Dear All,

Georg Schreckenbach wrote:
> looking through the ADF database, I notice that ZORA/TZP/U.5d and
> ZORA/TZ2P/U.5d are identical at the TZP (IV) level. (I haven't yet had
> the chance to check but I assume the same is true for other elements as
> well, and it might or might not apply to the regular basis sets, too.)
> This conforms to the way it used to be in the previous releases where
> 'IV' existed for all elements but 'V' only for the upper part of the
> periodic table. However, it contradicts the current documentation in the
> 'README' file, and I find it to be highly misleading in any case -- we
> found out about this by mere accident only.

Answer:
First, which basis sets are the same in TZP and TZ2P:
The basis sets for the all electron and frozen core lanthanides
from La to Gd are the same in TZP and TZ2P, namely:
La Ce Pr Nd Pm Sm Eu Gd
La.4d Ce.4d Pr.4d Nd.4d Pm.4d Sm.4d Eu.4d Gd.4d
La.5p Ce.5p Pr.5p Nd.5p Pm.5p Sm.5p Eu.5p Gd.5p
The basis sets for the frozen core actinides from Ac to Lr
are the same in TZP and TZ2P, namely:
Ac.5d Th.5d Pa.5d U.5d Np.5d Pu.5d Am.5d Cm.5d Bk.5d Cf.5d Es.5d Fm.5d Md.5d
No.5d Lr.5d
Ac.6p Th.6p Pa.6p U.6p Np.6p Pu.6p Am.6p Cm.6p Bk.6p Cf.6p Es.6p Fm.6p Md.6p
No.6p Lr.6p

As far as I know in all 298 other cases the basis sets in TZP and TZ2P differ.
The reason that sometimes basis sets are the same in the TZP and TZ2P
is the idea to have a complete set of possible basis sets
available at the DZ, TZP, and TZ2P level.
Especially for the lanthanides and actinides one could say
that the TZP is already of TZ2P quality, and it is difficult to add
a sensible extra function in the TZ2P basis sets, or to remove one in the
TZP basis sets. (note that ADF can not handle g-functions in the basis set).

Part of the table in README:
                        | valence | extra functions
                        | orbitals | DZ TZP TZ2P QZ4P
 --------------------------------------------------------------------------
 La-Gd Z = 57-64 | 4f 5d 6s | pf p p 2p2f
 Ac-Lr Z = 89-103 | 5f 6d 7s | pf p 2p 2p

For the lanthanides one could have deduced that the TZP and TZ2P
basis sets are the same if one looks in the table in the README.
For the actinides in the table in the README one can find
that there are two extra p-functions in the TZ2P basis set, and
only one extra p-function in the TZP basis set.
The table is not correct for the TZP frozen core actinide basis sets,
where it should have said that there are two extra p-functions.
The table is correct for the all-electron actinide basis sets.
One should note that the table summarizes the frozen core and all-electron
basis sets in one
line, for an exact description of the basis set one always has to look
in the atomicdata basis set file itself.

Some comments on U.5d.
In the case of U.5d the valence p-functions in the ZORA basis set are
(both in the TZP and TZ2P basis set):
6P 4.45, 6P 3.00, 6P 2.10, and 7P 1.55
Thus three 6P and one 7P function, which means two extra p-functions
compared to a double-zeta 6P description.
Compare this with the valence p-functions in the
non-relativistic TZP basis for U.5d:
6P 4.05, 6P 2.50, 7P 1.47
Here one only has one extra p-function compared to a double-zeta 6P
description.

>From calculations that I have done, I found out that it was important to
have three 6P functions in the U.5d basis set.
In the all-electron basis set this is not so important, because one already has
more flexibility in the core region.
I have chosen to put three 6P functions not only in the TZ2P actinide.5d
basis sets, but also in the TZP actinide.5d frozen core basis set.
This is a change in comparison with the previous adf2000 atomicdata basis sets,
where the TZP (old name IV) actinide.5d frozen core basis sets only had two 6P
functions.

Best regards,
Erik van Lenthe

----------------------------------------------------------------------
Dr. Erik van Lenthe SCIENTIFIC COMPUTING & MODELLING NV
Tel: +31 20 44 47615 Vrije Universiteit, Theoretische Chemie
secretary: 44 47519 De Boelelaan 1083
 fax: 44 47629 1081 HV Amsterdam, The Netherlands
 e-mail: vanlenthe@scm.com http://www.scm.com
----------------------------------------------------------------------
Received on 2002-05-17 11:35:18