AFTD Auto-Formalizing Theoretical Domains

one_strongly_euclidean_sphere_extension_scalar_bounds

theoremverified

For positive r and s with s squared plus twice r s equal to one, q at least one, and L equal to q divided by 2 r plus s q, the contraction factor t=sL satisfies the cross-layer identity and the inequalities needed for all positive Hamming distances.

Statement

theorem one_strongly_euclidean_sphere_extension_scalar_bounds (r s q L : ℝ)
    (hr : 0 < r) (hs : 0 < s) (hq : 1 ≤ q) (hid : s^2 + 2*r*s = 1)
    (hL : 0 ≤ L) (hdef : L * (2*r + s*q) = q) :
    let t := s*L
    q^2 * (1-t)^2 = 4*L^2*r^2 ∧ t ≤ L^2 ∧
      ∀ h : ℝ, 1 ≤ h → t^2*h ≤ L^2 * min (s^2*h^2) h
Source
arXiv:2611.08546, Section 6.2; algebra for the symmetric extension of its two-layer cube.
Verified
12 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Used by 1

The proof, explained

The proof establishes three properties for the contraction factor t = s*L: an exact cross-layer squared identity, the bound t ≤ L^2, and an upper bound on t^2*h for all h ≥ 1. The defining relation L*(2*r + s*q) = q rearranges to q*(1 - s*L) = 2*L*r, whose square gives the first identity. The sphere condition s^2 + 2*r*s = 1 with q ≥ 1 implies s ≤ L, which immediately yields t ≤ L^2 since L is non-negative. Finally, distributing L^2 over the minimum reduces the third bound to two separate inequalities: (s*L)^2*h ≤ (s*L)^2*h^2 (following from h ≥ 1) and (s*L)^2*h ≤ L^2*h (following from s^2 ≤ 1).

  1. The bounds s^2 ≤ 1 and 0 ≤ 2*r*s*(q-1) hold, and q*(s^2 + 2*r*s) = q.

    We simplify the goal by expanding the definition t = s*L. Since s^2 + 2*r*s = 1 and r, s > 0, we have s^2 ≤ 1. Because r, s > 0 and q ≥ 1, the factor 2*r*s*(q-1) is non-negative. Multiplying both sides of the identity s^2 + 2*r*s = 1 by q using congrArg yields q*(s^2 + 2*r*s) = q.

    lines 7–10

    uses congrArg

      dsimp only
      have hsq : s^2 ≤ 1 := by nlinarith
      have hp : 0 ≤ 2*r*s*(q-1) := by positivity
      have hmul := congrArg (fun x : ℝ => q*x) hid
  2. s ≤ L

    Since r, s > 0 and q ≥ 1, the denominator term 2*r + s*q is strictly positive. Expanding s*(2*r + s*q) = 2*r*s + s^2*q and bounding it by q*(2*r*s + s^2) = q shows that s*(2*r + s*q) ≤ q. Comparing this with the definition L*(2*r + s*q) = q and dividing by the positive quantity 2*r + s*q gives s ≤ L.

    lines 11–14

      have hsL : s ≤ L := by
        have hd : 0 < 2*r+s*q := by positivity
        have hq_le : s*(2*r+s*q) ≤ q := by nlinarith
        nlinarith
  3. q*(1 - s*L) = 2*L*r and (q*(1 - s*L))^2 = (2*L*r)^2

    Rearranging the definition L*(2*r + s*q) = q gives q - q*s*L = 2*L*r, which is q*(1 - s*L) = 2*L*r. Squaring both sides via congrArg gives (q*(1 - s*L))^2 = (2*L*r)^2.

    lines 15–16

    uses congrArg

      have hid2 : q*(1-s*L)=2*L*r := by nlinarith [hdef]
      have hsqeq := congrArg (fun x : ℝ => x^2) hid2
  4. The proof reduces to establishing the three conjuncts separately.

    We decompose the conjunction into three subgoals: the identity q^2 * (1-t)^2 = 4*L^2*r^2, the upper bound t ≤ L^2, and the inequality for all h ≥ 1.

    line 17

      refine ⟨?_, ?_, ?_⟩
  5. q^2 * (1 - s*L)^2 = 4*L^2*r^2

    This is the algebraic expansion of the squared relation (q*(1 - s*L))^2 = (2*L*r)^2.

    line 18

      · nlinarith [hsqeq]
  6. s*L ≤ L^2

    Since 0 ≤ L and s ≤ L, the difference L - s is non-negative by sub_nonneg.mpr. By mul_nonneg, the product L*(L - s) = L^2 - s*L is non-negative, which gives s*L ≤ L^2.

    line 19

    uses mul_nonneg, sub_nonneg.mpr

      · nlinarith [mul_nonneg hL (sub_nonneg.mpr hsL)]
  7. For any h ≥ 1, it suffices to show both (s*L)^2 * h ≤ L^2 * (s^2 * h^2) and (s*L)^2 * h ≤ L^2 * h.

