forum@abinit.org
Subject: The ABINIT Users Mailing List ( CLOSED )
List archive
- From: "Guillaume Dumont" <dumont.guillaume@gmail.com>
- To: forum@abinit.org
- Subject: Re: [abinit-forum] NaN with npband 2 and when nband > VB+2
- Date: Fri, 2 Feb 2007 15:34:26 +0100
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:references; b=ThQiGXZyRqhIppw446VERnYZwJBuwvCMaAqKiFmIY5iMJk5GAHn20ITnv4WzB1BWs4hJIWNhMaas7J9D5qmEBBP93523JkAZByiVEPtu3Mj9YLTrqd3ieIZ52ByTChTevtPeWc6WeADKS6/FfedJzkV8Ow4CLQjbqf8D0xuAE60=
Dear Joe,
You should be aware that the band/fft paralelisation for testing purpose only. I see in your input file that you are trying to do NON self-consistent (iscf<0) and this feature does not work yet with the band/fft parallelisation (not in 5.2.x nor 5.3.0)
Also if your density file that you get by getden -1 has been generated using the band/fft parallelisation it won't either cause the file isn't written properly. The plane-wave coeffs are distributed over the nodes and you only get a slice of it.
Hope this helps
Guillaume
On 2/2/07, geoffrey.stenuit@tyndall.ie <geoffrey.stenuit@tyndall.ie
> wrote:
Dear Abinit users,
Running abinit-5.2.4 compiled with pathscale (pathf90) on AMD Opteron (64bit), i experience some instabilities with the calculations of the band structure at k=0 when the npfft or npband input flags are different from 1 and when wfoptalg is set to 4. In particular, i found that the eigenvalues become "NaN" when the number of bands that i request at this Gamma point is larger than the number of Valence band + 2 (i.e 128+2=130 in my GaAsN 64-atom supercell). On the other hand, when i request nband= 129 or 130, the programm is giving the correct eigenvalues (but i need to analyse more than 130 bands).
Do you think it is a bug in abinit, or an error when i compiled it, or ?
Please find below the input file i submitted.
Thank you for your help,
Regards,
Joe
-------------------------------------------------------------
# Convergence with respect to the number of k points.
chkprim 0
# mgfft = 96 => nbr of processors: 1 or 2, 3, 4, 6, 8, 12, 16, 24,... ,48,96
### Parallel FLAGS:
fftalg 401
wfoptalg 4
fft_opt_lob 2
npband 2
npfft 1 ## mgfft=40 => 1,2,4,5,10,20 and 40 are OK
iprcch 0
#Definition of the unit cell
#Definition of the unit cell
acell 3*2.1203311987E+01 # This is equivalent to
rprim 1.0 0.0 0.0 # FCC primitive vectors (to be scaled by acell)
