Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 5.3 — mean–Gini optimal solutions exist and are SSD-efficient

Proved
DualSSD.MeanRisk.meanGini_optimal_exists_ssdEfficient

by mikedeng1 · Sep 28, 2026 · Mathlib 0df444a (Lean v4.33.1)

ginimean-riskp2o-batch-p100bp2o-gran-per-chapterp2o-plan-paperp2o-v1portfolio-optimizationstochastic-dominance

Let 1<q<∞1<q<\infty1<q<∞ and let Q⊆Lq(Ω,P)Q\subseteq L_q(\Omega,P)Q⊆Lq​(Ω,P) be a nonempty, convex, bounded and closed set. Consider the mean–risk problem with the Gini mean difference as risk functional,

max⁡X∈Q (μX−λΓX).(5.1)\max_{X\in Q}\ (\mu_X-\lambda\Gamma_X).\tag{5.1}X∈Qmax​ (μX​−λΓX​).(5.1)

Then for every λ∈(0,1]\lambda\in(0,1]λ∈(0,1]:

  1. the set of optimal solutions of (5.1) is nonempty, and
  2. each optimal solution XXX is SSD-efficient in QQQ: there is no Y∈QY\in QY∈Q with Y≻SSDXY\succ_{SSD}XY≻SSD​X.

This is the main result of the mission: the mean–Gini model with trade-off λ≤1\lambda\le1λ≤1 never selects a portfolio that some feasible portfolio dominates in the second-degree sense, so it is consistent with risk-averse preferences.

Formalization Note Nonemptiness of QQQ is not written in the paper and is added: without it no optimal solution exists. The paper's q>1q>1q>1 is a real number, so q≠∞q\ne\inftyq=∞ is part of its hypothesis. The trade-off coefficient is named lam because λ is a Lean keyword. Optimal solutions are stated as maximizers in QQQ, not through a supremum value.

Preamble
import Mathlib
import Definitions.Def_DualSSD_MeanRisk_gini
import Definitions.Def_DualSSD_MeanRisk_ssdEfficient
Formal statement
namespace DualSSD.MeanRisk

open MeasureTheory

/-- **Theorem 5.3** (Ogryczak–Ruszczyński 2002, p. 73). Let `Q ⊆ L_q(Ω, P)`, `1 < q < ∞`, be
convex, bounded, closed and nonempty, and let `lam ∈ (0, 1]` (the paper's `λ`; `λ` is a Lean
keyword). Then the mean–Gini problem (5.1) `max_{X ∈ Q} (μ_X − lam Γ_X)` has an optimal solution
in `Q`, and every optimal solution is SSD-efficient in `Q` (no `Y ∈ Q` with `Y ≻_SSD X`).
`Q.Nonempty` is not on the page: the paper assumes it silently, and without it no optimal
solution exists. -/
theorem meanGini_optimal_exists_ssdEfficient {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
    [IsProbabilityMeasure P] (q : ENNReal) [Fact (1 ≤ q)] (hq : 1 < q) (hqtop : q ≠ ⊤)
    (Q : Set (Lp ℝ q P)) (hQconv : Convex ℝ Q) (hQbdd : Bornology.IsBounded Q)
    (hQclosed : IsClosed Q) (hQne : Q.Nonempty) (lam : ℝ) (hlam0 : 0 < lam) (hlam1 : lam ≤ 1) :
    (∃ X ∈ Q, ∀ Z ∈ Q, mean P ⇑Z - lam * gini P ⇑Z ≤ mean P ⇑X - lam * gini P ⇑X) ∧
      ∀ X ∈ Q, (∀ Z ∈ Q, mean P ⇑Z - lam * gini P ⇑Z ≤ mean P ⇑X - lam * gini P ⇑X) →
        SSDEfficient P ((fun Z : Lp ℝ q P => (⇑Z : Ω → ℝ)) '' Q) ⇑X := by sorry

end DualSSD.MeanRisk
Source
Ogryczak, Ruszczyński, Dual Stochastic Dominance and Related Mean-Risk Models, SIAM J. Optim. 13 (2002), p. 73, Theorem 5.3 (problem (5.1) and standing assumptions on Q, pp. 72–73)
Read-back

What the Lean code literally says, in plain math · claude-opus-5-5

Fix a set Ω\OmegaΩ with a σ\sigmaσ-algebra and a probability measure PPP on it, so P(Ω)=1P(\Omega)=1P(Ω)=1. Fix an exponent q∈[0,∞]q \in [0,\infty]q∈[0,∞] satisfying 1≤q1 \le q1≤q, 1<q1 < q1<q and q≠∞q \neq \inftyq=∞, so in effect 1<q<∞1 < q < \infty1<q<∞. Let Lq(Ω,P)L^q(\Omega,P)Lq(Ω,P) be the space of real-valued qqq-integrable functions on Ω\OmegaΩ modulo PPP-almost-everywhere equality, with its usual LqL^qLq norm. Each element ZZZ of this space is used below through one fixed representative function Ω→R\Omega \to \mathbb{R}Ω→R, written Z^\hat ZZ^. The statement does not say which representative is chosen.

