AFTD Auto-Formalizing Theoretical Domains

standardized_laplacian_order_four_zero_mem_spectrum

theoremverified

For every standardized Laplacian matrix L of order 4, zero is an eigenvalue of L.

Statement

theorem standardized_laplacian_order_four_zero_mem_spectrum (L : Matrix (Fin 4) (Fin 4) ℝ)
    (hL : is_standardized_laplacian 4 L) :
    (0 : ℂ) ∈ spectrum ℂ (L.map (algebraMap ℝ ℂ))
Source
arXiv:2610.14100 Sec. 2
Verified
24 Oct 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 1
  • is_standardized_laplacianA standardized Laplacian matrix of order n is a real n x n matrix with zero row sums and off-diagonal entries in [-1/n, 0].
Used by 1

Read back from the Lean

For every $4 \times 4$ real matrix $L$, if $L$ satisfies is_standardized_laplacian 4 L (that is, all row sums of $L$ are zero, and for all distinct indices $i \neq j$, $-1/4 \le L_{i,j} \le 0$), then $0 \in \mathbb{C}$ is in the complex spectrum of $L$ (considered as a complex matrix via entrywise coercion from $\mathbb{R}$ to $\mathbb{C}$).

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 proceeds by contradiction. By definition, 0 belongs to the spectrum of L if and only if L is not invertible. If L were invertible, then the all-ones vector 1 would equal L⁻¹ (L *ᵥ 1); however, because the row sums of a standardized Laplacian are all zero, L *ᵥ 1 = 0, which would force 1 = 0, a contradiction.

  1. It suffices to show that L (as a complex matrix) is not invertible.

    By definition of spectrum in an algebra, 0 is in the spectrum of L if and only if 0 • I - L = -L is not invertible. Since a matrix is invertible if and only if its negation is invertible, this is equivalent to L not being invertible. We proceed by contradiction, assuming that L is invertible (hUnit).

    lines 5–8

    uses spectrum.mem_iff, map_zero, zero_sub, IsUnit.neg_iff

      open Matrix in
      rw [spectrum.mem_iff]
      simp only [map_zero, zero_sub, IsUnit.neg_iff]
      intro hUnit
  2. Multiplying L by the all-ones vector yields the zero vector.

    To show (L *ᵥ 1) = 0, we check each row i. The i-th component is the dot product of row i with the constant vector 1, which simplifies to the sum of entries in row i. By the first condition of a standardized Laplacian, every row sum is zero in ℝ, so mapping this sum into ℂ yields 0.

    lines 9–13

    uses mulVec, dotProduct, Pi.zero_apply, mul_one, map_sum, map_zero

      have hMv : (L.map (algebraMap ℝ ℂ)) *ᵥ (fun _ => (1 : ℂ)) = 0 := by
        ext i
        simp only [mulVec, dotProduct, Pi.zero_apply, mul_one]
        change ∑ j : Fin 4, (algebraMap ℝ ℂ) (L i j) = 0
        rw [← map_sum, hL.1 i, map_zero]
  3. The all-ones vector is nonzero.

    If the constant vector with all entries 1 were the zero vector, evaluating both sides at index 0 would give 1 = 0, contradicting the fact that 1 ≠ 0 in ℂ.

    lines 14–17

    uses congr_fun, one_ne_zero

      have hv : (fun _ : Fin 4 => (1 : ℂ)) ≠ 0 := by
        intro h0
        have := congr_fun h0 0
        exact one_ne_zero this
  4. The invertibility of L leads to a contradiction.

    Since L is assumed invertible, write L as a unit u. Multiplying by u⁻¹ and using associativity shows that the all-ones vector satisfies 1 = u⁻¹ *ᵥ (u *ᵥ 1) = u⁻¹ *ᵥ (L *ᵥ 1). But L *ᵥ 1 = 0, so 1 = u⁻¹ *ᵥ 0 = 0, which contradicts that the all-ones vector is nonzero.

    lines 18–24

    uses mulVec_mulVec, Units.inv_mul, one_mulVec, mulVec_zero

      obtain ⟨u, hu⟩ := hUnit
      have h1 : (fun _ : Fin 4 => (1 : ℂ)) =
          ((u⁻¹ : (Matrix (Fin 4) (Fin 4) ℂ)ˣ) : Matrix (Fin 4) (Fin 4) ℂ) *ᵥ
          ((u : Matrix (Fin 4) (Fin 4) ℂ) *ᵥ (fun _ => 1)) := by
        rw [mulVec_mulVec, Units.inv_mul, one_mulVec]
      rw [hu, hMv, mulVec_zero] at h1
      exact hv h1
spectrum.mem_iff
An element r belongs to the spectrum of an algebra element a if and only if r • 1 - a is not invertible (not a unit).
map_zero
A zero-preserving map sends 0 to 0.
zero_sub
Subtracting an element from 0 equals its negation: 0 - a = -a.
IsUnit.neg_iff
An element is a unit if and only if its negation is a unit.
mulVec
The definition of matrix-vector multiplication, defined by taking dot products of matrix rows with the vector.
dotProduct
The definition of the dot product of two vectors as the sum of their componentwise products.
Pi.zero_apply
The zero function evaluated at any input yields zero.
mul_one
Multiplying any element by 1 leaves it unchanged.
map_sum
An additive monoid homomorphism distributes over finite sums.
congr_fun
If two functions are equal, their values at any argument are equal.
one_ne_zero
In any nontrivial structure with 0 and 1, 1 ≠ 0.
mulVec_mulVec
Matrix multiplication associates with matrix-vector multiplication: M *ᵥ (N *ᵥ v) = (M * N) *ᵥ v.
Units.inv_mul
For every invertible element u, u⁻¹ * u = 1.
one_mulVec
Multiplying a vector by the identity matrix leaves the vector unchanged.
mulVec_zero
Multiplying the zero vector by any matrix yields the zero vector: A *ᵥ 0 = 0.

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

/-- Zero is an eigenvalue of every standardized Laplacian matrix of order 4. -/theorem standardized_laplacian_order_four_zero_mem_spectrum (L : Matrix (Fin 4) (Fin 4) ℝ)    (hL : is_standardized_laplacian 4 L) :    (0 : ℂ) ∈ spectrum ℂ (L.map (algebraMap ℝ ℂ)) := by  open Matrix in  rw [spectrum.mem_iff]  simp only [map_zero, zero_sub, IsUnit.neg_iff]  intro hUnit  have hMv : (L.map (algebraMap ℝ ℂ)) *ᵥ (fun _ => (1 : ℂ)) = 0 := by    ext i    simp only [mulVec, dotProduct, Pi.zero_apply, mul_one]    change ∑ j : Fin 4, (algebraMap ℝ ℂ) (L i j) = 0    rw [← map_sum, hL.1 i, map_zero]  have hv : (fun _ : Fin 4 => (1 : ℂ)) ≠ 0 := by    intro h0    have := congr_fun h0 0    exact one_ne_zero this  obtain ⟨u, hu⟩ := hUnit  have h1 : (fun _ : Fin 4 => (1 : ℂ)) =      ((u⁻¹ : (Matrix (Fin 4) (Fin 4) ℂ)ˣ) : Matrix (Fin 4) (Fin 4) ℂ) *ᵥ      ((u : Matrix (Fin 4) (Fin 4) ℂ) *ᵥ (fun _ => 1)) := by    rw [mulVec_mulVec, Units.inv_mul, one_mulVec]  rw [hu, hMv, mulVec_zero] at h1  exact hv h1
See it in the graphDiscuss this resultChallenge it