When will a countable group be the inductive limit of its finite subgroups

abstract-algebracategory-theorygroup-theorylimits

It is proved in this post that every group is the inductive limit of the family of its finitely generated subgroups (where the partial order is the inclusion). Indeed, given $G_i, G_j$ two finitely generated subgroups of $G$, the group generated by generators of $G_i, G_j$ is another finitely generated subgroup that contains $G_i, G_j$. Definitions of the inductive limit of groups can be found in the same post.

My question is: given $G$ a countably infinite group, when can $G$ be written as an inductive limit of finite subgroups; or equivalently, when can the inclusion, as a partial order, is directed in the family of finite subgroups? In general, given a countably infinite family of finite groups $\{G_i\}_{i\in\mathbb{N}}$, for each $N\in\mathbb{N}$, define:

$$
H_N = \prod_{1\leq i \leq N}G_i
$$

and, $\phi_{N, N+1}$, the connecting group homomorphism as follows:

$$
\phi_{N, N+1}: H_N\rightarrow H_{N+1}, \hspace{0.5cm} (g_1, \cdots, g_N) \mapsto (g_1, \cdots, g_N, e_{N+1})
$$

where $e_n$ denotes the identity of $G_n$ for each $n\in\mathbb{N}$. Then define:

$$
H = \prod_{i\in\mathbb{N}} G_i
$$

and:

$$
\Phi_N: H_N\rightarrow H, \hspace{0.5cm}(g_1, \cdots, g_N)\mapsto (g_1, \cdots, g_N, e_{N+1}, e_{N+2}, \cdots)
$$

Then $(H, \{\Phi_i\}_{i\in\mathbb{N}})$ is an inductive limit of finite subgroups where the group operation is defined coordinate-wise. I believe this is one way to build the desired group, or one situation when the inclusion is a partial order in the family of finite subgroups. Any concrete examples, or ideally a necessary condition for the statement to be true will be highly appreciated.

Best Answer

Inclusion is always a partial order of the family of finite subgroups. I think you mean to ask when this partial order is directed and covers $G$ (i.e. every element is contained in some finite subgroup), so that $G$ is a directed colimit (aka direct limit, aka inductive limit) of its finite subgroups.

The answer is: exactly when $G$ is a locally finite group, i.e. when every finitely generated subgroup is finite. Countability of $G$ is not relevant here.

If $G$ is locally finite, then since you already know that $G$ is a directed colimit of its finitely generated subgroups, and each such subgroup is actually finite, then $G$ is a directed colimit of its finite subgroups.

Conversely, suppose $G$ is a directed colimit of its finite subgroups. Let $g_1,\dots,g_n$ be finitely many elements of $G$. Each $g_i$ is in some finite subgroup $G_i$, and by directeness there is a finite subgroup $G_*$ containing all of the $G_i$. Now the subgroup of $G$ generated by $g_1,\dots,g_n$ is contained in $G_*$ and hence is finite. Thus every finitely generated subgroup of $G$ is finite, and $G$ is locally finite.

The wikipedia article on locally finite groups that I linked to above contains lots of examples, as well as closure properties of the class of locally finite groups, which you can use to generate many more examples.

Related Question