1
$\begingroup$

If $A$ is a right $R$-module and $B$ is a left $R$-module, then the tensor product $A\otimes_R B$ is an object $X$ with a map $\theta:A\times B\rightarrow X$ such that $\theta$ is $R$-bilinear and $\theta(ar,b)=\theta(a,rb)$ for all $a\in A, b\in B, r\in R$; further the pair $(X,\theta)$ has following universal property. If $X'$ is an object and $\theta':A\times B\rightarrow X'$ is $R$-bilinear map with $\theta'(ar,b)=\theta'(a,rb)$ then there is unique $g:X\rightarrow X'$ such that $\theta\circ g=\theta'$.

In some books, the tensor product $A\otimes_R B$ is defined to be an abelian group $X$ with map $\theta$ satisfying universal properties as stated above. Whereas in some books, it is defined to be an $R$-module $X$ with a map $\theta:A\times B\rightarrow X$ satisfying above universal properties.

What is appropriate choice for tensor product, as module or just as abelian group?

If we accept it as abelian group, then we can give it a left $R$-module structure or right $R$-module structure; so it seems to me that it should considered as abelian group only (together with map $\theta$).

Whereas, the places, where the tensor product is defined to be a module, they do not mention whether it is left $R$-module or right $R$-module.

For definition of tensor product, we need to take one left $R$-module and one right $R$-module; so we can not define tensor product to be universal object within some category of left $R$-modules or within category of some right $R$-modules.


In above write-up, $\theta$ can be said to be balanced map; and tensor product is defined to be an object with a balanced map, satisfying universal property.

But my question is concerned about what the object there should be; whether it should be just abelian group of $R$-modules?

$\endgroup$
  • 1
    $\begingroup$ If $R$ is commutative, then $A\otimes_R B$ is an $R$-module. But when $R$ is not commutative you cannot define an $R$-module structure: for example, you need $A$ to be a bi-$R$-module then the left action of $R$ on $A\otimes_R B$ makes sense. $\endgroup$ – Eclipse Sun Apr 18 at 4:36
1
$\begingroup$

For commutative $R$, the usual construction gives an object $A\otimes_R B$ that's an $R$-module with action $r(a\otimes b) = (ra) \otimes b = a\otimes (rb)$. In the noncommutative case, $A\otimes_R B$ is still an abelian group, but the action above doesn't carry over; about the best you can do is to get a left $R$-module structure by setting $r(a\otimes b) = (ra)\otimes b$ if $A$ is an $R$-bimodule, or similarly (but distinctly) for $B$.

$\endgroup$

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.