From: nma@12000.org   
      
   I finished porting Albert Rich grading function from Maple to Python.   
      
   Tested it on random integrals, and it gives same grade as Maple for   
   same input in Python. Now running it on the whole test suite to   
   verify more the grades.   
      
   I found something strange doing this with sympy. (I've seen   
   this before I think).   
      
   Here is an example (sympy 1.5)   
      
   (base) >python   
   from sympy import *   
   from sympy.abc import z,x   
   integrand=1/(cos(z)+sin(z)+sqrt(2))   
   integrate(integrand,z)   
      
      
   -8764825322083891395107744408255764367335970881679568886584767372   
   06166480148877802131608460208841059870201798/(-105800800822030308   
   195426917760537664540035358159831204871954190267657005107678279521   
   4631108949695080001219479*tan(z/2) + 7481246371622488258548281990   
   09082431883576062841145246524390135639826645575610848140217669527   
   057804968160189*sqrt(2)*tan(z/2) - 309883371058054256099440978596   
   29421351677751875716680219515176703674340550117194707441343942263   
   7275033059290*sqrt(2) + 43824126610419456975538722041278821836679   
   8544083978444329238368603083240074438901065804230104420529935100899) +   
   61976674211610851219888195719258842703355503751433360439030353407   
   3486811002343894148826878845274550066118580*sqrt(2)/(-10580080082   
   20303081954269177605376645400353581598312048719541902676570051076   
   782795214631108949695080001219479*tan(z/2) + 7481246371622488258   
   48281990090824318835760628411452465243901356   
   39826645575610848140217669527057804968160189*sqrt(2)*tan(z/2) -    
   098833710580542560994409785962942135167775187571668021951517670   
   36743405501171947074413439422637275033059290*sqrt(2) + 438241266   
   0419456975538722041278821836679854408397844432923836860   
   3083240074438901065804230104420529935100899)   
      
      
   But if the integrand is simplified first, then   
      
   simplify(integrand)   
          sqrt(2)/(2*(sin(z + pi/4) + 1))   
      
      
   integrate(simplify(integrand),z)   
      
              -sqrt(2)/(tan(z/2 + pi/8) + 1)   
      
   Much better and shorter result, which will get better grade.   
      
   Since the build program does not add "simplify" to each integrand,   
   such strange result will happen with sympy on a number of tests.   
      
   I could add an automatic "simplify" for sympy to simplify the integrand   
   before calling integrate in order to work around this issue.   
      
   Should I do this (i.e. add explicit simplify for sympy?) or leave it   
   as is now? Since all other CAS system do not have this issue.   
      
   --Nasser   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)