The ABINIT Users Mailing List ( CLOSED )

[Francois Bottin <Francois.Bottin@cea.fr>]

Dear Joe,
As Guillaume said, this part is under development. Some functionalities 
of Abinit are not yet available in the framework of BandFFT as the 
Non-self-consistent calculations (thank you Guillaume for this point), 
but also I/O, spin polarization... It remains a lot of work. You can fix 
one of these points if you want :-) !
I don't really understand your problem. You use a number of band 
processors (npband) higher than the number of valence bands in your 
system (nband). Note that, if it is the case, the number of band and FFT 
processors (npband and npfft) HAVE TO BE COMMENSURATE with the number of 
bands and FFT planes in each direction (nband and ngfft).
Regards
Francois

Geoffrey Stenuit 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 is different from 1 and 
> wfoptalg set to 4. In particular, i found that the eigenvalues become 
> "NaN" when the number of band 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 in attachement 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
>
>  


--
##############################################################
Francois Bottin                    tel: 01 69 26 41 73
CEA/DIF                            fax: 01 69 26 70 77
BP 12 Bruyeres-le-Chatel         email: Francois.Bottin@cea.fr
##############################################################