[GAP Forum] A permutation group with huge finite orbits
Stefan Kohl
stefan at mcs.st-and.ac.uk
Mon Dec 10 13:44:14 GMT 2012
Dear Jason, dear Forum,
> The length of the orbit is 47610700792.
>
> Using C with GMP and OpenMP, the result can be found in roughly 3 days on a
> new core i7 machine. It is possible to compute the orbit in both directions
> at the same time on separate cores, stopping when they meet somewhere in
> the orbit. The maximum length of a digit in the orbit is 76,785.
>
> You can find the C code here: http://www.jasonbhill.com/residue-collatz/
Thanks. -- Great!!
Just two remarks:
1. In your C program you apply the mappings a, b and c separately.
It is possible to reduce the number of arithmetic operations by
evaluating the product g := abc and applying it as one mapping.
We have
gap> LoadPackage("rcwa");
gap> a := ClassTransposition(0,2,1,2);;
gap> b := ClassTransposition(0,5,4,5);;
gap> c := ClassTransposition(1,4,0,6);;
gap> g := a*b*c;;
gap> Display(g);
Rcwa permutation of Z with modulus 60
/
| n-1 if n in 3(30) U 9(30) U 17(30) U 23(30) U 27(30) U 29(30)
| 3n/2 if n in 0(20) U 12(20) U 16(20)
| n+1 if n in 2(20) U 6(20) U 10(20)
| (2n+1)/3 if n in 7(30) U 13(30) U 19(30)
| n+3 if n in 1(30) U 11(30)
| n-5 if n in 15(30) U 25(30)
n |-> < (3n+12)/2 if n in 4(20)
| (3n-12)/2 if n in 8(20)
| n+5 if n in 14(20)
| n-3 if n in 18(20)
| (2n-7)/3 if n in 5(30)
| (2n+9)/3 if n in 21(30)
|
\
2. There does not seem to be an obvious reason why the cycles of abc always
traverse entire orbits of G = <a,b,c>, and at least on a first glance
one doesn't see a pattern in which order a cycle of abc traverses an orbit
of G. Nevertheless, computational evidence supports so far the assumption
that cycles of abc correspond to orbits of <a,b,c>.
Thanks again and
Best regards,
Stefan
-----------------------------------------------------------------------------
http://www.gap-system.org/DevelopersPages/StefanKohl/
-----------------------------------------------------------------------------
More information about the Forum
mailing list