"Nasser M. Abbasi" schrieb:   
   >   
   > On 3/12/2017 6:55 PM, clicliclic@freenet.de wrote:   
   > >   
   > > I remember that Maxima solves only about 80% of the Timofeev   
   > > testsuite; Sympy should even be noticeably weaker as Sage chose   
   > > Maxima for its default integrator. The Giac/Xcas integrator (as   
   > > used in GeoGebra) presumably isn't stronger either - somebody   
   > > should really run the tests on Sympy and Giac/Xcas:   
   > >   
   >   
   > I tried to add Fricas and Sympy to the CAS integration tests,   
   > but my Fricas skills still not good enough yet to write a build   
   > script for it. For sympy, which I know little about, but when I   
   > tried it, the syntax seems much different, so can't Axiom files   
   > for sympy. It needs new syntax set of files for all the   
   > 63,000+ integrals currently at Albert's Rubi web site.   
   >   
   > For example, one has to use ** for exponent, but axiom uses ^ for   
   > exponent). One has to define all symbols using a=Symbol('a') and   
   > so on. So there is some syntax difference from Axiom. May be   
   > there is a way to automatically convery Axiom files to Sympy   
   > valid syntax?   
   >   
   > If all CAS systems used the same common computer programming   
   > language, life will be so much simpler, and one does not   
   > have to learn new syntax each time  :)   
   >   
      
   It should suffice to let them loose on the Timofeev textbook suite: the   
   percentages of integrals solved were pretty close to those for the full   
   gamut of tests from Albert's website.   
      
   As regards Sympy, the replacement of "^" by "**" is easily accomplished   
   with the help of an editor. The only symbols used in the Timofeev suite   
   are [A, B, a, b, b1, c, c1, k, l, m, n, r, s, x], which may simply be   
   defined for every integral regardless of actual use. Nonelementary   
   functions appearing in Albert's "optimal" evaluations are [Ei(x),   
   gamma(n, x), hypergeometric(a, b, x), polylog(n,x)]. My analysis here is   
   based on the Maxima file of the Timofeev problems.   
      
   I expect both Sympy and Giac/Xcas to succeed on somewhere between 50%   
   and 70% of these textbook problems. It would be nice if the results   
   could be graded as discussed earlier.   
      
   FriCAS should be taught to distinguish single-valued from multivalued   
   functions and correct its Risch-conditioning of integrands for a test on   
   the full suite to be meaningful.   
      
   Martin.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)