From: fateman@cs.berkeley.edu   
      
   On 3/10/2017 11:02 AM, clicliclic@freenet.de wrote:   
   > What you describe appears to be the back-end of an integrator only.   
      
   If you look at the description of Joel Moses' integration   
   program SIN, you see that after trying various methods   
   in 3 stages, a final routine based on lookup was suggested   
   (but not implemented in Macsyma.).  It was called ITALU   
   pronounced "I tell you".   I think the stages uses in SIN   
   have influenced other algorithmic approaches to integration   
   in other systems.   
      
    From some perspectives this may be the wrong place. If you   
   want an "optimal" form for the result, one of the earlier   
   methods may produce a non-optimal one, when Rubi would   
   do better, if it were called.   
      
   Possibly Rubi should be unleashed in parallel with other   
   methods.  The idea would be doing something like this "in the cloud"   
   and in concert with any number of different CAS running   
   on any number of different operating systems etc.   
      
   TILU, which I wrote some time ago, was intended to be   
   used in this way as a network-based resource. It was   
   a lookup program, not nearly as "curated" as RUBI.   
      
   Given several results, you still have to figure out   
   which is the optimal one. This would have to be both   
   small AND CORRECT.  (Do we know how to automatically   
   determine the "most continuous" etc.).  Remember that   
   two expressions can be non-equal yet both correct, if   
   they differ by a constant..   
      
   RJF   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)