Re: Bug 3! Gaussian integration in 5.0 is broken!
- To: mathgroup at smc.vnet.net
- Subject: [mg43816] Re: Bug 3! Gaussian integration in 5.0 is broken!
- From: Konstantin L Kouptsov <kouptsov at wsu.edu>
- Date: Tue, 7 Oct 2003 02:40:55 -0400 (EDT)
- Organization: Washington State University
- References: <blj66o$1i8$1@smc.vnet.net> <bllo8p$bsh$1@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Indeed works. But as you can see, it does not generate the message that convergence cannot be
checked. Thus, in this case Mathematica chooses a different evaluation route (using different
rules), and avoids the one that is broken.
Thanks for the hint.
Konstantin.
Paul Abbott wrote:
> In article <blj66o$1i8$1 at smc.vnet.net>,
> Konstantin L Kouptsov <kouptsov at wsu.edu> wrote:
>
>
>>Check this:
>>
>>Integrate[Exp[(I/h)*(x*Sqrt[B/2 + D] + x0)^2], {x, -Infinity, Infinity}]
>>
>>and observe the Erf[] functions that should not be there. Neither should x0.
>
>
> However, note the assumptions that Integrate makes. On the other hand,
>
> Assuming[x0 > 0 && B > 0 && D > 0 && h > 0,
> Integrate[E^((I*(Sqrt[B/2 + D]*x + x0)^2)/h),
> {x, -Infinity, Infinity}]]
>
> yields a reasonable result.
>
> Cheers,
> Paul
>