5.21.2 Tensor Product Associative Isomorphism

Tensor Product Associative Isomorphism is the specific canonical linear isomorphism α_{U,V,W} : (U ⊗ V) ⊗ W → U ⊗ (V ⊗ W), regarded as a mathematical object in its own right — a distinguished, uniquely determined map with precise algebraic properties (linearity, bijectivity, naturality) — rather than merely the informal claim that the two groupings "are the same." Studying this isomorphism as an object clarifies exactly what data it consists of, how it is proved to be well-defined and invertible, and how it interacts with linear maps between the underlying spaces, which is the technical content underlying all of tensor product associativity.

Formal Definition of the Map

Action on Simple Tensors

The associative isomorphism is the unique linear map satisfying

for all u ∈ U, v ∈ V, w ∈ W, indexed explicitly by the triple (U, V, W) to emphasize that a different isomorphism α_{U,V,W} exists for every choice of three vector spaces.

Existence via Nested Universal Properties

To construct α_{U,V,W} rigorously, fix w ∈ W and note that (u, v) ↦ u ⊗ (v ⊗ w) is bilinear in (u, v), inducing a linear map φ_w : U ⊗ V → U ⊗ (V ⊗ W). The assignment w ↦ φ_w is itself linear in w, so (t, w) ↦ φ_w(t) for t ∈ U ⊗ V is bilinear in (t, w), and applying the universal property once more to this bilinear map yields the single linear map α_{U,V,W} : (U ⊗ V) ⊗ W → U ⊗ (V ⊗ W).

Bijectivity of the Isomorphism

Constructing the Inverse Explicitly

By the mirror-image construction — fixing u ∈ U first and building a map from V ⊗ W into (U ⊗ V) ⊗ W — one obtains β_{U,V,W} : U ⊗ (V ⊗ W) → (U ⊗ V) ⊗ W satisfying β_{U,V,W}(u ⊗ (v ⊗ w)) = (u ⊗ v) ⊗ w.

Verifying Mutual Inverses on a Spanning Set

Both composites β_{U,V,W} ∘ α_{U,V,W} and α_{U,V,W} ∘ β_{U,V,W} act as the identity on all simple tensors of their respective domains, since composing the defining formulas gives back the original nested simple tensor unchanged. Because simple tensors span, and two linear maps agreeing on a spanning set are equal, both composites equal the respective identity maps, proving α_{U,V,W} is a bijection with inverse β_{U,V,W}.

Diagram of the Isomorphism and Its Inverse

Naturality of the Isomorphism

Naturality Square

The family {α_{U,V,W}}, ranging over all triples of vector spaces, is natural in each argument: for linear maps f : U → U′, g : V → V′, h : W → W′, the square

commutes. This means α is not just a single isomorphism for one choice of spaces but a natural transformation, connecting the associative isomorphism consistently across all vector spaces and all linear maps between them, a defining feature of α's role in the symmetric monoidal structure of vector spaces.

Why Naturality Matters

Naturality guarantees that regrouping a tensor product and then applying transformations to its factors gives the same result as applying the transformations first and regrouping afterward, so α behaves predictably under every linear map one might apply to U, V, or W, not merely for the identity maps.

Matrix Description in Finite Dimensions

Basis Correspondence

With bases {eᵢ}, {fⱼ}, {gₖ} of U, V, W, the isomorphism α_{U,V,W} sends the basis element (eᵢ ⊗ fⱼ) ⊗ gₖ of (U ⊗ V) ⊗ W to the basis element eᵢ ⊗ (fⱼ ⊗ gₖ) of U ⊗ (V ⊗ W). Relative to these correspondingly ordered bases, α_{U,V,W} is represented by a permutation matrix (in fact the identity matrix under the natural indexing), since it merely relabels the same triple of basis indices without altering any coefficients.

Pentagon Coherence for the Isomorphism

The Coherence Condition

For four vector spaces U, V, W, X, five distinct groupings exist, connected by instances of α. The pentagon coherence condition requires that the two paths of associative isomorphisms from ((U ⊗ V) ⊗ W) ⊗ X to U ⊗ (V ⊗ (W ⊗ X)) — one going through three applications of α in one order, the other through a different order — agree exactly, ensuring α's behavior is unambiguous when iterated.

Significance of the Associative Isomorphism

The Precise Object Behind Informal Associativity

Whereas "tensor product associativity" is often stated informally as (U ⊗ V) ⊗ W ≅ U ⊗ (V ⊗ W), the associative isomorphism α_{U,V,W} is the precise, natural, coherent map realizing that isomorphism, carrying all the structural information — linearity, invertibility, naturality, coherence — needed to use associativity safely in further constructions.

Building Block of Monoidal Category Structure

The associative isomorphism, together with the unit and symmetry isomorphisms, is one of the structural natural transformations that equips the category of vector spaces with its symmetric monoidal structure, the categorical setting in which tensor products and their coherence properties are formalized and studied.