0.0 1.0 0.0
0.0 0.0 1.0
xcart # This keyword indicate that the location of the atoms
-1.74374822972885E-19 1.80698275312510E-19 1.09334001556395E-19
-1.74374822972885E-19 1.80698275312510E-19 1.06016559933404E+01
-1.74374822972885E-19 1.06016559933404E+01 1.09334001556395E-19
-1.74374822972885E-19 1.06016559933404E+01 1.06016559933404E+01
1.06016559933404E+01 1.80698275312510E-19 1.09334001556395E-19
1.06016559933404E+01 1.80698275312510E-19 1.06016559933404E+01
1.06016559933404E+01 1.06016559933404E+01 1.09334001556395E-19
1.06016559933404E+01 1.06016559933404E+01 1.06016559933404E+01
9.68225628470105E-03 5.16468510169365E+00 5.16468510169366E+00
-9.68225628470105E-03 5.16468510169365E+00 1.60386268849871E+01
-9.68225628470105E-03 1.60386268849870E+01 5.16468510169366E+00
9.68225628470105E-03 1.60386268849870E+01 1.60386268849871E+01
1.06013045001119E+01 5.29123892677001E+00 5.29123892677001E+00
1.06020074865687E+01 5.29123892677001E+00 1.59120730599107E+01
1.06020074865687E+01 1.59120730599107E+01 5.29123892677001E+00
1.06013045001119E+01 1.59120730599107E+01 1.59120730599107E+01
5.16468510169366E+00 9.68225628470109E-03 5.16468510169366E+00
5.16468510169366E+00 -9.68225628470105E-03 1.60386268849871E+01
5.29123892677001E+00 1.06013045001119E+01 5.29123892677001E+00
5.29123892677001E+00 1.06020074865687E+01 1.59120730599107E+01
1.60386268849871E+01 -9.68225628470105E-03 5.16468510169366E+00
1.60386268849871E+01 9.68225628470109E-03 1.60386268849871E+01
1.59120730599107E+01 1.06020074865687E+01 5.29123892677001E+00
1.59120730599107E+01 1.06013045001119E+01 1.59120730599107E+01
5.16468510169366E+00 5.16468510169365E+00 9.68225628470109E-03
5.29123892677001E+00 5.29123892677001E+00 1.06013045001119E+01
5.16468510169366E+00 1.60386268849870E+01 -9.68225628470106E-03
5.29123892677001E+00 1.59120730599107E+01 1.06020074865687E+01
1.60386268849871E+01 5.16468510169365E+00 -9.68225628470106E-03
1.59120730599107E+01 5.29123892677001E+00 1.06020074865687E+01
1.60386268849871E+01 1.60386268849870E+01 9.68225628470109E-03
1.59120730599107E+01 1.59120730599107E+01 1.06013045001119E+01
7.88922678674774E+00 2.64420766012477E+00 7.88922678674772E+00
7.93032740139436E+00 2.64935519364036E+00 1.85539567930404E+01
7.95231990242752E+00 1.32509920842532E+01 7.95231990242752E+00
7.88922678674774E+00 1.33140851999330E+01 1.85591043265559E+01
1.85539567930404E+01 2.64935519364036E+00 7.93032740139436E+00
1.89681850257071E+01 2.23512696097355E+00 1.89681850257071E+01
1.85591043265559E+01 1.33140851999330E+01 7.88922678674772E+00
1.85539567930404E+01 1.32729845852864E+01 1.85539567930404E+01
2.23512696097355E+00 2.23512696097355E+00 2.23512696097355E+00
2.64935519364036E+00 2.64935519364036E+00 1.32729845852864E+01
2.64935519364036E+00 1.32729845852864E+01 2.64935519364037E+00
2.64420766012477E+00 1.33140851999330E+01 1.33140851999330E+01
1.32729845852864E+01 2.64935519364036E+00 2.64935519364037E+00
1.33140851999330E+01 2.64420766012477E+00 1.33140851999330E+01
1.33140851999330E+01 1.33140851999330E+01 2.64420766012476E+00
1.32509920842532E+01 1.32509920842532E+01 1.32509920842532E+01
2.64420766012477E+00 7.88922678674775E+00 7.88922678674772E+00
2.64935519364036E+00 7.93032740139436E+00 1.85539567930404E+01
2.64935519364036E+00 1.85539567930404E+01 7.93032740139436E+00
2.23512696097355E+00 1.89681850257071E+01 1.89681850257071E+01
1.32509920842532E+01 7.95231990242758E+00 7.95231990242752E+00
