We can form tensor products of a wide variety of objects. For example:
1.
The tensor product of an n dimensional vector u and an m dimensional vector v is an nm dimensional vector
.
2.
If A and B are operators on n and m dimensional vectors, respectively, then
is an operator on nm dimensional vectors.
3.
if
and
are Hilbert spaces, then
is also a Hilbert space. If
and
are finite dimensional with bases
and
respectively, then
has dimension nm with basis
.
Tensor products obey a number of nice rules.
For matrices A, B, C, D, U, vectors u, v, w, and scalars a, b, c, d the following hold:
Thus for matrices,
which specializes for scalars to
The conjugate transpose distributes over tensor products:
The tensor product of several matrices is unitary if and only if each one of the
matrices is unitary up to a constant. Let
.
Then
U is unitary if
and
.
Note that
.
This implies that
,
and therefore
must
be the zero vector of the tensor product Hilbert space.
This in turn implies (reminds us?) that the tensor product space is actually the equivalence
classes in a quotient space.
In particular, if A and B are vector spaces, F is the free abelian group on ,
and K is the subgroup of F generated by all elements of the following forms (where
a scalar):