Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Section 2, equations (2)–(3) — homogenization

Proved
WangMinimaxRegret.Reformulation.eq_2_3

by Shuze Chen · Oct 29, 2026 · Mathlib 0df444a (Lean v4.33.1)

linear-optimizationmechanism-designoperations-researchp2o-gran-per-chapterp2o-plan-paperp2o-v1

Fix consistent valuation data, a payment m∈Mm\in\mathcal Mm∈M, and a finite price p∈Ip\in Ip∈I. Let F0\mathcal F_0F0​ contain all probability laws supported on III with the specified finite moments; unlike the printed ambiguity class, displayed (2) imposes no condition F(a)=0F(a)=0F(a)=0. Let Hp\mathcal H_pHp​ contain the finite nonnegative measures supported on III with finite moments, centered moment equalities ∫(xk−μk) dH=0\int(x^k-\mu_k)\,dH=0∫(xk−μk​)dH=0, and tail mass H([p,∞))=1H([p,\infty))=1H([p,∞))=1. Then

inf⁡F∈F0, TF(p)>0∫m dFTF(p)=inf⁡H∈Hp∫m dH.\inf_{F\in\mathcal F_0,\ T_F(p)>0}\frac{\int m\,dF}{T_F(p)}=\inf_{H\in\mathcal H_p}\int m\,dH.F∈F0​, TF​(p)>0inf​TF​(p)∫mdF​=H∈Hp​inf​∫mdH.

Moreover each positive-tail F∈F0F\in\mathcal F_0F∈F0​ maps to a feasible H=F/TF(p)H=F/T_F(p)H=F/TF​(p), normalization of this HHH recovers FFF, and ∫m dH=(∫m dF)/TF(p)\int m\,dH=(\int m\,dF)/T_F(p)∫mdH=(∫mdF)/TF​(p). Conversely each H∈HpH\in\mathcal H_pH∈Hp​ has positive total mass, its normalization F=H/H(I)F=H/H(I)F=H/H(I) lies in F0\mathcal F_0F0​ and has positive tail, rescaling this FFF recovers HHH, and the same objective identity holds. The equality includes infeasible-price cases through empty infima of +∞+\infty+∞. This is the measure, rather than finite-vector, version of the displayed change of variables.

Differences from the source:

  • encoding: real interval, payment functions, probability laws, and moment orders 1 through K → real-line Borel measures supported on the interval, Stieltjes functions extended by zero below a and constantly above finite b, and Fin K exponents i+1; reason: extensions only fix values outside the domain; all atoms and payment jumps are retained, including the payment atom at a>0
  • encoding: the seller knows moments of an actual valuation distribution → Consistent d requires a nonempty printed ambiguity class, with finite specified moments; reason: makes the standing information-consistency assumption explicit; no interiority, positive variance, or density is assumed
  • encoding: nonnegative revenue integrals and optimization infima/suprema → nonnegative extended integrals and ENNReal/EReal extrema; zero-tail laws impose no fractional constraint; reason: avoids nonintegrable real integrals defaulting to zero and empty real extrema; empty infima are positive infinity
  • encoding: homogeneous rows k=0,...,K and scalar rescaling of H → finite nonnegative H, integrable centered rows k=1,...,K, and a tail indicator row; reason: finite mass and finite moments make rescaling and real moment equalities meaningful; the k=0 row is redundant because mu_0=1
  • reading: the domain [a,b], with infinity used in later applications → finite-valued valuations and prices on [a,infinity) when b=infinity; reason: retains those applications without a compact cutoff
Preamble
import Definitions.Def_WangMinimaxRegret_Reformulation_Programs

open MeasureTheory Set
open scoped ENNReal NNReal
set_option autoImplicit false
Formal statement
namespace WangMinimaxRegret.Reformulation

