On Mon, Sep 1, 2014 at 2:32 PM, Mike Marchywka <marchywka at hotmail.com> wrote:
[...] >> For that, you do _not_ need more digits for exactly the reason I gave >> you. If you output more digits, you are just adding random numbers. >> Literally. > > I suppose in general there could be offset problems with energies and you are > adding > some large fixed amount to a small change of interest. Or, you could want to > examine that would only make matters worse, since in that case you lose additional digits of precision due to the lower "density" of floating point numbers at large absolute values. > the numerical issues. what would be the point of computing a number with a significantly large systematic error with extreme numerical precision? not to mention that only because we treat atom cores as point particles, we can specify their positions with high precision. aiming to do accurate calculations is commendable, but one has to put things into perspective and be able to tell numerical precision from actual accuracy of the calculation. a lot of calculations we do look much more precise than what their accuracy is. axel. >> >>> >>> Please let me know if you or anyone know how to change the code >> showing more digits in pw.x. >> >> Many people know (including me). But why should anybody tell you, if >> what you want is a stupid thing? Have you talked about this with your >> adviser? >> >> In general, please spent a little time learning about the systematic >> errors of DFT, the numerical precision of floating point numbers, the >> accuracy of a self-consistent calculation and error propagation. >> >> You should quickly see the folly of your request. >> >>> Best regards >>> Haibei >>> >>> >>> On Sun, Aug 31, 2014 at 1:51 PM, Axel Kohlmeyer >> <akohlmey at gmail.com<mailto:akohlmey at gmail.com>> wrote: >>>> >>>> >>>> On Aug 30, 2014 10:57 PM, "Haibei Huang" >> <haibeih at student.unimelb.edu.au<mailto:haibeih at student.unimelb.edu.au>> >> wrote: >>>>> >>>>> Dear All, >>>>> >>>>> I am aware of that the QE will only show at most 8 digits after >> the decimal point in the output file, while more digits are needed in >> my calculation. >>>>> >>>>> Could any of you provide a way to increase the number digits after >> the decimal point? >>>> >>>> What is the point of using more digits, when the error is larger? >> You could just add random digits and have the same effect. >>>> >>>>> >>>>> Thanks for your time! >>>>> Looking forward to your reply! >>>>> Regards >>>>> Haibei >>>>> >>>>> >>>>> _______________________________________________ >>>>> Pw_forum mailing list >>>>> Pw_forum at pwscf.org<mailto:Pw_forum at pwscf.org> >>>>> http://pwscf.org/mailman/listinfo/pw_forum >>>> >>>> >>>> _______________________________________________ >>>> Pw_forum mailing list >>>> Pw_forum at pwscf.org<mailto:Pw_forum at pwscf.org> >>>> http://pwscf.org/mailman/listinfo/pw_forum >>> >>> >>> >>> _______________________________________________ >>> Pw_forum mailing list >>> Pw_forum at pwscf.org<mailto:Pw_forum at pwscf.org> >>> http://pwscf.org/mailman/listinfo/pw_forum >> >> _______________________________________________ Pw_forum mailing list >> Pw_forum at pwscf.org http://pwscf.org/mailman/listinfo/pw_forum > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://pwscf.org/mailman/listinfo/pw_forum -- Dr. Axel Kohlmeyer akohlmey at gmail.com http://goo.gl/1wk0 College of Science & Technology, Temple University, Philadelphia PA, USA International Centre for Theoretical Physics, Trieste. Italy.
