A058890

[Brendan McKay]

* Gordon Royle <gordon at cs.uwa.edu.au> [020609 14:03]:
> > to compute the next term.  (I don't think there's a feasible algorithm to
> > compute the automorphism group of a graph.)
> 
> As an aside, there is a feasible algorithm to compute the automorphism
> group of a graph (Brendan McKay's nauty) but of course there are too
> many graphs to check exhaustively for the next term of the sequence.
> Well, actually that's not true because the next term turns out to
> be 11, but in principle there are too many graphs to do anything
> more than another term or two in our lifetimes.

Not quite, since we only need to check the groups of those graphs
on which a particular cyclic group is already known to act.
My guess is that only very basic theory is needed to make an
exhaustive search feasible up to n=20 or maybe further.

Brendan.