Tensor product

(Redirected from Tensor products)

In mathematics, the tensor product of two vector spaces V and W (over the same field) is a vector space to which is associated a bilinear map that maps a pair to an element of denoted .

An element of the form is called the tensor product of v and w. An element of is a tensor, and the tensor product of two vectors is sometimes called an elementary tensor or a decomposable tensor. The elementary tensors span in the sense that every element of is a sum of elementary tensors. If bases are given for V and W, a basis of is formed by all tensor products of a basis element of V and a basis element of W.

The tensor product of two vector spaces captures the properties of all bilinear maps in the sense that a bilinear map from into another vector space Z factors uniquely through a linear map (see Universal property).

Tensor products are used in many application areas, including physics and engineering. For example, in general relativity, the gravitational field is described through the metric tensor, which is a tensor field with one tensor at each point of the space-time manifold, and each belonging to the tensor product of the cotangent space at the point with itself.

Definitions and constructions

edit

The tensor product of two vector spaces is a vector space that is defined up to an isomorphism. There are several equivalent ways to define it. Most consist of defining explicitly a vector space that is called a tensor product, and, generally, the equivalence proof results almost immediately from the basic properties of the vector spaces that are so defined.

The tensor product can also be defined through a universal property; see § Universal property, below. As for every universal property, all objects that satisfy the property are isomorphic through a unique isomorphism that is compatible with the universal property. When this definition is used, the other definitions may be viewed as constructions of objects satisfying the universal property and as proofs that there are objects satisfying the universal property, that is that tensor products exist.

From bases

edit

Let V and W be two vector spaces over a field F, with respective bases   and  .

The tensor product   of V and W is a vector space that has as a basis the set of all   with   and  . This definition can be formalized in the following way (this formalization is rarely used in practice, as the preceding informal definition is generally sufficient):   is the set of the functions from the Cartesian product   to F that have a finite number of nonzero values. The pointwise operations make   a vector space. The function that maps   to 1 and the other elements of   to 0 is denoted  .

The set   is then straightforwardly a basis of  , which is called the tensor product of the bases   and  .

We can equivalently define   to be the set of bilinear forms on   that are nonzero at only a finite number of elements of  . To see this, given   and a bilinear form  , we can decompose   and   in the bases   and   as:   where only a finite number of  's and  's are nonzero, and find by the bilinearity of   that:  

Hence, we see that the value of   for any   is uniquely and totally determined by the values that it takes on  . This lets us extend the maps   defined on   as before into bilinear maps   , by letting:  

Then we can express any bilinear form   as a (potentially infinite) formal linear combination of the   maps according to:   making these maps similar to a Schauder basis for the vector space   of all bilinear forms on  . To instead have it be a proper Hamel basis, it only remains to add the requirement that   is nonzero at an only a finite number of elements of  , and consider the subspace of such maps instead.

In either construction, the tensor product of two vectors is defined from their decomposition on the bases. More precisely, taking the basis decompositions of   and   as before:  

This definition is quite clearly derived from the coefficients of   in the expansion by bilinearity of   using the bases   and  , as done above. It is then straightforward to verify that with this definition, the map   is a bilinear map from   to   satisfying the universal property that any construction of the tensor product satisfies (see below).

If arranged into a rectangular array, the coordinate vector of   is the outer product of the coordinate vectors of   and  . Therefore, the tensor product is a generalization of the outer product, that is, an abstraction of it beyond coordinate vectors.

A limitation of this definition of the tensor product is that, if one changes bases, a different tensor product is defined. However, the decomposition on one basis of the elements of the other basis defines a canonical isomorphism between the two tensor products of vector spaces, which allows identifying them. Also, contrarily to the two following alternative definitions, this definition cannot be extended into a definition of the tensor product of modules over a ring.

As a quotient space

edit

A construction of the tensor product that is basis independent can be obtained in the following way.

Let V and W be two vector spaces over a field F.

One considers first a vector space L that has the Cartesian product   as a basis. That is, the basis elements of L are the pairs   with   and  . To get such a vector space, one can define it as the vector space of the functions   that have a finite number of nonzero values and identifying   with the function that takes the value 1 on   and 0 otherwise.

Let R be the linear subspace of L that is spanned by the relations that the tensor product must satisfy. More precisely, R is spanned by the elements of one of the forms:

 

where  ,   and  .

