[Pw_forum] How to set ecutwfc and ecutrho
Jiaye Li
jameslipd at gmail.com
Wed Nov 11 13:05:01 CET 2009
Dear Hui Wang and Gabriele Sclauzero
Thank you for your replies. I just start to learn how to use pwscf and there
are many knowledge and skills that I should learn. Anyway, I will read more
references and make more tests.
On Wed, Nov 11, 2009 at 7:04 PM, <pw_forum-request at pwscf.org> wrote:
> Send Pw_forum mailing list submissions to
> pw_forum at pwscf.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> http://www.democritos.it/mailman/listinfo/pw_forum
> or, via email, send a message with subject or body 'help' to
> pw_forum-request at pwscf.org
>
> You can reach the person managing the list at
> pw_forum-owner at pwscf.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Pw_forum digest..."
>
>
> Today's Topics:
>
> 1. Re: How to set ecutwfc and ecutrho (xirainbow)
> 2. Re: How to set ecutwfc and ecutrho (Gabriele Sclauzero)
> 3. graphene scf calculation (Vit)
> 4. Re: Descriptions of FHI and UPF formats? (Lorenzo Paulatto)
> 5. Re: Descriptions of FHI and UPF formats? (Paolo Giannozzi)
> 6. Re: graphene scf calculation (Eyvaz Isaev)
> 7. Re: graphene scf calculation (xirainbow)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Wed, 11 Nov 2009 08:38:26 +0000
> From: xirainbow <nkxirainbow at gmail.com>
> Subject: Re: [Pw_forum] How to set ecutwfc and ecutrho
> To: PWSCF Forum <pw_forum at pwscf.org>
> Message-ID:
> <21fbc4790911110038g1d398803o932dde721606d3e7 at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
> Dear Li?
>
>
> I am curious about the calculation speed related to "ecutwfc" and
> "ecutrho".
> > As indicated in the tutorial, ecutwfc represents the energy cutoff (Ry)
> and
> > ecutrho refers to kinetic energy cutoff for charge density and potential.
> I
> > cannot fully understand the meaning of that brief description.
>
> If you use Norm-concerving PP, you can just omit the parameter "ecutrho".
> If you use Ultra-soft PP, you must set "ecutrho" as (8~12)X"ecutwfc".
> You will fully understand the "ecutrho" after reading some paper about
> Ultra-soft PP?? (For example?PRB, 41 7892)
>
>
>
> > From my experience (vasp), the larger the energy cutoff, the slower the
> > calculation would be, but more accurate.
>
> Yes? it is true for PWscf, too.
>
> I just would like to know how to set the experience values of the two flags
> > for the case of metal bulk (and surface) and metal oxide bulk (and
> surface)?
> > Thank you.
> >
> I think, the value is different with different kind of metal.
> You can take a convergence test to get proper "ecutrho" and "ecutwfc" ??
>
> ____________________________________
> Hui Wang
> School of physics, Nankai University, Tianjin, China
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://www.democritos.it/pipermail/pw_forum/attachments/20091111/db895df0/attachment-0001.htm
>
> ------------------------------
>
> Message: 2
> Date: Wed, 11 Nov 2009 09:58:49 +0100
> From: Gabriele Sclauzero <sclauzer at sissa.it>
> Subject: Re: [Pw_forum] How to set ecutwfc and ecutrho
> To: PWSCF Forum <pw_forum at pwscf.org>
> Message-ID: <4AFA7CC9.6090108 at sissa.it>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> Dear Jiaye,
>
> Jiaye Li wrote:
> > Dear users
> >
> > I am curious about the calculation speed related to "ecutwfc" and
> > "ecutrho". As indicated in the tutorial, ecutwfc represents the energy
> > cutoff (Ry) and ecutrho refers to kinetic energy cutoff for charge
> > density and potential. I cannot fully understand the meaning of that
>
> More precisely, ecutwfc and ecutrho are the kinetic energy cutoff of the
> plane wave basis
> used to describe the wavefunctions and the charge density, respectively.
>
> > brief description. From my experience (vasp), the larger the energy
> > cutoff, the slower the calculation would be, but more accurate. I just
>
> That's right and I think should be a general consideration valid for any
> code using plane
> waves. I would say "longer", rather than "slower" (which I feel more
> appropriate with
> regard to the optimization of a code or the performance of a computer).
>
>
> > would like to know how to set the experience values of the two flags for
> > the case of metal bulk (and surface) and metal oxide bulk (and surface)?
>
> The choice of the cutoffs depends only on the pseudopotentials you are
> using and on which
> quantities you are interested in (for instance, magnetic properties are
> harder to converge
> wrt total energy). Once you've chosen your PPs you can test convergence of
> the energy wrt
> to cutoffs in the cheapest system (usually in the bulk). The threshold of
> convergence is
> somehow your choice: personally I don't go below 1mRy since this is the
> expected order of
> magnitude of the errors due to PPs.
