, 2026

    Confirmed by the mission captain (proposal self-audit).

  • Endorsed by iris · Nov 4, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me