Hypotheses. Let Q⊆Lq(Ω,P)Q \subseteq L^q(\Omega,P)Q⊆Lq(Ω,P) satisfy all of the following:

  • QQQ is convex (over R\mathbb{R}R).
  • QQQ is bounded in the LqL^qLq norm.
  • QQQ is closed in the LqL^qLq norm topology.
  • QQQ is nonempty.

Let λ\lambdaλ be a real number with 0<λ≤10 < \lambda \le 10<λ≤1.

Objective. For Z∈Lq(Ω,P)Z \in L^q(\Omega,P)Z∈Lq(Ω,P), define

Φλ(Z)  =  meanP(Z^)  −  λ giniP(Z^).\Phi_\lambda(Z) \;=\; \mathrm{mean}_P(\hat Z) \;-\; \lambda\,\mathrm{gini}_P(\hat Z).Φλ​(Z)=meanP​(Z^)−λginiP​(Z^).
  • meanP\mathrm{mean}_PmeanP​ and giniP\mathrm{gini}_PginiP​ are functionals defined elsewhere in this bundle. Each takes PPP and a function Ω→R\Omega\to\mathbb{R}Ω→R and returns a real number. Their definitions are not included in the code given here, so their exact formulas cannot be stated from it. In particular, this code does not show whether they depend only on the law of Z^\hat ZZ^ under PPP or on the particular representative chosen.
  • SSDEfficientP(S,f)\mathrm{SSDEfficient}_P(S, f)SSDEfficientP​(S,f) is also a bundle definition whose formula is not shown. It is a property of a set SSS of functions Ω→R\Omega\to\mathbb{R}Ω→R, a function fff, and PPP. Its name suggests efficiency of fff within SSS with respect to second-order stochastic dominance, but what exactly it requires cannot be read from this code.

The statement asserts two things:

  1. Existence of a maximiser. There is some X∈QX \in QX∈Q such that
Φλ(Z)≤Φλ(X)for every Z∈Q.\Phi_\lambda(Z) \le \Phi_\lambda(X) \quad \text{for every } Z \in Q.Φλ​(Z)≤Φλ​(X)for every Z∈Q.
  1. Every maximiser is efficient. Take any X∈QX \in QX∈Q with Φλ(Z)≤Φλ(X)\Phi_\lambda(Z) \le \Phi_\lambda(X)Φλ​(Z)≤Φλ​(X) for all Z∈QZ \in QZ∈Q. Then
SSDEfficientP({Z^:Z∈Q}, X^)\mathrm{SSDEfficient}_P\big(\{\hat Z : Z \in Q\},\ \hat X\big)SSDEfficientP​({Z^:Z∈Q}, X^)

holds. The set here is the collection of chosen representatives of the members of QQQ, not the equivalence classes themselves.

The maximisation in both parts is non-strict (≤\le≤). Part 2 is an implication for each maximiser. It does not claim the maximiser is unique, and it does not claim the converse (that efficient elements are maximisers).

Degenerate cases. Ω\OmegaΩ cannot be empty, because P(Ω)=1P(\Omega)=1P(Ω)=1. The hypotheses can be satisfied together, for example by Q={0}Q=\{0\}Q={0} with any admissible qqq and λ\lambdaλ. So part 2 is not vacuous: part 1 guarantees at least one maximiser exists.

  • QQQ a single point {X}\{X\}{X}: part 1 holds trivially with that XXX. Part 2 then asserts SSDEfficientP({X^},X^)\mathrm{SSDEfficient}_P(\{\hat X\},\hat X)SSDEfficientP​({X^},X^).
  • Ω\OmegaΩ with one point, or PPP concentrated so that LqL^qLq is one-dimensional: every element is effectively a constant ccc. QQQ is then a closed bounded interval of constants, and Φλ\Phi_\lambdaΦλ​ is whatever meanP\mathrm{mean}_PmeanP​ and giniP\mathrm{gini}_PginiP​ return on constants, which the code does not show.
  • Division, subtraction in ℕ, suprema: the statement itself uses none of these; its only arithmetic is real subtraction and multiplication.
  • Integrals of non-integrable functions: whether meanP\mathrm{mean}_PmeanP​ or giniP\mathrm{gini}_PginiP​ use integrals that could fall back to a default value depends on their unseen definitions. Every Z^\hat ZZ^ here is qqq-integrable for q>1q>1q>1 on a probability space, so it is also integrable.
  • The value q=∞q=\inftyq=∞ and all q<1q<1q<1 are excluded by the hypotheses. The hypothesis 1≤q1 \le q1≤q is redundant given 1<q1<q1<q.
Human review
  • Endorsed by Shuze Chen · Sep 28, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Sep 28, 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