4.24.4 Tensor Multilinear Map Tensor Product Boundary

Explore how tensor multilinear maps define tensor products and their boundary properties in algebraic structures.

Tensor Multilinear Map Tensor Product Boundary is the delineation of exactly which multilinear maps admit a factorization through a tensor product, marking the point at which a multilinear map out of a Cartesian product V_1 × V_2 × ⋯ × V_k is replaced by a linear map out of the tensor product V_1 ⊗ V_2 ⊗ ⋯ ⊗ V_k without loss of information. The boundary is not a restriction on which multilinear maps qualify — every multilinear map factors this way — but a precise statement of what the factorization preserves and what changes when a multilinear map is converted into its tensor product form.

The Factorization Statement

From Multilinear to Linear

Given a multilinear map

T : V_{1} \times \dots \times V_{k} \to F

the universal property of the tensor product guarantees a unique linear map

\tilde{T} : V_{1} \otimes \dots \otimes V_{k} \to F

such that

\tilde{T} (v_{1} \otimes \dots \otimes v_{k}) = T (v_{1}, \dots, v_{k})

on every decomposable element v_1 ⊗ ⋯ ⊗ v_k. This equation is the boundary in operational form: it identifies which linear map on the tensor product corresponds to a given multilinear map, and the uniqueness clause guarantees that no two distinct multilinear maps are ever assigned the same linear map on the tensor product.

Bijective Correspondence, Not Mere Existence

The boundary is strict in both directions: every multilinear map produces exactly one linear map on the tensor product, and every linear map on the tensor product, when restricted to decomposable elements and composed with the canonical map v_1, ..., v_k ↦ v_1 ⊗ ⋯ ⊗ v_k, recovers exactly one multilinear map. This two-sided correspondence is what makes the tensor product boundary a genuine identification of two categories of object rather than a one-way conversion that discards information.

What the Boundary Does Not Claim

It Does Not Extend to Non-Multilinear Maps

A map that fails the multilinear boundary condition — one that is not separately linear in each argument — is not assigned any linear map on the tensor product by this correspondence, because the universal property that produces T~ is stated only for multilinear T. Attempting to apply the factorization to a non-multilinear map produces no well-defined result, since the defining equation on decomposable elements would be inconsistent with the relations already imposed on the tensor product.

It Does Not Assign Meaning Outside Decomposable Elements Independently

The value of T~ on a general element of V_1 ⊗ ⋯ ⊗ V_k, which is a finite sum of decomposable elements rather than a single one, is fixed by linearity once the values on decomposable elements are fixed; the boundary does not permit an independent choice on sums, since any such choice would have to agree with the linear extension already forced by the equation above.

Role in the Tensor Framework

The Hinge Between Two Descriptions of the Same Data

This boundary is what allows multilinear algebra to be studied equivalently as a theory of multi-argument maps or as a theory of linear maps on a single tensor product space, and results proved on one side transfer automatically to the other. Every later identification of tensors with multilinear maps — of type (p, q) tensors with multilinear maps on p copies of V* and q copies of V, for instance — is an instance of this same factorization boundary applied to a specific choice of the spaces V_i.