# Name for a set of coset representatives which contains a transversal

A transversal for a quotient group $G/N$ is a subset of $G$ which contains precisely one element for each coset of $G/N$. This can be defined analogously for any collection of sets and not just cosets (e.g. $N$ need not be normal, so "left-transversal" and so on), but lets just talk about quotient groups because this is a terminology question and terminology often change between subject areas.

My question is: I have a subset $S$ of $G$ which contains a transversal of $G/N$. Is there any nice, snappy name (in the same vein as "transversal") I can use for this property of $S$: the set $S$ is a XXX for $G/N$?

"$S$ is a set of coset representatives" is not precise enough - it allows the possibility of cosets without a representative in $S$. "$S$ is a subset of $G$ which contains a transversal for $G/N$" or even "$S$ is a subset of $G$ which contains an element from each coset" are not snappy.

• I would probably say $S$ is a subset of $G$ that projects onto $G/N$, although it's not particularly snappy! – Derek Holt Oct 13 '15 at 21:25

Let $U \le G$ and $S \subseteq G$. Then $S$ is a right transversal of $U$ in $G$ if $S$ contains exactly one element from every right cosets $Ux$, $x \in G$; and $S$ is a left transversal of $U$ in $G$ if $S$ contains exactly one element of every left coset of $U$ in $G$.
Now you could define a transversal to be a set which is simultanously a left and a right transversal. In a finite group for each subgroup we can find such an transversal (which is a consequence of Hall's marriage theorem), but what makes normal subgroups special here is that they have the property that every left transversal is a right transversal (finite or not). So you could safely just say transversal in the normal case, i.e. let $S$ be a transversal for $N$ in $G$. But if you like to be more specific you could call it "normal transversal" or "transversal for normal subgroup".
• When I said "quotient group" in the first line, I was implicitly assuming $N$ was normal in $G$ (else it wouldn't be a quotient group). – user1729 Oct 13 '15 at 16:04
• Yes, of course. That was clear to me, but by the way this wording sounds very unusual to me, if is more common to speak of a transversals of some subgroup (normal or not), but when you speak of the quotient group someone might get the impression you are looking for transversals of subgroups of $G/N$. – StefanH Oct 13 '15 at 16:08