Then, the tensor product is defined as the quotient space:

 

and the image of   in this quotient is denoted  .

It is straightforward to prove that the result of this construction satisfies the universal property considered below. (A very similar construction can be used to define the tensor product of modules.)

Universal property

edit
 
Universal property of tensor product: if h is bilinear, there is a unique linear map ~h that makes the diagram commutative (that is, h = ~hφ).

In this section, the universal property satisfied by the tensor product is described. As for every universal property, two objects that satisfy the property are related by a unique isomorphism. It follows that this is a (non-constructive) way to define the tensor product of two vector spaces. In this context, the preceding constructions of tensor products may be viewed as proofs of existence of the tensor product so defined.

A consequence of this approach is that every property of the tensor product can be deduced from the universal property, and that, in practice, one may forget the method that has been used to prove its existence.

The "universal-property definition" of the tensor product of two vector spaces is the following (recall that a bilinear map is a function that is separately linear in each of its arguments):

The tensor product of two vector spaces V and W is a vector space denoted as  , together with a bilinear map   from   to  , such that, for every bilinear map  , there is a unique linear map  , such that   (that is,   for every   and  ).

Linearly disjoint

edit

Like the universal property above, the following characterization may also be used to determine whether or not a given vector space and given bilinear map form a tensor product.[1]

Theorem — Let  , and   be complex vector spaces and let   be a bilinear map. Then   is a tensor product of   and   if and only if[1] the image of   spans all of   (that is,  ), and also   and   are  -linearly disjoint, which by definition means that for all positive integers   and all elements   and   such that  ,

  1. if all   are linearly independent then all   are  , and
  2. if all   are linearly independent then all   are  .

Equivalently,   and   are  -linearly disjoint if and only if for all linearly independent sequences   in   and all linearly independent sequences   in  , the vectors   are linearly independent.

For example, it follows immediately that if   and   are positive integers then   and the bilinear map   defined by sending   to   form a tensor product of   and  .[2] Often, this map   will be denoted by   so that   denotes this bilinear map's value at  .

As another example, suppose that   is the vector space of all complex-valued functions on a set   with addition and scalar multiplication defined pointwise (meaning that   is the map   and   is the map  ). Let   and   be any sets and for any   and  , let   denote the function defined by  . If   and   are vector subspaces then the vector subspace   of   together with the bilinear map:   form a tensor product of   and  .[2]

Properties

edit

Dimension

edit

If V and W are vectors spaces of finite dimension, then   is finite-dimensional, and its dimension is the product of the dimensions of V and W.

This results from the fact that a basis of   is formed by taking all tensor products of a basis element of V and a basis element of W.

Associativity

edit

The tensor product is associative in the sense that, given three vector spaces  , there is a canonical isomorphism:

 

that maps   to  .

This allows omitting parentheses in the tensor product of more than two vector spaces or vectors.

Commutativity as vector space operation

edit

The tensor product of two vector spaces   and   is commutative in the sense that there is a canonical isomorphism:

 

that maps   to  .

On the other hand, even when  , the tensor product of vectors is not commutative; that is  , in general.

The map   from   to itself induces a linear automorphism that is called a braiding map. More generally and as usual (see tensor algebra), let   denote the tensor product of n copies of the vector space V. For every permutation s of the first n positive integers, the map:

 

induces a linear automorphism of  , which is called a braiding map.

Tensor product of linear maps

edit

Given a linear map  , and a vector space W, the tensor product:

 

is the unique linear map such that:

 

The tensor product   is defined similarly.

Given two linear maps   and  , their tensor product:

 

is the unique linear map that satisfies:

 

One has:

 

In terms of category theory, this means that the tensor product is a bifunctor from the category of vector spaces to itself.[3]

If f and g are both injective or surjective, then the same is true for all above defined linear maps. In particular, the tensor product with a vector space is an exact functor; this means that every exact sequence is mapped to an exact sequence (tensor products of modules do not transform injections into injections, but they are right exact functors).

By choosing bases of all vector spaces involved, the linear maps f and g can be represented by matrices. Then, depending on how the tensor   is vectorized, the matrix describing the tensor product   is the Kronecker product of the two matrices. For example, if V, X, W, and Y above are all two-dimensional and bases have been fixed for all of them, and f and g are given by the matrices:   respectively, then the tensor product of these two matrices is:  

The resultant rank is at most 4, and thus the resultant dimension is 4. rank here denotes the tensor rank i.e. the number of requisite indices (while the matrix rank counts the number of degrees of freedom in the resulting array).  .

