Let $R$ be a subring of $S$ containing $1$. $N$ an $R$-module. Then the construction of $S \otimes _R N$ (D&F p360) is by quotienting the free abelian group $(S \times N)^{ab}$ by the group $H$ with generating elements \begin{align*} & (s_1+s_2,n)-(s_1,n) -(s_2,n), \\ & (s,n_1+n_2) - (s,n_1) - (s,n_2), \\ & (sr,n) - s,rn). \quad \quad \quad \quad \quad \quad \quad (*) \end{align*} Let $s \otimes n$ be coset of $(s,n)$, then we can define left $S-$module by $$s ( \sum _{finite} s_i \otimes n_i ) = \sum_{finite} (ss_i) \otimes n_i . $$

To show the above left action is well defined, Dummit&Foote argues we also quotient out \begin{align*} &(s'(s_1+s_2),n)-(s's_1,n) -(s's_2,n), \\ &(s's,n_1+n_2) - (s's,n_1) - (s's,n_2), \\ & (s'(sr),n) - s's,rn). \quad \quad \quad \quad \quad \quad \quad (**) \end{align*} so

$\sum s_i \otimes n_i - \sum s_i' \otimes n_i' = \bar{0} \Rightarrow \sum ss_i \otimes n_i - \sum ss_i' \otimes n_i' = \bar{0}.$

I don't follow the boxed line. Yes, the LHS can be written as sums of elements of $H$, then we can replace those with $s$ within - but I couldn't write this out rigorously.

We have $LHS = \sum n_i h_i \in H$, $n_i \in \mathbb{Z}$, $h_i$ are elements of form $(*)$. Hence $\sum n_i h'_i \in H$ where $h'_i$ are elements of form $(**)$. Then why can we deduce $\sum n_i h'_i = RHS$ ? Here $LHS$ and $RHS$ denote left and right side of implication.

  • 1
    $\begingroup$ Pick a finite index for $i$ (like say $i = 1,2,\dots,n$), and write it out long-hand. The point is, the difference explicitly in the LHS, when you express these in terms of the cosets they represent, will be a sum of elements in $H$, which is again in $H$ by the definition of "generates". Remember that $\overline{0}$ "means" $H$. It's harder to be much clearer than that, because $H$ is a very large set, typically. What D&F are doing here, is "forcing" $S$-bilinearity. $\endgroup$ Apr 26, 2017 at 11:24
  • $\begingroup$ May you write an example? I am not completely sure what you mean: Yes, I understand we first write the cosets as a sum of elements in $H$. Then we "force" $s$ to the expressions - but how do we know this yields the same in turn is the representation of the RHS of implication? I want to know how one writes this out. $\endgroup$
    – Bryan Shih
    Apr 26, 2017 at 11:45

1 Answer 1


I will do this for two summands, it's the same for more, but it can get cumbersome to write out.

So, we want to show that the definition:

$s\cdot(s_1\otimes n_1 + s_2\otimes n_2) = (ss_1)\otimes n_1 + (ss_2)\otimes n_2$ is well-defined, that is, if:

$[(s_1,n_1)+(s_2,n_2)] + H = [(s_1',n_1') + (s_2',n_2')] + H$, then:

$[(ss_1,n_1) + (ss_2,n_2)] + H = [(ss_1',n_1') + (ss_2',n_2')] + H$, as well.

By assumption, we are given that:

$(s_1,n_1) + (s_2,n_2) - (s_1',n_1') - (s_2',n_2') \in H$.

Now, let's look a bit closer at what happens to generators of $H$, when you stick an arbitrary $s \in S$ in the "first coordinate":

If we take:

$(s_1+s_2,n_1) - (s_1,n_1) - (s_2,n_2) \in H$, we have:

$(s(s_1+s_2),n_1) - (ss_1,n_1) - (ss_2,n_1) = (ss_1 + ss_2,n_1) - (ss_1,n_1) - (ss_2,n_1)$

which is also a generator of $H$ (with $ss_1$ and $ss_2$ taking the place of $s_1$ and $s_2$, respectively).

Similarly, with $(s_1,n_1+n_2) - (s_1,n_1) - (s_1,n_2)$, if we replace $s_1$ with $ss_1$, we get the generator (which is, of course, also in $H$):

$(ss_1,n_1+n_2) - (ss_1,n_1) - (ss_1,n_2)$.

Finally, it is easy to see that for any $r \in R$:

$(s(s_1r),n_1) - (ss_1,rn_1) = ((ss_1)r,n_1) - (ss_1,rn_1)$, which is a generator of $H$ of the third type in your $(\ast)$.

Therefore, if a sum:

$\sum\limits_j (s_j,n_j) \in H$ (being some finite sum of generators), it follows that:

$\sum\limits_j (ss_j,n_j) \in H$ (for any $s \in S$), as well (being the same finite sum of generators with an $s$ left-multiplied to the first entry of each, which as we have seen above is also a sum of generators, so in $H$ by closure).

But this means, that:

$(s_1,n_1) + (s_2,n_2) - (s_1',n_1') - (s_2',n_2') \in H$ then implies:

$(ss_1,n_1) + (ss_2,n_2) - (ss_1',n_1') - (ss_2',n_2') \in H$ also, so that the two cosets:

$[(ss_1,n_1) + (ss_2,n_2)] + H$ and $[(ss_1',n_1') + (ss_2',n_2')] + H$ are equal and the (left) $S$-action is well-defined on $S\otimes_R N \stackrel{\text{def}}{=} (S \times N)^{ab}/H$.


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.