AFTD Auto-Formalizing Theoretical Domains

four_point_positive_kernel_mixed_direction

theoremverified

Let a,b be two integer rows on four coordinates and x a strictly positive integer vector annihilated by both rows. If the first two columns have nonzero determinant, then the common integer kernel contains a mixed-sign vector w whose coordinate sum is zero. The two zero row sums and the positive and negative entries are explicit conclusions.

Statement

theorem four_point_positive_kernel_mixed_direction
    (a b x : Fin 4 → ℤ) (hx : ∀ i, 0 < x i)
    (hxa : (∑ i, x i*a i) = 0) (hxb : (∑ i, x i*b i) = 0)
    (hind : a 0*b 1 - a 1*b 0 ≠ 0) :
    ∃ w : Fin 4 → ℤ, (∑ i, w i) = 0 ∧
      (∑ i, w i*a i) = 0 ∧ (∑ i, w i*b i) = 0 ∧
      (∃ i, 0 < w i) ∧ (∃ j, w j < 0)
Source
Supporting lemma or explicit route obstruction for OP-197 / arXiv:2610.16562 Conjecture1, deep session2. The abstract two-ray fact and finite obstruction are not claimed novel or research-level; no part is the full box conjecture. The five-point kernel has columns (1,1),(1,0),(0,1),(-2,0),(0,-2).
Verified
28 Oct 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 2
  • LinearOrder.toTotalPreorderEvery LinearOrder is a TotalPreorder.
  • TotalPreorderA total preorder: a preorder where ≤ is total (complete). This is weaker than LinearOrder — it does NOT require antisymmetry or…
Used by 1

The proof, explained

The proof defines an auxiliary vector c as the 3D cross product of the first three coordinates of a and b (with fourth coordinate 0), which is orthogonal to both a and b. Setting T = ∑ x i > 0 and S = ∑ c i, it forms the linear combination w = T*c - S*x, whose coordinate sum vanishes and which remains in the common kernel of a and b. Evaluating w at indices 3 and 2 shows that w ≠ 0 thanks to the nonzero 2×2 determinant and positivity of x; because its coordinates sum to zero and w is nonzero, w must have both a strictly positive and a strictly negative entry.

  1. The auxiliary vector c = (a 1*b 2 - a 2*b 1, a 2*b 0 - a 0*b 2, a 0*b 1 - a 1*b 0, 0) is orthogonal to both a and b.

    The vector c is defined as the 3D cross product of the first three coordinates of a and b, with zero in the fourth coordinate. The dot products ∑ i, c i * a i and ∑ i, c i * b i vanish because they expand to 3×3 determinants with duplicate rows; Lean expands the finite sum with `Fin.sum_univ_succ` and `norm_num`, then verifies the polynomial cancellation with `ring`.

    lines 12–14

    uses Fin.sum_univ_succ

      let c : Fin 4 → ℤ := ![a 1*b 2-a 2*b 1, a 2*b 0-a 0*b 2, a 0*b 1-a 1*b 0, 0]
      have hca : (∑ i, c i*a i) = 0 := by norm_num [Fin.sum_univ_succ,c]; ring
      have hcb : (∑ i, c i*b i) = 0 := by norm_num [Fin.sum_univ_succ,c]; ring
  2. With T = ∑ i, x i, S = ∑ i, c i, and candidate vector w = T*c - S*x, the scalar T is strictly positive.

    Since every coordinate x i is strictly positive by assumption (hx) and the index type Fin 4 is nonempty (containing 0), the sum T is strictly positive by `Finset.sum_pos`.

    lines 15–20

    uses Finset.sum_pos, Finset.mem_univ

      let T := ∑ i, x i
      let S := ∑ i, c i
      let w := fun i => T*c i-S*x i
      have hT : 0 < T := by
        have h0 := hx 0; have h1 := hx 1; have h2 := hx 2; have h3 := hx 3
        exact Finset.sum_pos (fun k _ => hx k) ⟨0,Finset.mem_univ 0⟩
  3. The sum of the coordinates of w is zero: ∑ i, w i = 0.

    By definition w i = T * c i - S * x i. Distributing the sum using `Finset.sum_sub_distrib` and factoring out the scalars with `Finset.mul_sum` rewrites ∑ i, w i as T * S - S * T, which is 0 by commutativity of integer multiplication.

    lines 21–24

    uses Finset.sum_sub_distrib, Finset.mul_sum

      have hsw : (∑ i, w i) = 0 := by
        simp only [w, Finset.sum_sub_distrib, ← Finset.mul_sum]
        change T*S-S*T=0
        ring
  4. The vector w is orthogonal to a: ∑ i, w i * a i = 0.

