Sun<-> PC
- To: mathgroup at smc.vnet.net
- Subject: [mg29540] Sun<-> PC
- From: bernd at bio.vu.nl (Bernd Brandt)
- Date: Sat, 23 Jun 2001 01:47:29 -0400 (EDT)
- Sender: owner-wri-mathgroup at wolfram.com
Dear group, We have Mathematica 4.0.2 for PC and SUN Solaris. During a minimization procedure of a numerical function defined by NDSolve the behavior is different: The PC (Pentium 500 Mhz) just returns a found minimum, the SUN (Ultra 10) produced a warning coming from NDSolve: NDSolve::"ndsz": "At t == 11.465276734029507, step size is effectively zero; singularity suspected." and then from the interpolation: InterpolatingFunction::"dmval": "Input value 11.989 lies outside the range of data in the interpolating function. Extrapolation will be used. The returned minimum on the SUN is smaller (0.03668 vs 0.03757), these values may not be compared because of the NDSolve errors. We recentely have upgraded to 4.1. The FindMinimum in this version is not able to converge within the used nr of 50 iterations, and also show these platform dependent 'errors'. This also means that Numerical Integration on a Intel / AMD/ SUN processors yield different outcomes (when values become very small probably) and are dependent on the systems FPU units. Has anyone come accross these platform dependent differences ? Regards, Bernd