qutrit_block_real_frame_product_decomposition
A real Parseval frame that simultaneously expands a qutrit block and its lower diagonal block yields an explicit four-term product decomposition.
Statement
theorem qutrit_block_real_frame_product_decomposition
(B C : Matrix (Fin 3) (Fin 3) ℂ) (O : Matrix (Fin 3) (Fin 4) ℝ) (lam : Fin 4 → ℂ)
(hO : O * O.transpose = 1)
(hB : ∀ i j, B i j = ∑ k, (O i k : ℂ) * lam k * (O j k : ℂ))
(hC : ∀ i j, C i j = ∑ k, (O i k : ℂ) * (lam k * star (lam k)) * (O j k : ℂ)) :
∃ (v : Fin 4 → Fin 3 → ℂ) (a b : Fin 4 → ℂ),
(∀ t i, star (v t i) = v t i) ∧
(1 : Matrix (Fin 3) (Fin 3) ℂ) =
∑ t, (a t * star (a t)) • vecMulVec (v t) (star (v t)) ∧
B = ∑ t, (a t * star (b t)) • vecMulVec (v t) (star (v t)) ∧
C = ∑ t, (b t * star (b t)) • vecMulVec (v t) (star (v t))- Source
- Helper for OP-693, in the normal-completion route related to https://arxiv.org/abs/0903.0090.
- Verified
- 12 Nov 2026
- Axioms
Classical.choiceQuot.soundpropext
The proof, explained
The proof proceeds by explicitly constructing the frame vectors and scalar weights from the columns of the Parseval frame O and the given spectrum lam. The vectors v_t are taken to be the columns of O (as complex vectors), while the scalar sequences are set to a_t = 1 and b_t = star (lam t). The reality of v_t follows because O has real entries, the identity matrix decomposition reduces entrywise to the orthogonality condition O * Oᵀ = 1, and the product decompositions of B and C reduce entrywise to the supplied expansion hypotheses hB and hC via the involution properties of complex conjugation.
There exist candidate witnesses: the vectors v_t are given by the columns of O, the coefficients a_t are all 1, and the coefficients b_t are the complex conjugates of lam t.
The existential claim is instantiated by setting v_t(i) = (O i t : ℂ), a_t = 1, and b_t = star (lam t) for each t ∈ Fin 4. This reduces the theorem to four subgoals: showing each v_t is real, verifying the decomposition of the identity matrix 1, verifying the decomposition of B, and verifying the decomposition of C.
refine ⟨fun t i => (O i t : ℂ), fun _ => 1, fun t => star (lam t), ?_, ?_, ?_, ?_⟩
Each vector v_t has real entries: star (v_t(i)) = v_t(i) for all t and i.
For arbitrary indices t and i, v_t(i) is the canonical real-to-complex embedding of O i t. Since the complex conjugate of any real number is itself, simplification proves the claim immediately.
· intro t i simp
The 3 × 3 identity matrix decomposes as ∑ t, (a_t * star (a_t)) • vecMulVec (v_t) (star (v_t)).
By matrix extensionality, it suffices to check the identity entrywise for every i and j. Applying congruence to the hypothesis O * Oᵀ = 1 yields the scalar equality ∑ k, O i k * O j k = (1 : Matrix (Fin 3) (Fin 3) ℝ) i j. Casting this equation to the complex numbers and performing a case analysis on whether i = j or i ≠ j reduces the (i, j)-entry of the sum to the (i, j)-entry of the identity matrix.
uses
congrArg,Matrix.mul_apply,Matrix.transpose_apply,Matrix.vecMulVec_apply,Matrix.one_apply,Matrix.sum_apply,Matrix.smul_apply,mul_comm,mul_left_comm,mul_assoc· ext i j have hr := congrArg (fun M : Matrix (Fin 3) (Fin 3) ℝ => M i j) hO simp only [Matrix.mul_apply, Matrix.transpose_apply] at hr have hc : (∑ k, (O i k : ℂ) * (O j k : ℂ)) = ((1 : Matrix (Fin 3) (Fin 3) ℝ) i j : ℂ) := by exact_mod_cast hr by_cases hij : i = j <;> simpa [hij, Matrix.vecMulVec_apply, Matrix.one_apply, Matrix.sum_apply, Matrix.smul_apply, mul_comm, mul_left_comm, mul_assoc] using hc.symm
The matrix B decomposes as ∑ t, (a_t * star (b_t)) • vecMulVec (v_t) (star (v_t)).
