Subject: integrate ( exp (x^3), x, 0, 1) still wrong
From: Robert Dodier
Date: Sat, 12 May 2012 08:42:57 +0000 (UTC)
On 2012-05-11, Edwin Woollett <woollett at charter.net> wrote:
> Using the lastest gcl 5.27.0-1, there is no longer
> a stack problem, but a wildly wrong answer.
>
> (%i1) (display2d:false,fpprintprec:8)$
>
> (%i2) integrate(exp(x^3),x,0,1);
> (%o2) (gamma_incomplete(1/3,-1)-gamma(1/3))/3
>
> (%i3) float(%);
> (%o3) 0.333333*(-3.4863699*%i-2.0128566)
>
> (%i4) expand(%);
> (%o4) -1.1621233*%i-0.670952
Well, I don't know that this is "wildly wrong". It seems that abs(%o4) =
1.34... so that makes me think there is some branch cut stuff going on
here. As such, reasonable choices can yield different results. Not sure
that there's really a universal resolution to that.
FWIW
Robert Dodier