OctoInformBeta

OpenAI preprint, September 2026 · §4. A matrix-cone homomorphism with a full corner

4.1 Algebraic independence of commuting factors: And for finite families in commuting factor von Neumann algebras, with injective tensor multiplication on finite-dimensional real spaces embedded into their selfadjoint parts

Multiplication preserves linear independence across a factor and its commutant.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

Let HH be a complex Hilbert space and B(H)B(H) be the algebra of bounded linear operators on HH, with identity operator 11. For a von Neumann algebra M⊆B(H)M \subseteq B(H), write M′={T∈B(H):TS=ST for all S∈M}M' = \{T \in B(H) : TS = ST \text{ for all } S \in M\} for its commutant. The question is how linear independence behaves under operator multiplication between MM and M′M', or across pairwise commuting von Neumann algebras on HH, when the intersection of each algebra with its commutant consists only of scalar multiples of 11, both over ℂ\mathbb{C} and on real algebraic tensor products of selfadjoint operator spaces.

Definitions

A factor von Neumann algebra (briefly, a factor) MM on HH is a von Neumann algebra M⊆B(H)M \subseteq B(H) satisfying M∩M′=ℂ1M \cap M' = \mathbb{C}1.✦

For n∈ℕn \in \mathbb{N} and real vector spaces V1,…,VnV_1, \dots, V_n, let ⨂ℝ,i=1nVi\bigotimes_{\mathbb{R}, i=1}^n V_i denote their real algebraic tensor product (understood as ℝ\mathbb{R} when n=0n = 0), and let AA be a unital associative ℝ\mathbb{R}-algebra with identity 1A1_A. Given a family f=(fi)i=1nf = (f_i)_{i=1}^n of ℝ\mathbb{R}-linear maps fi:Vi→Af_i \colon V_i \to A, the tensor multiplication map μf:⨂ℝ,i=1nVi→A\mu_f \colon \bigotimes_{\mathbb{R}, i=1}^n V_i \to A is the unique ℝ\mathbb{R}-linear map sending 1∈ℝ1 \in \mathbb{R} to 1A1_A when n=0n = 0 and acting for n≥1n \ge 1 on pure tensors v1⊗⋯⊗vnv_1 \otimes \cdots \otimes v_n (vi∈Viv_i \in V_i) by μf(v1⊗⋯⊗vn)=f1(v1)⋯fn(vn).\mu_f(v_1 \otimes \cdots \otimes v_n) = f_1(v_1) \cdots f_n(v_n).(1)✦

The Claim

Theorem 1. Let HH be a complex Hilbert space.

  1. Given a factor M⊆B(H)M \subseteq B(H), a finite index set II, a family (ai)i∈I(a_i)_{i \in I} in MM, and a ℂ\mathbb{C}-linearly independent family (bi)i∈I(b_i)_{i \in I} in M′M', ∑i∈Iaibi=0⟹ai=0for all i∈I.\sum_{i \in I} a_i b_i = 0 \implies a_i = 0 \quad \text{for all } i \in I.(2)✦

  2. For any factor M⊆B(H)M \subseteq B(H), arbitrary index sets II and KK, and ℂ\mathbb{C}-linearly independent families (ai)i∈I(a_i)_{i \in I} in MM and (bj)j∈K(b_j)_{j \in K} in M′M', (aibj)(i,j)∈I×K is ℂ-linearly independent in B(H).(a_i b_j)_{(i, j) \in I \times K} \text{ is } \mathbb{C}\text{-linearly independent in } B(H).(3)✦

  3. Let n∈ℕn \in \mathbb{N} and let M1,…,Mn⊆B(H)M_1, \dots, M_n \subseteq B(H) be pairwise commuting factors, meaning Mj⊆Mi′M_j \subseteq M_i' whenever i≠ji \neq j. For finite index sets K1,…,KnK_1, \dots, K_n and ℂ\mathbb{C}-linearly independent families (ai,j)j∈Ki(a_{i, j})_{j \in K_i} in MiM_i (i=1,…,ni = 1, \dots, n), with K=∏i=1nKiK = \prod_{i=1}^n K_i, (∏i=1nai,p(i))p∈K is ℂ-linearly independent in B(H).\Bigl(\prod_{i=1}^n a_{i, p(i)}\Bigr)_{p \in K} \text{ is } \mathbb{C}\text{-linearly independent in } B(H).(4)✦

  4. Whenever n∈ℕn \in \mathbb{N}, M1,…,Mn⊆B(H)M_1, \dots, M_n \subseteq B(H) are pairwise commuting factors, V1,…,VnV_1, \dots, V_n are finite-dimensional real vector spaces, and f=(fi)i=1nf = (f_i)_{i=1}^n consists of injective ℝ\mathbb{R}-linear maps fi:Vi→B(H)f_i \colon V_i \to B(H) taking selfadjoint values in MiM_i, the tensor multiplication map μf\mu_f of (1) satisfies μf:⨂ℝ,i=1nVi→B(H) is injective.\mu_f \colon \bigotimes_{\mathbb{R}, i=1}^n V_i \longrightarrow B(H) \text{ is injective}.(5)✦

In the paper, the result reads: for a factor M⊆B(H)M \subseteq B(H) with 1M=11_M = 1 (where 11 stands for the paper’s 1H1_H) and a unital subalgebra B⊆M′B \subseteq M', the multiplication map on the complex algebraic tensor product M⊙B→B(H)M \odot B \to B(H) is injective, and similarly for the algebraic tensor product of any finite family of commuting factors.

What Is Checked

The claims (2), (3), (4), and (5) are stated over a complex Hilbert space HH, factor von Neumann algebras M⊆B(H)M \subseteq B(H) and pairwise commuting factors M1,…,Mn⊆B(H)M_1, \dots, M_n \subseteq B(H) (n∈ℕn \in \mathbb{N}), operator families in MM, M′M', and M1,…,MnM_1, \dots, M_n (finitely indexed in (2) with ℂ\mathbb{C}-linear independence in M′M', ℂ\mathbb{C}-linearly independent in MM and M′M' for (3), and finitely indexed and ℂ\mathbb{C}-linearly independent in each MiM_i for (4)), and the tensor multiplication map (1) for injective ℝ\mathbb{R}-linear maps from finite-dimensional real vector spaces taking selfadjoint values in MiM_i for each i=1,…,ni = 1, \dots, n.