12

Can someone explain how the Schreier-Sims Algorithms works on a permutation group with a simple example? All the books I read have a dense notation that hard to comprehend but a simple and concrete example would help greatly!

Olexandr Konovalov
  • 7,186
  • 2
  • 35
  • 73

1 Answers1

15

The problem is that working through this algorithm by hand is very tedious and repetitive.

But, let's do $G = \langle x,y \rangle \le S_4$ with $x=(1,2,3)$, $y=(1,2,4)$.

Start with a single base point $1$, and (strong) generating set $S = [x,y]$.

The orbit of $1$ is $O_1=[1,2,3,4]$ with permutations $T_1=[1,x,x^{-1},xy]$ mapping $1$ to the orbit points.

Now we consider the Schreier generators of the stabilizer of $1$. We consider each product $ta$ in turn with $t \in T_1$ and $a \in S$, and then multiply it by the inverse of the appropriate element of $T$ to make it fix $1$.

$1xx^{-1} = 1$

$1yx^{-1} = (2,3,4)$.

This is not the identity, so we interrupt the loop through the $ta$, and adjoin a new base point $2$, and a new strong generator $z=(2,3,4)$.

So now $B=[1,2]$ and $S=[x,y,z]$ with $z$ fixing base point $1$.

Now we work with base point $2$ in the stabilizer of $1$, using strong generators $S_2=[z]$. The orbit of $2$ is $O_2=[2,3,4]$ with transversal $T_2=[1,z,z^{-1}]$. We consider the products $ta$ with $t \in T_2$ and $a \in S_2$ and get

$1zz^{-1}=1$, $zzz = 1$, $z^2z=1$. So we get the identity each time and we have verified that $S_2$ is a strong generating set for the subgroup of the stabilizer of $1$ generated by $S_2$.

Now we go back to base point $1$ and recheck the products $ta$ with $t \in T_1$, $a \in S$. I won't work through all of these, but let's just look at the one that failed last tiem, i.e. $t=1$, $a=y$. Again we multiply by $x^{-1}$ giving $1yx^{-1} = (2,3,4)$ which fixes $1$. But now we have the new base point $2$ and we have strong generator $z$ mapping $2$ to $3$. So we adjust the product to $1yx^{-1}z^{-1} = 1$.

You can verify that all $8$ products $ta$ reduce to the identity on mulitplying by inverses of elements of $T_1$ and $T_2$. So the algorithm stops, and we now know that the order of the group is the product of the basic orbit lengths i.e. $4 \times 3 = 12$.

Derek Holt
  • 96,726
  • "The orbit of 1 is $O_1$=[1,2,3,4] with permutations $T_1$=[1,x,x−1,xy] mapping 1 to the orbit points." I am confused here a bit because I am not sure how permutations $T_1$ map 1 to its orbits? I can see that orbit(1) = [1,2,3,4] because in x 1->2 and 1->3 and in y 1->2 and 1->4 but not sure how $T_1$ maps 1 to its orbits and not sure how you came up with these elements 1,x,x−1,xy for $T_1$ ? – user1870400 Feb 22 '16 at 12:09
  • $1$ maps $1$ to $1$, $x$ maps $1$ to $2$, $x^{-1}$ maps $1$ to $3$ and $xy$ maps $1$ to $4$. Sorry, but I don't think I can help you much more with this. – Derek Holt Feb 22 '16 at 12:15
  • Thanks a lot for all your responses! – user1870400 Feb 22 '16 at 12:21
  • In case this is causing confusion, I should have pointed out earlier that in all of the books that cover this algorithm the convention is that permutations are composed from left to right, and I am also using that convention. So $x$ maps $1$ to $2$ and $y$ maps $2$ to $4$, so $xy$ maps $1$ to $4$. – Derek Holt Feb 22 '16 at 12:40
  • Lot of books that I came across actually do it from right to left like a typical function composition in math. The ones that do from left to right I came across were Donald Knuth books! I still need to seek some help elsewhere to dissect this Algorithm! Since I already asked you too many questions I just want to say thank you so much for all your responses! I really feel lucky being able to get answers from someone like yourself given your background! – user1870400 Feb 22 '16 at 13:02
  • Thank you very much for this. I think z = (2,4,3) rather than (2,3,4) although this makes no material difference. – niemiro Mar 22 '19 at 19:04