theorem eq_2_3 (d : Data) (hc : Consistent d) (m : StieltjesFunction ℝ)
    (hm : IsPayment d m) (p : ℝ) (hp : p ∈ interval d) :
    fractionalValue {F | MomentLaw d F} m p = (homogenizedLP d m p).value ∧
    (∀ F : Measure ℝ, MomentLaw d F → 0 < tail F p →
      (homogenizedLP d m p).Feasible (scaleToTail F p) ∧
      normalizeMass (scaleToTail F p) = F ∧
      revenue m (scaleToTail F p) = revenue m F / tail F p) ∧
    (∀ H : Measure ℝ, (homogenizedLP d m p).Feasible H →
      0 < H univ ∧ MomentLaw d (normalizeMass H) ∧
      0 < tail (normalizeMass H) p ∧ scaleToTail (normalizeMass H) p = H ∧
      revenue m (normalizeMass H) / tail (normalizeMass H) p = revenue m H) := by sorry

end WangMinimaxRegret.Reformulation
Source
Wang, Liu and Zhang (2024), Minimax Regret Robust Screening with Moment Information, https://doi.org/10.1287/msom.2023.0072, Section 2, p. 8, equations (2)–(3)
Read-back

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

Setting. Fix a data record ddd:

  • reals a≥0a \ge 0a≥0 and b∈R∪{+∞}b\in\mathbb R\cup\{+\infty\}b∈R∪{+∞} with a≤ba\le ba≤b;
  • K∈NK\in\mathbb NK∈N and reals μ0,…,μK−1\mu_0,\dots,\mu_{K-1}μ0​,…,μK−1​;
  • the interval I={x:a≤x≤b}I = \{x : a\le x\le b\}I={x:a≤x≤b}.

A Borel measure FFF on R\mathbb RR is a moment law when:

  • FFF is a probability measure;
  • F(R∖I)=0F(\mathbb R\setminus I) = 0F(R∖I)=0;
  • xi+1x^{i+1}xi+1 is FFF-integrable with ∫xi+1 dF=μi\int x^{i+1}\,dF = \mu_i∫xi+1dF=μi​ for each i<Ki<Ki<K.

It is ambiguous when, in addition, F((−∞,a])=0F((-\infty,a]) = 0F((−∞,a])=0.

Hypotheses.

  1. ddd is consistent: some ambiguous measure exists.
  2. mmm is a Stieltjes function (nondecreasing, right-continuous, real-valued) that is a payment for ddd:
    • m(0)=0m(0)=0m(0)=0;
    • m=0m=0m=0 on (−∞,a)(-\infty,a)(−∞,a);
    • mmm is constant on {x≥b}\{x\ge b\}{x≥b};
    • m≥0m\ge 0m≥0 on III;
    • ∫I∩(0,x]t−1 dνm(t)≤1\int_{I\cap(0,x]} t^{-1}\,d\nu_m(t)\le 1∫I∩(0,x]​t−1dνm​(t)≤1 for x∈Ix\in Ix∈I, with νm\nu_mνm​ the Lebesgue–Stieltjes measure of mmm.
  3. p∈Ip \in Ip∈I.

Quantities used. All are in [0,∞][0,\infty][0,∞], with 0−1=∞0^{-1}=\infty0−1=∞, ∞−1=0\infty^{-1}=0∞−1=0 and 0⋅∞=00\cdot\infty = 00⋅∞=0.

