linear_flattening_testable_dim_zero
At dimension n = 0 the secant variety σ_r((ℂ⁰)⊗³) is never testable by the linear flattening method, for any r: the tensor space is a subsingleton, so every linear flattening has rank 0 on every tensor.
Statement
theorem linear_flattening_testable_dim_zero (r : ℕ) : ¬ linear_flattening_testable 0 r- Source
- step towards linear_flattening_testable_even_step
- Verified
- 23 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 3
- 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)…
- 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) =…
- tensor3_pureThe pure tensor x ⊗ y ⊗ z in ℂⁿ ⊗ ℂⁿ ⊗ ℂⁿ, with entries x_i y_j z_k.
The proof, explained
At dimension $n = 0$, the vector space $\mathbb{C}^0$ and the tensor space $(\mathbb{C}^0)^{\otimes 3}$ are trivial, so every tensor is identically zero. Consequently, any linear flattening $L$ maps both the witness tensor $T$ and any pure rank-one tensor to the zero matrix, which has rank $0$. Specializing the testability condition to the zero vectors then produces the impossible inequality $r \cdot 0 < 0$.
Assume for contradiction that the secant variety is testable at dimension 0.
We unpack the definition of `linear_flattening_testable 0 r`, which assumes there exist matrix dimensions $u, v$, a linear flattening map $L : (\mathbb{C}^0)^{\otimes 3} \to \mathrm{Matrix}_{u, v}(\mathbb{C})$, and a witness tensor $T$ satisfying $r \cdot \operatorname{rank}(L(x \otimes y \otimes z)) < \operatorname{rank}(L(T))$ for all vectors $x, y, z \in \mathbb{C}^0$.
intro ⟨u, v, L, T, hT⟩The witness tensor $T$ is the zero tensor ($T = 0$).
Tensors over $\mathbb{C}^0$ are functions on index triples from $\mathrm{Fin}\ 0$. By functional extensionality (`ext`), two tensors are equal if they agree on all indices. Since $\mathrm{Fin}\ 0$ has no elements, any assumed index is impossible (`i.elim0`), so $T = 0$ holds vacuously.
have hT0 : T = 0 := by ext i exact i.elim0
The pure tensor formed by the zero vectors is also the zero tensor.
Similarly, by functional extensionality, the rank-1 tensor $\mathrm{tensor3\_pure}(0, 0, 0)$ is identically zero because evaluating it at any triple of indices requires an element of $\mathrm{Fin}\ 0$, which is empty.
have hpure : tensor3_pure (fun (i : Fin 0) => (0 : ℂ)) (fun i => 0) (fun i => 0) = 0 := by ext i exact i.elim0
The testability condition reduces to the strict inequality $r \cdot 0 < 0$.
We specialize the potency hypothesis to the zero vectors. Since both $T$ and the pure tensor equal $0$, the linear map $L$ sends both to the zero matrix (`map_zero`). Because the rank of the zero matrix is $0$ (`Matrix.rank_zero`), the inequality simplifies to $r \cdot 0 < 0$.
uses
map_zero,Matrix.rank_zerohave key := hT (fun i => 0) (fun i => 0) (fun i => 0) rw [hT0, hpure, map_zero, Matrix.rank_zero] at key
A contradiction is reached, concluding the proof. automation
Linear integer arithmetic (`omega`) simplifies $r \cdot 0 < 0$ to $0 < 0$, which is false.
omega
map_zero- A linear map (or any zero-preserving map) maps zero to zero: $f(0) = 0$.
Matrix.rank_zero- The rank of the zero matrix is $0$.
- In Lean, elements of $(\mathbb{C}^0)^{\otimes 3}$ are represented as functions $\mathrm{Fin}\ 0 \to \mathrm{Fin}\ 0 \to \mathrm{Fin}\ 0 \to \mathbb{C}$. Equality with zero is proved vacuously via functional extensionality (`ext`) and the empty eliminator `Fin.elim0`.
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.
/-- Degenerate small instance: on the zero-dimensional tensor space everything is 0, so no order is testable; this pins down where the definition starts having content. -/theorem linear_flattening_testable_dim_zero (r : ℕ) : ¬ linear_flattening_testable 0 r := by intro ⟨u, v, L, T, hT⟩ have hT0 : T = 0 := by ext i exact i.elim0 have hpure : tensor3_pure (fun (i : Fin 0) => (0 : ℂ)) (fun i => 0) (fun i => 0) = 0 := by ext i exact i.elim0 have key := hT (fun i => 0) (fun i => 0) (fun i => 0) rw [hT0, hpure, map_zero, Matrix.rank_zero] at key omega