Possible errors in the supplied basis files (data/BASIS_SET and data/ALL_BASIS_SETS)

Vladimir Rybkin rybk... at gmail.com
Thu Nov 10 13:33:26 CET 2016


Dear Daniele,

thank you for the feedback.
We have checked the basis sets. Thus,

1) someone has added a non-optimized f function to the ANO basis in the 
BASIS_SET file: you can see that those are not optimized numbers, but a 
reasonable (?) guess.
0.8000000000 0.7000000000 
0.3000000000 0.4000000000
Thus, the official ANO is the one in the ALL_BASIS_SETS file, identical to 
the one in EMSL.

2) the one in ALL_BASIS_SETS for Li is not valid since it's once basis 
rather than 2. 

We will soon make a fix.

Yours,

Vladimir


среда, 9 ноября 2016 г., 16:05:41 UTC+1 пользователь Daniele Fontanari 
написал:
>
> Hello everyone,
> While browsing the various basis sets and I noticed something odd.
>
> We have the definition of "Si DZ-ANO" (Roos Augmented Double Zeta ANO 
> basis sets) in two files and they differs (they seem both valid but one has 
> 3 exponent sets and the other 4, but on the first 3 exponent sets they are 
> the same)
>
> https://github.com/cp2k/cp2k/blob/master/cp2k/data/BASIS_SET#L108
> https://github.com/cp2k/cp2k/blob/master/cp2k/data/ALL_BASIS_SETS#L155
>
> moreover we have ththe definition of "Li DZVP-GTH-BLYP" (DZVP basis sets 
> optimized for the GTH pseudopotential and the BLYP exchange-correlation 
> functional) in these two files and again they differs, in this case however 
> it seems that the one in ALL_BASIS_SET might not be valid, since it states 
> the existence of 2 exponent set, but it appears that only one exponent set 
> is present (where in the other file it looks fine, and give only one 
> exponent set. The exponents in the set are the same):
>
> https://github.com/cp2k/cp2k/blob/master/cp2k/data/BASIS_SET#L3244
> https://github.com/cp2k/cp2k/blob/master/cp2k/data/ALL_BASIS_SETS#L3972
>
> However I am no expert, so I might be wrong.
>
> Cheers
>
> Daniele
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20161110/5981eb18/attachment.html>


More information about the CP2K-user mailing list