Structure as an Abelian Group.
Let and be vector spaces over the field of real numbers, . The tensor product is formed by treating and only as abelian groups (i.e., -modules). The resulting object is, by definition, an abelian group whose elements are finite sums of simple tensors for . The relations are those of a -module tensor product:
Absence of a Natural R-Vector Space Structure.
There is no natural way to define scalar multiplication by an element on a simple tensor . A proposed definition such as is not well-defined because it must be consistent with . However, the axioms of the tensor product over do not guarantee that for . For example, in , the elements and are distinct. Since the proposed scalar multiplication is ambiguous, does not carry a natural -vector space structure.
Structure as a Q-Vector Space.
Definition (Torsion-Free and Divisible Modules). A -module is torsion-free if for any non-zero integer , the map is injective. A -module is divisible if for any non-zero integer , the map is surjective.
A -module is a vector space over the rational numbers if and only if it is both torsion-free and divisible. Any -vector space , when viewed as a -module, has these properties. It is torsion-free because if for , then multiplication by implies . It is divisible because for any and non-zero , the equation has the solution .
Theorem (Structure of the Tensor Product). Let and be vector spaces over . The abelian group is a vector space over .
Proof. Since and are -vector spaces, the tensor product is also a -vector space. Scalar multiplication by is well-defined on a simple tensor by . This is well-defined because for any such that , .
Isomorphism and Dimension.
Theorem (Isomorphism over Q). If and are vector spaces over , there is a canonical isomorphism of -vector spaces:
Proof. The -bilinear map sending to induces a -module homomorphism . This map is also -linear. The map is not only -bilinear but also -bilinear, since for , . This induces a homomorphism . The maps and are mutually inverse.
Theorem (Dimension of the Tensor Product). Let and be -vector spaces with dimensions and . The dimension of as a -vector space is:
where is the cardinality of the continuum.
Proof. The dimension of as a vector space over is . For any -vector space , its dimension over is . Using the isomorphism and the rule for the dimension of a tensor product of vector spaces:
The final step uses the fact that in cardinal arithmetic.
Summary and Examples.
Example (). Let and , so and . The tensor product is a -vector space with dimension:
Example (). Let and , so and . In contrast to the -vector space , which has dimension 6, the tensor product over is a -vector space with dimension: