Nickel cell optimization

Alex nedo... at gmail.com
Mon Sep 29 22:00:50 CEST 2014


Hi Matt,

Thanks for the reply. The way it is specified in CP2K is quite confusing to 
me, to be honest. I am used to unit cell + translation vectors + # 
repetitions combination. Incidentally, this setup was actually suggested by 
another CP2K user from here, and it yielded per-atom nickel cohesive energy 
quite close to the experimental value (-4.422 eV from my calculations). The 
same for aluminum yielded -2.9 eV, which isn't great, but also close.
Now I wonder how that's possible, if the setup is wildly incorrect...

On Monday, September 29, 2014 2:20:06 AM UTC-6, Matt W wrote:
>
> Hi,
>
> you aren't running bullk Ni, but a Ni atom in a large box. See the 
> definition of multiple unit cell in the CELL section
>
> MULTIPLE_UNIT_CELL {Integer} {Integer} {Integer}
> Specifies the numbers of repetition in space (X, Y, Z) of the defined 
> cell, assuming it as a unit cell. This keyword affects only the CELL 
> specification. The same keyword in SUBSYS%TOPOLOGY%MULTIPLE_UNIT_CELL 
> should be modified in order to affect the coordinates specification.  [Edit]
>
> Why you aren't getting any output, I'm not sure. Maybe the memory cost is 
> too high and your system is swapping?
>
> Matt
>
> On Monday, September 22, 2014 11:26:18 PM UTC+1, Alex wrote:
>>
>> Hi everyone. I'm trying to optimize a periodic (6x6x6) lattice of nickel 
>> using either CELL_OPT or GEO_OPT (input attached). The simulation's been 
>> running for several days on 8 cores, no SCF updates, it just hangs at
>>
>>
>>  *******************************************************************************
>>  ***                     STARTING GEOMETRY OPTIMIZATION                   
>>    ***
>>  ***                           CONJUGATE GRADIENTS                       
>>     ***
>>
>>  *******************************************************************************
>>
>>  Spin 1
>>
>>  Number of electrons:                                                     
>>      9
>>  Number of occupied orbitals:                                             
>>      9
>>  Number of molecular orbitals:                                           
>>      26
>>
>>  Spin 2
>>
>>  Number of electrons:                                                     
>>      9
>>  Number of occupied orbitals:                                             
>>      9
>>  Number of molecular orbitals:                                           
>>      26
>>
>>  Number of orbital functions:                                             
>>     26
>>  Number of independent orbital functions:                                 
>>     26
>>
>>  Extrapolation method: initial_guess
>>
>>
>>  SCF WAVEFUNCTION OPTIMIZATION
>>
>>   Step     Update method      Time    Convergence         Total energy   
>>  Change
>>   
>> ------------------------------------------------------------------------------
>>
>>
>>  BFGS minimizer doesn't seem like an option with this system size, so I 
>> used CG... Am I doing something seriously wrong?
>>
>> Thank you.
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140929/a9a5953f/attachment.html>


More information about the CP2K-user mailing list