1.8.2 Vector Space Tensor Abstraction

Vector Space Tensor Abstraction generalizes multilinear relationships, enabling manipulation of vectors through tensor operations in higher dimensions.

Vector Space Tensor Abstraction is the specific instance of tensor abstraction obtained when the underlying algebraic objects are vector spaces over a field, a setting in which the general tensor product construction acquires extra structure and extra canonical isomorphisms not available for modules over an arbitrary ring, because every vector space has a basis and every finite-dimensional vector space is reflexive. It identifies exactly which pleasant properties of tensors depend on working over a field with bases, marking the boundary between what abstraction preserves universally and what it gains specifically from vector space structure.

Why Vector Spaces Are a Privileged Case

The Existence of a Basis

Every vector space over a field has a basis, a fact that does not hold for general modules over a general ring. This guarantees that the tensor product of vector spaces can always be described concretely, once bases of the factors are chosen, as the space spanned by the tensor products of basis vectors, with no obstruction of the kind that can arise for modules lacking a basis.

V \otimes W = span (e_{i} \otimes f_{j})

Finite-Dimensional Reflexivity

For a finite-dimensional vector space V, the natural map from V to its double dual V^{**} is an isomorphism, a property called reflexivity, which holds automatically in finite dimensions but can fail for infinite-dimensional spaces and for general modules, and which underlies several canonical identifications available only in this setting.

V ≅ V^{* *}

Dimension Behaves Multiplicatively

The Dimension Formula

For finite-dimensional vector spaces V and W of dimensions m and n, the tensor product V ⊗ W has dimension exactly mn, a clean multiplicative formula that follows directly from the basis description of the tensor product and that does not hold in general for the tensor product of arbitrary modules.

dim (V \otimes W) = dim (V) \cdot dim (W)

Contrast with Modules Over a General Ring

For modules over a ring that is not a field, an analogous tensor product can have unexpected size, including collapsing entirely to the zero module even when the factors are nonzero, a phenomenon with no counterpart for vector spaces, illustrating how much of the well-behaved dimension theory is special to the field case.

Canonical Isomorphisms Available for Vector Spaces

Tensor Product and Linear Maps

For finite-dimensional vector spaces, the space of linear maps from V to W is canonically isomorphic to the tensor product of V^* and W, an identification that relies on reflexivity and that expresses every linear map as a formal sum of rank-one maps built from covectors and vectors.

Hom (V, W) ≅ V^{*} \otimes W

Duals of Tensor Products

For finite-dimensional vector spaces, the dual of a tensor product is canonically isomorphic to the tensor product of the duals, again an identification that depends on finite dimensionality and fails, or requires extra care, in the infinite-dimensional setting.

{(V \otimes W)}^{*} ≅ V^{*} \otimes W^{*}

The Iterated Tensor Product and Tensor Spaces

Building Type (p, q) Spaces from Vector Space Data

The tensor spaces T^p_q(V) used throughout tensor algebra are themselves instances of vector space tensor abstraction, built by repeated application of the tensor product to copies of a single finite-dimensional vector space V and its dual V^*, with the dimension formula giving each T^p_q(V) dimension n^{p+q} for V of dimension n.

Symmetric and Exterior Powers as Vector Space Constructions

The symmetric and exterior powers of a vector space, obtained by taking the tensor power V^{⊗k} and passing to the symmetric or antisymmetric part, are likewise constructions that depend on vector space structure for their standard dimension formulas, such as the binomial coefficient count of a basis for the exterior power.

dim (⋀^{k}, V) = (\binom{n}{k})

Where Vector Space Abstraction Meets the More General Theory

Specializing the Module Construction

Every statement made about the tensor product of vector spaces can be recovered as a special case of the tensor product of modules by regarding a vector space as a module over its field of scalars, so vector space tensor abstraction does not introduce a separate construction but rather identifies the additional theorems that become available once the ring in question is a field.

The Boundary of the Privileged Properties

Recognizing precisely which properties, existence of a basis, multiplicative dimension, reflexivity-based isomorphisms, depend on the field and finite-dimensionality assumptions is what allows a practitioner to know in advance which familiar vector space facts about tensors will fail to carry over when working with modules, infinite-dimensional spaces, or other more general settings.

Diagrammatic Summary

The diagram nests vector spaces inside the wider class of modules, marking the extra properties, multiplicative dimension and canonical reflexivity-based isomorphisms, that vector space tensor abstraction contributes beyond what holds for modules in general.