Let $R$ be a ring $M$ an $R$-module. How can I prove that if

  • $M\cong R^n/N$ for some $n\!\in\!\mathbb{N}$ and some submodule $N\leq R^n$

and if

  • $M\cong R^{(I)}/\langle u_1,\ldots,u_m\rangle$ for some set $I$ and some vectors $u_1,\ldots,u_m\in R^{(I)}$,


  • $M\cong R^k/\langle v_1,\ldots,v_l\rangle$ for some $k\!\in\!\mathbb{N}$ and some vectors $v_1,\ldots,v_l\in R^k$ ?
  • $\begingroup$ Is it even possible for $I$ to be infinite in this case? $\endgroup$ – tomasz Jul 26 '13 at 13:16
  • $\begingroup$ @tomasz: A module can be finitely generated and infinitely related, e.g. take $R=K[x_1,x_2,x_3,\ldots]$ and the ideal/submodule $\mathfrak{a}=\langle\langle x_1^1,x_2^2,x_3^3,\ldots\rangle\rangle$ and $M=R/\mathfrak{a}$, which is an finitely generated (by $1$) infinitely related (by $x_i^i$ with $i\in\mathbb{N}$) $R$-module. $\endgroup$ – Leo Jul 26 '13 at 15:22
  • $\begingroup$ But I'm asking if it's possible for a finitely generated module to be obtained as a quotient of a free module of infinite rank by a finitely generated submodule. It seems intuitively impossible, as the quotient will still have as a submodule (even a direct summand!) a free module of infinite rank. $\endgroup$ – tomasz Jul 26 '13 at 18:55
  • $\begingroup$ @tomasz: Well, by YACP's answer, it is not possible. Your situation implies (by Schanuel's lemma) that $R^{(I)}$ is finitely generated, so $|I|<\infty$. $\endgroup$ – Leo Jul 26 '13 at 20:37

Using Schanuel's Lemma we find that $N\oplus R^{(I)}\simeq N'\oplus R^n$, where $N'=\langle u_1,\ldots,u_m\rangle$, so we get that $N\oplus R^{(I)}$ is finitely generated which shows that $N$ is finitely generated.

| cite | improve this answer | |
  • $\begingroup$ How does "$N\oplus R^{(I)}$ finitely generated" imply "$N$ finitely generated"? $\endgroup$ – Leo Jul 18 '13 at 3:16
  • 2
    $\begingroup$ It's trivial that $M_1\oplus M_2/M_2\simeq M_1$ and any factor of a finitely generated module is also finitely generated. $\endgroup$ – user26857 Jul 18 '13 at 7:16
  • $\begingroup$ VERY elegant, just what I was looking for, thank you! Related question: does "finitely generated & finitely related $\Rightarrow$ finitely presented" also hold for groups, rings, algebras? One would think so, but I'm worried, since we don't have Shanuel's lemma there. Any alternative (more universal) proof? $\endgroup$ – Leo Jul 18 '13 at 13:59
  • $\begingroup$ Any ideas about group presentations $\langle\ldots|\ldots\rangle$ and $R$-algebra presentations $R\langle\ldots|\ldots\rangle$? $\endgroup$ – Leo Jul 26 '13 at 15:31
  • $\begingroup$ Dear @LeonLampret, maybe you want to post it as a separate question in order to find better answers. $\endgroup$ – user26857 Jul 26 '13 at 15:59

Every generating system $E$ of a finitely generated module $M$ contains a finite generating system (namely, look at those generators of $E$ which are needed to generate a finite generating system of $M$). Now assume that there are only finitely many relations between the generators of $E$. But these only use finitely many generators of $E$. It follows that every presentation of $M$ can be adjusted to a finite presentation.

| cite | improve this answer | |
  • $\begingroup$ Could you please be more specific/explicit. I don't understand the last two sentences. Ok, we have elements $x_1,\ldots,x_n$ that generate $R^{(I)}/\langle u_1,\ldots,x_m\rangle$. Now what? $\endgroup$ – Leo Jul 17 '13 at 7:27
  • $\begingroup$ Is it clearer now? $\endgroup$ – Martin Brandenburg Jul 17 '13 at 7:46
  • 1
    $\begingroup$ No, it's even more confusing. I agree with the first sentence. Don't understand the second sentence: the number of relations is always infinite (when $R$ is inifnite), since it constitutes a submodule. It would really help if you were more explicit, with actual elements and concrete isomorphisms and submodules given by generators. You can assume known that any module has a presentation, which follows from the universal property of the free module $R^{(I)}$. $\endgroup$ – Leo Jul 17 '13 at 15:13
  • $\begingroup$ Of course I mean that finitely many relations suffice (to generate all other relations), as in your question! $\endgroup$ – Martin Brandenburg Aug 5 '13 at 16:36

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.