We aim to "fix" the following two issues: GL and SL are not 2-transitive because they can't linearly dependent vectors to linearly independent ones. SL is not simple (even though it's perfect) because it has a center. Let V=Vn(q) and n≥2 throughout.
We define an equivalence relation R on V# by vRw iff v=λw for some nonzero λ∈Fq.
Definition We then have the projective space P(V) of projective vectors. Write Pn−1(q).
For a subspace U⊆V the set of equivalence classes (projective vectors) [U] (the image of U#) is a subspace of P(V) so it inherits geometric structure. The dimension of [U] is the dimension of U minus 1. A point is a class [v] for some vector v, and a line is the projective class of a 2D subspace.
Given g∈GL(V), v∈V# and λ∈F#q we have (λv)g=λ(vg)∈[vg] so we can (well) define an action by [v]g=[vg]. In this way GL(V) and SL(V) act on P(V), but not faithfully.
Lemma Let G be GL(V) or SL(V), the kernel of the action of G on P(V) is Z(G).
proof: From the previous post we have seen what Z(G) is: scalar multiples of the identity. If gs=1 then clearly g acts trivially on P(V). Conversely if g∈GL(V) is in the kernel of the action then [vg]=[v] for all [v]∈P(V), then for every vector V we have vg=λvv for some λv∈F# and the proof concludes in the same way as before.
Definition The projective general linear group and projective special linear group are defined by PGL(V)=GL(V)/Z(GL(V)) and PSL(V)=SL(V)/Z(SL(V)). They have faithful action on P(V). (Note: PSL might not be a subgroup of PGL anymore).
We call [g]∈PGL or PSL the image (???) if g∈GL or SL if [v][g]=[vg] for v∈P(V). PGLn(q) and PSLn(q) act on Pn−1(q) by the order calculations in the previous post we find |PGLn(q)|=qn(n−1)/2∏ni=2(qi−1) and |PSLn(q)|=qn(n−1)/2(n,q−1)∏ni=2(qi−1). Thus |PGL2(q)|=(q+1)q(q−1).
Proposition The permutation group PGL2(q) is sharply 3-transitive on P1(q).
proof: For g=(abcd)∈GL2(q) and v∈V2(q)#, vg=(aλ1+cλ2,bλ1+dλ2). These work out exactly as the mobius transformations when regarding P1(q) as Fq∪{∞}. We have already shown this one point extension is genrated by the mobius transforms!
Proposition Both PGL(V) and PSL(V) act 2-transitively on P(V).
proof: Given ([e1],[e2]), ([e′1],[e′2]) distinct then extend to a basis and find a map between them (why doesn't this work for all n-transitivity?)
No comments:
Post a Comment