Skip to main content

Section 7.3 Finitely Generated Abelian Groups

“The classification of the constituents of a chaos, nothing less is here essayed.”
―Herman Melville
In this section we see that we can classify finitely generated abelian groups into isomorphism classes.

Proof.

Using Proposition 2.27, we let \(\psi: \mathbb{Z}\to \mathbb{Z}/(p_1^{e_1}) \times \cdots \times \mathbb{Z}/(p_l^{e_l})\) be the unique homomorhism that sends \(1\) to \(([1]_{{p_1}^{e_1}}, \cdots, [1]_{{p_l}^{e_l}})\text{.}\) Then
\begin{equation*} \psi(j) = ([j]_{{p_1}^{e_1}}, \cdots, [j]_{{p_l}^{e_l}}). \end{equation*}
We see \(m \in \operatorname{ker}(\psi)\) and so \(\langle m \rangle \subseteq \operatorname{Ker}(\psi)\text{.}\) Conversely, if \(\psi(n) = 0\text{,}\) then \(p_i^{e_i} \mid n\) for all \(i\) and since \(p_1^{e_1}, \dots, p_l^{e_l}\) are pairwise relatively prime, it follows that \(m \mid n\text{.}\) This proves \(\operatorname{ker}(\psi) = \langle m \rangle\text{.}\) The claim follows by the First Isomorphism Theorem.

Remark 7.25.

Sunzi’s Remainder Theorem frequently goes by the name of The Chinese Remainder Theorem, but it has been remarked that this is somewhat like referencing the Pythagorean Theorem as The Greek Triangle Theorem or Fermat’s Little Theorem as The French Power Postulate. In the interest of giving credit where credit is due, we include it here with the name of the 3rd-century Chinese mathematician, Sunzi, to whom the result is often attributed.
The following is the classification theorem for finitely generated abelian groups. We present it without proving it for now. The full proof will be given in the spring semester.

Proof.

It suffices prove that for a given group \(G\text{,}\) we can recover its invariant factor form from its elementary divisor form, and vice versa. We will be a bit hand-wavey for this following the ideas from the above examples.
\begin{equation*} G \cong \mathbb{Z}^r \times \mathbb{Z}/p_1^{e_1} \times \cdots \times \mathbb{Z}/p_l^{e_l}. \end{equation*}
by applying Sunzi’s Remainder Theorem we have
\begin{equation*} G \cong \mathbb{Z}^r \times \mathbb{Z}/d_1 \times \cdots \times \mathbb{Z}/d_n \end{equation*}
where \(d_1\) is the product of the elementary divisors of highest power for each distinct prime in the list \(p_1, \dots, p_l\text{,}\) \(d_2\) is the product of the next highest possible prime powers, and so on. We will have that \(d_2\mid d_1\) and in general that \(d_{i+1}\mid d_i\) since by definition the exponent of \(p_j\in d_i\) is greater or equal to the exponent of \(p_j\in d_{i+1}\text{.}\)
Conversely, given
\begin{equation*} G \cong \mathbb{Z}^r \times \mathbb{Z}/d_1 \times \cdots \times \mathbb{Z}/d_n \end{equation*}
with \(d_1 \mid d_2 \mid \cdots \mid d_n\text{,}\) we may apply Sunzi’s Remainder Theorem to each \(\mathbb{Z}/d_i\) to find its elementary divisor form.

Example 7.27. FTFGAG in Action.

For \(G\cong\mathbb{Z}/3\times \Z/5\times \Z/5\) we have \(Q_1=\mathbb{Z}/3\text{,}\) \(Q_2=\mathbb{Z}/5\times \Z/5\text{.}\)

Definition 7.28. Rank, Invariant Factors, Elementary Divisors.

In Theorem 7.26, the number \(r\) is the rank of \(G\text{,}\) the \(p_i^{a_{i,k}}\) are the elementary divisors of \(G\text{,}\) and the decomposition of \(G\) in parts (1-2) is called the elementary divisor decomposition of \(G\text{.}\) The decomposition in part (1) is also called a primary decomposition.
In Theorem 7.26, the number \(r\) is the rank of \(G\text{,}\) the numbers \(n_1,\ldots,n_t\) are the invariant factors of \(G\text{,}\) and the decomposition of \(G\) in part (1) is the invariant factor decomposition of \(G\text{.}\)

Example 7.29. Elementary Divisor Form to Invariant Factor Form.

