These are chat archives for symengine/symengine

18th
Jan 2015
Isuru Fernando
@isuruf
Jan 18 2015 06:24
@certik and also, what should be done for expand of negative powers? For eg. (1/(x+y)).expand() gives an error that is not really helpful
Ondřej Čertík
@certik
Jan 18 2015 06:33
@isuruf see here: sympy/csympy#370
I think negative powers should be left as they are
As to (3/2)**(4/3), I don't think there is a reason. The only rule is to keep the exponent between (0, 1], and I guess our code currently removes the fraction in order to do that.
Ondřej Čertík
@certik
Jan 18 2015 06:43
Finally, for the negative powers, I reported the behavior as a bug: sympy/csympy#371
Isuru Fernando
@isuruf
Jan 18 2015 07:23
I think we should keep (3/2)**(4/3) as in the master and change (3/2)**(1/3) to do the same
Isuru Fernando
@isuruf
Jan 18 2015 14:39
Now that #369 is merged, I ran the Problem R1 in http://wiki.sagemath.org/symbench and CSymPy (Using only expand) is almost 200 times faster than SymPy. (0.978ms vs 195ms)