[Date Index] [Thread Index] [Author Index]

Re: Mathematica 8 Integration Bug

• To: mathgroup at smc.vnet.net
• Subject: [mg116012] Re: Mathematica 8 Integration Bug
• From: Daniel Lichtblau <danl at wolfram.com>
• Date: Sat, 29 Jan 2011 05:26:05 -0500 (EST)

ChiaraB wrote:
> Hello,
> 
> I needed to manipulate some probability distributions, so I ran the
> following code:
> 
> ccdfPL[x_, alpha_, xmin_] = (xmin/(xmin + x))^(alpha)
> 
> Assuming[alpha \[Element] Reals && alpha > 0 &&
>   xmin \[Element] Reals && xmin > 0, \!\(
> \*SubsuperscriptBox[\(\[Integral]\), \(0\), \(+\[Infinity]\)]\(D[
>     1 - ccdfPL[x, alpha, xmin], x] \[DifferentialD]x\)\)]
> 
> According to Mathematica 8, this integral does not converge on [0,+\
> [Infinity]), while it does converge using Mathematica 6 and
> Mathematica 7 (it converges using paper and pencil as well :) ). Any
> idea on how this is possible?
> 
> Thanks,
> 
> Chiara

Naughty integration code? (Wouldn't be the first time we needed to put 
it in the time out room.)

I filed a bug report for this. Will investigate further.

Daniel Lichtblau
Wolfram Research