1.33140851999330E+01 7.88922678674775E+00 1.85591043265559E+01
1.33140851999330E+01 1.85591043265559E+01 7.88922678674772E+00
1.32729845852864E+01 1.85539567930404E+01 1.85539567930404E+01
7.88922678674774E+00 7.88922678674775E+00 2.64420766012476E+00
7.95231990242752E+00 7.95231990242758E+00 1.32509920842532E+01
7.93032740139436E+00 1.85539567930404E+01 2.64935519364037E+00
7.88922678674774E+00 1.85591043265559E+01 1.33140851999330E+01
1.85539567930404E+01 7.93032740139436E+00 2.64935519364037E+00
1.85591043265559E+01 7.88922678674775E+00 1.33140851999330E+01
1.89681850257071E+01 1.89681850257071E+01 2.23512696097355E+00
1.85539567930404E+01 1.85539567930404E+01 1.32729845852864E+01
# Second: The BS...
#getcell2 -1 # start from the previous acell values
#getxred2 -1 # start from the previous xred values
#prtwf 0 # No wavefunction output is provided
iscf -2 # no self-consistent calculations
getden -1 # because start with the previous computed density
kptopt 0 # read directly nkpt, kpt, kptnrm and wtk (corresponds to the usage before version 2.1)
nband 132 # 6*32=192 ( 2atoms: 6 = 4 VB + 2 CB) en plus pour avoir la conduction
nkpt 1 #5 #nbre of kpts
kpt
0.00 0.00 0.00
tolwfr 1.0d-12
enunit 1 # Will output the eigenenergies in eV
#Definition of the atom types in the cristal
ntypat 3 # There are three types of atom
znucl 7 33 31 # The keyword "znucl" refers to the atomic number of the
# possible type(s) of atom. The pseudopotential(s)
# mentioned in the "files" file must correspond
# to the type(s) of atom. Here, the only type is Silicon.
#Definition of the atoms
natom 64 # There are 64 atoms
typat 1 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2
3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3
3 3 3 3 3 3 3 3
#Definition of the planewave basis set
#ecut 20.0 # Maximal kinetic energy cut-off, in Hartree
ecut 25.0
#ecut 27.5625
#ecut3 30.0
#ecut4 35.0
#ecut 32.0
#Definition of the SCF procedure
nstep 60 # Maximal number of SCF cycles
#toldff 5.0d-6 # A value ten times smaller than tolmxf is suggested (for example 5.0d-6 hartree/bohr)
toldfe 1.0d-10 # Will stop when, twice in a row, the difference
diemac 12.0 # Although this is not mandatory, it is worth to
--
Guillaume Dumont
=========================
guillaume.dumont.1@umontreal.ca
dumont.guillaume@gmail.com
(514) 341 5298
(514) 343 6111 ext. 13279
- NaN with npband 2 and when nband > VB+2, geoffrey . stenuit, 02/02/2007
- Re: [abinit-forum] NaN with npband 2 and when nband > VB+2, Guillaume Dumont, 02/02/2007
- ecutsm\=0 in relax and RF calculations, Marcel Mohr, 02/02/2007
- Re: [abinit-forum] ecutsm\=0 in relax and RF calculations, Anglade Pierre-Matthieu, 02/02/2007
- Re: [abinit-forum] ecutsm\=0 in relax and RF calculations, Marcel Mohr, 02/03/2007
- Re: [abinit-forum] ecutsm\=0 in relax and RF calculations, Xavier Gonze, 02/03/2007
- Re: [abinit-forum] ecutsm\=0 in relax and RF calculations, Marcel Mohr, 02/05/2007
- Re: [abinit-forum] ecutsm\=0 in relax and RF calculations, Anglade Pierre-Matthieu, 02/02/2007
- <Possible follow-up(s)>
- NaN with npband 2 and when nband > VB+2, Geoffrey Stenuit, 02/02/2007
- Re: [abinit-forum] NaN with npband 2 and when nband > VB+2, Francois Bottin, 02/02/2007
- Re: ***SPAM*** Re: [abinit-forum] NaN with npband 2 and when nband > VB+2, Geoffrey Stenuit, 02/02/2007
- Re: [abinit-forum] NaN with npband 2 and when nband > VB+2, Francois Bottin, 02/02/2007
Archive powered by MHonArc 2.6.16.