[Wien] QTL-B message in scf2 after "x lapw2 -qtl"

Peter Blaha pblaha at theochem.tuwien.ac.at
Wed Mar 13 07:29:13 CET 2013


Hi,

Yes, you have got a "ghostband" at 2.1 Ry, which is way above the occupied states.
Therefore you did not get any warnings during scf (only occupied states are needed)
and the scf cycle is fine.
Only the DOS at energies above 2 Ry will be affected and may be wrong (not very accurate).

You can probably fix this, when you set the L=0 energy parameters of the second atom
to -1.46 (as before) and 1.3 (1 Ry higher than before, maybe even at 2.0). It should hardly affect
the valence bands (O-2s), but improve the description of the "O-3s" states at very
high energies.

PS: Remember: the "linearization" is good only near the expansion energy, so a state at
2 Ry is not well described by a radial wave function expanded at 0.3 Ry, even when
one adds the energy derivative (like in a Taylor series).

Regards

Am 13.03.2013 04:00, schrieb Sanae Fujita:
> Dear WIEN2k-users,
>
> I'm a beginner of ab-initio calculation.
>
> I calculated MgO's DOS by WIEN2k_11.1, and DOS's graph looks well for me,
>
> but I wonder whether my calculation works well.
>
> Because I found message as below in scf2 file (below).
>
> ------
>
> QTL-B VALUE .EQ.84.31874 in Band of energy2.15443ATOM=2L=0
>
> Check for ghost bands or EIGENVALUES BELOW XX messages
>
> Adjust your Energy-parameters for this ATOM and L (or use -in1new switch), check RMTs!!!
>
> ------
>
> I consulted User's guide(U.G.) or mails in mailing list, but I can't find exact answer.
>
> My question is:
>
> (1) My process (I wrote below),in which I evaluated ghost band may consist of eigenvalues from EIG00014 to EIG00016, is right? Or any other good method? Or is it true ghost
> band?
>
> (2) I think that if ghost band is below Fermi energy, I can judge it is true ghost band by checking that number of eigenvalue is equal to NE(electron number in valence
> bond) in in2 file. Am I right?
>
> (3) If (1) is 'Yes', how should I adjust energy parameter in in1 or any other parameter?
>
> The condition of this calculation is as below:
>
> ------
>
> 225_Fm-3m,7.96bohr cubic,
>
> Atom1(Mg)-Pos. is (0,0,0),Atom2(O)-Pos. is (0.5,0.5,0.5),
>
> RMT(Mg)=1.88, RMT(O)=1.88(this is the value set by w2web's RMT page, so I think it's appropriate),
>
> RKmax=7,Gmax=12,Emin/Emax=-9/2.5Ry,
>
> (4,4,4)k-points-div(in scf1, :KPT=8)
>
> No spin-polarized calc.,
>
> Core: Mg's 1s, O's 1s (number of electrons in semi-core and valence band=16).
>
> ------
>
> My process is below:
>
> After "run_lapw -ec 0.0001 -NI "
>
> and before "x lapw2 -qtl", I had no QTL-B message in outpput2 file.
>
> But after "x lapw2 -qtl" to make data of DOS,
>
> I met themessage implying ghost band in scf2 or output2 file, as cited above.
>
> In help032 file, I found some strange point cited below.
>
> I think Q(UE)=84.319 in Atom2's L=0 line in the band of 2.15443eV implying ghost band.
>
> I don't know exact criteria of Q(UE)'s value to become ghost band,
>
> but it's 400% of Atom2's s-like charge.
>
> ------
>
> BAND#14E=2.15443WEIGHT= 0.0000000
>
> L= 020.122931042.55284.3191355.684 -1110.907-120.309