> For normconserving PPs you don't need to specify ecutrho, since the default
> (4*ecutwfc) is
> good (you can try to figure out why, writing down the charge density in
> terms of the PW
> expansion of the wavefunctions).
> For ultrasoft PPs you will need higher ecutrho due to the augmentation
> terms in the charge.
>
> Try to read some literature about pseudopotential and ask to some
> experienced people
> working next door and, of course, experience yourself by trying.
>
> HTH
>
> GS
>
>
> > Thank you.
> >
> > --
> > Sincerely yours
> >
> > Jiaye Li
> >
> >
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> > http://www.democritos.it/mailman/listinfo/pw_forum
>
> --
>
>
> o ------------------------------------------------ o
> | Gabriele Sclauzero, PhD Student |
> | c/o: SISSA & CNR-INFM Democritos, |
> | via Beirut 2-4, 34014 Trieste (Italy) |
> | email: sclauzer at sissa.it |
> | phone: +39 040 3787 511 |
> | skype: gurlonotturno |
> o ------------------------------------------------ o
>
>
> ------------------------------
>
> Message: 3
> Date: Wed, 11 Nov 2009 15:42:13 +0600
> From: Vit <vitruss at gmail.com>
> Subject: [Pw_forum] graphene scf calculation
> To: "PWSCF Forum" <pw_forum at pwscf.org>
> Message-ID: <op.u2748ll9ijpzuf at arch.che.intra.net>
> Content-Type: text/plain; charset="utf-8"
>
> Dear all!
> I'm trying to plot graphene band structure as a part of computations of
> carbon based composites. But scf calculation for graphene can't achieve
> convergence. I've tried different mixing_beta, mixing_mode, degaus ant
> ecutwfc. The example input file is attached. Could you please tell me
> which parameters I should tune to achieve convergence?
>
> Thanks in advance,
> Koroteev Victor
> PhD student,
> Nikolaev Institute of Inorganic Chemistry SB RAS
> Novosibirsk, Russian Federation.
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://www.democritos.it/pipermail/pw_forum/attachments/20091111/f8afcec1/attachment-0001.htm
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: graphite.scf.in
> Type: application/octet-stream
> Size: 605 bytes
> Desc: not available
> Url :
> http://www.democritos.it/pipermail/pw_forum/attachments/20091111/f8afcec1/attachment-0001.obj
>
> ------------------------------
>
> Message: 4
> Date: Wed, 11 Nov 2009 10:46:03 +0100
> From: "Lorenzo Paulatto" <paulatto at sissa.it>
> Subject: Re: [Pw_forum] Descriptions of FHI and UPF formats?
> To: "PWSCF Forum" <pw_forum at pwscf.org>
> Message-ID: <op.u275e10ta8x26q at paulax>
> Content-Type: text/plain; charset=utf-8; format=flowed; delsp=yes
>
> In data 11 novembre 2009 alle ore 02:17:23, Brad Malone
> <brad.malone at gmail.com> ha scritto:
> > Is
> >
> http://quantum-espresso.org/wiki/index.php/Unified_pseudopotential_format
> > up to date?
>
> Yes, it is. There is a new UPF format which is quite different (the
> description is currently sitting in my hard disk), but fhi2upf converts to
> the one described on the web page.
>
> > If so, any explanation for FHI then? It would be nice to not
> > simply get an understanding of the format by looking at fhi2upf.f90,
> > since I then might not be able to tell what assumptions result in it
> > occasionally
> > not working.
>
> I see you point, but I don't know of any such description. You may be
> luckier on the abinit web site, although a quick google search didn't find
> anything for me.
>
> regards
>
> --
> Lorenzo Paulatto
> SISSA & DEMOCRITOS (Trieste)
> phone: +39 040 3787 511
> skype: paulatz
> www: http://people.sissa.it/~paulatto/<http://people.sissa.it/%7Epaulatto/>
>
> *** save italian brains ***
> http://saveitalianbrains.wordpress.com/
>
>
> ------------------------------
>
> Message: 5
> Date: Wed, 11 Nov 2009 10:50:14 +0100
> From: Paolo Giannozzi <giannozz at democritos.it>
> Subject: Re: [Pw_forum] Descriptions of FHI and UPF formats?
> To: PWSCF Forum <pw_forum at pwscf.org>
> Message-ID: <4AFA88D6.9060205 at democritos.it>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> Brad Malone wrote:
>
> > If so, any explanation for FHI then?
>
> FHI = Fritz Haber Institute...some time ago I looked for
> documentation of their format and found nothing
>
> P.
