[gpaw-users] Funny electronic band structure

Wed Feb 4 16:35:18 CET 2015

On 02/04/2015 04:09 PM, Darío Fdez-Pello Lois wrote:
> Dear Jens Jørgen, dear all,
>
> Yes, the primitive unit cell solved the inconsistency!
please note also that e_f = calc.get_fermi_level() should be taken from 
the first (regular SCF) calculations,
not the fixdensity one. Another thing is that the old Mo setup 
(setups={'Mo': '6'}) won't produce the correct
band structure and it should not be used. Another hint: converging 
unoccupied bands is faster with the eigensolver='cg'
than with the default eigensolver='dav'.

Best regards,

Marcin
>
> Many thanks to you all!!
>
> Cheers,
>
>    Darío.
>
> 2015-02-04 14:12 GMT+01:00 Jens Jørgen Mortensen <jensj at fysik.dtu.dk 
> <mailto:jensj at fysik.dtu.dk>>:
>
>     On 02/03/2015 10:59 AM, Darío Fdez-Pello Lois wrote:
>
>         Dear all,
>
>         Thanks for your replies, indeed it should be a colon after
>         'size', I mistyped it somehow.
>
>         The requested figures are attached, please let me say that
>         Mo_band.png comes from the QE and ELK calculations while
>         MoBandstructure.png comes from GPAW.
>
>
>     It looks like you have a cubic cell with two atoms.  Try with the
>     primitive unit cell:
>
>     >>> from ase.lattice import bulk
>     >>> mo = bulk('Mo')
>
>     See here:
>
>     https://wiki.fysik.dtu.dk/ase/ase/structure.html#ase.lattice.bulk
>
>     Jens Jørgen
>
>
>         In addition, here are the scripts:
>         -------SCF
>         import numpy as NP
>         from ase.lattice.cubic import BodyCenteredCubic
>         from gpaw import GPAW,PW,MethfesselPaxton,Mixer
>         from ase.io <http://ase.io> <http://ase.io> import read
>
>         k = 10
>         ecut = 600
>         N = 1
>         configs = read('MoStrain.traj@:')
>         aa = NP.array([config.cell[0, 0] for config in configs])
>         a = aa[-1]
>
>         mo = BodyCenteredCubic(size=(N,N,N),symbol='Mo',latticeconstant=a)
>         mo.calc = GPAW(mode=PW(ecut), xc='PBE', kpts={'size':
>         (k,k,k),'gamma': True}, setups={'Mo': '6'},
>         occupations=MethfesselPaxton(0.01), mixer=Mixer(0.1, 1, 100),
>         txt='bulk.txt')
>         mo.get_potential_energy()
>         mo.calc.write('bulk.gpw')
>
>         --------BAND
>         import numpy as NP
>         from ase.io <http://ase.io> <http://ase.io> import read, write
>
>         from ase.lattice.cubic import BodyCenteredCubic
>         from ase.dft.kpoints import ibz_points, get_bandpath
>         from gpaw import GPAW, PW, MethfesselPaxton
>         import matplotlib
>         matplotlib.use('Agg')
>         import matplotlib.pyplot as plt
>
>         num_bands=12
>
>         calc = GPAW('bulk.gpw', fixdensity=True, usesymm=None,
>         convergence={'bands': num_bands}, maxiter=600, txt='band.txt')
>         points = ibz_points['bcc']
>         G = points['Gamma']
>         H = points['H']
>         N = points['N']
>         P = points['P']
>         kpts, x, X = get_bandpath([G, H, N, G, P], calc.atoms.cell, 200)
>         calc.set(kpts=kpts)
>         calc.get_potential_energy()
>         e_kn = NP.array([calc.get_eigenvalues(k) for k in
>         range(len(kpts))])
>
>         # Plot the band structure
>
>         e_f = calc.get_fermi_level()
>         e_kn -= e_f
>         emin = -10 #e_kn.min() - 1.0
>         emax = 10 #e_kn[:, num_bands].max() + 1.0
>
>         plt.figure(figsize=(5, 6))
>         for n in range(num_bands):
>             plt.plot(x, e_kn[:, n])
>         for p in X:
>             plt.plot([p, p], [emin, emax], 'k-')
>         plt.plot([0, X[-1]], [0, 0], 'k-')
>         plt.xticks(X, ['$%s$' % n for n in ['\Gamma', 'H', 'N',
>         '\Gamma', 'P']])
>         plt.axis(xmin=0, xmax=X[-1], ymin=emin, ymax=emax)
