tr_pauli_of_ne_one
If a Pauli string on $n$ qubits contains at least one non-identity single-qubit Pauli operator, then its tensor-product matrix has trace zero.
Statement
lemma tr_pauli_of_ne_one (n : ℕ) (p : PauliString n)
(hp : ∃ i : Fin n, p i ≠ PauliBasis.I) :
(pauliMatrix p).trace = 0- Source
- arXiv:2610.37807, Lemma 10 proof
- Verified
- 08 Nov 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 4
- pauliMatrixFor a Pauli string $p$ on $n$ qubits, the $2^n\times 2^n$ matrix obtained as the tensor product of the single-qubit Pauli matrices…
- PauliStringA Pauli string of length $n$ is a function $p : \mathrm{Fin}\,n \to \mathrm{PauliBasis}$.
- PauliBasisThe four Pauli basis elements $\{I,X,Y,Z\}$, represented as an inductive type.
- PauliBasis.toMatrixMaps each Pauli basis element $\{I,X,Y,Z\}$ to its corresponding $2\times 2$ matrix $\{\sigma_I,\sigma_X,\sigma_Y,\sigma_Z\}$.
- Used by 1
- stabilizer_subspace_finrank_eq_two_powLet $k \le n$ and let $g_1, \dots, g_{n-k}$ be $n$-qubit Pauli operators $g_l = \varepsilon_l P_l$ with signs…
Read back from the Lean
For every natural number n and every Pauli string p on n qubits (that is, p : Fin n → PauliBasis), if there exists an index i : Fin n such that p i ≠ PauliBasis.I, then the trace of the Pauli matrix associated with p ((pauliMatrix p).trace) is equal to 0 in 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 trace of an $n$-qubit Pauli tensor product matrix factors into the product of the traces of its individual single-qubit Pauli factors via generalized distributivity. By assumption, at least one factor at some qubit index $k$ is a non-identity Pauli operator ($X, Y,$ or $Z$). Since the single-qubit Pauli matrices $X, Y,$ and $Z$ all have trace zero, that factor vanishes, which makes the entire product zero.
There exists a qubit index $k$ such that the Pauli operator $p(k)$ is not the identity.
We instantiate the existential hypothesis ($hp$) to fix a specific qubit index $k$ and the fact that $p(k) \ne I$.
obtain ⟨k, hk⟩ := hpThe trace of the Pauli string matrix equals the sum over all binary multi-indices of the product of the diagonal entries of each single-qubit factor.
By definition, the Pauli string matrix is the tensor product of the individual Pauli matrices, with matrix entries given by the product of the corresponding single-qubit entries. Since the trace is the sum of diagonal entries, this identity holds directly by definition (`rfl`).
have hkey : (pauliMatrix p).trace = ∑ i : Fin n → Fin 2, ∏ k, (p k).toMatrix (i k) (i k) := rfl
The sum of products of diagonal entries factors into the product over all qubits of the traces of the individual Pauli matrices.
Distributing multiplication over addition across the function space (using `Finset.sum_prod_piFinset`) converts the sum over multi-indices of products into a product over qubits of sums. Congruence (`congr 1`) equates the domain of summation with the universe of functions from $\text{Fin } n$ to $\text{Fin } 2$.
uses
Finset.sum_prod_piFinsethave hfac : (∑ i : Fin n → Fin 2, ∏ k, (p k).toMatrix (i k) (i k)) = ∏ k, ∑ a : Fin 2, (p k).toMatrix a a := by rw [← Finset.sum_prod_piFinset (s := Finset.univ) (g := fun k (a : Fin 2) => (p k).toMatrix a a)] congr 1
It suffices to show that the trace of the single-qubit operator $p(k)$ is zero.
Rewriting the trace of `pauliMatrix p` using the previous identities expresses it as the product $\prod_{j} \operatorname{Tr}(p(j))$. By `Finset.prod_eq_zero`, a finite product vanishes if at least one factor is zero, so it suffices to show that the $k$-th factor $\sum_{a=0}^1 (p(k))_{a,a}$ is zero.
uses
Finset.prod_eq_zero,Finset.mem_univrw [hkey, hfac] apply Finset.prod_eq_zero (Finset.mem_univ k)
The single-qubit Pauli matrix $p(k)$ has trace zero.
We perform case analysis on the Pauli basis operator $p(k)$: - If $p(k) = I$, this contradicts the hypothesis $p(k) \ne I$ via `absurd`. - If $p(k) = X$, the matrix $\sigma_X$ has zeros on the diagonal, so its trace is $0 + 0 = 0$. - If $p(k) = Y$, the matrix $\sigma_Y$ has zeros on the diagonal, so its trace is $0 + 0 = 0$. - If $p(k) = Z$, the matrix $\sigma_Z$ has diagonal entries $1$ and $-1$, which sum to $0$. In each non-identity case, simplifying using the definitions of the matrices confirms the trace evaluates to zero.
uses
absurd,PauliBasis.toMatrix,sigmaX,sigmaY,sigmaZcases hpk : p k with | I => exact absurd hpk hk | X => simp [PauliBasis.toMatrix, sigmaX] | Y => simp [PauliBasis.toMatrix, sigmaY] | Z => simp [PauliBasis.toMatrix, sigmaZ]
Finset.sum_prod_piFinset- A sum of products over functions with values in a finite set factors as the product over the domain of the sums over the codomain: $\sum_{f \in \prod s} \prod_i g(i, f(i)) = \prod_i \sum_{j \in s} g(i, j)$.
Finset.prod_eq_zero- A finite product is zero if at least one factor in the product is zero.
Finset.mem_univ- Every element of a finite type belongs to its universal finite set `Finset.univ`.
absurd- From a proposition $P$ and its negation $\neg P$, any conclusion follows (principle of explosion).
PauliBasis.toMatrix- Maps each Pauli basis element ($I, X, Y, Z$) to its corresponding 2×2 complex matrix.
sigmaX- The 2×2 Pauli X matrix, $\begin{pmatrix} 0 & 1 \\ 1 & 0 \end{pmatrix}$.
sigmaY- The 2×2 Pauli Y matrix, $\begin{pmatrix} 0 & -i \\ i & 0 \end{pmatrix}$.
sigmaZ- The 2×2 Pauli Z matrix, $\begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix}$.
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
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
/-- Any Pauli string having at least one non-identity single-qubit factor has trace zero. -/lemma tr_pauli_of_ne_one (n : ℕ) (p : PauliString n) (hp : ∃ i : Fin n, p i ≠ PauliBasis.I) : (pauliMatrix p).trace = 0 := by obtain ⟨k, hk⟩ := hp have hkey : (pauliMatrix p).trace = ∑ i : Fin n → Fin 2, ∏ k, (p k).toMatrix (i k) (i k) := rfl have hfac : (∑ i : Fin n → Fin 2, ∏ k, (p k).toMatrix (i k) (i k)) = ∏ k, ∑ a : Fin 2, (p k).toMatrix a a := by rw [← Finset.sum_prod_piFinset (s := Finset.univ) (g := fun k (a : Fin 2) => (p k).toMatrix a a)] congr 1 rw [hkey, hfac] apply Finset.prod_eq_zero (Finset.mem_univ k) cases hpk : p k with | I => exact absurd hpk hk | X => simp [PauliBasis.toMatrix, sigmaX] | Y => simp [PauliBasis.toMatrix, sigmaY] | Z => simp [PauliBasis.toMatrix, sigmaZ]