> --
> Paolo Giannozzi, Democritos and University of Udine, Italy
>
>
> ------------------------------
>
> Message: 6
> Date: Wed, 11 Nov 2009 03:04:08 -0800 (PST)
> From: Eyvaz Isaev <eyvaz_isaev at yahoo.com>
> Subject: Re: [Pw_forum] graphene scf calculation
> To: PWSCF Forum <pw_forum at pwscf.org>
> Message-ID: <304105.15826.qm at web65702.mail.ac4.yahoo.com>
> Content-Type: text/plain; charset=us-ascii
>
> Dear Victor,
>
>
> > But scf calculation for graphene can't
> > achieve convergence. I've tried different mixing_beta,
> > mixing_mode, degaus ant ecutwfc. The example input file is
> > attached. Could you please tell me which parameters I should
> > tune to achieve convergence?
>
> There are several issues (according to your input file) that can affect on
> convergence in your case. But the main reason, most likely, is k-mesh grid,
> using 16x16x1 grid the convergence was OK.
>
> Please pay attention, 16x16x1, because you simulate 2D case.
>
> For Ultrasoft PsP you should use ecutrho about (8-12) times larger than
> ecutwfc.
>
> You have 2 carbon atoms with 4 valence electrons, so, you do not need 8
> bands (only 4 is enough), if graphene is semiconducting. In this case you
> can omit degauss (not sure it is used, if you did not specify "occupations"
> keyword). Otherwise use keywords specific for metallic case
> (occupations='smearing', smearing='XYZ', degauss=LKJ).
>
> I used mixing_beta=0.1 and mixing_mode-'local-TF', too.
>
> Bests,
> Eyvaz.
>
>
> -------------------------------------------------------------------
> Prof. Eyvaz Isaev,
> Theoretical Physics Department, Moscow State Institute of Steel & Alloys,
> Russia,
> Department of Physics, Chemistry, and Biology (IFM), Linkoping University,
> Sweden
> Condensed Matter Theory Group, Uppsala University, Sweden
> Eyvaz.Isaev at fysik.uu.se, isaev at ifm.liu.se, eyvaz_isaev at yahoo.com
>
>
> --- On Wed, 11/11/09, Vit <vitruss at gmail.com> wrote:
>
>
>
>
>
>
>
>
>
>
> ------------------------------
>
> Message: 7
> Date: Wed, 11 Nov 2009 11:04:17 +0000
> From: xirainbow <nkxirainbow at gmail.com>
> Subject: Re: [Pw_forum] graphene scf calculation
> To: PWSCF Forum <pw_forum at pwscf.org>
> Message-ID:
> <21fbc4790911110304g39d379fam4e3f69b79065d9ed at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
> Dear Victor:
> I look at your input file and notice :"K_POINTS {automatic} 24 24 24 0 0 0"
> I think there is no need to set kz=24, because you deal with a two
> dimension
> graphene.
> Of couse, setting kz=24 is not wrong.
>
> The following is my graphene.rx.in (not scf.in) for graphene.
> I calculated it about half a year ago, and it went well??
> I also put my graphene's results at
> http://blog.sina.com.cn/s/blog_5f15ead20100c9fu.html
> I hope it can help you??
>
> &CONTROL
> calculation = 'vc-relax' ,
> restart_mode = 'from_scratch' ,
> outdir = '/home/raman/graphene/g-monolayer/' ,
> pseudo_dir = '/home/raman/espresso-4.0.4/pseudo/' ,
> etot_conv_thr = 1.0e-7 ,
> forc_conv_thr = 1.0e-7 ,
> tstress = .true. ,
> tprnfor = .true. ,
> /
> &SYSTEM
> ibrav = 4,
> celldm(1) = 4.608737,
> celldm(3) = 4.536666,
> nat = 2,
> ntyp = 1,
> ecutwfc = 60.D0 ,
> nosym = .false. ,
> nbnd = 8,
> nelec = 8,
> tot_charge = 0.000000,
> occupations = 'fixed' ,
> /
> &ELECTRONS
> conv_thr = 1.D-8 ,
> mixing_mode = 'local-TF' ,
> mixing_beta = 0.7D0 ,
> /
> &IONS
> ion_dynamics = 'bfgs' ,
> pot_extrapolation = 'second_order' ,
> wfc_extrapolation = 'second_order' ,
> /
> &CELL
> cell_dynamics = 'bfgs' ,
> press_conv_thr = 0.2 ,
> cell_dofree = 'xyz' ,
> /
> ATOMIC_SPECIES
> C 12.00000 C.pz-rrkjus.UPF
> ATOMIC_POSITIONS crystal
> C 0.000000000 0.000000000 0.000000000 1 1 1
> C 0.333333333 -0.333333333 0.000000000 1 1 1
> K_POINTS automatic
> 23 23 1 0 0 0
>
>
> ____________________________________
> Hui Wang
> School of physics, Nankai University, Tianjin, China
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://www.democritos.it/pipermail/pw_forum/attachments/20091111/b99d5b23/attachment.htm
>
> ------------------------------
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
> End of Pw_forum Digest, Vol 29, Issue 31
> ****************************************
>
--
Sincerely yours
Jiaye Li
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20091111/2a8878f2/attachment.html>
More information about the users
mailing list