AFTD Auto-Formalizing Theoretical Domains

op_1693_noon_scaling_surface_symmetric

theoremverifiedformalization

For any non-negative real loss a, at balanced amplitude q = 1/2 the asymptotic scaling surface satisfies noon_scaling_surface (1 / 2) a a = Real.exp (-2 * a) * (bessel_I0 (2 * a) + 1).

Statement

theorem op_1693_noon_scaling_surface_symmetric (a : ℝ) (ha : 0 ≤ a) :
    noon_scaling_surface (1 / 2) a a = Real.exp (-2 * a) * (bessel_I0 (2 * a) + 1)
Formalizes
arXiv:2611.17021
Verified
15 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 2
  • bessel_I0The real-valued modified Bessel function of the first kind of order zero is the function bessel_I0 : Real -> Real mapping a real number z…
  • noon_scaling_surfaceFor real balance parameter q and real loss parameters a, b, the asymptotic realignment scaling surface is defined as F_q(a, b) = q^2 *…

Read back from the Lean

For every non-negative real number $a$ (i.e., $a : \mathbb{R}$ with $0 \le a$), evaluating noon_scaling_surface at $q = 1/2$, $a$, and $b = a$ equals $\exp(-2a) \cdot (\mathrm{bessel\_I0}(2a) + 1)$.

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 evaluates the scaling surface definition at balanced amplitude q = 1/2 and symmetric loss b = a. Using the non-negativity of a, the cross-term radical √(a * a) simplifies to a, bringing all exponential and Bessel function arguments to -2a and 2a respectively. Ring normalization then sums the rational weights to 1, matching the target expression exp(-2a) * (bessel_I0(2a) + 1).

  1. The definition of noon_scaling_surface is expanded at q = 1/2 and b = a.

    Unfolding the definition of noon_scaling_surface substitutes q = 1/2 and b = a into the formula, yielding a sum of four terms involving powers of 1/2, exponentials, and Bessel functions.

    line 4

    uses noon_scaling_surface

      simp only [noon_scaling_surface]
  2. The cross-term square root √(a * a) simplifies to a.

    Because a is non-negative (ha : 0 ≤ a), Real.sqrt_mul_self simplifies √(a * a) to a, turning the Bessel argument 2 * √(a * a) into 2 * a.

    line 5

    uses Real.sqrt_mul_self

      rw [Real.sqrt_mul_self ha]
  3. The terms are collected and normalized to match the right-hand side.

    The exponential arguments -(a + a) simplify to -2 * a. The coefficients of exp(-2 * a) * bessel_I0 (2 * a) are (1/2)² + (1 - 1/2)² + 2 * (1/2) * (1 - 1/2) = 1/4 + 1/4 + 1/2 = 1, and the coefficient of exp(-2 * a) is 4 * (1/2) * (1 - 1/2) = 1. The tactic ring_nf normalizes these polynomial ring expressions, showing both sides are identical and completing the proof.

    line 6

      ring_nf
noon_scaling_surface
The asymptotic scaling surface function defined by S(q, a, b) = q² exp(-2a) I₀(2a) + (1-q)² exp(-2b) I₀(2b) + 2q(1-q) exp(-(a+b)) I₀(2√(ab)) + 4q(1-q) exp(-(a+b)).
Real.sqrt_mul_self
For any non-negative real number x ≥ 0, √(x * x) = x.

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

/-- Simplification of the asymptotic scaling surface at balanced amplitude q = 1/2 and symmetric loss a = b. -/theorem op_1693_noon_scaling_surface_symmetric (a : ℝ) (ha : 0 ≤ a) :    noon_scaling_surface (1 / 2) a a = Real.exp (-2 * a) * (bessel_I0 (2 * a) + 1) := by  simp only [noon_scaling_surface]  rw [Real.sqrt_mul_self ha]  ring_nf
See it in the graphDiscuss this resultChallenge it