Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Consider the following problem.

Problem: Suppose $M$ is a module over $\mathbb{Q}[x]$ such that $M$ is finitely generated over $\mathbb{Q}$. Prove that there is a non-zero polynomial $p(x) \in \mathbb{Q}[x]$ and a non-zero $m \in M$ such that $p(x)*m = 0$.

My attempt: Since $M$ is finitely generated over a field, $M \simeq \mathbb{Q}^n$ as a $\mathbb{Q}$-module. If $M$ is torsion-free over $\mathbb{Q}[x]$ too, then $M \simeq \mathbb{Q}[x]^m$ as $\mathbb{Q}[x]$-modules. Now restricting the action to $\mathbb{Q}$ again, we get $\mathbb{Q}^n \simeq \mathbb{Q}[x]^m$ as $\mathbb{Q}$-modules. At this point, I feel this is absurd, but can't quite explain why.

My questions: Is the above reasoning correct? How do I finish the argument? If not, what would be the right way to prove this?


share|cite|improve this question
Consider the action of $x$ on a set of generators of $M$ over $\mathbb{Q}$ and write them as linear combinations of the generators. Form the matrix equation, then the characteristic polynomial will be the what you want. – Ehsan M. Kermani Jul 30 '12 at 20:47
In your attempt: before the second sentence, “Since $M$ is also finitely generated as a $\mathbb{Q}[x]$-module,” For the finish, calculate the vector space dimension of the two sides of your $\simeq$ and conclude there are very few solutions (just one by my count). – Jack Schmidt Jul 30 '12 at 20:55
Is there anything special about $\Bbb{Q}$ in this argument? It seems that using $\Bbb{R}$ would work as well.... – user38268 Jul 31 '12 at 0:31
@JackSchmidt Dear Jack, thanks for pointing that out. Could you elaborate on your suggestion? I can see that $m = n$, but how do I show then that this is impossible? Can't, for instance, $\mathbb{Q}$ act in a weird way on $\mathbb{Q}[x]$ such that they are isomorphic vector spaces? – student Jul 31 '12 at 17:50
@student: The dimension of the right hand side may be confusing: $$\dim\left({\mathbb{Q}[x]}^m\right) = \begin{cases} 0 & \text{ if } m = 0 \\ \infty & \text{ if } m > 0 \end{cases}$$ – Jack Schmidt Jul 31 '12 at 17:53
up vote 6 down vote accepted

$m, xm, x^2m,...$ cannot all be linearly independent over $\mathbb{Q}$. hence for every $m\in M$ there is a $p_m(x)\in\mathbb{Q}[x]$ with $p_m(x)m=0$

share|cite|improve this answer

A variation on your attempt and other comments: the map $\mathbb Q[x]$ to $\mathbb Q$-linear endomorphisms of $M$ must have a non-trivial kernel, since $M$ is finite-dimensional as $\mathbb Q$-module and $\mathbb Q[x]$ is not. The kernel is thus a non-zero ideal in the PID $\mathbb Q[x]$, generated by the minimal polynomial of $x$ acting on $M$.

(Although it is interesting to understand Cayley-Hamilton and the characteristic polynomial, I think it is also useful to understand the minimal poly.)

Edit: In response to the comment, the map I'm thinking of just sends $x$ to the map $m\to x\cdot m$ for $m\in M$.

share|cite|improve this answer
Dear Paul, could you explain a little bit what is the map you're talking about? I can't see any natural map from $\mathbb{Q}[x]$ to $\mathbb{Q}$-linear endomorphisms of $M$, unless I'm given some fixed $T : M \rightarrow M$, and then interpret each $p(x) \in \mathbb{Q}[x]$ as $p(T)$ formally. – student Jul 31 '12 at 18:27

You can build upon what you already wrote to finish:

Although it's not stated explicitly, you must be requiring that $M \neq 0,$ otherwise it contains no non-zero elements $m$ at all.

But if $M \neq 0$ and $M = \mathbb Q[x]^m$, then $m \geq 1$, and so $M$ contains a coyp of $\mathbb Q[x]$. Now thinking, as you suggest, over $\mathbb Q$, you have an inclusion of $\mathbb Q$-vector spaces $\mathbb Q[x] \subset \mathbb Q^n$. Is this possible?

[This is a minor variation of the reasoning suggested by Jack Schmidt in his comments.]

Incidentally, it's not only true that $M$ is not torsion-free; in fact if $M$ is finite-dimensional over $\mathbb Q$ it will be torsion over $\mathbb Q[x]$. The same argument as above proves this (if $M$ is not torsion, it contains a copy of $\mathbb Q[x]$), as do the arguments in the other answers.

share|cite|improve this answer
Dear Matt, thanks for elaborating this point. I just have one question, if you allow me, which is the same objection to Jack's comments: in this reasoning, it seems we're assuming that $\mathbb{Q}$ "acts" on elements of $\mathbb{Q}[x]$ in the natural way; is this necessarily the case? Couldn't $\mathbb{Q}$ act on elements of $\mathbb{Q}[x]$ in some different way, thus enabling the dimension of $\mathbb{Q}[x]$ to be finite? – student Jul 31 '12 at 18:32
@student: Dear student, The embeding $\mathbb Q[x] \subset M$ is an inclusion of $\mathbb Q[x]$-modules, so you are giving your self that $\mathbb Q[x]$ acts on itself in the natural way (by multiplication of polynomials). In particular the $\mathbb Q$-action is given as the natural one. Regards, – Matt E Jul 31 '12 at 19:05

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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