linear_flattening_testable_tensor_rank_gt
If the secant variety σ_r((ℂⁿ)⊗³) is testable by the linear flattening method, then there exists a tensor in (ℂⁿ)⊗³ that cannot be written as a sum of r pure tensors.
Statement
theorem linear_flattening_testable_tensor_rank_gt {n r : ℕ}
(h : linear_flattening_testable n r) :
∃ T : Fin n → Fin n → Fin n → ℂ,
∀ x y z : Fin r → Fin n → ℂ, T ≠ ∑ i, tensor3_pure (x i) (y i) (z i)- Source
- arXiv:2610.08504 Sec. 1
- Verified
- 20 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 4
- linear_flattening_testableThe secant variety σ_r((ℂⁿ)⊗³) is testable by the linear flattening method if there exists a linear flattening L : (ℂⁿ)⊗³ → Mat_{u×v}(ℂ)…
- not_sum_pure_of_linear_flattening_potency_gtIf a linear flattening L on (ℂⁿ)⊗³ and a tensor T satisfy (L T).rank > r * (L(x ⊗ y ⊗ z)).rank for all x, y, z ∈ ℂⁿ, then T cannot be…
- tensor3_pureThe pure tensor x ⊗ y ⊗ z in ℂⁿ ⊗ ℂⁿ ⊗ ℂⁿ, with entries x_i y_j z_k.
- linear_flattening_potency_gtA linear flattening L : ℂⁿ ⊗ ℂⁿ ⊗ ℂⁿ → Mat_{u×v}(ℂ) has potency greater than r: some tensor T has rank L(T) > r · ρ(L), where ρ(L) =…
Read back from the Lean
Given natural numbers $n$ and $r$ (implicit), if there exist natural numbers $u$ and $v$ and a $\mathbb{C}$-linear map $L$ from the space of 3-tensors $(\operatorname{Fin} n \to \operatorname{Fin} n \to \operatorname{Fin} n \to \mathbb{C})$ to $u \times v$ complex matrices such that there exists a tensor $T_0$ satisfying $r \cdot \operatorname{rank}(L(x \otimes y \otimes z)) < \operatorname{rank}(L(T_0))$ for all vectors $x, y, z : \operatorname{Fin} n \to \mathbb{C}$, then there exists a 3-tensor $T : \operatorname{Fin} n \to \operatorname{Fin} n \to \operatorname{Fin} n \to \mathbb{C}$ such that for all families of $r$ vectors $x, y, z : \operatorname{Fin} r \to \operatorname{Fin} n \to \mathbb{C}$, $T$ is not equal to $\sum_{i \in \operatorname{Fin} r} (x(i) \otimes y(i) \otimes z(i))$ (i.e., $T$ cannot be decomposed into a sum of $r$ pure rank-1 tensors).
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
To show that there exists a tensor of rank strictly greater than r, we unpack the hypothesis that order r is testable by linear flattenings. This yields a linear map L to a matrix space and a tensor T whose flattened rank strictly exceeds r times the flattened rank of any pure tensor. Applying not_sum_pure_of_linear_flattening_potency_gt confirms that T cannot be decomposed into a sum of r pure tensors, completing the proof.
Unpack the testability assumption into a linear flattening map L, a tensor T, and a rank inequality.
By definition, r being testable by linear flattenings means there exist matrix dimensions u and v and a linear map L : (ℂⁿ)⊗³ → Mat_{u × v}(ℂ) with potency greater than r. In turn, potency greater than r provides a tensor T such that rank(L(T)) > r · rank(L(x ⊗ y ⊗ z)) for all pure tensors x ⊗ y ⊗ z.
rcases h with ⟨u, v, L, T, hT⟩
The tensor T cannot be written as a sum of r pure tensors, satisfying the theorem's claim.
We provide T as the witness. By not_sum_pure_of_linear_flattening_potency_gt, any tensor whose image under L has rank strictly exceeding r times the rank of any pure tensor under L cannot be a sum of r pure tensors.
uses
not_sum_pure_of_linear_flattening_potency_gtexact ⟨T, not_sum_pure_of_linear_flattening_potency_gt L r T hT⟩
not_sum_pure_of_linear_flattening_potency_gt- if a linear map L sends every pure tensor to a matrix of rank strictly less than (1/r) rank(L(T)), then T cannot be expressed as a sum of r pure tensors
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.
/-- If order r is testable by linear flattenings, there exists a tensor in (ℂⁿ)⊗³ with tensor rank strictly greater than r. -/theorem linear_flattening_testable_tensor_rank_gt {n r : ℕ} (h : linear_flattening_testable n r) : ∃ T : Fin n → Fin n → Fin n → ℂ, ∀ x y z : Fin r → Fin n → ℂ, T ≠ ∑ i, tensor3_pure (x i) (y i) (z i) := by rcases h with ⟨u, v, L, T, hT⟩ exact ⟨T, not_sum_pure_of_linear_flattening_potency_gt L r T hT⟩