1

Let $\Bbb F$ be a field and $M$ a finitely generated $\Bbb F[x]$-module. The structure theorem for modules over a PID says that $$ M\cong \Bbb F[x]^r\oplus\biggl(\bigoplus_{j=0}^s\Bbb F[x]/(f_j(x))\biggr)\text, $$ where $r$ and $s$ are non-negative integers and $f_j(x)$ are polynomials in $\Bbb F[x]$.

Now, because the ring $\Bbb F[x]$ is graded by the standard grading $\Bbb F[x]=\sum_{i=0}^\infty x^i\Bbb F$, according to this paper (Theorem 4.8.), $M$ can be written also in the form $$ M\cong\biggl(\bigoplus_{i=1}^M(x^{a_i})\biggr)\oplus\biggl(\bigoplus_{j=1}^N(x^{b_j})/(x^{c_j})\biggr) $$ for some non-negative integers $M,N$, $a_i,b_j,c_j$.

I can't understand this sentence from the proof:

The free component is composed of graded rings of the form $\bigoplus_{i\geq q}x^i\Bbb F$ , which are isomorphic to ideals of the form $(x^q)$.

The free part is $\Bbb F[x]^r$: How can I write this as a direct sum of things of the form $\bigoplus_{i\geq q}x^i\Bbb F$?

User
  • 1,173

1 Answers1

1

The free part is a direct sum of shifted copies of $\mathbb F[X]$. That is, $\mathbb F[X](-k)$ for some $k\ge 0$. This is nothing but $X^k\mathbb F[X]$.

user26857
  • 52,094
  • What do you mean exactly by $\Bbb Fx$? I can't understand how the sum could be direct if the summands overlap – User Feb 02 '16 at 16:56
  • Notation is explained e.g. here. – user26857 Feb 02 '16 at 16:57
  • Btw, you got a long explanation of the shifting phenomenon in a former question: http://math.stackexchange.com/questions/1610978/decomposition-of-a-graded-module – user26857 Feb 02 '16 at 17:00
  • Now I understand the notation $\Bbb Fx$ which is the same as $\Sigma^k\Bbb F[x]$ of my other question, but I still can't understand the direct sum. If the free part has rank 1, then it is $\Bbb F[x]$ which is the same as $\Bbb Fx$. But if, for example, the free part has rank 2, $\Bbb F[x]\oplus \Bbb F[x]$, how can I write it in terms of $\Bbb Fx$'s? What are the values of $k$? – User Feb 02 '16 at 17:22
  • The shifts appear as follows: if $M$ is free of rank one and the homogeneous generator is $z$ with $\deg z=d>0$, then $M\simeq \mathbb FX$. Why? Because the (iso)morphisms of graded modules are usually graded, that is, send $n$th degree part into $n$th degree part. In the example above, $1\mapsto e$ is not a graded morphism since $\deg 1=0$ and $\deg z=d$. Then you have to make $\deg 1=d$ by moving it $d$ places to the right. This means to consider $\mathbb FX$ which has $\mathbb FXn=\mathbb F[X]{n-d}$, that is, – user26857 Feb 02 '16 at 17:31
  • $\mathbb FX=0\oplus\cdots\oplus 0\oplus F\oplus FX\oplus\cdots$ ($0$ appears $d$ times). – user26857 Feb 02 '16 at 17:34
  • In the above comment should be $1\mapsto z$ instead of $1\mapsto e$. – user26857 Feb 02 '16 at 17:41
  • Consider $M\cong \Bbb F[x]^r$, where $M$ is a graded module, and call $\psi:\Bbb F[x]^r\to M$ the isomorphism (a normal isomorphism, not a graded one). Let's call $m_i:=\psi(e_i)$, where $e_i$ are the standard basis elements of $\Bbb F[x]^r$. From what I can understand, in order to grade $\Bbb F[x]^r$ and write it as a sum of shifted copies of $\Bbb F[x]$ one have to assign to each $e_i$ the degree of the corresponding element $m_i$. But how can I know each $m_i$ is homogeneous? – User Feb 02 '16 at 18:15
  • In fact you can always choose a homogeneous basis for $M$. This is proved here in a more general setting. (As you can see, everything you need is on this site. Just need to search for this a little.) – user26857 Feb 02 '16 at 21:38
  • Thank you! Now I think I understand the decomposition. When it is possible I will bounty your answer, for the effort you put in answering all my comments – User Feb 03 '16 at 10:22