Group extension problem: Difference between revisions
| Line 33: | Line 33: | ||
The set of group extensions upto automorphisms of <math>N</math> and <math>H</math> is a quotient of the set of all group extensions, by an equivalence relation. | The set of group extensions upto automorphisms of <math>N</math> and <math>H</math> is a quotient of the set of all group extensions, by an equivalence relation. | ||
==Classifying group extensions for an | ==Classifying group extensions for an abelian normal subgroup== | ||
If <math>N</math> is an [[abelian group]], then there is a procedure to classify all group extensions with normal subgroup <math>N</math>. | If <math>N</math> is an [[abelian group]], then there is a procedure to classify all group extensions with normal subgroup <math>N</math>. | ||
Revision as of 18:20, 23 December 2012
Statement
The group extension problem for two groups and , is the problem of finding all groups with as a normal subgroup of , and the quotient group isomorphic to .
Congruence classes formulation
In this formulation, we're thinking of and as specific groups, and looking at short exact sequences:
where two short exact sequences:
and:
are equivalent if there is an isomorphism from to that induces the identity map on and on .
Formulation upto automorphisms
This is a more general formulation, where we declare two short exact sequences:
and:
are equivalent if there is an isomorphism from to that induces an automorphism on and an automorphism on .
The set of group extensions upto automorphisms of and is a quotient of the set of all group extensions, by an equivalence relation.
Classifying group extensions for an abelian normal subgroup
If is an abelian group, then there is a procedure to classify all group extensions with normal subgroup .
Finding a list of possible actions
The first step is to note that the quotient group acts on the normal subgroup. In other words, given any group with a specified normal subgroup and a quotient group , there is a homomorphism:
This automorphism is determined by the congruence class of the extensions, so we can partition all congruence classes of extensions as a disjoint union of congruence classes of extensions corresponding to elements of .
If we're looking at extensions modulo equivalence upto automorphisms, then an extension doesn't define a unique map . Rather, the map is defined uniquely up to pre-composition with automorphisms of and conjugation by automorphisms of . Thus, when classifying equivalence classes of extensions in this fashion, it suffices to consider equivalence classes of elements in .
Finding all congruence classes for a given action
Having fixed a homomorphism:
the next step is to find all congruence classes of extensions that give rise to this homomorphism. This is, in fact, described using the second cohomology group . There are a number of shortcuts to computing this group when or have special structure (for instance second cohomology group for trivial group action of finite cyclic group on finite cyclic group).
Multiple congruence classes of extensions may again be equivalent upto automorphisms, so the set of equivalence classes of extensions is not necessarily a group.