A dyadic product is the special case of the tensor product between two vectors of the same dimension.

General tensors

edit

For non-negative integers r and s a type   tensor on a vector space V is an element of:   Here   is the dual vector space (which consists of all linear maps f from V to the ground field K).

There is a product map, called the (tensor) product of tensors:[4]  

It is defined by grouping all occurring "factors" V together: writing   for an element of V and   for an element of the dual space:  

If V is finite dimensional, then picking a basis of V and the corresponding dual basis of   naturally induces a basis of   (this basis is described in the article on Kronecker products). In terms of these bases, the components of a (tensor) product of two (or more) tensors can be computed. For example, if F and G are two covariant tensors of orders m and n respectively (i.e.   and  ), then the components of their tensor product are given by:[5]  

Thus, the components of the tensor product of two tensors are the ordinary product of the components of each tensor. Another example: let U be a tensor of type (1, 1) with components  , and let V be a tensor of type   with components  . Then:   and:  

Tensors equipped with their product operation form an algebra, called the tensor algebra.

Evaluation map and tensor contraction

edit

For tensors of type (1, 1) there is a canonical evaluation map:   defined by its action on pure tensors:  

More generally, for tensors of type  , with r, s > 0, there is a map, called tensor contraction:   (The copies of   and   on which this map is to be applied must be specified.)

On the other hand, if   is finite-dimensional, there is a canonical map in the other direction (called the coevaluation map):   where   is any basis of  , and   is its dual basis. This map does not depend on the choice of basis.[6]

The interplay of evaluation and coevaluation can be used to characterize finite-dimensional vector spaces without referring to bases.[7]

Adjoint representation

edit

The tensor product   may be naturally viewed as a module for the Lie algebra   by means of the diagonal action: for simplicity let us assume  , then, for each  ,   where   is the transpose of u, that is, in terms of the obvious pairing on  ,  

There is a canonical isomorphism   given by:  

Under this isomorphism, every u in   may be first viewed as an endomorphism of   and then viewed as an endomorphism of  . In fact it is the adjoint representation ad(u) of  .

Linear maps as tensors

edit

Given two finite dimensional vector spaces U, V over the same field K, denote the dual space of U as U*, and the K-vector space of all linear maps from U to V as Hom(U,V). There is an isomorphism:   defined by an action of the pure tensor   on an element of  ,  

Its "inverse" can be defined using a basis   and its dual basis   as in the section "Evaluation map and tensor contraction" above:  

This result implies:   which automatically gives the important fact that   forms a basis of   where   are bases of U and V.

Furthermore, given three vector spaces U, V, W the tensor product is linked to the vector space of all linear maps, as follows:   This is an example of adjoint functors: the tensor product is "left adjoint" to Hom.

Tensor products of modules over a ring

edit

The tensor product of two modules A and B over a commutative ring R is defined in exactly the same way as the tensor product of vector spaces over a field:   where now   is the free R-module generated by the cartesian product and G is the R-module generated by these relations.

More generally, the tensor product can be defined even if the ring is non-commutative. In this case A has to be a right-R-module and B is a left-R-module, and instead of the last two relations above, the relation:   is imposed. If R is non-commutative, this is no longer an R-module, but just an abelian group.

The universal property also carries over, slightly modified: the map   defined by   is a middle linear map (referred to as "the canonical middle linear map"[8]); that is, it satisfies:[9]  

The first two properties make φ a bilinear map of the abelian group  . For any middle linear map   of  , a unique group homomorphism f of   satisfies  , and this property determines   within group isomorphism. See the main article for details.

Tensor product of modules over a non-commutative ring

edit

Let A be a right R-module and B be a left R-module. Then the tensor product of A and B is an abelian group defined by:   where   is a free abelian group over   and G is the subgroup of   generated by relations:  

The universal property can be stated as follows. Let G be an abelian group with a map   that is bilinear, in the sense that:  

Then there is a unique map   such that   for all   and  .

