CP2K energy very different from Gaussian03

Megha Anand meghaan... at gmail.com
Mon Sep 5 21:22:59 CEST 2016


Thank you Matt and Davide! I will try your suggestions and hope to get more 
from the experts.

Best regards,
Megha 

On Monday, September 5, 2016 at 1:35:07 PM UTC-4, Matt W wrote:
>
> Hi,
>
> you certainly should get a similar number for the observable reaction 
> energy Fe + S -> FeS. 
>
> But the energies of the individual species will be different. Potentially 
> totally different if using different pseudopotentials. So just looking at 
> the Fe atom energy is not a good idea.
>
> On an actual CP2K note your basis set is too small - change SZV to DZVP 
> for something OK. 
>
> As David T says, best to compare like with like where possible, PADE 
> functional in CP2K is LDA, so probably like SVWN keywords for exchange 
> correlation in Gaussian for comparison.
>
> Matt
>
> On Monday, September 5, 2016 at 6:03:07 PM UTC+1, Megha Anand wrote:
>>
>> Dear All,
>>
>> I am pretty new to CP2K and still learning the basics of the program. I 
>> want to get the energy of the energy of the reaction: Fe(g) + S(g) -> 
>> FeS(g). To begin with, I simply did single point calculation of iron atom 
>> using Gaussian03 and CP2K. Browsing through the previous posts, I figured 
>> out that the ECP basis sets used by most quantum chemists cannot be used in 
>> CP2K. Even if I use different level of theory and basis sets across the two 
>> programs, I would anticipate roughly similar energies if not exactly same. 
>> I used basis set B3LYP/SDD in Gaussian03 and SZV-MOLOPT-SR-GTH in CP2K. The 
>> energies I get from Gaussian is: -123.85372 while in CP2K: 
>> -19.96605887771384. Even if the two programs results in two spin states I 
>> would not expect such large different. 
>>
>> I do not know what I am doing wrong. Any help will be greatly 
>> appreciated. I have included the input below. Apart from these, I have few 
>> more general questions:
>>
>> (a) Is there any recommended combination of pseudopotential and basis set 
>> for complex compounds containing metal iron. There are also: carbon, 
>> oxygen, sulfur, and nitrogen in the system.
>>
>> (b) Should I use GPW or do all electron calculation using GAPW. I would 
>> have used ECP basis set, if I had to use Gaussian03. But I am not sure what 
>> is right way proceed with CP2K.
>>
>> (c) Any suggestion for the functional available in CP2K for studying iron 
>> complexes of high spin multiplicity. 
>>
>> (d) I saw ECP_POTENTIALS file on CP2K Github, is it possible to use ECPs 
>> like SDD, LANL2DZ, etc. in CP2K.
>>
>>
>> Here is my CP2K input:
>>
>> &GLOBAL
>>
>>   PROJECT Fe
>>
>>   RUN_TYPE ENERGY
>>
>>   PRINT_LEVEL MEDIUM
>>
>>   WALLTIME 86400
>>
>> &END GLOBAL
>>
>>
>> &FORCE_EVAL
>>
>>   METHOD Quickstep
>>
>>   &DFT
>>
>>     BASIS_SET_FILE_NAME /share/apps/cp2k/cp2k/tests/QS/BASIS_MOLOPT
>>
>>     POTENTIAL_FILE_NAME /share/apps/cp2k/cp2k/tests/QS/GTH_POTENTIALS
>>
>>     CHARGE 0
>>
>>     MULTIPLICITY 5
>>
>>     UKS T
>>
>>
>>     &MGRID
>>
>>       CUTOFF 400
>>
>>       REL_CUTOFF 60
>>
>>     &END MGRID
>>
>>
>>     &QS
>>
>>       METHOD GPW
>>
>>       EPS_DEFAULT 1.0E-12
>>
>>     &END QS
>>
>>
>>     &SCF
>>
>>       MAX_SCF 500
>>
>>       &OUTER_SCF
>>
>>         EPS_SCF 1.0E-6
>>
>>         MAX_SCF 60
>>
>>       &END OUTER_SCF
>>
>>     &END SCF
>>
>>
>>     &POISSON
>>
>>       PERIODIC NONE
>>
>>       PSOLVER WAVELET
>>
>>     &END POISSON
>>
>>
>>
>>     &XC
>>
>>       &XC_FUNCTIONAL PADE
>>
>>       &END XC_FUNCTIONAL
>>
>>     &END XC
>>
>>
>>   &END DFT
>>
>>
>>   &SUBSYS
>>
>>     &CELL
>>
>>       ABC 7.00 7.00 7.00
>>
>>       PERIODIC NONE
>>
>>     &END CELL
>>
>>
>>     &COORD
>>
>>       Fe    0.0000    0.0000    0.0000
>>
>>     &END COORD
>>
>>
>>     &KIND Fe
>>
>>       BASIS_SET SZV-MOLOPT-SR-GTH
>>
>>       POTENTIAL GTH-PADE-q8
>>
>>     &END KIND
>>
>>   &END SUBSYS
>>
>> &END FORCE_EVAL
>>
>>
>>
>> Thanks,
>>
>> Megha
>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20160905/ae103ca0/attachment.html>


More information about the CP2K-user mailing list