AFTD Auto-Formalizing Theoretical Domains

b3_core_ineq

theoremverified

For naturals a ≥ 2, 1 ≤ b ≤ a, c ≥ 1: (a^2 + b^2 - c^2 + 2(a+b+c)c + 3 - 2(a+b+c)) (a+c)^2 ≤ (a+b+c)^2 (a+c-1)^2.

Statement

theorem b3_core_ineq (a b c : ℕ) (hb : 1 ≤ b) (hc : 1 ≤ c) (hab : b ≤ a) (h2 : 2 ≤ a) :
    ((a:ℝ)^2 + b^2 - c^2 + 2*(a+b+c)*c + 3 - 2*(a+b+c)) * ((a:ℝ)+c)^2
      ≤ ((a:ℝ)+b+c)^2 * ((a:ℝ)+c-1)^2
Source
arXiv:2610.21388 (Stronger spectral bound for clique number), proof of Theorem 3.1 (the case n_1 ≥ n_2, (n_1,n_2) ≠ (1,1), cleared of denominators); serves b3_sqrt_trace_quotient_sq_le_clique_bound
Verified
18 Oct 2026
Axioms
Classical.choiceQuot.soundpropext
Used by 1
  • b3_sq_leFor positive naturals a, b, c with (a, b) ≠ (1, 1): sqrt(a^2 + b^2 - c^2 + 2(a+b+c)c + 3 - 2(a+b+c)) ≤ (a+b+c)(1 - 1/max(a+c, b+c)).

The proof, explained

The proof establishes the polynomial inequality over the real numbers by first casting the integer constraints $a \ge 2$, $1 \le b \le a$, and $c \ge 1$ into $\mathbb{R}$. It then forms several non-negative products from the differences $(a - 2)$, $(b - 1)$, $(c - 1)$, and $(a - b)$. Finally, nonlinear arithmetic (`nlinarith`) verifies that the difference between the right-hand side and left-hand side is non-negative as a conic combination of these witness products.

  1. The given inequalities $a \ge 2$, $b \ge 1$, $c \ge 1$, and $a \ge b$ hold when the natural numbers are cast into the real numbers.

    The hypotheses $h2$, $hb$, $hc$, and $hab$ are cast from $\mathbb{N}$ to $\mathbb{R}$, establishing $2 \le a$, $1 \le b$, $1 \le c$, and $b \le a$ as real inequalities.

    lines 4–7

      have ha' : (2:ℝ) ≤ a := by exact_mod_cast h2
      have hb' : (1:ℝ) ≤ b := by exact_mod_cast hb
      have hc' : (1:ℝ) ≤ c := by exact_mod_cast hc
      have hab' : (b:ℝ) ≤ a := by exact_mod_cast hab
  2. $((a^2 + b^2 - c^2 + 2(a+b+c)c + 3 - 2(a+b+c)) (a+c)^2 \le (a+b+c)^2 (a+c-1)^2$. automation

    From the real bounds, the linear differences $a - 2$, $b - 1$, $c - 1$, and $a - b$ are all non-negative by `sub_nonneg.2`. Applying `mul_nonneg` to these differences yields seven non-negative quadratic and cubic products: $(a-2)(b-1)$, $(c-1)(b-1)$, $(a-b)(b-1)$, $(a-b)(c-1)$, $(c-1)^2(b-1)$, $(a-2)(b-1)(c-1)$, and $(a-b)(b-1)(c-1)$. The nonlinear arithmetic solver `nlinarith` uses these non-negative products together with the linear inequalities to deduce that the right-hand side minus the left-hand side is non-negative.

    lines 8–12

    uses mul_nonneg, sub_nonneg.2

      nlinarith [mul_nonneg (sub_nonneg.2 ha') (sub_nonneg.2 hb'), mul_nonneg (sub_nonneg.2 hc') (sub_nonneg.2 hb'),
        mul_nonneg (sub_nonneg.2 hab') (sub_nonneg.2 hb'), mul_nonneg (sub_nonneg.2 hab') (sub_nonneg.2 hc'),
        mul_nonneg (mul_nonneg (sub_nonneg.2 hc') (sub_nonneg.2 hb')) (sub_nonneg.2 hc'),
        mul_nonneg (mul_nonneg (sub_nonneg.2 ha') (sub_nonneg.2 hb')) (sub_nonneg.2 hc'),
        mul_nonneg (mul_nonneg (sub_nonneg.2 hab') (sub_nonneg.2 hb')) (sub_nonneg.2 hc')]
mul_nonneg
The product of two non-negative elements is non-negative ($0 \le x$ and $0 \le y$ imply $0 \le x \cdot y$).
sub_nonneg.2
If $y \le x$, then the difference is non-negative ($0 \le x - y$).

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

theorem b3_core_ineq (a b c : ℕ) (hb : 1 ≤ b) (hc : 1 ≤ c) (hab : b ≤ a) (h2 : 2 ≤ a) :    ((a:ℝ)^2 + b^2 - c^2 + 2*(a+b+c)*c + 3 - 2*(a+b+c)) * ((a:ℝ)+c)^2      ≤ ((a:ℝ)+b+c)^2 * ((a:ℝ)+c-1)^2 := by  have ha' : (2:ℝ) ≤ a := by exact_mod_cast h2  have hb' : (1:ℝ) ≤ b := by exact_mod_cast hb  have hc' : (1:ℝ) ≤ c := by exact_mod_cast hc  have hab' : (b:ℝ) ≤ a := by exact_mod_cast hab  nlinarith [mul_nonneg (sub_nonneg.2 ha') (sub_nonneg.2 hb'), mul_nonneg (sub_nonneg.2 hc') (sub_nonneg.2 hb'),    mul_nonneg (sub_nonneg.2 hab') (sub_nonneg.2 hb'), mul_nonneg (sub_nonneg.2 hab') (sub_nonneg.2 hc'),    mul_nonneg (mul_nonneg (sub_nonneg.2 hc') (sub_nonneg.2 hb')) (sub_nonneg.2 hc'),    mul_nonneg (mul_nonneg (sub_nonneg.2 ha') (sub_nonneg.2 hb')) (sub_nonneg.2 hc'),    mul_nonneg (mul_nonneg (sub_nonneg.2 hab') (sub_nonneg.2 hb')) (sub_nonneg.2 hc')]
See it in the graphDiscuss this resultChallenge it