[MUG] Guaranteed multiple-precision results

Search email archive for  

[MUG] Guaranteed multiple-precision results
Author: Vincent Lefevre    Posted: Tue, 14 Jan 2003 15:31:49 +0100

>> From: Vincent Lefevre />
I'm searching for a way to obtain guaranteed multiple-precision
results with Maple.

A solution is true interval arithmetic. I used to use the intpak
package, but

* it has been said that the authors made wrong assumptions on the
accuracy of Maple functions (I could then use intpakX though,
but is it really reliable or does it also make assumptions based
on unknown facts?);

* intpak seems to be written for Maple 5, and Maple 5 doesn't run
on the most recent Linux machines. The intpakX package would be
affected by the same problem, as it is an intpak extension and
needs the intpak package.

Is there a working interval arithmetic package (and reliable enough)
for Maple 8?

Is the accuracy of Maple functions known? Note that when I say
"known", I want something that could be regarded as guaranteed,
based on proofs. Having guaranteed results for particular cases
such as sin(x) where x is close to a multiple of pi is important.

Thanks in advance,

Vincent Lefhvre.

[MUG] AW: Guaranteed multiple-precision results
Author: Thomas Richard    Posted: Fri, 31 Jan 2003 09:08:52 +0100

>> From: "Thomas Richard" />
> >> From: Vincent Lefevre /> >
> [...]
> * intpak seems to be written for Maple 5, and Maple 5 doesn't run
> on the most recent Linux machines. The intpakX package would be
> [...]

Maple V (Releases 4, 5 and 5.1) for Linux were linked with libc5,
while recent Linux systems are based on libc6 (aka glibc2). This
results in a very misleading error message when you lauch Maple.

But you can install a compatibility package providing the old libs.
How that package is called depends on the Linux distribution. I've
never encountered trouble with this in Red Hat, Debian, and SuSE.
(It is not included in Red Hat >= 7.0 anymore, but you can use an
older RPM package of the libc5 files.)

Some hints are provided in the Maple V troubleshooting FAQ at WMI's
web site.

Mit freundlichen Gruessen / best regards
Thomas Richard Tel.: +49-241-40008-52, Fax: -13
Maple Support /> Scientific Computers GmbH <http://www.scientific.de>

Previous by date: [MUG] Problem with SWP Implementation of Maple, Jim Moser
Next by date: [MUG] binomial coefficients, Maple User Group
Previous thread: [MUG] bug in dsolve with piecewise, Sussman
Next thread: [MUG] binomial coefficients, Maple User Group

For the time being we are unable to offer the following product ranges although we are currently working hard to increase the number of products we can offer in the future. Please contact us to talk about alternative products that we may be able to offer you.