2
$\begingroup$

I have studied direct products. I know a few applications of direct products, like group isomorphism, etc. What are some applications of sub-direct product of groups?

$\endgroup$
1
  • $\begingroup$ Exercise: a group is isomorphic to a (nontrivial) subdirect product of two groups iff it has two nontrivial normal subgroups with trivial intersection. (By nontrivial subdirect, I mean none of the two projections is injective on the subgroup.) $\endgroup$
    – YCor
    Commented Jan 24, 2019 at 3:46

3 Answers 3

2
$\begingroup$

Subdirect products arise naturally as intransitive subgroups of $S_n$, where they are subdirect products of the induced actions of the group on its orbits. Similarly for completely reducible linear groups.

$\endgroup$
2
  • $\begingroup$ I was interested in the application in computer science. $\endgroup$ Commented Jan 23, 2019 at 5:41
  • $\begingroup$ Also, centralizers in the alternating group (e.g., of a 3-cycle) are usually naturally subdirect (fibre) products. $\endgroup$
    – YCor
    Commented Jan 24, 2019 at 3:47
2
$\begingroup$

A powerful tool in group theory is something called the fibre product. This is a special type of subdirect product of the group with itself.

Associated to a short exact sequence $1\rightarrow N\rightarrow H\rightarrow Q\rightarrow 1$, where $\pi(H)=Q$ is the natural map, is the fibre product $P\subset H\times H$, $$ P:=\{(h_1, h_2)\mid \pi(h_1)=\pi(h_2)\}. $$

Clearly the diagonal component $\Delta:=\{(h, h)\mid h\in H\}$ is contained in $P$, so $P$ is a subdirect product.

The first application of fibre products which is know of is that there exist finitely generated subgroups of $F_2\times F_2$ which have insoluble membership problem: Firstly, let $Q$ be finitely presented and have insoluble word problem, then $P$ has insoluble membership problem. To see that $P$ is finitely generated, since $Q$ is finitely presented we have that $N \subset H$ is finitely generated as a normal subgroup, and then to obtain a finite generating set for $P$ one chooses a finite normal generating set for $N\times\{1\}$ and then appends a generating set for the diagonal $\Delta\cong H=F_2$.

A surprisingly powerful result, called the $1$-$2$-$3$ theorem, gives conditions on the fibre products to be finitely presentable, see: G. Baumslag, M.R. Bridson, C.F. Miller III, H. Short, Fibre products, non-positive curvature, and decision problems, Comm. Math. Helv. 75 (2000), 457–477.

A super-powerful application of fibre products is the following (the application is the main result of a paper in the Annals of Mathematics, one of the top journals - undisputed top 4 journal, disputed no. 1). If $\Gamma$ is a group then $\widehat{\Gamma}$ denotes its profinite completion.

Question (Grothendieck, 1970). Let $\Gamma_1$ and $\Gamma_2$ be finitely presented, residually finite groups and let $u :\Gamma_1\rightarrow \Gamma_2$ be a homomorphism such that $\widehat{u} :\widehat{\Gamma}_1\rightarrow \widehat{\Gamma}_2$ is an isomorphism of profinite groups. Does it follow that $u$ is an isomorphism from $\Gamma_1$ onto $\Gamma_2$?

Theorem (Bridson, Grunewald, 2004, Ann. Math., link). There exists a short exact sequence $1\rightarrow N\rightarrow\Gamma\rightarrow Q\rightarrow 1$ where $\Gamma$ has a whole host of nice properties, and where $P$ and $\Gamma$ are a counter-example to Grothendieck's question: $P$ and $\Gamma$ are finitely presentable, with $\widehat{P}\cong \widehat{\Gamma}$ but $P\not\cong \Gamma$.

$\endgroup$
4
  • $\begingroup$ @Shaun regarding your numerals edit, you might be interested in this tex.SE question. $\endgroup$
    – user1729
    Commented Jan 22, 2019 at 19:10
  • $\begingroup$ Thank you, @user1729; that'll save me some time in future! It's nice to know a useful convention :) $\endgroup$
    – Shaun
    Commented Jan 22, 2019 at 19:45
  • $\begingroup$ "A powerful tool in group theory, developed over the last 20 or so years but has a longer history". This sounds weird. I'd certainly not say that the tool of fibre products was developed in this period. Also if I prove a result about direct products I'm not developing the tool of direct product. $\endgroup$
    – YCor
    Commented Jan 24, 2019 at 3:43
  • $\begingroup$ @YCor I'll remove the statement with the dates. However, I can only think of three results pre-Baumslag, Bridson, Miller, Short which use fibre products (but I think they invented the name?). So do you think it would be correct to say that they "came to the fore" in the past 20 years? $\endgroup$
    – user1729
    Commented Jan 24, 2019 at 9:41
0
$\begingroup$

Typically, when you know that an object is a subdirect product of other objects, you know that it inherits all properties that are passed from products of those objects to their subsets. For example, a subdirect product of abelian groups must be abelian (and the same for any other group identity).

$\endgroup$
1
  • 1
    $\begingroup$ This seems to just be a property of a "subgroup of the direct product", which is a weaker property than being a subdirect product. $\endgroup$
    – Ted
    Commented Jan 22, 2019 at 17:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .