[Axiom-developer] A terrible bug: romberg+simplify/expand: slowdown of 125/300 times

[Vladimir Bondarenko]

Hi *,

Any comment on layout of AXIOM's computational flow here?

.........................................................................

Maple 9.5.2

> restart; time(evalf(Int(1, z=0..1)));
> restart; time(evalf(Int(simplify(1), z=0..1)));
> restart; time(evalf(Int(expand(1), z=0..1)));

.310
.281
.149

Same with Mathematica 5.1.

.........................................................................

-- About the stuff below, I even have no idea reading the doc.
-- It IS an eerie bug, period. (If the doc claims the opposite,
-- there is a bug in the doc :)


-- AXIOM Computer Algebra System, Version of Wednesday December 15, 2004

-> )time on

-> romberg(z+->expand(1), 0, 1, 0.1, 0.1, 10, 12)

  [value= 1.0,error= 0.0,totalpts= 2049,success= true]
  Time: 0.05 (EV) = 0.05 sec

-> romberg(z+->1, 0, 1, 0.1, 0.1, 10, 12)

   [value= - 1.0,error= 0.0,totalpts= 2049,success= true]
   Time: 0.02 (EV) = 0.02 sec

-----------------------
-- applying simplify...
-----------------------

-> romberg(z+->simplify(1), 0, 1, 0.1, 0.1, 10, 12)

   [value= - 1.0,error= 0.0,totalpts= 2049,success= true]
   Time: 1.00 (IN) + 1.22 (EV) + 0.07 (OT) + 0.20 (GC) = 2.48 sec

-----------------------
-- even worse
-----------------------

-> romberg(z+->simplify(%i^2), 0, 1, 0.1, 0.1, 10, 12)

   [value= - 1.0,error= 0.0,totalpts= 2049,success= true]
   Time: 3.07 (IN) + 2.25 (EV) + 0.18 (OT) + 0.48 (GC) = 5.98 sec

.........................................................................


A bloodcurdling guesswork, Does AXIOM each time call simplify/expand ?!

Or... or what causes this user's nightmare?


Best,

Vladimir