[Octave-bug-tracker] [bug #30114] perms generates different results to matlab

[David Bateman]

Follow-up Comment #6, bug #30114 (project octave):

Neither Rik or I stated that the matlab perms order wasn't consistent between
versions. However, matlab doesn't guarantee won't change and there is no logic
in the current order at all. If you can reproduce the exact ordering of matlab
with a code that is no slower than the existing octave code then I see no
reason not to accept it into Octave. But I think you're wasting your time.

Take a good look at the results of perms(1:4) that you supplied. The ordering
is clearly not lexicographical as you seem to imply.

As for the speed, neither perms or sortrows should take more than a second or
so on most modern machines before you run out of memory. I suspect that what
you're using this mesh for will take a much larger portion of your computation
time


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?30114>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/