Furthermore, we can give   a module structure under some extra conditions:

  1. If A is a (S,R)-bimodule, then   is a left S-module, where  .
  2. If B is a (R,S)-bimodule, then   is a right S-module, where  .
  3. If A is a (S,R)-bimodule and B is a (R,T)-bimodule, then   is a (S,T)-bimodule, where the left and right actions are defined in the same way as the previous two examples.
  4. If R is a commutative ring, then A and B are (R,R)-bimodules where   and  . By 3), we can conclude   is a (R,R)-bimodule.

Computing the tensor product

edit

For vector spaces, the tensor product   is quickly computed since bases of V of W immediately determine a basis of  , as was mentioned above. For modules over a general (commutative) ring, not every module is free. For example, Z/nZ is not a free abelian group (Z-module). The tensor product with Z/nZ is given by:  

More generally, given a presentation of some R-module M, that is, a number of generators   together with relations:   the tensor product can be computed as the following cokernel:  

Here  , and the map   is determined by sending some   in the jth copy of   to   (in  ). Colloquially, this may be rephrased by saying that a presentation of M gives rise to a presentation of  . This is referred to by saying that the tensor product is a right exact functor. It is not in general left exact, that is, given an injective map of R-modules  , the tensor product:   is not usually injective. For example, tensoring the (injective) map given by multiplication with n, n : ZZ with Z/nZ yields the zero map 0 : Z/nZZ/nZ, which is not injective. Higher Tor functors measure the defect of the tensor product being not left exact. All higher Tor functors are assembled in the derived tensor product.

Tensor product of algebras

edit

Let R be a commutative ring. The tensor product of R-modules applies, in particular, if A and B are R-algebras. In this case, the tensor product   is an R-algebra itself by putting:   For example:  

A particular example is when A and B are fields containing a common subfield R. The tensor product of fields is closely related to Galois theory: if, say, A = R[x] / f(x), where f is some irreducible polynomial with coefficients in R, the tensor product can be calculated as:   where now f is interpreted as the same polynomial, but with its coefficients regarded as elements of B. In the larger field B, the polynomial may become reducible, which brings in Galois theory. For example, if A = B is a Galois extension of R, then:   is isomorphic (as an A-algebra) to the  .

Eigenconfigurations of tensors

edit

Square matrices   with entries in a field   represent linear maps of vector spaces, say  , and thus linear maps   of projective spaces over  . If   is nonsingular then   is well-defined everywhere, and the eigenvectors of   correspond to the fixed points of  . The eigenconfiguration of   consists of   points in  , provided   is generic and   is algebraically closed. The fixed points of nonlinear maps are the eigenvectors of tensors. Let   be a  -dimensional tensor of format   with entries   lying in an algebraically closed field   of characteristic zero. Such a tensor   defines polynomial maps   and   with coordinates:  

Thus each of the   coordinates of   is a homogeneous polynomial   of degree   in  . The eigenvectors of   are the solutions of the constraint:   and the eigenconfiguration is given by the variety of the   minors of this matrix.[10]

Other examples of tensor products

edit

Topological tensor products

edit

Hilbert spaces generalize finite-dimensional vector spaces to arbitrary dimensions. There is an analogous operation, also called the "tensor product," that makes Hilbert spaces a symmetric monoidal category. It is essentially constructed as the metric space completion of the algebraic tensor product discussed above. However, it does not satisfy the obvious analogue of the universal property defining tensor products;[11] the morphisms for that property must be restricted to Hilbert–Schmidt operators.[12]

In situations where the imposition of an inner product is inappropriate, one can still attempt to complete the algebraic tensor product, as a topological tensor product. However, such a construction is no longer uniquely specified: in many cases, there are multiple natural topologies on the algebraic tensor product.

Tensor product of graded vector spaces

edit

Some vector spaces can be decomposed into direct sums of subspaces. In such cases, the tensor product of two spaces can be decomposed into sums of products of the subspaces (in analogy to the way that multiplication distributes over addition).

Tensor product of representations

edit

Vector spaces endowed with an additional multiplicative structure are called algebras. The tensor product of such algebras is described by the Littlewood–Richardson rule.

Tensor product of quadratic forms

edit

Tensor product of multilinear forms

edit

Given two multilinear forms   and   on a vector space   over the field   their tensor product is the multilinear form:  [13]

This is a special case of the product of tensors if they are seen as multilinear maps (see also tensors as multilinear maps). Thus the components of the tensor product of multilinear forms can be computed by the Kronecker product.