Say I tell you
\begin{equation*} G \cong \mathbb{Z}^{3} \times \mathbb{Z}/4 \times \mathbb{Z}/8 \times \mathbb{Z}/9 \times \mathbb{Z}/27 \times \mathbb{Z}/25. \end{equation*}
\begin{equation*} \mathbb{Z}/8 \times \mathbb{Z}/27 \times \mathbb{Z}/25 \cong \mathbb{Z}/(8 \cdot 27 \cdot 25) \end{equation*}
and
\begin{equation*} \mathbb{Z}/4\times \mathbb{Z}/9 \cong \mathbb{Z}/(4 \cdot 9) \end{equation*}
so that
\begin{equation*} G \cong \mathbb{Z}^{3} \times \mathbb{Z}/(4 \cdot 9) \times \mathbb{Z}/(8 \cdot 27 \cdot 25). \end{equation*}
Since \(n_2:=(4 \cdot 9) \mid n_1:=(8 \cdot 27 \cdot 25)\text{,}\) this is in invariant factor form, and hence the rank of \(A\) is \(3\) and the invariant factors of \(A\) are \(4 \cdot 9\) and \(8 \cdot 27 \cdot 25\text{.}\)

Example 7.30. Invariant Factor Form to Elementary Divisor Form.

Suppose now I tell you
\begin{equation*} G \cong \mathbb{Z}^{4} \times \mathbb{Z}/6 \times \mathbb{Z}/36 \times \mathbb{Z}/180. \end{equation*}
\begin{equation*} G \cong \mathbb{Z}^{4} \times \mathbb{Z}/2 \times \mathbb{Z}/3 \times \mathbb{Z}/4 \times \mathbb{Z}/9 \times \mathbb{Z}/4 \times \mathbb{Z}/5 \times \mathbb{Z}/9, \end{equation*}
given the elementary divisor form.
The Fundamental Theorem of Finitely Generated Abelian Groups (FTFGAG) makes classification of finite abelian groups a very quick matter.

Example 7.31. Abelian Groups of Order \(75\).

We classify the abelian groups of order \(75\) up to isomorphism.
Let \(G\) be an abelian group of order \(75\text{.}\) Since \(G\) is finite the rank of \(G\) is \(r=0\text{.}\) Let’s determine the possible elementary divisors \(p_i^{a_{i,j}}\) so that
\begin{equation*} G\cong \prod_{k=1}^s \prod_{j=1}^{i_k} \mathbb{Z}/p_i^{a_{k,j}}. \end{equation*}
The above equation gives \(75=|G|= \prod_{k=1}^s\prod_{j=1}^{i_k}p_i^{a_{k,j}}\) and the possibilities for factoring \(75\) as a product of prime powers are \(75=3\cdot 5 \cdot 5\) or \(75=3\cdot 25\) which gives
\begin{equation*} G\cong \mathbb{Z}/25 \times \mathbb{Z}/3 \text{ or } G\cong \mathbb{Z}/5 \times \mathbb{Z}/5 \times \mathbb{Z}/3. \end{equation*}
Note that the two groups above are not isomorphic. To see this, note that there is an element of order \(25\) in \(\mathbb{Z}/25 \times \mathbb{Z}/3\text{,}\) namely \(([1]_{25},[0]_3)\) whereas every element \((a,b,c)\in \mathbb{Z}/5 \times \mathbb{Z}/5 \times \mathbb{Z}/3\) has order
\begin{equation*} |(a,b,c)=\operatorname{lcm}(|a|, |b|, |c|)\leq 15 \end{equation*}
since \(|a|, |b|\in\{1,5\}\) and \(|c|\in\{1,3\}\text{.}\)
Alternatively we could argue that the uniqueness of the Fundamental Theorem of Finitely Generated Abelian Groups (FTFGAG) tells us that \(G\) uniquely determines the elementary divisors, so two groups with distinct elementary divisors cannot be isomorphic.