[Wien] MBJ, case.in0_grr
tran at theochem.tuwien.ac.at
tran at theochem.tuwien.ac.at
Wed Dec 16 09:56:02 CET 2015
Hi,
MSR1 and PRATT (not PRRAT) should obviously lead to the same results
as long as you let the calculations run until they are sufficiently
well converged. By default a calculation stops after 40 iterations
even if the convergence criteria are not reached. Execute run_lapw with
option "-i 100" to increase to 100 for instance.
Usually, I start with MSR1. If the calculation
does not seem to converge with MSR1 (infinite oscillations or diverge),
I stop the calculation. Then, I restart the calculation with
PRATT with a very small mixing factor like 0.05 and regenerate
case.clmsum with dstart or from previous PBE calculation.
Then, if the calculation seems to converge, I switch back to default
parameters in case.inm to reach convergence a little bit faster.
F. Tran
On Wed, 16 Dec 2015, shima pourrad wrote:
>
> Dear F.Tran
>
> I appreciate a lot because of your explanation. I ran the calculations with -ec 0.00001 convergence criteria. But unfortunately I
> found that I made a bad mistake : for the second calculation I forgot to change the mixing scheme to PRRAT , and it ran with the
> default mixing scheme which is MSR1. And this mistake causes two different results.(When I corrected mixing scheme , both
> calculations gave me the same results ). I apologize because of this carelessness.
>
> So I have another question, which of the two mixing schemes gives me reliable results? I mean, is it possible my first calculation
> (with PRRAT and mixing factor=0.2) converged to the ghost band?? In user guide it was written that: in most cases it is possible
> to switch back to MSR1 after some initial scf-cycles. Can we use MSR1 from the first of the calculation? What kind of mixing scheme
> we can use for the mbj calculations?
>
> Thanks a lot.
>
> Shima M.pourrad
>
>
> On Mon, Dec 14, 2015 at 7:39 PM, <tran at theochem.tuwien.ac.at> wrote:
> For the moment, I suspect that your calculations are not very well
> converged. Did you run the calculations with good energy and charge
> convergence criteria? If not, run the two calculations with
> -ec 0.0001 -cc 0.0001
> which is fairly good criteria. Then, maybe the two calculations
> give same results.
>
>
> On Mon, 14 Dec 2015, shima pourrad wrote:
>
>
> Hi
>
> Thanks for your prompt answer.
>
> Yes, for the first calculation with “ EX_GRR VX_GRR” I have :
>
> Case.in0 :
>
> TOT XC_MBJ (XC_LDA,XC_PBESOL,XC_WC,XC_MBJ,XC_REVTPSS)
>
> R2V IFFT (R2V)
>
> 120 120 120 1.00 1 min IFFT-parameters, enhancement factor, iprint
>
>
>
> And case.in0_grr :
>
> TOT EX_GRR VX_GRR (XC_LDA,XC_PBESOL,XC_WC,XC_MBJ,XC_REVTPSS)
>
> R2V IFFT (R2V)
>
> 120 120 120 1.00 1 min IFFT-parameters, enhancement factor, iprint
>
>
>
> This calculation give me these results by Analysis :
>
> --- ENE -----------
>
> in 1 files:
>
> case.scf::ENE : *WARNING** TOTAL ENERGY IN Ry = -79386.93188247
>
> --- FER -----------
>
> in 1 files:
>
> case.scf::FER : F E R M I - ENERGY(TETRAH.M.)= 0.4117108915
>
> --- GAP -----------
>
> in 1 files:
>
> case.scf::GAP : 0.0564 Ry = 0.767 eV (provided you have a proper k-mesh)
>
>
>
> For the second calculation with “ EX_GRR EC_NONE VX_GRR VC_NONE” I have:
>
> Case.in0 :
>
> TOT XC_MBJ ( (XC_LDA,XC_PBESOL,XC_WC,XC_MBJ,XC_REVTPSS)
>
> R2V IFFT (R2V)
>
> 120 120 120 1.00 1 min IFFT-parameters, enhancement factor, iprint
>
>
>
> And case.in0_grr :
>
> TOT EX_GRR EC_NONE VX_GRR VC_NONE ( (XC_LDA,XC_PBESOL,XC_WC,XC_MBJ,XC_REVTPSS)
>
> R2V IFFT (R2V)
>
> 120 120 120 1.00 1 min IFFT-parameters, enhancement factor, iprint
>
>
>
>
>
> And its results :
>
> --- ENE -----------
>
> in 1 files:
>
> case.scf::ENE : *WARNING** TOTAL ENERGY IN Ry = -79386.93585459
>
> --- FER -----------
>
> in 1 files:
>
> case.scf::FER : F E R M I - ENERGY(TETRAH.M.)= 0.4269866858
>
> --- GAP -----------
>
> in 1 files:
>
> case.scf::GAP : 0.0423 Ry = 0.575 eV (provided you have a proper k-mesh)
>
>
>
> what is your opinion?
