All of the ordinals are most definitely comparable (by definition). I implemented quicksort and it worked very well, so I will stick with that.
Marshall On Wed, Aug 17, 2011 at 4:13 PM, Henry Rich <[email protected]> wrote: > If you have a partial ordering quicksort is fastest. I thought some > numbers might be incomparable, which would be a topological sort. > > > > Henry Rich > > On 8/17/2011 3:46 PM, Henry Rich wrote: > > All I know about Cantor normal form is what I read in Wikipedia, but > > from what I see there you should just be able to box the normal forms > > and sort using the standard ordering. > > > > More generally, what you are doing is called topological sorting. There > > are algorithms for it. I've never done it in J. > > > > Henry Rich > > > > On 8/17/2011 3:34 PM, Marshall Lochbaum wrote: > >> I have a list of things--specifically, ordinal numbers in Cantor normal > >> form--and an ordering relation that tells me whether the difference of > two > >> of them is positive, negative, or zero (analogous to *@-~ for J > numbers). Is > >> there an efficient J way to perform this sort? > >> > >> Marshall > >> ---------------------------------------------------------------------- > >> For information about J forums see http://www.jsoftware.com/forums.htm > >> > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