>
> L= 10.000000.0000.0000.0000.0000.000
>
> L= 24.429304.3540.0760.0000.0000.000
>
> D-EG:4.429304.3540.0760.0000.0000.000
>
> D-T2G:0.000000.0000.0000.0000.0000.000
>
> L= 30.000000.0000.0000.0000.0000.000
>
> L= 40.322950.3220.0010.0000.0000.000
>
> L= 50.000000.0000.0000.0000.0000.000
>
> L= 60.001640.0020.0000.0000.0000.000
>
> ------
>
> I tried to find another evidence that BAND14 is ghost band. But I can't find.
>
> Dateils are below:
>
> In scf2 or scf file, I found Fermi energy and eigenvalue of last iterarion.
>
> -------
>
> :FER: F E R M I - ENERGY(TETRAH.M.)=0.29301
>
> ------
>
> -------
>
> :EIG00001:-5.0222432-2.5763741-2.5763741-2.5763741-0.9749722
>
> :EIG00006:0.29301420.29301420.29301420.64139791.4552772
>
> :EIG00011:1.45527721.45527721.74233101.87492471.8749247
>
> :EIG00016:2.08474702.08474702.0847470
>
> ------
>
> I also found band14's energy range 1.77603~2.15443 in output2.
>
> ------
>
> band141.77602844732.15443355350.000000D+00
>
> ------
>
> So I think eigenvalue from EIG00014 to EIG00018 may belong to ghoat band.
>
> But I can't find any strange points from these eigenvalues, because EIG00014-EIG00018 are unoccupied states (higher than Fermi energy 0.29301Ry) as I described at my
> question (2).
>
> So I wonder whether band 14 is ghost band.
>
> Anyway, I tried to delete the "QTL-B" massage in scf2 file after "x lapw2 -qtl".
>
> My in1_st file is cited below, I had not changed any parameters from default setting.
>
> -------
>
> WFFILEF= 0.50000(WFFIL, WFPRI, ENFIL, SUPWF)
>
> 7.00104 (R-MT*K-MAX; MAX L IN WF, V-NMT
>
> 0.3040(GLOBAL E-PARAMETER WITH n OTHER CHOICES, global APW/LAPW)
>
> 00.300.000 CONT 1
>
> 0-5.570.001 STOP 1
>
> 10.300.000 CONT 1
>
> 1-3.120.001 STOP 1
>
> 0.3030(GLOBAL E-PARAM...
>
> 0-1.460.002 CONT 1
>
> 00.300.000 CONT 1
>
> 10.300.000 CONT 1
>
> K-VECTORS FROM UNIT:4-9.02.537emin/emax/nband
>
> -----
>
>  From Atom2's L=0's APW-lo line, I can find the energy parameter -1.46 is appropriate(; less than 0.2Ry from EF).
>
> Energy parameter of Atom2's L=0's APW-lo and LO are -1.46,0.3,each,
>
> so they are enough apart from each other,I think.
>
> (ref; http://www.wien2k.at/reg_user/faq/qtlb.html)
>
> As an experiment, I tried to change Atom2's L=0's LO's energy parameter to 0.09 from 0.3,
>
> butafter "x lapw2 -qtl", I got almost same message as below .
>
> -----
>
> QTL-B VALUE .EQ.85.28334 in Band of energy2.15371ATOM=2L=0
>
> Check for ghostbands or EIGENVALUES BELOW XX messages
>
> Adjust your Energy-parameters for this ATOM and L (or use -in1new switch), check RMTs!!!
>
> -----
>
> I am confused and can't be sure where to alter in in1 files.
>
> If you need more information, I will send.
>
> Any advice I appreciate. Please help me.
>
> Best regards,
>
> S.Fujita
>
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>

-- 
-----------------------------------------
Peter Blaha
Inst. Materials Chemistry, TU Vienna
Getreidemarkt 9, A-1060 Vienna, Austria
Tel: +43-1-5880115671
Fax: +43-1-5880115698
email: pblaha at theochem.tuwien.ac.at
-----------------------------------------


More information about the Wien mailing list