Subject: integrate ( exp (x^3), x, 0, 1) still wrong
From: Raymond Toy
Date: Sun, 13 May 2012 15:41:53 -0700
On 5/11/12 4:02 PM, Edwin Woollett wrote:
> --------------------------------------------------
> case exp(x^5) over (1,2)
> domain:complex greatly improves answer
> but large incorrect imaginary part
> and real part only accurate to two sig. figures.
I've written a bug report for that. I have a possible solution, but
need to test it some more.
I don't think your solution in nint of using bfloats is the correct
solution. It will work for many cases, but not all. For example,
gamma_incomplete(10.3b0,-100b0) is not very accurate with fpprec=16.
It's better with fpprec = 32.
Ray