[Pw_forum] setting grid size and memory management for pw scf
On Jan 12, 2013, at 17:21 , Paolo Giannozzi wrote: > > On Jan 12, 2013, at 14:47 , Mike Marchywka wrote: > >> Could not bracket Fermi level though > > may you need a few more Kohn-Sham states I meant: "maybe" you need ... P. --- Paolo Giannozzi, Dept of Chemistry&Physics&Environment, Univ. Udine, via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
[Pw_forum] setting grid size and memory management for pw scf
On Jan 12, 2013, at 14:47 , Mike Marchywka wrote: > Could not bracket Fermi level though may you need a few more Kohn-Sham states P. --- Paolo Giannozzi, Dept of Chemistry&Physics&Environment, Univ. Udine, via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
[Pw_forum] Relax Calculation for ZnO
On 10 January 2013 09:01, Angga Fauzi wrote: > Dear all QE users, > > I am trying to run relax calculation for my ZnO system. There are > something I want to ask. > >1. In running relax calculation, what occupations should I use? I am >trying to use occupations = 'tetrahedra', but it doesn't work. > > What do you mean by "it does not work"? Which symptom make you suspect that it may not be working? > >1. I am afraid if I use occupations = 'smearing', my result >calculation doesn't valid since ZnO is a semiconductor, not a metal. > > If it is not a metal, you can normally use fixed occupations. >1. After my relax calculation is done, should I run a scf calculation >again with atomic positions I get from relax calculation? > > You only need to for variable cell calculations, and only for code older that 5.0. cheers -- Dr. Lorenzo Paulatto IdR @ IMPMC -- CNRS & Universit? Paris 6 phone: +33 (0)1 44275 084 / skype: paulatz www: http://www-int.impmc.upmc.fr/~paulatto/ mail: 23-24/4?16 Bo?te courrier 115, 4 place Jussieu 75252 Paris C?dex 05 -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20130112/0d98baf6/attachment.html
[Pw_forum] setting grid size and memory management for pw scf
On 11 January 2013 17:06, Mike Marchywka wrote: > I'm try do dft on about 24 atoms using laptop and the thing runs out of > real memory. > I set diagnolaization to cg which may have solved the immediate problem > but curious > on what other things I can do. > Thanks. > > > You should not change the size of the fft grid by hand, except in some very specific case. What you change are the wavefunctions and charge density cutoffs, which are system independent and have some sort of physical meaning. These cutoffs in turn control the fft grid. Of course you cannot just reduce the cutoff, you need to test convergence and use a converged value. If you are running out of memory you can either but a bigger computer or study a smaller system. In some specific case changing the parallelism scheme or some computation parameter can help. bests -- Dr. Lorenzo Paulatto IdR @ IMPMC -- CNRS & Universit? Paris 6 phone: +33 (0)1 44275 084 / skype: paulatz www: http://www-int.impmc.upmc.fr/~paulatto/ mail: 23-24/4?16 Bo?te courrier 115, 4 place Jussieu 75252 Paris C?dex 05 -- next part -- An HTML attachment was scrubbed... URL: http://pwscf.org/pipermail/pw_forum/attachments/20130112/5b45ff63/attachment.html
[Pw_forum] setting grid size and memory management for pw scf
> From: giannozz at democritos.it > Date: Fri, 11 Jan 2013 23:01:13 +0100 > To: pw_forum at pwscf.org > Subject: Re: [Pw_forum] setting grid size and memory management for pw scf > > > On Jan 11, 2013, at 17:06 , Mike Marchywka wrote: > > > Sorry if there is a simple answer but I could not immediately > > figure out the > > issues with setting fft grid size or the actual parameters to set > > in the input file > > which issues? http://www.quantum-espresso.org/wp-content/uploads/Doc/pw_user_guide/node19.html Manually setting the FFT grids to a desired value is possible, but slightly tricky, using input variables?nr1,?nr2,?nr3?and?nr1s,?nr2s,?nr3s. The code will still increase them if not acceptable. Automatic FFT grid dimensions are slightly overestimated, so one may try?very carefully?to reduce them a little bit. The code will stop if too small values are required, it will waste CPU time and memory for too large values. In any case using CG diagonalization got it to run much faster, presumably making the point that memory can be more important in speed than just counting operations ( I understand the default is faster with infinite memory). ?Could not bracket Fermi level though, I'll have to check the other settings for this. The system is just all trans retinoic acid with a single Fe atom either near the COOH or out of plane of the chain. It I started with a a random placement and did structure optimization with jdftx but wanted to compare to QE and get more insight. I'm not even sure this system has physical meaning ( the pseudopotentials are the first ones I found without a lot of consideration and I have to come up with some "known" results for comparison yet).? Thanks. > > P. > --- > Paolo Giannozzi, Dept of Chemistry&Physics&Environment, > Univ. Udine, via delle Scienze 208, 33100 Udine, Italy > Phone +39-0432-558216, fax +39-0432-558222 > > > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://pwscf.org/mailman/listinfo/pw_forum