is_bipartite_locally_encodable_norm_one
For nonempty finite types m and n, every locally encodable state ψ in EuclideanSpace ℂ (m × n) satisfies ‖ψ‖ = 1.
Statement
theorem is_bipartite_locally_encodable_norm_one {m n : Type*} [Fintype m] [DecidableEq m] [Fintype n] [DecidableEq n] [Nonempty m] [Nonempty n] {ψ : EuclideanSpace ℂ (m × n)} (h : is_bipartite_locally_encodable ψ) : ‖ψ‖ = 1
- Formalizes
- arXiv:2611.16921 Sec. II.1
- Verified
- 14 Nov 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 1
- is_bipartite_locally_encodableA bipartite quantum state ψ in EuclideanSpace ℂ (m × n) is locally encodable if there exists an orthonormal basis B of EuclideanSpace ℂ (m…
Read back from the Lean
For any finite, nonempty types $m$ and $n$ with decidable equality, and any vector $\psi \in \mathrm{EuclideanSpace}\ \mathbb{C}\ (m \times n)$, if $\psi$ is bipartite locally encodable, then the Euclidean norm of $\psi$ is equal to 1 ($\|\psi\| = 1$).
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
By definition of local encodability, there is an orthonormal basis B such that every basis element is the image of ψ under some local unitary operator uA ⊗ uB. Since the index sets m and n are nonempty, we choose an index j, giving a basis vector B(j) = (uA ⊗ uB) ψ. The Kronecker product of unitary matrices is unitary, and the matrix-to-operator representation preserves the star-algebra structure, making the induced operator unitary and therefore isometric. Because B is orthonormal, ‖B(j)‖ = 1, which implies ‖ψ‖ = ‖(uA ⊗ uB) ψ‖ = ‖B(j)‖ = 1.
Obtain an orthonormal basis B, an index j, and local unitaries uA and uB expressing B(j) in terms of ψ.
By definition of local encodability, there is an orthonormal basis B whose elements are all local unitary transforms of ψ. Because m and n are nonempty, the product type m × n is nonempty, so we choose an index j. Specializing the local encodability hypothesis at j yields unitary matrices uA on ℂᵐ and uB on ℂⁿ such that B(j) = (uA ⊗ uB) ψ.
obtain ⟨B, hB⟩ := h obtain ⟨j⟩ := (inferInstance : Nonempty (m × n)) obtain ⟨uA, uB, hBj⟩ := hB j
The Kronecker product M = uA ⊗ uB is unitary.
We define M as the Kronecker product of uA and uB. Since both uA and uB are unitary matrices, their Kronecker product is unitary by Matrix.kronecker_mem_unitary.
uses
Matrix.kronecker_mem_unitaryset M : Matrix (m × n) (m × n) ℂ := Matrix.kronecker (uA : Matrix m m ℂ) (uB : Matrix n n ℂ) with hM have hmem : M ∈ Matrix.unitaryGroup (m × n) ℂ := Matrix.kronecker_mem_unitary uA.2 uB.2
The matrix-to-operator representation preserves adjoints, products, and identity.
The mapping Matrix.toEuclideanCLM from matrices to continuous linear operators on Euclidean space is an equivalence of star-algebras. Consequently, it commutes with taking adjoints (hstar), preserves multiplication (hmul), and sends the identity matrix to the identity operator (hone).
have hstar : star (Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) M) = Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) (star M) := ((Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ)).map_star' M).symm have hmul : ∀ A B : Matrix (m × n) (m × n) ℂ, Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) A * Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) B = Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) (A * B) := fun A B => ((Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ)).map_mul A B).symm have hone : Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) (1 : Matrix (m × n) (m × n) ℂ) = 1 := (Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ)).map_one
The continuous linear map induced by M is unitary.
By Unitary.mem_iff, an operator is unitary if its product with its adjoint on either side is the identity. Rewriting using the preservation of adjoints and products reduces this to the fact that M is a unitary matrix (M* M = 1 and M M* = 1), so the induced operator is unitary.
uses
Unitary.mem_iffhave hU : Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) M ∈ unitary (EuclideanSpace ℂ (m × n) →L[ℂ] EuclideanSpace ℂ (m × n)) := by rw [Unitary.mem_iff] constructor · rw [hstar, hmul, hmem.1, hone] · rw [hstar, hmul, hmem.2, hone]
The induced operator preserves vector norms and maps ψ to B(j).
