AFTD Auto-Formalizing Theoretical Domains

mul_pow_le_add_div_two_pow_two_mul

lemmaverified

For any nonnegative real numbers u and v and any natural number k, (u * v)^k is at most ((u + v) / 2)^(2 * k).

Statement

lemma mul_pow_le_add_div_two_pow_two_mul (u v : ℝ) (hu : 0 ≤ u) (hv : 0 ≤ v) (k : ℕ) :
    (u * v) ^ k ≤ ((u + v) / 2) ^ (2 * k)
Source
arXiv:2611.15027 Section 6.1
Verified
15 Nov 2026
Axioms
Classical.choiceQuot.soundpropext

Read back from the Lean

For all real numbers $u$ and $v$ such that $0 \le u$ and $0 \le v$, and for any natural number $k$, $(u \cdot v)^k \le \left(\frac{u + v}{2}\right)^{2k}$.

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

Lean source view module on GitHub

/-- The AM-GM inequality for power products: (u * v)^k ≤ ((u + v) / 2)^(2 * k) for nonnegative reals. -/
lemma mul_pow_le_add_div_two_pow_two_mul (u v : ℝ) (hu : 0 ≤ u) (hv : 0 ≤ v) (k : ℕ) :
    (u * v) ^ k ≤ ((u + v) / 2) ^ (2 * k) := by
  have h : u * v ≤ ((u + v) / 2) ^ 2 := by nlinarith [sq_nonneg (u - v)]
  have h0 : 0 ≤ u * v := mul_nonneg hu hv
  calc (u * v) ^ k ≤ (((u + v) / 2) ^ 2) ^ k := pow_le_pow_left₀ h0 h k
    _ = ((u + v) / 2) ^ (2 * k) := by rw [pow_mul]
See it in the graphDiscuss this resultChallenge it