Availability of MOLOPT basis sets

S Ling lingsa... at gmail.com
Sun Mar 22 22:32:29 UTC 2015

To generate a new pseudopotential for Ca, you may try the ATOMIC code, but 
I guess the current pseudopotential for Ca which was shipped with CP2K 
distribution should be good enough for you. What you need is to produce a 
new basis set for Ca, which is more complete than the DZVP-SR basis set 
provided by CP2K distribution. To do that, you can try with the 
OPTIMIZE_BASIS module. First, you perform a series of molecular 
calculations with a basis set which is close to the complete basis set 
limit. These accurate molecular calculations will serve as your reference. 
Then you choose a proper format of the basis set that you would like to 
optimize and provide initial guess for the Gaussian exponents and/or 
contraction coefficients, and you use the OPTIMIZE_BASIS module to optimize 
these variables by minimizing the electron density difference between the 
basis set that you would like to fit and your reference calculations. I am 
not sure if this explanation is clear enough. If not, I can try to produce 
a basis set for you, either using the same Gaussian exponents in DZVP-SR, 
or using more (and thus more diffused) Gaussian exponents if you believe 
your specific system will really benefit from that. Note the DZVP-SR basis 
set for Ca is already very diffused, with a small exponent of 0.055, which 
I think should be good enough for majority extended systems. Also keep in 
mind that a more complete basis set means much higher computational cost.


On Sunday, March 22, 2015 at 8:52:48 PM UTC, Christopher O'Brien wrote:
> Does anyone know how to generate an optimized basis set/pseudopotential 
> for Ca (specifically of the TZV2P type)? I noticed that the ATOM module may 
> permit me to do this by looking at the test files. Is there instructions 
> for generating a basis/pseudo?  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150322/3191f06d/attachment.htm>

More information about the CP2K-user mailing list