Firefly and PC GAMESS-related discussion club

 
Learn how to ask questions correctly  
 
 

 

Thomas
thomaspijper@hotmail.com

As with any calculation, the choice of basis set influences the result. This also goes for the Hessian -- force constants may vary with different basis sets.

The correlation consistent basis sets were designed to recover dynamic electron correlation energy in post-Hartree-Fock methods. As Masoud said, they were also designed to provide systematic convergence to the CBS limit. As to their quality and relation to Pople basis sets, many people have found that cc-pVDZ usually isn’t as accurate as 6-311G(d,p) or 6-31G(2d,p), though you may find differently depending on the system under investigation. cc-pVTZ on the other hand usually preforms better than an equally 'heavy' Pople basis set such as 6-311G(3df,2pd). Comparisons can be found in the literature. And, as Masoud said, you can compare the sets yourself using a smaller system similar to the system you’d like to study.

If your system is small enough to make it possible to use cc-pVTZ, you can do so. If you need a lighter basis set, a Pople basis set is often a better choice. Alternatively, you can use a different basis set. One possibility is Ahlrich’s and Weigand’s def2-SVP or def2-TZVP which are economical while accurate, but which lack general purpose diffuse functions (instead, they are sometimes augmented with the diffuse functions from the correlation consistent sets). Another possibility are the polarization consistent (pc) basis sets from Jensen, which are somewhat similar to the correlation consistent (cc) sets in design but which are optimized for use in HF and DFT methods (as opposed to the correlation consistents sets which are optimized for use in post-HF methods). pc-2 is comparable to cc-pVTZ in terms of computation time, but can be more accurate than cc-pVTZ at the HF or DFT level. pc-1 is one 'step' smaller and is comparable to cc-pVDZ in terms of computation time. pc-0 is even smaller, though probably too small to produce meaningful results.

Of course, there are many other sets you can consider.

Hope this gives you some ideas for choosing a basis set.


Kind regards,
Thom



On Tue Jan 15 '13 7:39pm, Siddheshwar Chopra wrote
--------------------------------------------------
>Dear Sir,
>I have been busy understanding the methods to speed up the Hessian calculations. One of the information I came across is the use of CORRELATION CONSISTENT BASIS SETS. Sir, is there any information available that indicates the sort of error induced due to their use; when compared to Pople's basis set like say 6-31G?
>Sir, another thing I wanted to know is that IS the use of basis set needed in HESSIAN calculation too? I mean does the choice of basis set play any role in Hessian calculation?

>Regards,
>S. Chopra