You are here
Homefinitely generated modules over a principal ideal domain
Primary tabs
finitely generated modules over a principal ideal domain
Let $R$ be a principal ideal domain and let $M$ be a finitely generated $R$ module.
Lemma.
Let $M$ be a submodule of the $R$module $R^{n}$. Then $M$ is free and finitely generated by $s\leq n$ elements.
Proof.
For $n=1$ this is clear, since $M$ is an ideal of $R$ and is generated by some element $a\in R$. Now suppose that the statement is true for all submodules of $R^{m},1\leq m\leq n1$.
For a submodule $M$ of $R^{n}$ we define $f\colon M\to R$ by $(k_{1},\ldots,k_{n})\mapsto k_{1}$. The image of $f$ is an ideal $\mathfrak{I}$ in $R$. If $\mathfrak{I}=\{0\}$, then $M\subseteq\ker(f)=(0)\times R^{{n1}}$. Otherwise, $\mathfrak{I}=(g),g\neq 0$. In the first case, elements of $\ker(f)$ can be bijectively mapped to $R^{{n1}}$ by the function $\ker(f)\to R^{{n1}}$ given by $(0,k_{1},\ldots,k_{{n1}})\mapsto(k_{1},\ldots,k_{{n1}})$; so the image of $M$ under this mapping is a submodule of $R^{{n1}}$, which by the induction hypothesis is finitely generated and free.
Now let $x\in M$ such that $f(x)=gh$ and $y\in M$ with $f(y)=g$. Then $f(xhy)=f(x)f(hy)=0$, which is equivalent to $xhy\in\ker(f)\cap R^{n}:=N$ which is isomorphic to a submodule of $R^{{n1}}$. This shows that $Rx+N=M$.
Let $\{g_{1},\ldots,g_{s}\}$ be a basis of $N$. By assumption, $s\leq n1$. We’ll show that $\{x,g_{1},\ldots,g_{s}\}$ is linearly independent. So let $rx+\sum_{{i=1}}^{s}r_{i}g_{i}=0$. The first component of the $g_{i}$ are 0, so the first component of $rx$ must also be 0. Since $f(x)$ is a multiple of $g\neq 0$ and $0=r\cdot f(x)$, then $r=0$. Since $\{g_{1},\ldots,g_{s}\}$ are linearly independent, $\{x,g_{1},\ldots,g_{s}\}$ is a generating set of $M$ with $s+1\leq n$ elements. ∎
Corollary.
If $M$ is a finitely generated $R$module over a PID generated by $s$ elements and $N$ is a submodule of $M$, then $N$ can be generated by $s$ or fewer elements.
Proof.
Let $\{g_{1},\ldots,g_{s}\}$ be a generating set of $M$ and $f\colon R^{s}\to M$, $(r_{1},\ldots,r_{s})\mapsto\sum_{{i=1}}^{s}r_{i}g_{i}$. Then the inverse image $N^{{{}^{{\prime}}}}$ of $N$ is a submodule of $R^{s}$, and according to lemma finitely generated modules over a principal ideal domain can be generated by $s$ or fewer elements. Let $n_{1},\ldots,n_{t}$ be a generating set of $n^{{{}^{{\prime}}}}$; then $t\leq s$, and since $f$ is surjective, $f(n_{1}),\ldots,f(n_{t})$ is a generating set of $N$. ∎
Theorem.
Let $M$ be a finitely generated module over a principal ideal domain $R$.
 (I)

Note that $M/\operatorname{tor}(M)$ is torsionfree, that is, $\operatorname{tor}(M/\operatorname{tor}(M))=\{0\}$. In particular, if $M$ is torsionfree, then $M$ is free.
 (II)

Let $\operatorname{tor}(M)$ be a proper submodule of $M$. Then there exists a finitely generated free submodule $F$ of $M$ such that $M=F\oplus\operatorname{tor}(M)$.
