AFTD Auto-Formalizing Theoretical Domains

energy_filter_abs_convolution_control

theoremverified

For a real integer sequence a in ℓ^(4/3), each absolute convolution sum is finite and the squares of those sums form a summable sequence.

Statement

theorem energy_filter_abs_convolution_control
    (a : ℤ → ℝ) (ha_lp : Memℓp a (4 / 3)) :
    (∀ m : ℤ, Summable (fun j : ℤ => |a j * a (m-j)|)) ∧
    Summable (fun m : ℤ => (∑' j : ℤ, |a j * a (m-j)|)^2)
Source
OP-892: convolution and quartic-cycle summability.
Verified
14 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 1
Used by 3

The proof, explained

The proof reduces the problem for an ℓ^(4/3) sequence a to an existing control theorem for non-negative ℓ^4 sequences by setting u(i) = |a(i)|^(1/3). Under this substitution, u is non-negative and satisfies u(i)^4 = |a(i)|^(4/3), placing u in ℓ^4 because a is in ℓ^(4/3). Furthermore, u(i)^3 = |a(i)|, which identifies the absolute convolution |a(j) a(m-j)| with the convolution u(j)^3 u(m-j)^3. Applying energy_filter_nonnegative_fourth_power_convolution_control to u then immediately yields both claims.

  1. The auxiliary sequence u(i) = |a(i)|^(1/3) is defined and non-negative everywhere: u(i) ≥ 0 for all i ∈ ℤ.

    We define u(i) = |a(i)|^(1/3). Since |a(i)| ≥ 0 by abs_nonneg, its real power is non-negative by Real.rpow_nonneg.

    lines 5–6

    uses Real.rpow_nonneg, abs_nonneg

      let u : ℤ → ℝ := fun i => |a i| ^ (1/3:ℝ)
      have hu0 (i : ℤ) : 0 ≤ u i := Real.rpow_nonneg (abs_nonneg _) _
  2. The cube of u reproduces the absolute value: u(i)^3 = |a(i)| for all i ∈ ℤ.

    Unfolding the definition of u, (|a(i)|^(1/3))^3 equals |a(i)|^((1/3) * 3) by Real.rpow_mul_natCast. Evaluating (1/3) * 3 = 1 gives |a(i)|^1 = |a(i)|.

    lines 7–10

    uses Real.rpow_mul_natCast, abs_nonneg

      have hu3 (i : ℤ) : u i ^ 3 = |a i| := by
        dsimp [u]
        rw [← Real.rpow_mul_natCast (abs_nonneg _) (1/3:ℝ) 3]
        norm_num
  3. The fourth power of u satisfies u(i)^4 = |a(i)|^(4/3) for all i ∈ ℤ.

    Similarly, (|a(i)|^(1/3))^4 equals |a(i)|^((1/3) * 4) by Real.rpow_mul_natCast, which simplifies by arithmetic to |a(i)|^(4/3).

    lines 11–14

    uses Real.rpow_mul_natCast, abs_nonneg

      have hu4 (i : ℤ) : u i ^ 4 = |a i| ^ (4/3:ℝ) := by
        dsimp [u]
        rw [← Real.rpow_mul_natCast (abs_nonneg _) (1/3:ℝ) 4]
        norm_num
  4. The sequence i ↦ u(i)^4 is summable, so u ∈ ℓ^4(ℤ).

    Because a ∈ ℓ^(4/3) and 4/3 > 0, the sum of ‖a(i)‖^(4/3) converges. Since the norm on ℝ is the absolute value (Real.norm_eq_abs) and u(i)^4 = |a(i)|^(4/3), the series ∑ u(i)^4 converges.

    lines 15–21

    uses Real.norm_eq_abs

      have hp : 0 < ((4/3:ENNReal)).toReal := by norm_num
      have hnorm := ha_lp.summable hp
      have hu : Summable (fun i => u i ^ 4) := by
        convert hnorm using 1
        funext i
        rw [hu4, Real.norm_eq_abs]
        norm_num
  5. The convolution of u^3 with itself converges at every index, and the sequence of squared convolution sums is summable.

    We apply the previously established theorem energy_filter_nonnegative_fourth_power_convolution_control to the non-negative sequence u ∈ ℓ^4(ℤ).

    line 22

    uses energy_filter_nonnegative_fourth_power_convolution_control

      obtain ⟨hconv, hconv2⟩ := energy_filter_nonnegative_fourth_power_convolution_control u hu0 hu
  6. For every m ∈ ℤ, the absolute convolution sum j ↦ |a(j) * a(m-j)| is summable.

