5
$\begingroup$

Let $M, N$ be finitely generated modules over a commutative Noetherian ring $R$. There are at least three ways to compute the Tor-modules $\operatorname{Tor}^R_i(M, N)$ all of which gives isomorphic modules hence the construction is unambiguous:

(1) Take a free resolution of $M$, tensor with $N$ and compute homologies.

(2) Take a free resolution of $N$, tensor with $M$ and compute homologies.

(3) Take free resolutions of $M$ and $N$, build the total tensor product complex and compute homologies.

Now assume $R$ is $\mathbb Z$-graded (or more generally, graded by an abelian group) and $M, N$ are graded $R$-modules. In each of the above three constructions, we can take resolution by finitely generated graded free modules which in turn produce gradings on the Tor modules. My question is: Are all these grading on the Tor modules coming from the above constructions are same? I.e., do all these constructions produce graded isomorphic modules?

I feel like there should be some standard literature discussing these, but I have been unable to locate any (perhaps I don't know the correct sources). Any pointers to literature would also be very appreciated. Thank you

$\endgroup$
3
  • 1
    $\begingroup$ Yes, the usual comparison maps preserve the internal grading. $\endgroup$ Commented Jun 8 at 11:34
  • $\begingroup$ @DaveBenson: do you have a reference for this? $\endgroup$ Commented Jun 8 at 13:40
  • $\begingroup$ @uno In Cartan and Eilenberg's 1956 book "Homological Algebra," Proposition 1.1a in chapter 6 asserts that the tensor product is a left balanced functor (which basically amounts to having the isomorphisms you ask about in your question). I believe their argument applies in the graded setting, so that is a reference you could use. $\endgroup$ Commented Jun 8 at 16:35

1 Answer 1

5
$\begingroup$

I believe this is for the straightforward reason pointed out by Dave Benson in the comments, which I will elaborate upon. Noetherian and finite generation hypotheses are not relevant.

Let $R$ be a graded commutative ring and $P_\bullet\to M$, $Q_\bullet\to N$ free resolutions of the modules $M,N$ (by graded module homomorphisms and graded differentials). In the ungraded category, the canonical disambiguation of $\mathsf{Tor}$ comes from the quasiisomorphisms: $P_\bullet\otimes_RN\leftarrow P_\bullet\otimes^{\mathsf{total}}_R Q_\bullet\rightarrow M\otimes_RQ_\bullet$, where the maps involved are quite straightforward - e.g. the map $Q_0\to N$ makes a map $\oplus_{j+k=n}P_j\otimes Q_k\to P_n\otimes Q_0\to P_n\otimes N$ which defines the $n$th component of $P_\bullet\otimes Q_\bullet\to P_\bullet\otimes N$.

Any sensible convention of graded tensor product will have that if $Q_0\to N$ is graded, then so is $P_n\otimes Q_0\to P_n\otimes N$. And so on. Therefore, we see these quasiisomorphisms are all graded maps of complexes of graded $R$-modules, so that the resulting identification (on taking homology and inverting) is a graded isomorphism.

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.