From: Albert_Rich@msn.com   
      
   On Monday, March 21, 2016 at 6:27:02 AM UTC-10, clicl...@freenet.de wrote:   
       
   > This sounds like much of the growth is caused by optimizing Rubi's   
   > results, and not by expanding Rubi's coverage of integrands.   
      
   To objectively determine if successive versions of Rubi expand its coverage of   
   integrands, one (i.e. Nasser) could run OLDER versions of Rubi through the   
   CURRENT test suite to see what Rubi 4.9 can now integrate that she couldn't   
   before.   
      
   > In any case, to be healthy, the growth must level off at some point.   
      
   As to whether any finite set of integration rules could produce optimal   
   antiderivatives for all possible integrands is, I think, an open and   
   interesting question.  But as Rubi shows, even if rule-based systems are   
   necessarily incomplete, they can produce    
   vastly superior results over the class of integrands they are designed to   
   handle...   
      
   Albert   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)