  • Tail. TF=F([p,∞))T_F = F([p,\infty))TF​=F([p,∞)).
  • Revenue. Rev(F)=∫max⁡(m,0) dF\mathrm{Rev}(F) = \int\max(m,0)\,dFRev(F)=∫max(m,0)dF.
  • Rescalings. scale(F)=TF−1⋅F\mathrm{scale}(F) = T_F^{-1}\cdot Fscale(F)=TF−1​⋅F and norm(H)=H(R)−1⋅H\mathrm{norm}(H) = H(\mathbb R)^{-1}\cdot Hnorm(H)=H(R)−1⋅H.
  • LP feasibility. A measure HHH is LP-feasible when:
    • HHH is finite;
    • H(R∖I)=0H(\mathbb R\setminus I)=0H(R∖I)=0;
    • H([p,∞))=1H([p,\infty)) = 1H([p,∞))=1;
    • for each j<Kj<Kj<K, the function xj+1−μjx^{j+1}-\mu_jxj+1−μj​ is HHH-integrable with ∫(xj+1−μj) dH=0\int(x^{j+1}-\mu_j)\,dH = 0∫(xj+1−μj​)dH=0.
  • LP value.
V=inf⁡H LP-feasibleRev(H).V = \inf_{H \text{ LP-feasible}}\mathrm{Rev}(H).V=H LP-feasibleinf​Rev(H).
  • Fractional value.
Φ=inf⁡{Rev(F)TF:F a moment law, TF>0}.\Phi = \inf\Big\{\frac{\mathrm{Rev}(F)}{T_F} : F \text{ a moment law},\ T_F>0\Big\}.Φ=inf{TF​Rev(F)​:F a moment law, TF​>0}.

This is computed over moment laws, not over ambiguous measures.

Conclusions. All three of the following hold.

  1. Values agree. Φ=V\Phi = VΦ=V.
  2. From moment laws to LP-feasible measures. For every moment law FFF with TF>0T_F > 0TF​>0:
    • scale(F)\mathrm{scale}(F)scale(F) is LP-feasible;
    • norm(scale(F))=F\mathrm{norm}(\mathrm{scale}(F)) = Fnorm(scale(F))=F;
    • Rev(scale(F))=Rev(F)/TF\mathrm{Rev}(\mathrm{scale}(F)) = \mathrm{Rev}(F)/T_FRev(scale(F))=Rev(F)/TF​.
  3. From LP-feasible measures to moment laws. For every LP-feasible HHH:
    • H(R)>0H(\mathbb R) > 0H(R)>0;
    • norm(H)\mathrm{norm}(H)norm(H) is a moment law;
    • the tail of norm(H)\mathrm{norm}(H)norm(H) at ppp is positive;
    • scale(norm(H))=H\mathrm{scale}(\mathrm{norm}(H)) = Hscale(norm(H))=H;
    • Rev(norm(H))/norm(H)([p,∞))=Rev(H)\mathrm{Rev}(\mathrm{norm}(H))\big/\mathrm{norm}(H)([p,\infty)) = \mathrm{Rev}(H)Rev(norm(H))/norm(H)([p,∞))=Rev(H).

Degenerate cases.

  • Consistency is not used by the conclusions' wording. The consistency hypothesis concerns ambiguous measures (no mass at or below aaa), while every conclusion concerns moment laws, which may have an atom at aaa.
  • Finite rescaling of moment laws. For a moment law, TF≤1T_F \le 1TF​≤1, so TF>0T_F>0TF​>0 makes TF−1T_F^{-1}TF−1​ finite. The rescalings in conclusion 2 never meet 0−10^{-1}0−1 or ∞−1\infty^{-1}∞−1.
  • Positive mass of LP-feasible measures. For an LP-feasible HHH, H(R)≥H([p,∞))=1H(\mathbb R)\ge H([p,\infty)) = 1H(R)≥H([p,∞))=1, so its positive mass is automatic.
  • K=0K=0K=0. Moment laws are all probability measures on III, and the LP has only the finiteness, support and unit-tail rows.
  • Empty infima. If no moment law has positive tail at ppp, then Φ=+∞\Phi = +\inftyΦ=+∞. Conclusion 1 then asserts V=+∞V = +\inftyV=+∞.
  • Infinite revenue. Revenues may be infinite, and divisions follow [0,∞][0,\infty][0,∞] arithmetic.
Human review
  • Endorsed by Community (Bot) · Oct 29, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Shuze Chen · Oct 29, 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