>
> Sincerely
>
> Shima M.Pourrad
>
>
> On Mon, Dec 14, 2015 at 6:11 PM, <tran at theochem.tuwien.ac.at> wrote:
> Hi,
>
> "XC_MBJ" in the 1st line of case.in0 indicates that the mBJ method
> will be used. For mBJ, it is necessary to have also the file case.in0_grr
> such that the average of grad(rho)/rho in the unit cell
> [used for Eq. (3) in PRL 102, 226401 (2009)] is calculated.
>
> According to a test that I've just made it does not matter which one
> of these two is specified in case.in0_grr:
> "EX_GRR EC_NONE VX_GRR VC_NONE"
> "EX_GRR VX_GRR"
>
> So, I don't understand why you got two different results.
> Can you show us the input files case.in0 and case.in0_grr
> that you used for the two calculations?
>
> F. Tran
>
> On Mon, 14 Dec 2015, shima pourrad wrote:
>
>
> Dear P.Blaha and F.Tran and Wien2k Users
>
> I am running wien version14.2 .The purpose of my calculations is to get accurate gap and band
> structure. Hence I am performing MBJ calculations with the parameters special to semiconductors.
>
> I would like to ask how we should edit case.in0_grr for a mbj calculation in wien2k14.2 ?
>
> I read the user guide and checked the mailing list before, I didn’t find any obvious
> instruction.
>
> In user guide, it was written that: when you perform init_mbj_lapw for the second time, some
> steps must do automatically:
>
> *
> edit case.in0 and change the functional to option XC_MBJ. (ok this step is done).
>
> *
> cp case.in0 case.in0_grr and choose EX_GRR VX_GRR in case.in0_grr.
>
> But this step, when I open the case.in0_grr to check that, at first line, it was written that:
> EX_GRR EC_NONE VX_GRR VC_NONE. Is it a correct edition? Should I remove “EC_NONE VC_NONE” from the first
> line?
>
> I think “EX_GRR EC_NONE VX_GRR VC_NONE” means that correlation energy and potential are
> ignored! But “EX_GRR VX_GRR” alone, means there is a kind of correlation energy and potential as a default
> for usual mbj calculation. Did I realize
> correctly??
>
> You answered the question about the use of the PBE instead of LDA for the energy before:
> http://www.mail-archive.com/wien%40zeus.theochem.tuwien.ac.at/msg13395.html
>
> And it is stated that how we must change case.in0 .
>
> But there is no instruction for case.in0_grr. What should be done for this issue?
>
> (What kind of edition is correct for case.in0_grr? “EX_GRR VX_GRR”? Or “ EX_GRR EC_NONE VX_GRR
> VC_NONE”?? I performed both of them for one structure, and got very different result: different Gaps,
> different total energies and different
> Fermi-energies!!! )
>
> Please help me.
>
> Sincerely
>
> Shima M.Pourrad
>
> PhD student of physics in condensed matter
>
> Science and Research Branch
> Islamic Azad University
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>
>
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>
>
>
>
More information about the Wien
mailing list