Consider actions of rank 2 (meaning that there are two suborbits of Gα), Ω∖{α} forms a single Gα orbit (because one its other orbit is αGα={α}). G acts on Ω2 non-transitively because g(β,β)=(γ,γ) but if we define the diagonal Delta={(β,β)∈Ω} this is a single orbit due to transitivity and so Ω2∖Δ is the interesting part:
Lemma The action of G on Ω is of rank 2 iff Ω2∖Δ is a single orbit.
proof: If the action has rank 2 then let (β1,β2),(γ1,γ2) lie off the diagonal and pick x,y∈G such that αx=β1, αy=γ1 then neither of β2x−1 and γy−1 are equal to α (otherwise β1=β2 or γ1=γ2) so there exists h∈Gα which maps one to the other β2x−1h=γ2y−1. Let g=x−1hy and compute (β1,β2)g=(γ1,γ2).
In the other direction if Ω2∖Δ is a single orbit the action is at least 2 (since we can fix any one element β in the first component and map any other element γ not equal to beta to any other element not equal to beta). So suppose the rank were larger than 2, then pick β,γ in different Gα orbits of Ω∖{α} and take (α,β),(α,γ)Ω2∖Δ there's clearly no way to map from one to the other.
We can generalize this to Ωt for any natural t≤|Ω|. Again Δ={(α,α,…)} is a single orbit, but the interesting part is Ω(t)={(α1,α2,…)|αi≠αj}.
Definition The action of G on Ω is t-transitive if the induced action on Ω(t) is transitive. This is equivalent to saying it can simultaneously map any t distinct points to any other t distinct points.
Lemma In terms of cosets, a group action is 2-transitive iff G=Gα∪GαgGα for any g∖Gα.
proof: We saw previous that double cosets correspond to suborbits, write Gα=Gα1Gα to see this is the same as rank 2.
Lemma The action of G on Ω is t-transitive iff the action of Gα on Ω∖{α} is (t−1)-transitive.
proof: write this out.
Corollary If G acts t-transitively on Ω and |Ω|=n then |G| is divisible by n(n−1)⋯(n−t+1). WHICH THEOREM DOES THIS DEPEND ON? CHECK OTHER BOOK
Theorem In the natural action Sn is n-transitive while An is n−2 transitive and not n−1 transitive.
proof: This is obvious from the fact Sn contains every permutation. For An we use induction: it clearly holds for A3. For n≥3 the stabilizer of any point of An is An−1 so by the lemma we complete the induction.
No comments:
Post a Comment