MathGroup Archive 1999

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

Search the Archive

Mathematica 3.0 - 4.0 discrepancies.

  • To: mathgroup at smc.vnet.net
  • Subject: [mg20377] Mathematica 3.0 - 4.0 discrepancies.
  • From: Jordi <jignes at mailhost.tcs.tulane.edu>
  • Date: Sun, 17 Oct 1999 02:45:48 -0400
  • Organization: Tulane University
  • Sender: owner-wri-mathgroup at wolfram.com

Just recently I have switched to Mathematica 4.0 from 3.0 and, to my
amazement, I have realized that some calculations I developed whith v
3.0 yield different results (quite different!) when evaluated with 4.0. 

But there is more: v 3.0 gave the same result regardless of the platform
it was running (Win9x, Linux, MacOS), which one would expect. HOWEVER, v
4.0 generates DIFFERENT results DEPENDING on the platform it is running! 
Only on MacOS I get the same results I got using 3.0. Win9x and Linux
generate different results running v 3.0 and v 4.0.

I am using the default installation (I am not changing any internal
precision parameters, etc). The computations in question involve mostly
some vector and matrix manipulation, which are used to generate a
numerical result.

Any help or suggestion would be much appreciated. I can provide the
notebook in question to anyone interested (it is only a couple dozen
lines long).


Thanks.

Jordi.


  • Prev by Date: Re: Books for:Elementary Numerical Computing with Mathematica
  • Next by Date: Font problems running remotely
  • Previous by thread: Re: Classes fail in baffling manner
  • Next by thread: RE: Mathematica 3.0 - 4.0 discrepancies.