Linear algebra (Osnabrück 2024-2025)/Part II/Lecture 55
Let a field. We consider two finite index sets and , and the mapping spaces and , which are both finite-dimensional -vector spaces. Is there a relation with the mapping set
For mappings and , we easily obtain a mapping on , called ( tensor ) that is defined by
For the standard vectors, we have
Every element can be written as a linear combination of elements of the form but not every has this simple form. We have
and accordingly in the second component.
In this lecture, we introduce an important construction for vector spaces, the so-called tensor product. In the special case just considered, the tensor product yields the mapping space on the product set; that is, we have
Here, the properties of the constructed object are more important than the construction itself. The construction is quite abstract, and rests on the construction of residue class spaces, and the following construction.
For an arbitrary set of symbols , and a field , we can construct the vector space that consists of all mappings
that have everywhere except for finitely many places the value . If denotes the mapping that has the value at the position , and everywhere else the value , then consists of all finite sums
Hence, the form a basis of this space.
This construction is a special case of the direct sum of (in general) infinitely many -vector spaces; we take the direct sum of the vector spaces with itself as often as the set tells us.
- The tensor product of vector spaces
Let be a field, and let be -vector spaces. We recall that a multilinear mapping in another -vector space is a mapping
that is -linear in every component, that is, when all the other components are fixed. We want to construct a vector space , together with a multilinear mapping
such that for every multilinear mapping as above, there exists a linear mapping
fulfilling . By this construction, multilinear mappings are transformed to linear mappings on a new vector space.
Let be a field, and let be -vector spaces. Let be the -vector space generated by all symbols (with , we write the basis elements as ). Let be the -linear subspace of generated by all elements of the form
- ,
- .
Then the residue class space is called the tensor product of the , . It is denoted by
We usually just write . The images of in are denoted by
This is the equivalence class of for the equivalence relation given by the linear subspace . Every element in has a (not unique) representation of the form
(with , and ). In particular, the decomposable tensors form a -generating system of the tensor product. The defining generators of the linear subspace become equations in the tensor product; they express the multilinearity. In particular, we have
for arbitrary , and
For and and , the elements of (in the sense of Definition) are linear combinations like
Using the standard vectors of and of , this is
Because these tuples are different, we can not simplify this expression in . The image of this element in is
This expression can be simplified as a linear combination of the family , , .
More important than the construction of the tensor product is the following universal property.
Theorem
Let be a field, and let denote vector spaces
over .- The
mapping
is -multilinear.
- Let be another
-vector space,
and let
denote a multilinear mapping. Then there exists a uniquely determined -linear mapping
fulfilling .
Proof
- follows immediately from the definition of the tensor product.
- Since the are a
-generating system
of , and because
must hold, there can exist at most one such a linear mapping. To show existence, we consider the -vector space from the construction of the tensor product. The form a basis of ; therefore, the assignment
defines a linear mapping
Because of the multilinearity of , the linear subspace is mapped by to . Hence, according to the factorization theorem, this mapping induces a -linear mapping
Corollary
Let be a field, and let and be vector spaces over . Then there exists a natural isomorphism
Proof
From Theorem 55.4 (2), we may conclude immediately that we have a bijection.
Corollary
Let be a field, and let denote vector spaces over . Then there exists a natural isomorphism
Proof
This follows immediately from Corollary 55.5 , applied to .
For , and , the multilinear mappings from to are just the bilinear forms on . Corollary 55.6 says in this situation that the dual space of represents all bilinear forms. For , the standard inner product (this term is only for correct) corresponds to the linear form given by
The tensor product is determined up to (unique) isomorphy by this universal property; this means the following.
Theorem
Let be a field, and let denote vector spaces over . Let be a -vector space, together with a multilinear mapping
and suppose that fulfills the universal property from Theorem 55.4 (2). Then there exists a uniquely determined isomorphism
Proof
Because is multilinear, there exists, due to the universal property of the tensor product, a uniquely determined linear mapping
Because of the universal property of and the multilinearity of , there is also a linear mapping
Because of the universal property, they are inverse to each other.
Therefore, this universal property is more important than the explicit construction of the tensor product.
Lemma
Let be a field, and let denote vector spaces
over . Then the following calculation rules hold.- For vectors
,
and
,
we have
- For vectors
,
we have
- Let
,
and
.
Then
Proof
(1) follows immediately from the construction. (2) follows from (1). (3) follows from the distributive law for multilinear mappings.
In , we have
Lemma
Let be a field, and let denote a
-vector space. Then the following properties hold.- We have
- We have
where corresponds to the vector
Proof
(1) follows from Lemma 55.9 (2).
(2). The scalar multiplication
is multilinear; therefore, there exists, according to Theorem 55.4 , a linear mapping
This mapping is surjective, because is mapped to . An element in the tensor product has the form
If this is mapped to , then we have
but then also the tensor element is , and the mapping is injective as well.
Theorem
Let be a field, and let denote vector spaces over . Let be index sets, and let
- If each family is a
generating system
of , then the family
is a generating system of .
- If each family is
linearly independent
in , then the family
is linearly independent in .
- If each family is a
basis
of , then the family
is a basis of .
Proof
(1). Due to the construction, the decomposable tensors form a generating system of the tensor product. Hence, it is enough to show that they are linear combinations of the given family. But this follows from Lemma 55.9 (3).
(2). We can restrict to finite families. We want to apply Lemma 14.8 . Let be fixed. Because of the linear independence of the families , , in , there exist linear forms
with and with for . Therefore,
is, according to Exercise 16.39 , a multilinear mapping. Due to Corollary 55.6 , we have a corresponding linear mapping
which sends to
and all other elements of the family to .
(3) follows from (1) and (2).
Corollary
Let be a field, and let denote finite-dimensional vector spaces over . Then the dimension of the tensor product is
Proof
This follows immediately from Theorem 55.12 (3).
We relate the motivating Example 55.1 with the general construction of the tensor product. The mapping (with the direct meaning of from the example)
is, due to Exercise 55.5 , multilinear. Because of Theorem 55.4 (2), there exists a uniquely determined linear mapping
where the tensor products correspond to each other. The surjectivity follows from the fact that the basis elements are in the image. The injectivity follows from Corollary 55.13 .
| << | Linear algebra (Osnabrück 2024-2025)/Part II | >> PDF-version of this lecture Exercise sheet for this lecture (PDF) |
|---|