These are chat archives for symengine/symengine

23rd
Dec 2015
Ralf Stephan
@rwst
Dec 23 2015 05:58
@certik @isuruf At some time I wanted to make Sage behave like that (expanding series(ex,n1)^2) but the bug ticket was turned down because I found that GiNaC explicitly disabled it, and they provided the expansion if one gave the command series(series(ex, n1)^2, n2). So you get more control on the precision. And noone wanted to change the GiNaC default. I'm divided on the matter. I doubt the benefit is so important.
Ralf Stephan
@rwst
Dec 23 2015 06:34
Automatic conversion to series is not even possible for all functions so there needs to be a way to recognize it, a superclass that is inherited from, or a member function that exists or not.
@certik I'm a fan of github's issue label system. I'll add these now for all issues without.
Ralf Stephan
@rwst
Dec 23 2015 06:49
@rijuldhir in the issue list at github there is a label menu where the issues are categorized. Maybe this is better?
Ondřej Čertík
@certik
Dec 23 2015 17:40
@rwst which function cannot be converted to series? And what does Mathematica do in that case?
The way I see it is that if we start the Series to live outside of Basic, then things are clear. Then later we can figure out if we want to move it under Basic and what the behavior should be. In the meantime, we can concentrate on the Series class to deliver good performance.
Ralf Stephan
@rwst
Dec 23 2015 19:13
@certik +1. Examples are abs(x),floor(x),dirac_delta(x). However, they can be differentiated, so there could be a series (although I have never seen it)
Ralf Stephan
@rwst
Dec 23 2015 19:58
After searching a bit, the last two can certainly not have a series expansion. Floor has infinitely many discontinuities, and dirac is not even a function.