Since the operator induced by M is unitary, it preserves norms by ContinuousLinearMap.norm_map_of_mem_unitary, giving ‖M ψ‖ = ‖ψ‖. Furthermore, Matrix.coe_toEuclideanCLM_eq_toEuclideanLin equates the continuous linear map with the linear map, so evaluating it on ψ yields B(j) via congrArg.
uses
ContinuousLinearMap.norm_map_of_mem_unitary,Matrix.coe_toEuclideanCLM_eq_toEuclideanLin,congrArghave hnorm : ‖(Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) M) ψ‖ = ‖ψ‖ := ContinuousLinearMap.norm_map_of_mem_unitary hU ψ have happ : (Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) M) ψ = B j := by rw [hBj] exact congrArg (fun f => f ψ) (Matrix.coe_toEuclideanCLM_eq_toEuclideanLin M)
The norm of ψ is 1.
Rewriting with ‖ψ‖ = ‖M ψ‖ and M ψ = B(j) reduces the goal to showing ‖B(j)‖ = 1. Since B is an orthonormal basis, every basis vector has norm 1, which closes the goal.
rw [← hnorm, happ] exact B.orthonormal.1 j
Matrix.kronecker_mem_unitary- The Kronecker product of two unitary matrices is itself a unitary matrix.
Unitary.mem_iff- An element U of a star monoid is unitary if and only if star U * U = 1 and U * star U = 1.
ContinuousLinearMap.norm_map_of_mem_unitary- A unitary continuous linear map on a Hilbert space preserves the norm of every vector.
Matrix.coe_toEuclideanCLM_eq_toEuclideanLin- The continuous linear map induced by a matrix coincides with the linear map induced by that matrix.
congrArg- Applying a function to equal inputs yields equal outputs (congruence of arguments).
- The formal proof distinguishes between the continuous linear map `toEuclideanCLM` (used to establish unitarity and norm preservation) and the bare linear map `toEuclideanLin` (used in the definition of local encodability); on paper, both are the standard linear action of a matrix on a finite-dimensional space.
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 locally encodable bipartite state in a nonempty system has norm 1. -/theorem is_bipartite_locally_encodable_norm_one {m n : Type*} [Fintype m] [DecidableEq m] [Fintype n] [DecidableEq n] [Nonempty m] [Nonempty n] {ψ : EuclideanSpace ℂ (m × n)} (h : is_bipartite_locally_encodable ψ) : ‖ψ‖ = 1 := by obtain ⟨B, hB⟩ := h obtain ⟨j⟩ := (inferInstance : Nonempty (m × n)) obtain ⟨uA, uB, hBj⟩ := hB j -- the local unitary `uA ⊗ₖ uB` is unitary set M : Matrix (m × n) (m × n) ℂ := Matrix.kronecker (uA : Matrix m m ℂ) (uB : Matrix n n ℂ) with hM have hmem : M ∈ Matrix.unitaryGroup (m × n) ℂ := Matrix.kronecker_mem_unitary uA.2 uB.2 have hstar : star (Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) M) = Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) (star M) := ((Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ)).map_star' M).symm have hmul : ∀ A B : Matrix (m × n) (m × n) ℂ, Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) A * Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) B = Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) (A * B) := fun A B => ((Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ)).map_mul A B).symm have hone : Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) (1 : Matrix (m × n) (m × n) ℂ) = 1 := (Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ)).map_one -- hence the induced operator on `EuclideanSpace ℂ (m × n)` is unitary, so it is an isometry have hU : Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) M ∈ unitary (EuclideanSpace ℂ (m × n) →L[ℂ] EuclideanSpace ℂ (m × n)) := by rw [Unitary.mem_iff] constructor · rw [hstar, hmul, hmem.1, hone] · rw [hstar, hmul, hmem.2, hone] have hnorm : ‖(Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) M) ψ‖ = ‖ψ‖ := ContinuousLinearMap.norm_map_of_mem_unitary hU ψ have happ : (Matrix.toEuclideanCLM (n := m × n) (𝕜 := ℂ) M) ψ = B j := by rw [hBj] exact congrArg (fun f => f ψ) (Matrix.coe_toEuclideanCLM_eq_toEuclideanLin M) rw [← hnorm, happ] exact B.orthonormal.1 j