On 04.09.2012, at 14:03, marco atzeri <marco.atz...@gmail.com> wrote:
> On 9/3/2012 12:03 PM, Lukas Reichlin wrote: >> On 03.09.2012, at 10:39, Rafael Laboissiere <raf...@laboissiere.net> wrote: > >> Hi Rafael, >> >> You can check whether the observed and expected results are equivalent >> state-space models (i.e. state-transformation, see command prescale for >> formulae). This can be done, e.g., by inspection of the Hankel singular >> values (command hsvd), time response (step, impulse) or frequency response >> (sigma). >> If they are the same, there should be nothing to worry about. If you want >> the same results, use Reference BLAS (and LAPACK) from www.netlib.org >> instead of ATLAS which you are probably using. The SLICOT authors recommend >> the use of the reference implementations. Correct results are more important >> than minor speed advantages of automatically tuned linear algebra software, >> aren't they? :-) >> >> Regards, >> Lukas >> > > Lukas, > there is no guarantee to have the "expected" result > also with the reference lib. > > Using Lapack 3.4.1 on cygwin > > > test /home/marco/octave/control-2.3.53/@lti/minreal.m > ***** assert (Ar, Ae, 1e-4); > !!!!! test failed > assert (Ar,Ae,1e-4) expected > 1.00000 -0.03930 -0.09800 -0.10660 0.07810 -0.23300 0.07770 > 0.00000 1.03120 0.27170 0.26090 -0.15330 0.67580 -0.35530 > 0.00000 0.00000 1.38870 0.66990 -0.42810 1.63890 -0.76150 > 0.00000 0.00000 0.00000 -1.21470 0.24230 -0.97920 0.47880 > 0.00000 0.00000 0.00000 0.00000 -1.05450 0.50350 -0.27880 > 0.00000 0.00000 0.00000 0.00000 0.00000 1.63550 -0.43230 > 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 1.00000 > but got > 1.00000 0.03935 0.09802 -0.10656 0.07805 -0.23302 0.07773 > 0.00000 1.03122 0.27169 -0.26093 0.15327 -0.67578 0.35529 > 0.00000 0.00000 1.38869 -0.66991 0.42806 -1.63894 0.76153 > 0.00000 0.00000 0.00000 1.21467 -0.24232 0.97919 -0.47884 > 0.00000 0.00000 0.00000 0.00000 1.05454 -0.50348 0.27881 > 0.00000 0.00000 0.00000 0.00000 0.00000 1.63547 -0.43234 > 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 1.00000 > maximum absolute error 3.27784 exceeds tolerance 0.0001 > > so we need to define a different test methodology > > Regards > Marco Hi Marco LAPACK 3.4.1 is not enough. You also need reference BLAS from Netlib. LAPACK + ATLAS doesn't count here. Do you have a different test methodology in mind? Regards, Lukas ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Octave-dev mailing list Octave-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/octave-dev
