I want the outputs, though as the \:~@(#/.~)@,"2 idiom expresses, I am primarily concerned with the structure of the underlying graph. (Any of the nodes with the same \:~@(#/.~)@,"2 signature here represent the same graph. And, I personally do not have reason to prefer one over another)
Thanks, -- Raul On Fri, Jun 19, 2020 at 7:08 PM 'Michael Day' via Programming <programm...@jsoftware.com> wrote: > > Do you wish to construct the outputs, or just need their sizes? > > Assuming the former, I managed to make the thing more complicated by > either > calling a function to add a column to a combination matrix, or embedding > its code in the graphs routine. This saves repeating in (n+1)comb m what > you've already done in n comb m, but the time improvement is marginal if > anything, and my versions use slightly more space! So I won't bore you > with the details. > > There appear to be rather complicated algorithms for deriving their sizes, > eg Wang at https://arxiv.org/pdf/1604.04148.pdf > > Best of luck! > Mike > > On 19/06/2020 19:59, Raul Miller wrote: > > https://oeis.org/A004251 > > > > require'stats' > > > > graphs=:verb define > > r=. ,a: > > if. 1 < y do. > > m=. #pairs=. 2 comb y > > for_n.1+i.m do. > > raw=. pairs{~ n comb m > > ndx=. (i. ~.) \:~@(#/.~)@,"2 raw > > r=. r, ndx <@{ raw > > end. > > end. > > ) > > > > #@graphs"0 i.8 > > 1 1 2 4 11 31 102 342 > > > > This implementation is too inefficient, in my opinion, to go further than > > that. > > > > Does anyone here have a more efficient implementation? > > > > Thanks, > > > > -- > This email has been checked for viruses by Avast antivirus software. > https://www.avast.com/antivirus > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