    Distributing the sum gives ∑ i, w i * a i = T * (∑ i, c i * a i) - S * (∑ i, x i * a i). Since c is orthogonal to a (hca) and x is orthogonal to a (hxa), both terms vanish, yielding T * 0 - S * 0 = 0.

    lines 25–31

    uses Finset.mul_sum, Finset.sum_sub_distrib, Finset.sum_congr

      have hwa : (∑ i, w i*a i) = 0 := by
        calc
          _ = T*(∑ i, c i*a i)-S*(∑ i, x i*a i) := by
            simp only [Finset.mul_sum, ← Finset.sum_sub_distrib]
            apply Finset.sum_congr rfl
            intro i hi; dsimp [w]; ring
          _ = 0 := by rw [hca,hxa]; ring
  5. The vector w is orthogonal to b: ∑ i, w i * b i = 0.

    Similarly, ∑ i, w i * b i = T * (∑ i, c i * b i) - S * (∑ i, x i * b i). Since c and x are both orthogonal to b (hcb and hxb), this evaluates to T * 0 - S * 0 = 0.

    lines 32–38

    uses Finset.mul_sum, Finset.sum_sub_distrib, Finset.sum_congr

      have hwb : (∑ i, w i*b i) = 0 := by
        calc
          _ = T*(∑ i, c i*b i)-S*(∑ i, x i*b i) := by
            simp only [Finset.mul_sum, ← Finset.sum_sub_distrib]
            apply Finset.sum_congr rfl
            intro i hi; dsimp [w]; ring
          _ = 0 := by rw [hcb,hxb]; ring
  6. The vector w is not the zero vector: w ≠ 0.

    Assuming w = 0, evaluating at index 3 via `congrFun` gives T * 0 - S * x 3 = 0, which implies S = 0 since x 3 > 0. Evaluating at index 2 with S = 0 then reduces to T * (a 0 * b 1 - a 1 * b 0) = 0. By `mul_eq_zero.mp`, either T = 0 (contradicting T > 0) or a 0 * b 1 - a 1 * b 0 = 0 (contradicting the hypothesis hind). Both branches are impossible, so w ≠ 0.

    lines 39–50

    uses congrFun, zero_mul, sub_zero, mul_eq_zero.mp

      have hwne : w ≠ 0 := by
        intro hh
        have hh3 := congrFun hh 3
        have hh2 := congrFun hh 2
        change T*0-S*x 3=0 at hh3
        change T*(a 0*b 1-a 1*b 0)-S*x 2=0 at hh2
        have hS : S=0 := by have := hx 3; nlinarith
        rw [hS] at hh2
        simp only [zero_mul,sub_zero] at hh2
        rcases mul_eq_zero.mp hh2 with h | h
        · omega
        · exact hind h
  7. There exists an index i such that w i > 0.

    Suppose, for contradiction, that no such index exists, so w i ≤ 0 for all i. Because ∑ i, w i = 0 and each coordinate is nonpositive, each w i must be 0 (verified coordinate-by-coordinate using `fin_cases` and `omega`). By `funext`, this would mean w = 0, contradicting w ≠ 0.

    lines 51–64

      have hps : ∃ i, 0 < w i := by
        by_contra hh
        push_neg at hh
        have h0 := hh 0; have h1 := hh 1; have h2 := hh 2; have h3 := hh 3
        have hsum := hsw
        change w 0+(w 1+(w 2+(w 3+0)))=0 at hsum
        have heq : w=0 := by
          funext i
          fin_cases i
          · change w 0=0; omega
          · change w 1=0; omega
          · change w 2=0; omega
          · change w 3=0; omega
        exact hwne heq
  8. There exists an index j such that w j < 0.

    Suppose, for contradiction, that no coordinate is negative, so w j ≥ 0 for all j. Since ∑ j, w j = 0 and all terms are nonnegative, each w j must be 0 (checked case-by-case with `fin_cases` and `omega`). This again implies w = 0, contradicting w ≠ 0.

    lines 65–78

      have hns : ∃ i, w i < 0 := by
        by_contra hh
        push_neg at hh
        have h0 := hh 0; have h1 := hh 1; have h2 := hh 2; have h3 := hh 3
        have hsum := hsw
        change w 0+(w 1+(w 2+(w 3+0)))=0 at hsum
        have heq : w=0 := by
          funext i
          fin_cases i
          · change w 0=0; omega
          · change w 1=0; omega
          · change w 2=0; omega
          · change w 3=0; omega
        exact hwne heq
  9. There exists a vector w satisfying all five conditions of the theorem.

