5.19 Tensor Product of Vector Spaces Structure

Tensor Product of Vector Spaces Structure is the overall organization of V ⊗ W as a vector space in its own right, encompassing its basis, dimension, subspace decompositions, and the ways it relates functorially to its two factor spaces V and W. Where the universal property characterizes the tensor product by what it does, this structural perspective examines what V ⊗ W is once built: a concrete vector space whose internal organization, from its basis tensors down to its rank stratification, can be studied using the ordinary tools of linear algebra.

Basic Vector Space Structure

Addition and Scalar Multiplication

V ⊗ W is a vector space over the shared field F, with addition and scalar multiplication inherited from the quotient construction, satisfying all the standard vector space axioms, associativity, commutativity of addition, distributivity, and the existence of an additive identity and inverses.

Dimension from the Factor Spaces

For finite-dimensional V and W, the dimension of the tensor product is determined multiplicatively:

a formula that distinguishes the tensor product structurally from the direct sum, whose dimension is additive rather than multiplicative.

Basis Structure

Basis Tensors from Factor Bases

Given bases {eᵢ} of V and {fⱼ} of W, the collection {eᵢ ⊗ fⱼ} forms a basis of V ⊗ W, providing a canonical, though basis-choice-dependent, coordinate system in which every tensor has a unique expansion as a formal linear combination.

Independence from the Specific Basis Choice up to Change of Basis

While the specific basis tensors depend on the chosen bases of V and W, the underlying dimension and the abstract vector space structure of V ⊗ W do not; changing bases in V or W induces a corresponding change of basis in V ⊗ W, following the standard tensor transformation rules familiar from multilinear algebra.

Subspace and Rank Stratification

Stratification by Tensor Rank

As detailed under the formal sum structure, V ⊗ W is stratified by tensor rank, ranging from the single zero element (rank zero), through the simple tensors (rank one), up to the maximal possible rank min(dim V, dim W), giving a layered internal organization not visible from the dimension count alone.

The Cone of Simple Tensors

The set of simple tensors, those of rank at most one, forms a cone (closed under scalar multiplication but not under addition) within V ⊗ W, rather than a linear subspace, reflecting the earlier observation that elementary outputs are not closed under addition.

Diagram of the Overall Structure

Functorial Structure Relating V ⊗ W to V and W

Bifunctoriality

As established through the factor order and vector-space-pair perspectives, V ⊗ W depends functorially on both V and W simultaneously: linear maps φ: V → V' and ψ: W → W' induce a linear map φ ⊗ ψ: V ⊗ W → V' ⊗ W', and this assignment respects composition and identities in each variable.

The Swap and Associativity Isomorphisms

Beyond the basic vector space axioms, V ⊗ W participates in a network of canonical isomorphisms, the swap isomorphism V ⊗ W ≅ W ⊗ V and the associativity isomorphism (U ⊗ V) ⊗ W ≅ U ⊗ (V ⊗ W), that together endow the collection of all vector spaces, equipped with the tensor product, with the structure of a symmetric monoidal category.

Distinguishing Structural Layers

Algebraic Layer versus Coordinate Layer

The algebraic layer of the structure, dimension, universal property, and functorial behavior, is basis-independent and holds abstractly for any vector spaces V and W; the coordinate layer, specific basis tensors and their transformation rules, depends on chosen bases and is the layer most directly used in explicit computation.

Rank Layer as an Intermediate, Basis-Independent Invariant

Sitting between these two layers, the rank stratification is basis-independent (rank is a property of the tensor itself, not of any chosen coordinate expression) yet finer-grained than the overall dimension count, providing a structural invariant that captures how "spread out" across the basis tensors a given element is.

Broader Significance

A Template for Understanding Any Constructed Vector Space

Examining V ⊗ W through these layers, basic vector space axioms, basis and dimension, rank stratification, and functorial behavior, provides a template applicable to any vector space built by a universal construction, including symmetric powers, exterior powers, and Hom-spaces, each of which can be analyzed along the same structural dimensions.

Bridge Between Abstract Universal Property and Concrete Computation

Ultimately, the structural perspective on V ⊗ W is what makes the abstract universal-property characterization of the tensor product usable in practice: knowing the dimension, having an explicit basis, and understanding the rank stratification are precisely the ingredients needed to perform concrete calculations with tensors while remaining assured that these calculations are faithful to the abstract construction.