[gpaw-users] 'TypeError' error in plane-wave mode calculation

[Marcin Dulak]

Hi,

On 07/17/2014 04:31 PM, Alfredo Palace Carvalho wrote:
> Hi,
>
> [wrong thread, but useful answer nevertheless...] :)
> Thanks. It seems that's it! The timing differences are reasonable.
> gpaw 0.10 does take a bit more time (the SCF part takes ~ 1.25 x 
> longer than in gpaw 0.9, and the statistics are not very good (2 
> calculations of each case)) which may be reasonable.
> The changes between version 0.10 and 0.9 did seem to have made the 
> calculations a bit slower for this particular case (I didn't check if 
> this behavior extends to other different systems) but I wouldn't find 
> these differences noteworthy, I know that some changes sometimes 
> worsen some calculations for the benefit of improving a larger number 
> of others.
yes, this was the goal for changing the default eigensolver and mixer.
For me your case makes 28 steps with rmm-diis and 30 with dav (in 
serial), though per scf iterations they will be slower than the old 
rmm-diis.

Best regards,

Marcin
> It was the very striking difference in the number of iterations that I 
> thought it was abnormal. But what you said explains it now.
>
> Regards,
> Alfredo
>
> On 2014-07-17 12:58, heineahansen wrote:
>> Hi,
>>
>> It seems the default number of wave function updates per scf step
>> changed from two to one at some point.
>> I think you should compare the time required to converge the density
>> and wave functions rather than the number of scf steps.
>> Point 3 here may be of interest:
>> https://wiki.fysik.dtu.dk/gpaw/documentation/convergence.html.
>>
>> Best Regards,
>> Heine
>>
>> On Jul 17, 2014, at 1:05 PM, gpaw-users-request at listserv.fysik.dtu.dk 
>> wrote:
>>
>>> Date: Thu, 17 Jul 2014 11:56:15 +0100
>>> From: Alfredo Palace Carvalho <ajpalace at uevora.pt>
>>> To: Tristan Maxson <tgmaxson at gmail.com>
>>> Cc: gpaw-users at listserv.fysik.dtu.dk
>>> Subject: Re: [gpaw-users] 'TypeError' error in plane-wave mode
>>>     calculation
>>> Message-ID: <5407374a65f5d6eabbf31ede122e50f6 at uevora.pt>
>>> Content-Type: text/plain; charset="utf-8"; Format="flowed"
>>>
>>> Well, it doesn't even get to the point to start the SCF iteration
>>> procedure. It blows just before that part begins.
>>> I'm attaching the output.
>>>
>>> On 2014-07-16 19:48, Tristan Maxson wrote:
>>>> Check your output, ?is it failing to converge? ?The code is failing
>>>> when you attempt to write the calculator which means it may be
>>>> converging but there could be an issue with writing a PW calculation
>>>> to a file. ?
>>>>
>>>> On Wed, Jul 16, 2014 at 9:16 AM, Alfredo Palace Carvalho
>>>> <ajpalace at uevora.pt [3]> wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> I've been using gpaw in 'fd' mode mostly, but now I was attempting
>>>>> the 'PW' mode. However I haven't been able to. I get the following
>>>>> error message:
>>>>>
>>>>> Traceback (most recent call last):
>>>>> ? File "./sp_PBE_mol_PW.py", line 26, in <module>
>>>>> ? ? system.calc.write(name + '.gpw')
>>>>> ? File "/home/programs/gpaw-0.10.0.11364/lib/python/gpaw/paw.py",
>>>>> line 810, in write
>>>>> ? ? gpaw.io.write(self, filename, mode, cmr_params=cmr_params,
>>>>> **kwargs)
>>>>> ? File
>>>>> "/home/programs/gpaw-0.10.0.11364/lib/python/gpaw/io/__init__.py",
>>>>> line 162, in write
>>>>> ? ? w.add('PotentialEnergy', (), hamiltonian.Etot + 0.5 *
>>>>> hamiltonian.S,
>>>>> TypeError: unsupported operand type(s) for *: 'float' and
>>>>> 'NoneType'
>>>>> GPAW CLEANUP (node 2): <type 'exceptions.TypeError'> occurred.
>>>>> ?Calling MPI_Abort!
>>>>>
>>>>> I'm using the 0.10.0.11364 version of gpaw and the 0.9.11271
>>>>> version of setups.
>>>>> I'm attaching the script which produces the error. I can't
>>>>> understand what's wrong with it.
>>>>> BTW, in regard to parallelization issues, I'm running in 16 cores.
>>>>> I'm assuming that the choice of nbands=64 is safe!
>>>>>
>>>>> Thanks in advance.
>>>>> Cheers,
>>>>> Alfredo
>>>>> _______________________________________________
>>>>> gpaw-users mailing list
>>>>> gpaw-users at listserv.fysik.dtu.dk [1]
>>>>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users [2]
>>>>
>>>>
>>>>
>>>> Links:
>>>> ------
>>>> [1] mailto:gpaw-users at listserv.fysik.dtu.dk
>>>> [2] https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>>>> [3] mailto:ajpalace at uevora.pt
>>> -------------- next part --------------
>>> An embedded and charset-unspecified text was scrubbed...
>>> Name: alprazolam.out
>>> URL: 
>>> <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20140717/71851d5b/attachment-0001.ksh>
>>>
>>> ------------------------------
>
> _______________________________________________
> 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

***********************************