Proof of (I): Let $T=\operatorname{tor}(M)$. For $m\in M$, $\overline{m}$ denotes the coset modulo $T$ generated by $m$. Let $m$ be a torsion element of $M/T$, so there exists $\alpha\in R\setminus\{0\}$ such that $\alpha\cdot\overline{m}=0$, which means $\alpha\cdot\overline{m}\subseteq T$. But then $\alpha\cdot m$ is a member of $T$, and this implies that $M/T$ has no nonzero torsion elements (which is obvious if $M=\operatorname{tor}(M)$).
Now let $M$ be a finitely generated torsionfree $R$module. Choose a maximal linearly independent subset $S$ of $M$, and let $F$ be the submodule of $M$ generated by $S$. Let $\{m_{1},\dots,m_{n}\}$ be a set of generators of $M$. For each $i=1,\dots,n$ there is a nonzero $r_{i}\in R$ such that $r_{i}\cdot m_{i}\in F$. Put $r=\prod_{{i=1}}^{n}r_{i}$. Then $r$ is nonzero, and we have $r\cdot m_{i}\in F$ for each $i=1,\dots,n$. As $M$ is torsionfree, the multiplication by $r$ is injective, so $M\cong r\cdot M\subseteq F$. So $M$ is isomorphic to a submodule of a free module, and is therefore free.
Proof of (II): Let $\pi\colon M\to M/T$ be defined by $a\mapsto a+T$. Then $\pi$ is surjective, so $m_{1},\ldots,m_{t}\in M$ can be chosen such that $\pi(m_{i})=n_{i}$, where the $n_{i}$’s are a basis of $M/T$. If $0_{M}=\sum_{{i=1}}^{t}a_{i}m_{i}$, then $0_{n}=\sum_{{i=1}}^{t}a_{i}n_{i}$. Since $n_{1},\ldots,n_{t}$ are linearly independent in $N$ it follows $0=a_{1}=\ldots=a_{t}$. So the submodule spanned by $m_{1},\ldots,m_{t}$ of $M$ is free.
Now let $m$ be some element of $M$ and $\pi(m)=\sum_{{i=1}}^{t}a_{i}n_{i}$. This is equivalent to $m\left(\sum_{{i=1}}^{t}a_{i}n_{i}\right)\in\ker(\pi)=T$. Hence, any $m\in M$ is a sum of the form $f+t$, for some $f\in F$ and $t\in T$. Since $F$ is torsionfree, $F\cap T=\{0\}$, and it follows that $M=F\oplus T$.
Mathematics Subject Classification
13E15 no label found Forums
 Planetary Bugs
 HS/Secondary
 University/Tertiary
 Graduate/Advanced
 Industry/Practice
 Research Topics
 LaTeX help
 Math Comptetitions
 Math History
 Math Humor
 PlanetMath Comments
 PlanetMath System Updates and News
 PlanetMath help
 PlanetMath.ORG
 Strategic Communications Development
 The Math Pub
 Testing messages (ignore)
 Other useful stuff
Recent Activity
new correction: Error in proof of Proposition 2 by alex2907
Jun 24
new question: A good question by Ron Castillo
Jun 23
new question: A trascendental number. by Ron Castillo
Jun 19
new question: Banach lattice valued Bochner integrals by math ias
Jun 13
new question: young tableau and young projectors by zmth
Jun 11
new question: binomial coefficients: is this a known relation? by pfb
Jun 6
new question: difference of a function and a finite sum by pfb
Comments
Addendum
The statement could be made more precise in the following manner:
If M is a finitely generated module over a PID R, then
M \cong R^n \oplus R/(q_1) \oplus ... \oplus R/(q_r)
where q_1  q_1  ... q_r, the ideals (q_i) are uniquely determined.
Here, R^n is the free module F which is the torsion free part of M. Therefore, n is uniquely determined too. The other factors represent a direct sum decomposition of M_{tor} (the torsion submodule of M).