Comparison with Gaussian DFT calculation

Matt W MattWa... at gmail.com
Thu Jan 26 19:44:18 CET 2017


Hi,

you need to turn off periodicity in both the CELL section (which controls 
the generation of atoms), and in the POSSION section, where electrostatics 
are calculated.

&POISSON
   PSOLVER WAVELET
   PERIODIC NONE
&END

and periodic none in the CELL section. Your molecule needs to be in the 
center of the cell, not near the edges.

For a molecule that is not very polar, the artifacts from running 
periodically vs isolated should go very quickly as the box size increases.

Matt

On Thursday, January 26, 2017 at 5:42:03 PM UTC, Babgen Manookian wrote:
>
> Hey Matt,
>
> Thank you for the answer. I am working on running the calculations with 
> bigger cell size at the moment. In the meantime, I am wondering if there is 
> a way to turn off the periodicity in CP2K. I tried using the keyword 
> PERIODIC NONE but got the exact same results.
>
> Any help will be much appreciated it.
>
> Thank you,
> Bobby
>
> On Monday, January 23, 2017 at 4:59:51 PM UTC-5, Matt W wrote:
>>
>> Hi,
>>
>> it looks like you cell is very small. Unlike Gaussian CP2K tends to run 
>> with atoms periodically repeated inside a box. Your's is about 5x5x5 
>> Angstrom^3. Probably you have some steric clashes between images of the 
>> molecule, and likely these are different between the conformers.
>>
>> Matt
>>
>> On Monday, January 23, 2017 at 8:05:24 PM UTC, Babgen Manookian wrote:
>>>
>>> Hello all,
>>>
>>> I am new to CP2K and am working on making the transfer from Gaussian09. 
>>> I am trying to match energy calculations between the two programs but am 
>>> having trouble so I hope someone can help me.
>>>
>>> System: Just for testing purposes, I want to calculate the energy 
>>> difference between chair and boat conformations for a cyclohexane molecule.
>>>
>>> Gaussian09: for my Gaussian calculations I am using DFT with BLYP 
>>> functional and a 6-31G* basis set. Using this setup I obtain an energy 
>>> difference of ~34 kJ/mol which is in fair agreement with a textbook value 
>>> of ~30 kJ/mol.
>>>
>>> CP2K: For CP2K I am using DZVP-GTH-BLYP basis sets and GTH-BLYP 
>>> pseudopotentials for both carbon and hydrogen atoms. Pretty much all other 
>>> input parameters are set to default values or in accordance to the ones 
>>> given in the Si test files found on the CP2K website. The input file is 
>>> attached for further detail if needed. With this set-up I am obtaining an 
>>> energy difference of ~288 kJ/mol. Clearly this value is way off the 
>>> expected value and my Gaussian-obtained value and I am totally lost as to 
>>> why.
>>>
>>> I have done these calculations with different values for cutoffs and 
>>> different basis sets and in all cases I get a similar result where the 
>>> difference between the two methods is roughly ten-fold. I am wondering if 
>>> there may be something trivial that I am doing that is causing this.
>>>
>>> Any sort of insight will be truly appreciated.
>>>
>>> Thank you very much in advance.
>>>
>>> Babgen Manookian
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20170126/f91c7867/attachment.html>


More information about the CP2K-user mailing list