AFTD Auto-Formalizing Theoretical Domains

isotropic_chain_cubic_ineq

theoremverified

For every natural number m, m - 1 ≤ (1/6) m (m^2 - 1).

Statement

theorem isotropic_chain_cubic_ineq (m : ℕ) :
    (m : ℝ) - 1 ≤ (1 / 6 : ℝ) * (m : ℝ) * ((m : ℝ) ^ 2 - 1)
Source
arXiv:2611.04644, Lemma A.4 proof step 3
Verified
05 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Used by 1
  • isotropic_intersection_size_boundLet A ⊂ ℝ^d be a finite, non-empty set in isotropic position (so that ∑_{a ∈ A} a = 0 and ∑_{a ∈ A} (a · u)^2 = |A| ‖u‖^2 for all u ∈…

Read back from the Lean

For every natural number $m$, $(m : \mathbb{R}) - 1 \le \frac{1}{6} (m : \mathbb{R}) ((m : \mathbb{R})^2 - 1)$ in the real numbers.

Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.

The proof, explained

The proof proceeds by case analysis on m ∈ ℕ. The base cases m = 0 and m = 1 are verified directly by numerical computation. For m ≥ 2, substituting m + 2 allows the difference between the right-hand side and the left-hand side to be factored algebraically as (1/6) m (m + 1) (m + 5). Because m ≥ 0, each factor is nonnegative, which implies the difference is nonnegative and completes the inequality.

  1. The inequality holds for the base cases m = 0 and m = 1.

    The tactic `rcases` splits the natural number m into three cases: 0, 1, and m + 2. When m = 0, the inequality simplifies to -1 ≤ 0; when m = 1, it simplifies to 0 ≤ 0. Both base cases are discharged directly by numerical evaluation (`norm_num`).

    lines 4–6

      rcases m with _ | _ | m
      · norm_num
      · norm_num
  2. For m ≥ 2, the difference between the right-hand side and the left-hand side factors as (1/6) m (m + 1) (m + 5).

    Writing the natural number as m + 2 and distributing the coercion to ℝ across additions (`push_cast`), expanding both sides of the identity (1/6) (m + 2) ((m + 2)^2 - 1) - ((m + 2) - 1) = (1/6) m (m + 1) (m + 5) confirms that they are equal as polynomials (`ring`).

    lines 7–9

      push_cast
      have hpoly : (1 / 6 : ℝ) * ((m : ℝ) + 1 + 1) * (((m : ℝ) + 1 + 1) ^ 2 - 1) - (((m : ℝ) + 1 + 1) - 1) =
          (1 / 6 : ℝ) * (m : ℝ) * ((m : ℝ) + 1) * ((m : ℝ) + 5) := by ring
  3. The factored difference (1/6) m (m + 1) (m + 5) is nonnegative.

    Since m is a natural number, m ≥ 0. Each factor in the product—1/6, m, m + 1, and m + 5—is nonnegative, so their product is nonnegative, as established by the `positivity` tactic.

    line 10

      have hpos : 0 ≤ (1 / 6 : ℝ) * (m : ℝ) * ((m : ℝ) + 1) * ((m : ℝ) + 5) := by positivity
  4. The inequality holds for all m ≥ 2.

    Knowing that RHS - LHS = (1/6) m (m + 1) (m + 5) and that this expression is at least 0, linear arithmetic (`linarith`) concludes that LHS ≤ RHS, completing the proof.

    line 11

      linarith

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

/-- Elementary inequality (1/6) m (m^2 - 1) >= m - 1 for all natural numbers m. -/theorem isotropic_chain_cubic_ineq (m : ℕ) :    (m : ℝ) - 1 ≤ (1 / 6 : ℝ) * (m : ℝ) * ((m : ℝ) ^ 2 - 1) := by  rcases m with _ | _ | m  · norm_num  · norm_num  push_cast  have hpoly : (1 / 6 : ℝ) * ((m : ℝ) + 1 + 1) * (((m : ℝ) + 1 + 1) ^ 2 - 1) - (((m : ℝ) + 1 + 1) - 1) =      (1 / 6 : ℝ) * (m : ℝ) * ((m : ℝ) + 1) * ((m : ℝ) + 5) := by ring  have hpos : 0 ≤ (1 / 6 : ℝ) * (m : ℝ) * ((m : ℝ) + 1) * ((m : ℝ) + 5) := by positivity  linarith
See it in the graphDiscuss this resultChallenge it