For any entry (i, j), the coefficient simplifies to a_t * star (b_t) = 1 * star (star (lam t)) = lam t. Expanding the matrix operations on the right-hand sum gives ∑ t, O i t * lam t * O j t, which matches the assumption hB i j by definition.
uses
Matrix.vecMulVec_apply,Matrix.sum_apply,Matrix.smul_apply,mul_comm,mul_left_comm,mul_assoc· ext i j simpa [Matrix.vecMulVec_apply, Matrix.sum_apply, Matrix.smul_apply, mul_comm, mul_left_comm, mul_assoc] using hB i j
The matrix C decomposes as ∑ t, (b_t * star (b_t)) • vecMulVec (v_t) (star (v_t)).
Similarly, for any entry (i, j), the coefficient simplifies to b_t * star (b_t) = star (lam t) * lam t = lam t * star (lam t). The resulting entrywise sum ∑ t, O i t * (lam t * star (lam t)) * O j t coincides with the given formula hC i j.
uses
Matrix.vecMulVec_apply,Matrix.sum_apply,Matrix.smul_apply,mul_comm,mul_left_comm,mul_assoc· ext i j simpa [Matrix.vecMulVec_apply, Matrix.sum_apply, Matrix.smul_apply, mul_comm, mul_left_comm, mul_assoc] using hC i j
congrArg- Applying a function f to both sides of an equality preserves the equality: if a₁ = a₂, then f(a₁) = f(a₂).
Matrix.mul_apply- The (i, k)-th entry of a matrix product M * N is given by the sum ∑ j, M i j * N j k.
Matrix.one_apply- The (i, j)-th entry of the identity matrix is 1 when i = j and 0 otherwise.
Matrix.smul_apply- The (i, j)-th entry of a scalar multiple r • A is r * A i j.
Matrix.sum_apply- The (i, j)-th entry of a finite sum of matrices is the sum of their (i, j)-th entries.
Matrix.transpose_apply- The (i, j)-th entry of the transpose matrix Mᵀ is M j i.
Matrix.vecMulVec_apply- The (i, j)-th entry of the outer product vecMulVec w v is w i * v j.
mul_assoc- Multiplication is associative: (a * b) * c = a * (b * c).
mul_comm- Multiplication is commutative: a * b = b * a.
mul_left_comm- Multiplication satisfies left-commutativity: a * (b * c) = b * (a * c).
- In Lean, Matrix.vecMulVec (v t) (star (v t)) denotes the outer product v_t v_t^*; because each v_t is real, this is simply the rank-one matrix v_t v_t^T.
- The sequence b is set to the complex conjugates star (lam t) so that a_t * star (b_t) = 1 * lam t = lam t and b_t * star (b_t) = lam t * star (lam t).
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.
open Matrixopen scoped BigOperators Matrix ComplexOrderset_option backward.isDefEq.respectTransparency falseset_option linter.unusedSimpArgs falsetheorem qutrit_block_real_frame_product_decomposition (B C : Matrix (Fin 3) (Fin 3) ℂ) (O : Matrix (Fin 3) (Fin 4) ℝ) (lam : Fin 4 → ℂ) (hO : O * O.transpose = 1) (hB : ∀ i j, B i j = ∑ k, (O i k : ℂ) * lam k * (O j k : ℂ)) (hC : ∀ i j, C i j = ∑ k, (O i k : ℂ) * (lam k * star (lam k)) * (O j k : ℂ)) : ∃ (v : Fin 4 → Fin 3 → ℂ) (a b : Fin 4 → ℂ), (∀ t i, star (v t i) = v t i) ∧ (1 : Matrix (Fin 3) (Fin 3) ℂ) = ∑ t, (a t * star (a t)) • vecMulVec (v t) (star (v t)) ∧ B = ∑ t, (a t * star (b t)) • vecMulVec (v t) (star (v t)) ∧ C = ∑ t, (b t * star (b t)) • vecMulVec (v t) (star (v t)) := by refine ⟨fun t i => (O i t : ℂ), fun _ => 1, fun t => star (lam t), ?_, ?_, ?_, ?_⟩ · intro t i simp · ext i j have hr := congrArg (fun M : Matrix (Fin 3) (Fin 3) ℝ => M i j) hO simp only [Matrix.mul_apply, Matrix.transpose_apply] at hr have hc : (∑ k, (O i k : ℂ) * (O j k : ℂ)) = ((1 : Matrix (Fin 3) (Fin 3) ℝ) i j : ℂ) := by exact_mod_cast hr by_cases hij : i = j <;> simpa [hij, Matrix.vecMulVec_apply, Matrix.one_apply, Matrix.sum_apply, Matrix.smul_apply, mul_comm, mul_left_comm, mul_assoc] using hc.symm · ext i j simpa [Matrix.vecMulVec_apply, Matrix.sum_apply, Matrix.smul_apply, mul_comm, mul_left_comm, mul_assoc] using hB i j · ext i j simpa [Matrix.vecMulVec_apply, Matrix.sum_apply, Matrix.smul_apply, mul_comm, mul_left_comm, mul_assoc] using hC i j