    Let h ≥ 1 be given. Since L^2 is non-negative by sq_nonneg, multiplication by L^2 distributes over the minimum by mul_min_of_nonneg, rewriting L^2 * min (s^2*h^2) h as min (L^2*(s^2*h^2)) (L^2*h). By le_min, the inequality holds if (s*L)^2 * h is bounded by both terms of the minimum.

    lines 20–22

    uses le_min, mul_min_of_nonneg, sq_nonneg

      · intro h hh
        rw [mul_min_of_nonneg _ _ (sq_nonneg L)]
        apply le_min
  8. (s*L)^2 * h ≤ L^2 * (s^2 * h^2)

    From h ≥ 1, we have h^2 - h ≥ 0. Multiplying this by the non-negative square (s*L)^2 using mul_nonneg and sq_nonneg gives (s*L)^2 * (h^2 - h) ≥ 0, which expands and rearranges to (s*L)^2 * h ≤ L^2 * s^2 * h^2.

    lines 23–25

    uses mul_nonneg, sq_nonneg

        · have hhh : 0 ≤ h^2-h := by nlinarith
          have := mul_nonneg (sq_nonneg (s*L)) hhh
          nlinarith
  9. (s*L)^2 * h ≤ L^2 * h

    Since L^2 ≥ 0, h ≥ 1 > 0, and s^2 ≤ 1 (so 1 - s^2 ≥ 0), the product L^2 * h * (1 - s^2) is non-negative. Expanding this product yields L^2 * h - (s*L)^2 * h ≥ 0, which rearranges to (s*L)^2 * h ≤ L^2 * h.

    lines 26–27

        · have hp2 : 0 ≤ L^2*h*(1-s^2) := by positivity
          nlinarith
congrArg
Applying a function to both sides of an equality preserves the equality: if a = b, then f(a) = f(b).
le_min
A value is at most the minimum of two values if and only if it is at most each of them: c ≤ min a b holds whenever c ≤ a and c ≤ b.
mul_min_of_nonneg
Multiplication by a non-negative number distributes over the minimum: if 0 ≤ a, then a * min b c = min (a * b) (a * c).
mul_nonneg
The product of two non-negative numbers is non-negative: if 0 ≤ a and 0 ≤ b, then 0 ≤ a * b.
sq_nonneg
The square of any real number is non-negative: 0 ≤ a^2 for all a.
sub_nonneg.mpr
A difference is non-negative if the subtrahend is at most the minuend: if b ≤ a, then 0 ≤ a - b.

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 one_strongly_euclidean_sphere_extension_scalar_bounds (r s q L : ℝ)    (hr : 0 < r) (hs : 0 < s) (hq : 1 ≤ q) (hid : s^2 + 2*r*s = 1)    (hL : 0 ≤ L) (hdef : L * (2*r + s*q) = q) :    let t := s*L    q^2 * (1-t)^2 = 4*L^2*r^2 ∧ t ≤ L^2 ∧      ∀ h : ℝ, 1 ≤ h → t^2*h ≤ L^2 * min (s^2*h^2) h := by  dsimp only  have hsq : s^2 ≤ 1 := by nlinarith  have hp : 0 ≤ 2*r*s*(q-1) := by positivity  have hmul := congrArg (fun x : ℝ => q*x) hid  have hsL : s ≤ L := by    have hd : 0 < 2*r+s*q := by positivity    have hq_le : s*(2*r+s*q) ≤ q := by nlinarith    nlinarith  have hid2 : q*(1-s*L)=2*L*r := by nlinarith [hdef]  have hsqeq := congrArg (fun x : ℝ => x^2) hid2  refine ⟨?_, ?_, ?_⟩  · nlinarith [hsqeq]  · nlinarith [mul_nonneg hL (sub_nonneg.mpr hsL)]  · intro h hh    rw [mul_min_of_nonneg _ _ (sq_nonneg L)]    apply le_min    · have hhh : 0 ≤ h^2-h := by nlinarith      have := mul_nonneg (sq_nonneg (s*L)) hhh      nlinarith    · have hp2 : 0 ≤ L^2*h*(1-s^2) := by positivity      nlinarith
See it in the graphDiscuss this resultChallenge it