[gpaw-users] AttributeError: 'NoneType' object has no attribute 'bd'

Thu Nov 21 17:51:01 CET 2013

Ask and Jens --- I just sent you the files off list.

Marcin --- that's one thing I noticed.  When I ran over, say 3 cores, it
would increase my bands by 3 (so from 321 to 324 --- so still divisible by
3), but it would still throw the error.

So for PW mode, I'll just have to do --dry-run, get the number of A.O.s,
set nbands to that, and adjust my parallel keyword dict accordingly.

Thanks guys :)

On Thu, Nov 21, 2013 at 8:25 AM, Ask Hjorth Larsen <asklarsen at gmail.com>wrote:

> Hello
>
> This reminds me that the ability to construct the FD wavefunctions
> from existing LCAO ones when doing set(mode='fd') seems to have been
> lost once more.  But I really should be minding my various current
> projects.
>
> Regards
> Ask
>
> 2013/11/21 Marcin Dulak <Marcin.Dulak at fysik.dtu.dk>:
> > On 11/21/2013 07:44 AM, Jens Jørgen Mortensen wrote:
> >
> > Den 21-11-2013 06:10, Glen Jenness skrev:
> >
> > Ask,
> > Setting nbands=<# of A.O.s> seems to have done the trick!  It's at least
> > started in on the SCF.
> >
> > I went up to 15 cores, and had to play with the parallel options, but
> it's
> > working.
> >
> >
> > Anyway, I'd like to have your script so I can try to reproduce the
> original
> > error.
> >
> > i remember cases where the new automatic algorithm for changing the
> number
> > of bands
> > sets the number of bands to be larger than the number of available lcao
> > orbitals -
> > that's probably the case here. An example below to be run on 4 cores:
> >
> > from ase.lattice import bulk
> > from gpaw import GPAW, PW
> >
> > b = bulk('Ca', 'fcc')
> > b.set_calculator(GPAW(mode=PW(), kpts=(4,4,4)))
> > b.get_potential_energy()
> >
> > Best regards,
> >
> > Marcin
> >
> >
> > Jens Jørgen
> >
> >
> > Glen
> >
> >
> > On Wed, Nov 20, 2013 at 9:12 PM, Ask Hjorth Larsen <asklarsen at gmail.com>
> > wrote:
> >>
> >> Hello Glen
> >>
> >> It could very well be that the band addition mechanism is involved in
> >> this.  But if it is, it would only be covering for some other error.
> >> In the previous LCAO calculation, how many bands did you have?  Was it
> >> equal to the number of atomic orbitals (which can be a bit high but
> >> would not necessarily be a problem in itself) or did you set it
> >> separately?  If it was equal to the number of atomic orbitals, you may
> >> be able to force the number to be the same in the two calculations and
> >> avoid the error.  But I am not sure.  You may wish to send the input
> >> scripts and the output text files.
> >>
> >> Best regards
> >> Ask
> >>
> >> 2013/11/20 Glen Jenness <glenjenness at gmail.com>:
> >> > Dear users,
> >> >  I'm trying to figure out what the following arcane error means:
> >> >
> >> >  File
> >> >
> >> >
> "/home/vlachos/glenjenness/programs/gpaw/build/lib.linux-x86_64-2.7/gpaw/wavefunctions/fdpw.py",
> >> > line 75, in initialize_wave_functions_from_basis_functions
> >> >     lcaoksl, lcaobd = self.initksl, self.initksl.bd
> >> > AttributeError    lcaoksl, lcaobd = self.initksl, self.initksl.bd
> >> >     lcaoksl, lcaobd = self.initksl, self.initksl.bd
> >> > AttributeError: AttributeError: 'NoneType' object has no attribute
> 'bd'
> >> > 'NoneType' object has no attribute 'bd'
> >> >
> >> > I am running a single hydrogen on a Al2O3 surface in PW mode.  I have
> >> > successfully optimized this system with lcao mode, and would like to
> do
> >> > a
> >> > single point in PW mode for a BE comparison.  I am trying this on 16
> >> > cores
> >> > (our nodes have 8 cores each, so two nodes total), and --dry-run=16
> >> > doesn't
> >> > show any errors.  However, when I go to run it, it spits out the
> above.
> >> >
> >> > I can get this message to vanish by varying the number of cores so
> that
> >> > the
> >> > number of atomic orbitals are divisible by my number of cores.
>  However,
> >> > this means using a sub-optimal set of cores based on our cluster setup
> >> > (I'd
> >> > be wasting ~60% of my node resources).
> >> >
> >> > Now, whenever I do PW mode with dry-run, it will show that it adds on
> a
> >> > number of empty bands to fit with the number of cores.  However, there
> >> > is
> >> > now a mismatch between the A.O.'s and bands, and I believe this is
> what
> >> > causes the above error.
> >> >
> >> > So my questions are:
> >> >
> >> > 1)  Is my above statement correct, and if not, then what causes this
> >> > error
> >> > to occur?
> >> > and
> >> > 2)  Is there a way around this?
> >> >
> >> > Thanks!
> >> > Glen
> >> >
> >> >
> >> > --
> >> > Dr. Glen Jenness
> >> > Catalysis Center for Energy Innovation (CCEI)
> >> > University of Delaware
> >> >
> >> > _______________________________________________
> >> > gpaw-users mailing list
> >> > gpaw-users at listserv.fysik.dtu.dk
> >> > https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
> >
> >
> >
> >
> > --
> > Dr. Glen Jenness
> > Catalysis Center for Energy Innovation (CCEI)
> > University of Delaware
> >
> >
> > _______________________________________________
> > gpaw-users mailing list
> > 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
> >
> >
> >
> > --
> > ***********************************
> >
> > Marcin Dulak
> > Technical University of Denmark
> > Department of Physics
> > Building 307, Room 229
> > DK-2800 Kongens Lyngby
> > Denmark
> > Tel.: (+45) 4525 3157
> > Fax.: (+45) 4593 2399
> > email: Marcin.Dulak at fysik.dtu.dk
> >
> > ***********************************
>

-- 
Dr. Glen Jenness
Catalysis Center for Energy Innovation (CCEI)
University of Delaware
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20131121/b98b1196/attachment.html>