[Pw_forum] how to consider the "dt" flag during vc-relax

Cesar R.S. da Silva cesards at msi.umn.edu
Tue Oct 23 00:21:26 CEST 2007


Dear Chao,

Your dt setting should work OK. wmass could be reduced to somthing like
0.002, provided atomic masses are all set to the average mass (amu), but
your value is not that bad. The only thing I think could produce
convergence problems is your conv_thr. Try conv_thr = 1e-9. Variable cell
requires precise stress calculation, and a more stringent convergence
criteria for the SCF cycle is required.

Cesar.


> ------------------------------
>
> Message: 5
> Date: Mon, 22 Oct 2007 15:56:16 +0200
> From: "Chaohao Hu" <chaohao.mse at gmail.com>
> Subject: [Pw_forum] how to consider the "dt" flag during vc-relax
> 	calculations
> To: "PWSCF Forum" <pw_forum at pwscf.org>
> Message-ID:
> 	<6502ee650710220656o5c8ac9b4s29eda137dc8d5686 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> Dear Pwscf users,
>
> I want to perform a vc-relax calculation for a low-symmetry structure
> (P-1),
> but the calculation is not always convergenced. Maybe this structure is
> far
> away from its equilibrium. From the help files I know the time step flag
> "dt"  is a important parameter for variable cell relaxing calculation. In
> the manual, it is said the good values of "dt" is 50/70 or 100/150, but
> the
> default value in INPUT_PW file is 20.0D0. How to understand them?  Here, a
> part of my input file is attached. The initial atomic position is
> determined
> by performing "relax" calculation.
>
> &CONTROL
>    calculation =   "vc-relax"   ,
>    restart_mode = 'from_scratch' ,
>    outdir='~/tmp/' ,
>    pseudo_dir = '~/espresso32/pseudo' ,
>    disk_io = 'default' ,
>    verbosity = 'default' ,
>    tstress = .true. ,
>    tprnfor = .true. ,
>    nstep =  55  ,
>                etot_conv_thr = 1.0E-5  ,
>                forc_conv_thr = 1.0D-4 ,
>                           dt = 50 ,
> /
> &SYSTEM
>   ibrav     = 0,
>   celldm(1) = 12.2409,
>   nosym     = .true.,
>   nat       = 30,
>   ntyp      = 3,
>   ecutwfc   = 25.D0,
>   ecutrho   = 150.D0,
> /
> &ELECTRONS
>   electron_maxstep = 500,
>   conv_thr    = 1.D-7,
>   mixing_beta = 0.7D0,
>   diagonalization   =   'david'   ,
> /
> &IONS
> /
> &CELL
>   cell_dynamics = 'damp-w',
>           press = 0.00,
>       wmass = 0.007,
> /
> ...
> ...
> K_POINTS automatic
> 5 5 5 1 1 1
>
> Best regards,
> C. H. Hu
>
> --
> ============================
> C.H. Hu
> Postdoctoral fellow
> Chimie et Physico-Chimie appliquees
> Institut Francais du Petrole (IFP)
> Rueil-Malmaison, France
> E_mail: chaohao.mse at gmail.com
> ============================
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://www.democritos.it/pipermail/pw_forum/attachments/20071022/e9b4550a/attachment-0001.htm
>
> ------------------------------
>
> Message: 6
> Date: Mon, 22 Oct 2007 16:08:37 +0200
> From: "Chaohao Hu" <chaohao.mse at gmail.com>
> Subject: Re: [Pw_forum] how to consider the "dt" flag during vc-relax
> 	calculations
> To: "PWSCF Forum" <pw_forum at pwscf.org>
> Message-ID:
> 	<6502ee650710220708gb4d71aftb8dedd39ab37a46e at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> I am very sorry for sending this mail again. But I am still awaiting your
> response.
>
> Regards,
> C. H. Hu
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://www.democritos.it/pipermail/pw_forum/attachments/20071022/82d63c6f/attachment-0001.htm
>
> ------------------------------
>
> Message: 7
> Date: Mon, 22 Oct 2007 16:23:50 +0200
> From: Xunlei Ding <ding at sissa.it>
> Subject: Re: [Pw_forum] How to calculate the properties of  free OH
> To: PWSCF Forum <pw_forum at pwscf.org>
> Message-ID: <471CB276.7060900 at sissa.it>
> Content-Type: text/plain; charset=x-gbk; format=flowed
>
> Dear Ou,
>
> I think you need to use nspin=2 for HO calculation, because it has 7
> electrons.
> So I think you need to add these in &SYSTEM section:
> nspin = 2, starting_magnetization(1)=0.1,
> occupations='fixed',
> nelup=4, neldw=3, nelec=7
>
> And as others have said, you need also decide:
> 1. whether to use pseudo of Cu for H
> 2. Use Gamma kpoint instead of what you used now.
>
> Best wishes,
> Ding
>
> oulihui666 wrote:
>
>> Dear pwscf users:
>>
>> Because of the need of research, I tried to calculate the properties
>> of free OH with pwscf, however,when calculating, error information was
