5.15 Tensor Product Equivalence Relation Structure

Tensor Product Equivalence Relation Structure is the equivalence relation imposed on the free module F(V × W) during the construction of the tensor product, under which two elements are declared equivalent exactly when their difference lies in the submodule generated by the bilinear relations. This equivalence relation is the precise mechanism that converts an unstructured collection of formal symbols indexed by pairs (v, w) into the algebraically meaningful space V ⊗ W, and understanding its structure clarifies exactly which formal sums represent the same tensor.

Defining the Equivalence Relation

The Relation Submodule

Let F(V × W) denote the free module generated by all pairs (v, w), and let R be the submodule generated by all elements of the four families:

The Induced Equivalence Relation

Two elements x, y ∈ F(V × W) are declared equivalent, written x ∼ y, precisely when:

Verifying the Equivalence Relation Axioms

Reflexivity

For any x ∈ F(V × W), x − x = 0 ∈ R, since R is a submodule and therefore contains the zero element. Hence x ∼ x.

Symmetry

If x ∼ y, then x − y ∈ R. Since R is a submodule, it is closed under scalar multiplication by −1, so y − x = −(x − y) ∈ R, giving y ∼ x.

Transitivity

If x ∼ y and y ∼ z, then x − y ∈ R and y − z ∈ R. Because R is closed under addition, (x − y) + (y − z) = x − z ∈ R, so x ∼ z.

Equivalence Classes as Tensors

The Tensor Product as a Set of Classes

The tensor product V ⊗ W is formally the set of equivalence classes of this relation:

and the simple tensor v ⊗ w denotes the equivalence class containing the basis element (v, w) of the free module.

Compatibility with the Module Structure

Because R is a submodule rather than an arbitrary subset, the equivalence relation is compatible with addition and scalar multiplication, meaning that if x ∼ x' and y ∼ y', then x + y ∼ x' + y' and cx ∼ cx' for any scalar c. This compatibility is exactly what allows the set of equivalence classes to inherit a well-defined module (or vector space) structure from F(V × W).

Visualizing the Structure

Consequences of the Equivalence Structure

Multiple Representatives for the Same Tensor

Since equivalence classes generally contain more than one element of the free module, a given tensor in V ⊗ W can be represented by more than one formal expression in F(V × W), and two such expressions represent the same tensor exactly when their difference lies in R.

Well-Definedness of Operations Depends on This Structure

Every operation defined on V ⊗ W, whether it is addition, scalar multiplication, or a linear map induced by a bilinear map, must be checked for well-definedness with respect to this equivalence relation, meaning the operation must give the same result regardless of which representative of an equivalence class is used.

Relation to the Universal Property

The Quotient Map as the Canonical Bilinear Map

The natural quotient map F(V × W) → F(V × W)/∼ restricted to the generating pairs (v, w) is exactly the canonical bilinear map ⊗: V × W → V ⊗ W, and the well-definedness of this restriction on equivalence classes is what allows the universal property of the tensor product to be established from the equivalence relation structure alone.

Generalization to Multilinear Constructions

The same equivalence relation pattern, generated by an appropriately enlarged relation submodule, extends directly to constructing iterated tensor products and other multilinear universal objects, following the identical logic of quotienting a free module by the relations needed to enforce the desired multilinear behavior.