    Combining the constructed vector w with the five established properties (coordinate sum zero, orthogonality to a, orthogonality to b, a strictly positive coordinate, and a strictly negative coordinate) completes the proof.

    line 79

      exact ⟨w,hsw,hwa,hwb,hps,hns⟩
Fin.sum_univ_succ
A sum of a function over `Fin (n + 1)` equals the value at `0` plus the sum of the remaining terms over `Fin n`.
Finset.sum_pos
The sum of a strictly positive function over a non-empty finite set is strictly positive.
Finset.mem_univ
Every element of a finite type belongs to its universal finset.
Finset.sum_sub_distrib
The sum of a difference over a finite set is the difference of the sums.
Finset.mul_sum
A constant factor can be distributed into or factored out of a finite sum.
Finset.sum_congr
If two index sets are equal and two functions agree at each element of the index set, their sums are equal.
congrFun
If two functions are equal, their values at any argument are equal.
zero_mul
Multiplying zero by any element yields zero.
sub_zero
Subtracting zero from any element leaves it unchanged.
mul_eq_zero.mp
If the product of two integers is zero, at least one of the factors is zero.

Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.

Lean source view module on GitHub

open Finsetset_option maxHeartbeats 1000000set_option maxRecDepth 4000theorem four_point_positive_kernel_mixed_direction    (a b x : Fin 4 → ℤ) (hx : ∀ i, 0 < x i)    (hxa : (∑ i, x i*a i) = 0) (hxb : (∑ i, x i*b i) = 0)    (hind : a 0*b 1 - a 1*b 0 ≠ 0) :    ∃ w : Fin 4 → ℤ, (∑ i, w i) = 0 ∧      (∑ i, w i*a i) = 0 ∧ (∑ i, w i*b i) = 0 ∧      (∃ i, 0 < w i) ∧ (∃ j, w j < 0) := by  let c : Fin 4 → ℤ := ![a 1*b 2-a 2*b 1, a 2*b 0-a 0*b 2, a 0*b 1-a 1*b 0, 0]  have hca : (∑ i, c i*a i) = 0 := by norm_num [Fin.sum_univ_succ,c]; ring  have hcb : (∑ i, c i*b i) = 0 := by norm_num [Fin.sum_univ_succ,c]; ring  let T := ∑ i, x i  let S := ∑ i, c i  let w := fun i => T*c i-S*x i  have hT : 0 < T := by    have h0 := hx 0; have h1 := hx 1; have h2 := hx 2; have h3 := hx 3    exact Finset.sum_pos (fun k _ => hx k) ⟨0,Finset.mem_univ 0⟩  have hsw : (∑ i, w i) = 0 := by    simp only [w, Finset.sum_sub_distrib, ← Finset.mul_sum]    change T*S-S*T=0    ring  have hwa : (∑ i, w i*a i) = 0 := by    calc      _ = T*(∑ i, c i*a i)-S*(∑ i, x i*a i) := by        simp only [Finset.mul_sum, ← Finset.sum_sub_distrib]        apply Finset.sum_congr rfl        intro i hi; dsimp [w]; ring      _ = 0 := by rw [hca,hxa]; ring  have hwb : (∑ i, w i*b i) = 0 := by    calc      _ = T*(∑ i, c i*b i)-S*(∑ i, x i*b i) := by        simp only [Finset.mul_sum, ← Finset.sum_sub_distrib]        apply Finset.sum_congr rfl        intro i hi; dsimp [w]; ring      _ = 0 := by rw [hcb,hxb]; ring  have hwne : w ≠ 0 := by    intro hh    have hh3 := congrFun hh 3    have hh2 := congrFun hh 2    change T*0-S*x 3=0 at hh3    change T*(a 0*b 1-a 1*b 0)-S*x 2=0 at hh2    have hS : S=0 := by have := hx 3; nlinarith    rw [hS] at hh2    simp only [zero_mul,sub_zero] at hh2    rcases mul_eq_zero.mp hh2 with h | h    · omega    · exact hind h  have hps : ∃ i, 0 < w i := by    by_contra hh    push_neg at hh    have h0 := hh 0; have h1 := hh 1; have h2 := hh 2; have h3 := hh 3    have hsum := hsw    change w 0+(w 1+(w 2+(w 3+0)))=0 at hsum    have heq : w=0 := by      funext i      fin_cases i      · change w 0=0; omega      · change w 1=0; omega      · change w 2=0; omega      · change w 3=0; omega    exact hwne heq  have hns : ∃ i, w i < 0 := by    by_contra hh    push_neg at hh    have h0 := hh 0; have h1 := hh 1; have h2 := hh 2; have h3 := hh 3    have hsum := hsw    change w 0+(w 1+(w 2+(w 3+0)))=0 at hsum    have heq : w=0 := by      funext i      fin_cases i      · change w 0=0; omega      · change w 1=0; omega      · change w 2=0; omega      · change w 3=0; omega    exact hwne heq  exact ⟨w,hsw,hwa,hwb,hps,hns⟩
See it in the graphDiscuss this resultChallenge it