>> found, it displayed:
>>
>> ==============================================
>>
>> * **from electrons : error # 1*
>>
>> * charge is wrong*
>>
>> ==============================================
>>
>> I am really newbie in this thing, sorry about these naive questions...
>>
>> Can you please suggest some solutions?
>>
>> And this is the input file:
>>
>> Thanks a lot
>>
>> best regards
>>
>> Lihui Ou
>>
>> &CONTROL
>>
>> calculation = 'relax' ,
>>
>> restart_mode = 'from_scratch' ,
>>
>> outdir = '/home/olh/tmp/' ,
>>
>> pseudo_dir = '/home/olh/pseudo/' ,
>>
>> prefix = 'OH' ,
>>
>> forc_conv_thr = 1.0D-4 ,
>>
>> tstress = .true. ,
>>
>> tprnfor = .true. ,
>>
>> /
>>
>> &SYSTEM
>>
>> ibrav = 14,
>>
>> A = 10.000 ,
>>
>> B = 10.000 ,
>>
>> C = 10.000 ,
>>
>> cosAB = 0 ,
>>
>> cosAC = 0 ,
>>
>> cosBC = 0 ,
>>
>> nat = 2,
>>
>> ntyp = 2,
>>
>> ecutwfc = 60 ,
>>
>> ecutrho = 240 ,
>>
>> /
>>
>> &ELECTRONS
>>
>> conv_thr = 1.D-7 ,
>>
>> mixing_beta = 0.7D0 ,
>>
>> /
>>
>> &IONS
>>
>> ion_dynamics = 'bfgs' ,
>>
>> pot_extrapolation = 'second_order' ,
>>
>> wfc_extrapolation = 'second_order' ,
>>
>> /
>>
>> ATOMIC_SPECIES
>>
>> O 16.00000 O.pz-rrkjus.UPF
>>
>> H 1.00000 Cu.pz-d-rrkjus.UPF
>>
>> ATOMIC_POSITIONS angstrom
>>
>> O 5.000000000 5.000000000 5.000000000
>>
>> H 5.000000000 5.000000000 6.000000000
>>
>> K_POINTS automatic
>>
>> 3 3 1 1 1 1
>>
>>
>>
>> ------------------------------------------------------------------------
>> ???? Yeah.net ?? ?? ?? ?? ?? ?? ?????? ?? ?? ?? ?? ?? ?? ?? ?? ?? ??
>> ?? ?? ?? >>
>> <%20http://event.mail.163.com/chanel/click.htm?from=NO_16&domain=126>
>>
>>------------------------------------------------------------------------
>>
>>_______________________________________________
>>Pw_forum mailing list
>>Pw_forum at pwscf.org
>>http://www.democritos.it/mailman/listinfo/pw_forum
>>
>>
>
>
> ------------------------------
>
> Message: 8
> Date: Mon, 22 Oct 2007 11:23:54 -0400 (EDT)
> From: Axel Kohlmeyer <akohlmey at cmm.chem.upenn.edu>
> Subject: Re: [Pw_forum] how to consider the "dt" flag during vc-relax
> 	calculations
> To: Chaohao Hu <chaohao.mse at gmail.com>
> Cc: PWSCF Forum <pw_forum at pwscf.org>
> Message-ID: <Pine.LNX.4.64.0710221059180.2512 at localhost.localdomain>
> Content-Type: TEXT/PLAIN; charset=US-ASCII
>
> On Mon, 22 Oct 2007, Chaohao Hu wrote:
>
> CHH> I am very sorry for sending this mail again. But I am still
> CHH> awaiting your response.
>
> ...and you think that by not even waiting a single day for
> a response and reposting, you will increase your chances of
> getting a (useful) reply? hmmmmm...
>
> please keep in mind that people responding to questions do
> this voluntarily and are not getting paid for it (as you
> didn't have to pay for the code), so nobody is obliged to
> answer, even more so in a timely fashion.
>
> if you don't get an answer to a question right away,
> can have a lot of reasons. for example:
> - nobody knows an answer.
> - nobody understood your question.
> - nobody feels like answering your question.
> - a person that would answer is in a different time zone,
>   e.g. 7-8 hours behind you.
> - a person that would answer feels that you should search
>   the mailing list archive fist.
> - a person that would answer does not have the time to do
>   it right now.
> - ...
>
> the best chance for you would be to have a closer look at
> the mailing list archives (there have been a lot of questions
> about vc-relax) and see if you can find something applicable
> to you. IIRC, the main reason for non-convergence are too
> tight convergence criteria on the forces.
>
> cheers,
>    axel.
>
>
> CHH> Regards,
> CHH> C. H. Hu
> CHH>
>
> --
> =======================================================================
> Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
>    Center for Molecular Modeling   --   University of Pennsylvania
> Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
> tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
> =======================================================================
> If you make something idiot-proof, the universe creates a better idiot.
>
>
> ------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
> End of Pw_forum Digest, Vol 4, Issue 29
> ***************************************
>





More information about the users mailing list