From: nma@12000.org   
      
   On 7/14/2017 1:59 PM, clicliclic@freenet.de wrote:   
      
   > If timeout effects are indeed negligible here, more than half of the   
   > roughly 5000 new integrals cannot be evaluated by either Rubi or   
   > Mathematica or Maple. One may consequently wonder how the roughly 3000   
   > corresponding model antiderivatives were obtained.   
   >   
   > Martin.   
   >   
      
   The difference in number of tests between current build test suite   
   Rubi 4.11 and earlier one 4.9.8 is (5,179) new integrals added.   
      
   If half of the new integrals could not be solved, then indeed the new   
   result makes sense (5% drop in overall score).   
      
   Looking at random number of integrals that could not be solved by Rubi,   
   I see some have _no_ corresponding model antiderivatives associated with   
   them.   
      
   It just shows the input itself as the optimal.   
      
   It looks like these new integrals were put there as place holders   
   by Albert until optimal anti-derivative is found?   
      
   Here is one example   
      
   http://12000.org/my_notes/CAS_integration_tests/reports/rubi_4_1   
   /test_cases/8_Special_functions/8.1Errorfunctions/resu7.htm#7   
      
   Int[Erf[b*x]^2/x,x]   
      
   No optimal anti-derivative there.   
      
   This explains why the new test results have lower score than the   
   last one. But again, Albert would know for sure, as he wrote the test   
   input files.   
      
   Regards,   
   --Nasser   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)