AFTD Auto-Formalizing Theoretical Domains

energy_filter_delta_lattice_sums

theoremverifiedoriginal

For the Kronecker delta filter delta : ℤ → ℝ defined by delta 0 = 1 and delta j = 0 for j ≠ 0, the second-moment lattice sum energy_filter_sequence_lattice_s2 delta is 1, the fourth-moment convolution sum energy_filter_sequence_lattice_s4 delta is 1, and the asymptotic purity constant energy_filter_sequence_purity_constant delta is 2 / Real.sqrt Real.pi.

Statement

theorem energy_filter_delta_lattice_sums :
    let delta : ℤ → ℝ := fun j => if j = 0 then 1 else 0
    energy_filter_sequence_lattice_s2 delta = 1 ∧
    energy_filter_sequence_lattice_s4 delta = 1 ∧
    energy_filter_sequence_purity_constant delta = 2 / Real.sqrt Real.pi
Related work
arXiv:2405.02158 Energy-filtered quantum states and the emergence of non-local correlations
Verified
04 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 3
Used by 1

Read back from the Lean

Let $\delta : \mathbb{Z} \to \mathbb{R}$ be defined by $\delta(j) = 1$ if $j = 0$ and $0$ if $j \neq 0$. The theorem asserts that: 1. energy_filter_sequence_lattice_s2 delta = 1, 2. energy_filter_sequence_lattice_s4 delta = 1, and 3. energy_filter_sequence_purity_constant delta = 2 / Real.sqrt Real.pi.

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 computes the second- and fourth-moment lattice sums for the Kronecker delta filter δ(j) = [j = 0] by direct summation. The second-moment sum S₂(δ) = ∑ j, δ(j)² collapses to 1² = 1 because only j = 0 is non-zero. The discrete convolution (δ * δ)(m) collapses via `tsum_eq_single` to δ(m), which makes the fourth-moment sum S₄(δ) = ∑ m, δ(m)² = 1. Finally, substituting S₂ = 1 and S₄ = 1 into the definition of the purity constant (2 / √π) · S₄ / S₂² immediately yields 2 / √π.

  1. Unpack the definition of the delta filter and establish that its second-moment sum ∑' j, δ(j)² equals 1. automation

    The proof unfolds the local definition δ(j) = if j = 0 then 1 else 0 across the goal. It then proves the auxiliary lemma `hsum2` stating that the sum of squares ∑' (j : ℤ), δ(j)² equals 1, which the simplifier closes since the summand is 1 at j = 0 and 0 everywhere else.

    lines 7–8

      dsimp only
      have hsum2 : (∑' (j : ℤ), (if j = 0 then (1:ℝ) else 0) ^ 2) = 1 := by simp
  2. For every integer m, the discrete convolution (δ * δ)(m) = ∑' j, δ(j) δ(m - j) equals δ(m).

    Fix an integer m. For any index j ≠ 0, δ(j) = 0, so the product δ(j) δ(m - j) vanishes. Applying `tsum_eq_single` at j = 0 reduces the infinite sum over ℤ to the single term at j = 0, which evaluates to δ(0) δ(m - 0) = 1 · δ(m) = δ(m).

    lines 9–16

    uses tsum_eq_single

      have hinner : ∀ m : ℤ,
          (∑' (j : ℤ), (if j = 0 then (1:ℝ) else 0) * (if m - j = 0 then (1:ℝ) else 0))
            = (if m = 0 then (1:ℝ) else 0) := by
        intro m
        rw [tsum_eq_single (0:ℤ)]
        · simp
        · intro b hb
          simp [hb]
  3. The fourth-moment lattice sum ∑' m, ((δ * δ)(m))² equals 1.