    By multiplicativity of the absolute value (abs_mul) and the identity |a(i)| = u(i)^3, the term |a(j) * a(m-j)| equals u(j)^3 * u(m-j)^3. Summability is therefore inherited directly from that of the convolution of u^3.

    lines 23–26

    uses abs_mul

      have hconv_abs (m : ℤ) : Summable (fun j : ℤ => |a j * a (m-j)|) := by
        convert hconv m using 1
        funext j
        rw [hu3, hu3, abs_mul]
  7. The sequence of squared convolution sums m ↦ (∑_j |a(j) * a(m-j)|)^2 is summable.

    Because the series for |a(j) * a(m-j)| and u(j)^3 * u(m-j)^3 agree termwise, their infinite sums are equal by tsum_congr, and so their squares are equal. The summability of their squares follows from the corresponding result for u^3.

    lines 27–33

    uses tsum_congr, abs_mul

      have hconv2' : Summable (fun m : ℤ => (∑' j : ℤ, |a j * a (m-j)|)^2) := by
        convert hconv2 using 1
        funext m
        congr 1
        apply tsum_congr
        intro j
        rw [hu3, hu3, abs_mul]
  8. Both conclusions of the theorem hold.

    The proof concludes by assembling the pointwise summability of the convolution sums and the summability of their squares into a single conjunction.

    line 34

      exact ⟨hconv_abs, hconv2'⟩
abs_nonneg
The absolute value of any element in an ordered group is non-negative.
abs_mul
For real numbers, the absolute value of a product is the product of the absolute values: |a * b| = |a| * |b|.
energy_filter_nonnegative_fourth_power_convolution_control
For any non-negative sequence u in ℓ^4(ℤ), each convolution sum of u^3 with itself is finite, and the squares of these sums form a summable sequence.
Real.norm_eq_abs
The standard norm on the real numbers coincides with the absolute value: ‖r‖ = |r|.
Real.rpow_mul_natCast
For x ≥ 0, real exponent y, and natural number n, (x^y)^n = x^(y * n).
Real.rpow_nonneg
Any real power of a non-negative real number is non-negative: if x ≥ 0, then x^y ≥ 0.
tsum_congr
If two sequences have identical terms, their infinite sums are equal.

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 energy_filter_abs_convolution_control    (a : ℤ → ℝ) (ha_lp : Memℓp a (4 / 3)) :    (∀ m : ℤ, Summable (fun j : ℤ => |a j * a (m-j)|)) ∧    Summable (fun m : ℤ => (∑' j : ℤ, |a j * a (m-j)|)^2) := by  let u : ℤ → ℝ := fun i => |a i| ^ (1/3:ℝ)  have hu0 (i : ℤ) : 0 ≤ u i := Real.rpow_nonneg (abs_nonneg _) _  have hu3 (i : ℤ) : u i ^ 3 = |a i| := by    dsimp [u]    rw [← Real.rpow_mul_natCast (abs_nonneg _) (1/3:ℝ) 3]    norm_num  have hu4 (i : ℤ) : u i ^ 4 = |a i| ^ (4/3:ℝ) := by    dsimp [u]    rw [← Real.rpow_mul_natCast (abs_nonneg _) (1/3:ℝ) 4]    norm_num  have hp : 0 < ((4/3:ENNReal)).toReal := by norm_num  have hnorm := ha_lp.summable hp  have hu : Summable (fun i => u i ^ 4) := by    convert hnorm using 1    funext i    rw [hu4, Real.norm_eq_abs]    norm_num  obtain ⟨hconv, hconv2⟩ := energy_filter_nonnegative_fourth_power_convolution_control u hu0 hu  have hconv_abs (m : ℤ) : Summable (fun j : ℤ => |a j * a (m-j)|) := by    convert hconv m using 1    funext j    rw [hu3, hu3, abs_mul]  have hconv2' : Summable (fun m : ℤ => (∑' j : ℤ, |a j * a (m-j)|)^2) := by    convert hconv2 using 1    funext m    congr 1    apply tsum_congr    intro j    rw [hu3, hu3, abs_mul]  exact ⟨hconv_abs, hconv2'⟩
See it in the graphDiscuss this resultChallenge it