Date: Jul 2, 2013 6:05 PM
Author: clicliclic@freenet.de
Subject: Re: An independent integration test suite


Waldek Hebisch schrieb:
>
> FriCAS can do all 12 examples from chapter 3 and 9 elementary
> examples from chapter 7. The two nonelementary ones (numner 8
> and 9) can be done using 'complexIntegrate' or if you
> complicate them a bit (for example add 'exp(x)' to the
> integral), but return unevaliated as-is.
>
> BTW1: FriCAS is now at 1.2.1
>
> BTW2: FriCAS answers are at:
>
> http://axiom-wiki.newsynthesis.org/SandboxFriCASTimofeev37
>
> RTW2: If you want "winner", then examples are not challenging enough
> (in particular 12 rational examples).
>


Yes, I didn't really expect FriCAS to fail for one of these rational
integrands. Still it is nice to see there are no more surprises like the
failure of version 1.2.0 on Timofeev's example 21 from Chapter 1:
INT(1/(x*LN(x/a)+x)), x). The two non-elementary problems from Chapter 7
must be counted as "done" anyway; FriCAS 1.2.1 then succeeds (neglecting
the occasional branch-cut memory loss) on all problems from Chapters 1,
3, and 7, it appears.

Similarly I do not expect the problems from Chapter 2 (rational
integrands) to cause problems for Maple and Mathematica. More
interesting should be Chapter 5 (trigonometric integrands), Chapter 8
(exponential and hyperbolic integrands), and Chapter 9 (inverse
trigonometric integrands) - not to mention Chapter 4 (algebraic
integrands), which I am in the process of typing in.

So only Mathematica's performance on the problems from Chapters 3 and 7
remains unknown now.

Martin.