(All vector spaces are over a fixed field $F$).

Universal Property of Tensor Product. Given two finite dimensional vector spaces $V$ and $W$, the tensor product of $V$ and $W$ is a vector space $V\otimes W$, along with a multilinear map $\pi:V\times W\to V\otimes W$ such that whenever there is multilinear map $A:V\times W\to X$ to any vector space $X$, there exists a unique linear map $\bar A:V\otimes W\to X$ such that $\bar A\circ \pi = A$.

Notation. We write $\pi(v,w)$ as $v\otimes w$.

First Question: Is the map $\pi :V\times W\to V\otimes W$ sujective?

I think it should be surjective since $\pi(V\times W)$ also satisfies the universal property of tensor product of $V$ and $W$. Since $V\otimes W$ is given by a universal property, it is unique upto a unique isomorphism and thus $\pi$ should be surjective.

Please correct me if I am wrong anywhere in the the above said things.

Main Question: Assuming all is well so far, I am confused in the following:

Let $(e_1,\ldots, e_m)$ and $(f_1,\ldots, f_n)$ be bases for $V$ and $W$ repectively.

I was wondering what member of $V\times W$ maps to $e_1\otimes f_1 + e_2\otimes f_2$.

Since $\pi$ is surjective, some member should map to it.

But I am lost as to how to find it.

Can somebody help?


The map $\pi$ is not linear, so the image of $\pi$ is not a vector space and can not satisfy the universal property.

We have $\pi((e_1,f_1)+(e_2,f_2)) \ne \pi((e_1,f_1))+\pi((e_2,f_2))$.

A multilinear map is just linear in a single component if we fix all other components.

  • $\begingroup$ This clears a lot of things. Thanks. $\endgroup$ – caffeinemachine Feb 25 '15 at 15:01
  • 1
    $\begingroup$ Maybe I should add that we know $\pi( V \times W) $ spans $ V \otimes W $ since it contains a basis namely $\{ e_i \otimes f_j : 1 \le i \le m, 1 \le j \le n \}$. Using this we know that $V \otimes W$ has dimension $mn$, so we know that $\pi$ can not be surjective because $V \otimes W$ has dimension $m+n$. $\endgroup$ – user60589 Feb 25 '15 at 15:10
  • $\begingroup$ @user60589 You wrote $\otimes$ instead of $\times$. $\endgroup$ – Oscar Cunningham Feb 26 '15 at 8:21

In general it is not possible for $\pi$ to be surjective because the domain has a smaller dimension than the range provided then dimension of each factor is at least $2$. If $V$ has dimension $m$ and $W$ has dimension $n$, then $V\times W$ has dimension $m+n$ and $V\otimes W$ has dimension $mn$.

$\pi(V\times W)$ is not a subspace of $V\otimes W$, but it contains a basis, hence its span is the whole space.

  • $\begingroup$ @caffeinemachine You're welcome. $\endgroup$ – Matt Samuel Feb 25 '15 at 15:04

I am new to this stuff, so take my answer with a grain of salt. Your map $\pi$ is usually thought of as the inclusion map, is my understanding. So it need not be surjective. But $\pi$ maps the image of $V\times W$ into $V\otimes W$ injectively.

  • $\begingroup$ Am I wrong in thinking that $\pi(V\times W)$ satisfies the universal property? $\endgroup$ – caffeinemachine Feb 25 '15 at 14:26
  • $\begingroup$ I think if $\pi$ is surjective then it does. But we don't know necessarily know that $\pi$ is surjective, right? $\endgroup$ – Laars Helenius Feb 25 '15 at 14:29
  • $\begingroup$ $\pi$ is not surjective (usually) indeed (nor is it injective). But what do you mean by $\pi$ mapping the image of $V \times W $ into $V \otimes W$ injectively? The image of $V \times W$ is already a subset of $V \otimes W$. $\endgroup$ – darij grinberg Feb 25 '15 at 15:02
  • $\begingroup$ I meant that every element in the image of $\pi$ has a unique preimage. That is $\forall\,y\in\pi(V\times W)\,\exists!\,x\inV\times X$ such that $\pi(x)=y$. $\endgroup$ – Laars Helenius Feb 25 '15 at 15:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.