>         plt.xlabel('k-vector')
>         plt.ylabel('E - E$_F$ (eV)')
>         plt.title('Bandstructure of Mo')
>         plt.savefig('MoBandstructure.png')
>
>         --------------------
>         END of scripts
>
>
>         Best regards,
>
>            Darío
>
>
>         2015-02-03 8:19 GMT+01:00 Jussi Enkovaara
>         <jussi.enkovaara at csc.fi <mailto:jussi.enkovaara at csc.fi>
>         <mailto:jussi.enkovaara at csc.fi <mailto:jussi.enkovaara at csc.fi>>>:
>
>
>             Hi Dario,
>             could you be a bit more specific what's wrong with the band
>             stucture (preferrably with a picture of correct /
>         incorrect one),
>             e.g are some bands fine but others not?
>
>             In principle, your script below looks fine, but could you also
>             provide the full scripts for SCF and band structure
>         calculations?
>
>             Best regards,
>             Jussi Enkovaara
>
>
>             On 2015-02-03 00:38, Darío Fdez-Pello Lois wrote:
>
>                 Dear all,
>
>                 My question is related to the calculation of the
>         electronic
>                 band structure
>                 of molybdenum, which I was successfully able to
>         reproduce both
>                 using a full
>                 potential code (ELK) and a plane wave one (Quantum
>         Espesso).
>                 Unfortunately,
>                 the same doesn't happen when I use GPAW (with PW mode)
>                 resulting a quite
>                 different structure.
>                 Here are the relevant (I guess) lines, of course the
>                 obligatory parameter's
>                 convergence were peviouslly done:
>
>                 [ ... ]
>                
>         mo=BodyCenteredCubic(size(N,N,N),symbol='Mo',latticeconstant=a)
>                 mo.calc = GPAW(mode=PW(ecut),
>                        xc='PBE',
>                        kpts={'size'(k,k,k),'gamma': True},
>                        setups={'Mo': '6'},
>                        occupations=MethfesselPaxton(0.01),
>                        mixer=Mixer(0.1, 1, 100),
>                        txt='bulk.txt')
>                 [ ... ]
>
>                 thereafter I run the band structure script similar to
>         the one
>                 given in the
>                 tutorial (with obvious changes: bcc instead of fcc, etc.)
>
>                 I am very confused at this point and I hope some of
>         you would
>                 kindly give
>                 me some insight.
>
>                 Best,
>
>                    Darío.
>
>
>
>
>                 _______________________________________________
>                 gpaw-users mailing list
>         gpaw-users at listserv.fysik.dtu.dk
>         <mailto:gpaw-users at listserv.fysik.dtu.dk>
>                 <mailto:gpaw-users at listserv.fysik.dtu.dk
>         <mailto:gpaw-users at listserv.fysik.dtu.dk>>
>         https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
>
>
>             _______________________________________________
>             gpaw-users mailing list
>         gpaw-users at listserv.fysik.dtu.dk
>         <mailto:gpaw-users at listserv.fysik.dtu.dk>
>             <mailto:gpaw-users at listserv.fysik.dtu.dk
>         <mailto:gpaw-users at listserv.fysik.dtu.dk>>
>         https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
>
>
>
>         _______________________________________________
>         gpaw-users mailing list
>         gpaw-users at listserv.fysik.dtu.dk
>         <mailto:gpaw-users at listserv.fysik.dtu.dk>
>         https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
>
>
>
>
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20150204/9eb9eabd/attachment.html>