Tensor product of sheaves of modules

edit

Tensor product of line bundles

edit

Tensor product of fields

edit

Tensor product of graphs

edit

It should be mentioned that, though called "tensor product", this is not a tensor product of graphs in the above sense; actually it is the category-theoretic product in the category of graphs and graph homomorphisms. However it is actually the Kronecker tensor product of the adjacency matrices of the graphs. Compare also the section Tensor product of linear maps above.

Monoidal categories

edit

The most general setting for the tensor product is the monoidal category. It captures the algebraic essence of tensoring, without making any specific reference to what is being tensored. Thus, all tensor products can be expressed as an application of the monoidal category to some particular setting, acting on some particular objects.

Quotient algebras

edit

A number of important subspaces of the tensor algebra can be constructed as quotients: these include the exterior algebra, the symmetric algebra, the Clifford algebra, the Weyl algebra, and the universal enveloping algebra in general.

The exterior algebra is constructed from the exterior product. Given a vector space V, the exterior product   is defined as:  

When the underlying field of V does not have characteristic 2, then this definition is equivalent to:  

The image of   in the exterior product is usually denoted   and satisfies, by construction,  . Similar constructions are possible for   (n factors), giving rise to  , the nth exterior power of V. The latter notion is the basis of differential n-forms.

The symmetric algebra is constructed in a similar manner, from the symmetric product:  

More generally:  

That is, in the symmetric algebra two adjacent vectors (and therefore all of them) can be interchanged. The resulting objects are called symmetric tensors.

Tensor product in programming

edit

Array programming languages

edit

Array programming languages may have this pattern built in. For example, in APL the tensor product is expressed as ○.× (for example A ○.× B or A ○.× B ○.× C). In J the tensor product is the dyadic form of */ (for example a */ b or a */ b */ c).

J's treatment also allows the representation of some tensor fields, as a and b may be functions instead of constants. This product of two functions is a derived function, and if a and b are differentiable, then a */ b is differentiable.

However, these kinds of notation are not universally present in array languages. Other array languages may require explicit treatment of indices (for example, MATLAB), and/or may not support higher-order functions such as the Jacobian derivative (for example, Fortran/APL).

See also

edit

Notes

edit
  1. ^ a b Trèves 2006, pp. 403–404.
  2. ^ a b Trèves 2006, pp. 407.
  3. ^ Hazewinkel, Michiel; Gubareni, Nadezhda Mikhaĭlovna; Gubareni, Nadiya; Kirichenko, Vladimir V. (2004). Algebras, rings and modules. Springer. p. 100. ISBN 978-1-4020-2690-4.
  4. ^ Bourbaki (1989), p. 244 defines the usage "tensor product of x and y", elements of the respective modules.
  5. ^ Analogous formulas also hold for contravariant tensors, as well as tensors of mixed variance. Although in many cases such as when there is an inner product defined, the distinction is irrelevant.
  6. ^ "The Coevaluation on Vector Spaces". The Unapologetic Mathematician. 2008-11-13. Archived from the original on 2017-02-02. Retrieved 2017-01-26.
  7. ^ See Compact closed category.
  8. ^ Hungerford, Thomas W. (1974). Algebra. Springer. ISBN 0-387-90518-9.
  9. ^ Chen, Jungkai Alfred (Spring 2004), "Tensor product" (PDF), Advanced Algebra II (lecture notes), National Taiwan University, archived (PDF) from the original on 2016-03-04{{citation}}: CS1 maint: location missing publisher (link)
  10. ^ Abo, H.; Seigal, A.; Sturmfels, B. (2015). "Eigenconfigurations of Tensors". arXiv:1505.05729 [math.AG].
  11. ^ Garrett, Paul (July 22, 2010). "Non-existence of tensor products of Hilbert spaces" (PDF).
  12. ^ Kadison, Richard V.; Ringrose, John R. (1997). Fundamentals of the theory of operator algebras. Graduate Studies in Mathematics. Vol. I. Providence, R.I.: American Mathematical Society. Thm. 2.6.4. ISBN 978-0-8218-0819-1. MR 1468229.
  13. ^ Tu, L. W. (2010). An Introduction to Manifolds. Universitext. Springer. p. 25. ISBN 978-1-4419-7399-3.

References

edit