On Friday, July 27, 2012 12:25:01 PM UTC-7, MartinX wrote:
>
> I have had a go at implementing trac #3401, symbolic Li, and would like to
> override the behaviour of integrate(), but I can't see how to do it. We
> have the _derivative_ method in BuiltinFunction to define the derivative
> o
Yes, I have simply tweaked log_integral, to get Li and it is the linear
term that is my headache. I can't see Li in any of the integrators, but I
can get a valid integration if I return Li as li(x)-li(2) in _eval(). But
to me that seems untidy. It would be nice if _eval returned Li(x) but pas
On Jul 27, 3:25 pm, MartinX wrote:
> I have had a go at implementing trac #3401, symbolic Li, and would like to
> override the behaviour of integrate(), but I can't see how to do it. We
> have the _derivative_ method in BuiltinFunction to define the derivative
> of a function but no equivale