    Rewriting the inner convolution sum with the previously established identity `hinner` simplifies the expression to ∑' (m : ℤ), δ(m)², which Lean's simplifier evaluates to 1 just as in the second-moment calculation.

    lines 17–20

      have hsum4 : (∑' (m : ℤ), (∑' (j : ℤ),
          (if j = 0 then (1:ℝ) else 0) * (if m - j = 0 then (1:ℝ) else 0)) ^ 2) = 1 := by
        simp only [hinner]
        simp
  4. Split the main goal into three conjuncts and prove the first: S₂(δ) = 1. automation

    The proof splits the 3-part conjunction using `refine`. For the first conjunct, unfolding the definition `energy_filter_sequence_lattice_s2` reveals the sum ∑' j, δ(j)², which Lean's simplifier evaluates to 1.

    lines 21–22

    uses energy_filter_sequence_lattice_s2

      refine ⟨?_, ?_, ?_⟩
      · simp [energy_filter_sequence_lattice_s2]
  5. Prove the second conjunct: S₄(δ) = 1.

    Unfolding the definition `energy_filter_sequence_lattice_s4` transforms the second conjunct into the fourth-moment lattice sum, which matches the auxiliary lemma `hsum4` established earlier.

    lines 23–24

    uses energy_filter_sequence_lattice_s4

      · simp only [energy_filter_sequence_lattice_s4]
        exact hsum4
  6. Prove the third conjunct: the asymptotic purity constant of δ equals 2 / Real.sqrt Real.pi.

    Unfolding the definition `energy_filter_sequence_purity_constant` expresses the constant as (2 / √π) * S₄(δ) / (S₂(δ))². Substituting the known values S₂(δ) = 1 and S₄(δ) = 1 reduces the expression to (2 / √π) * 1 / 1², which `norm_num` simplifies to 2 / √π.

    lines 25–27

    uses energy_filter_sequence_purity_constant

      · simp only [energy_filter_sequence_purity_constant]
        rw [hsum2, hsum4]
        norm_num
tsum_eq_single
An infinite sum ∑' b, f b collapses to the single term f b₀ if f(b') = 0 for all b' ≠ b₀.
energy_filter_sequence_lattice_s2
The second-moment lattice sum of a sequence a : ℤ → ℝ, defined as ∑' j : ℤ, (a j)^2.
energy_filter_sequence_lattice_s4
The fourth-moment convolution lattice sum of a sequence a : ℤ → ℝ, defined as ∑' m : ℤ, (∑' j : ℤ, a j * a (m - j))^2.
energy_filter_sequence_purity_constant
The asymptotic purity constant of a sequence a : ℤ → ℝ, defined as (2 / Real.sqrt Real.pi) * S_4(a) / (S_2(a))^2.

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

/-- For the single-energy projector delta filter at zero, S_2 = 1, S_4 = 1, and the universal purity constant is 2/sqrt(pi). -/theorem energy_filter_delta_lattice_sums :    let delta : ℤ → ℝ := fun j => if j = 0 then 1 else 0    energy_filter_sequence_lattice_s2 delta = 1 ∧    energy_filter_sequence_lattice_s4 delta = 1 ∧    energy_filter_sequence_purity_constant delta = 2 / Real.sqrt Real.pi := by  dsimp only  have hsum2 : (∑' (j : ℤ), (if j = 0 then (1:ℝ) else 0) ^ 2) = 1 := by simp  have hinner : ∀ m : ℤ,      (∑' (j : ℤ), (if j = 0 then (1:ℝ) else 0) * (if m - j = 0 then (1:ℝ) else 0))        = (if m = 0 then (1:ℝ) else 0) := by    intro m    rw [tsum_eq_single (0:ℤ)]    · simp    · intro b hb      simp [hb]  have hsum4 : (∑' (m : ℤ), (∑' (j : ℤ),      (if j = 0 then (1:ℝ) else 0) * (if m - j = 0 then (1:ℝ) else 0)) ^ 2) = 1 := by    simp only [hinner]    simp  refine ⟨?_, ?_, ?_⟩  · simp [energy_filter_sequence_lattice_s2]  · simp only [energy_filter_sequence_lattice_s4]    exact hsum4  · simp only [energy_filter_sequence_purity_constant]    rw [hsum2, hsum4]    norm_num
See it in the graphDiscuss this resultChallenge it