Loomis.loomis_value_IJ_2
theoremverifiedformalizationimported Lean
Base case of the Loomis induction: a 1×1 matrix pair has the single ratio A i₀ j₀ / B i₀ j₀ as the common Loomis value.
Statement
set_option autoImplicit true in set_option relaxedAutoImplicit true in open Finset BigOperators in set_option linter.unusedSectionVars false in variable {I J : Type*} [Fintype I] [Fintype J] [Nonempty I] [Nonempty J] in theorem Loomis.loomis_value_IJ_2 (Hn : 2 = Fintype.card I + Fintype.card J) {A B : I → J → ℝ} (_hB : IsPositive B) : lamB0 A B = muB0 A B
- Formalizes
- EconCSLib, `Loomis.loomis_value_IJ_2`
- Lean proof
- By xbei (from the file's git history), from github.com/gametheoryinlean/EconCSLib at
1a88f809b538(Copyright (c) 2026 EconCSLib contributors. All rights reserved, Apache-2.0). Compiled here against leanprover/lean4:v4.33.1, Mathlib v4.33.1; written for leanprover/lean4:v4.30.0.Loomis.loomis_value_IJ_2← Loomis.loomis_value_IJ_2 (EconCSLib/Math/Minimax/Loomis.lean:303): verbatim
- Verified
- 08 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 14
- Loomis.ByColumn-vector product
(By)_i = ∑ⱼ Bᵢⱼ yⱼ. - Loomis.AyColumn-vector product
(Ay)_i = ∑ⱼ Aᵢⱼ yⱼ. - wsumWeighted sum of
fwith weights from a simplex elementx. Thinabbrevover Mathlib's finite dot product⬝ᵥ: definitionally equal,… - Loomis.colRatioRow player's per-column Loomis ratio
(xA)_j / (xB)_j. - Loomis.lamB.auxPlayer I's guaranteed Loomis ratio under mixed strategy
x: infimum over pure columns. - Loomis.muB0Minmax Loomis scalar
μ₀ = inf_y μ_aux(y). - wsum_pure_applyWeighted sum at a point mass evaluates the chosen coordinate.
- Loomis.xBRow-vector product
(xB)_j = ∑ᵢ xᵢ Bᵢⱼ. - Loomis.xARow-vector product
(xA)_j = ∑ᵢ xᵢ Aᵢⱼ. - MinimaxLoomis.singleton_of_card_oneWhen the row index type has cardinality 1, every simplex point is the unique pure strategy and
Finset.univis a singleton. - Loomis.IsPositiveA matrix
B : I → J → ℝis entrywise positive if every entry is> 0. This is the hypothesis driving the general Loomis theorem, and… - Loomis.rowRatioColumn player's per-row Loomis ratio
(Ay)_i / (By)_i. - Loomis.muB.auxPlayer II's Loomis-ratio cap under mixed strategy
y: supremum over pure rows. - Loomis.lamB0Maxmin Loomis scalar
λ₀ = sup_x λ_aux(x).
- Loomis.ByColumn-vector product
- Used by 1
- Loomis.loomis_value_eq_auxLoomis induction:
lamB0 A B = muB0 A Bfor any finite positive-Bmatrix pair with2 ≤ |I| + |J|, by strong induction on the…
- Loomis.loomis_value_eq_auxLoomis induction:
Lean source view module on GitHub
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
set_option autoImplicit true inset_option relaxedAutoImplicit true inopen Finset BigOperators inset_option linter.unusedSectionVars false invariable {I J : Type*} [Fintype I] [Fintype J] [Nonempty I] [Nonempty J] in/-- Base case of the Loomis induction: a 1×1 matrix pair has the single ratio `A i₀ j₀ / B i₀ j₀` as the common Loomis value. -/theorem Loomis.loomis_value_IJ_2 (Hn : 2 = Fintype.card I + Fintype.card J) {A B : I → J → ℝ} (_hB : IsPositive B) : lamB0 A B = muB0 A B := by classical have ⟨HSI, HSJ⟩ : Fintype.card I = 1 ∧ Fintype.card J = 1 := by have p1 := @Fintype.card_pos I _ _ have p2 := @Fintype.card_pos J _ _ refine ⟨?_, ?_⟩ <;> omega obtain ⟨i0, hi⟩ := MinimaxLoomis.singleton_of_card_one HSI obtain ⟨j0, hj⟩ := MinimaxLoomis.singleton_of_card_one HSJ -- On a singleton simplex every distribution puts mass 1 at the single point. have Hxx0 : ∀ x : stdSimplex ℝ I, x.val i0 = 1 := by intro x have hsum : (∑ i : I, x.val i) = 1 := x.property.2 have hcol : (∑ i : I, x.val i) = x.val i0 := by rw [show (Finset.univ : Finset I) = {i0} from hi, Finset.sum_singleton] linarith have Hyy0 : ∀ y : stdSimplex ℝ J, y.val j0 = 1 := by intro y have hsum : (∑ j : J, y.val j) = 1 := y.property.2 have hrow : (∑ j : J, y.val j) = y.val j0 := by rw [show (Finset.univ : Finset J) = {j0} from hj, Finset.sum_singleton] linarith -- Both ratios reduce to A i0 j0 / B i0 j0 regardless of the strategy. have HlamB : ∀ x, lamB.aux A B x = A i0 j0 / B i0 j0 := by intro x simp only [lamB.aux, hj, Finset.inf'_singleton] show colRatio A B x j0 = A i0 j0 / B i0 j0 unfold colRatio xA xB have hxA : wsum x (fun i => A i j0) = A i0 j0 := by change (∑ i, x.val i * A i j0) = A i0 j0 rw [show (Finset.univ : Finset I) = {i0} from hi, Finset.sum_singleton, Hxx0 x, one_mul] have hxB : wsum x (fun i => B i j0) = B i0 j0 := by change (∑ i, x.val i * B i j0) = B i0 j0 rw [show (Finset.univ : Finset I) = {i0} from hi, Finset.sum_singleton, Hxx0 x, one_mul] rw [hxA, hxB] have HmuB : ∀ y, muB.aux A B y = A i0 j0 / B i0 j0 := by intro y simp only [muB.aux, hi, Finset.sup'_singleton] show rowRatio A B y i0 = A i0 j0 / B i0 j0 unfold rowRatio Ay By have hAy : wsum y (fun j => A i0 j) = A i0 j0 := by change (∑ j, y.val j * A i0 j) = A i0 j0 rw [show (Finset.univ : Finset J) = {j0} from hj, Finset.sum_singleton, Hyy0 y, one_mul] have hBy : wsum y (fun j => B i0 j) = B i0 j0 := by change (∑ j, y.val j * B i0 j) = B i0 j0 rw [show (Finset.univ : Finset J) = {j0} from hj, Finset.sum_singleton, Hyy0 y, one_mul] rw [hAy, hBy] rw [lamB0, iSup_congr HlamB, ciSup_const, muB0, iInf_congr HmuB, ciInf_const]