Firefly and PC GAMESS-related discussion club


 
Learn how to ask questions correctly  
 
 
We are NATO-free zone
 


Re^5: CASSCF and basis set extrapolation

sanya
sanya@photonics.ru

>I'm attaching 5 input files for steps: single point S0 (for orbitals), S0 opt, S1 opt, conical intersection search, scan. It's a research in progress, so I don't want to upload the exact molecule. Yet without it I know it may be hard to optimize the options. It's a conjugated system with C1 symmetry, and I am studying it's E/Z isomerization, after photo-excitation. I think I can rely on sto-3g at least as far as not excluding a barrier of rotation, when the molecule is going through isomerization by it's S1 PES ? Or not finding a barrier, when there is none ? The exact size of the barrier is not of importance.

My experience in the study of E/Z photoisomerization says that minimal basis set is never sufficient, especially for conical intersections. At conical intersections, the shape of the molecule is severely distorted, and the basis set should be sufficiently flexible to allow for these distortions; otherwise, you can miss the necessary stationary point or conical intersection.
For example, a comprehensive CASSCF/XMCQDPT study of photoisomerization of stilbene was performed by Alex (http://pubs.acs.org/doi/abs/10.1021/ct400647w). The authors use cc-pVTZ basis set. For qualitative results, double-zeta basis set with polarization functions is OK.

Now, your inputs. For active space (8 electrons, 8 orbitals), CISTEP=ALDET in $mcscf and DISTCI= in $DET is better than CISTEP=GUGA. Your molecule is really small (134 electrons), but you need some important keywords to deal with it.

If you use a multicore computer, set:

$smp load=0 call64=.t. smppar=.t. $end

For cluster add this line (actually, I do not remove this line in single-node calculations, it does no harm):

$p2p p2p=.t. dlb=.t. $end (you can add mixed=.t. if the nodes of your cluster differ from each other)

This line is mandatory:

$trans mptran=2 dirtrf=.t. aoints=dist altpar=.t. mode=112 $end

This line speeds up CASCI step:

$ciinp castrf=.t. $end

This line is for CISTEP=GUGA if you prefer to use it:

$gugem pack2=.t. $end

For CASSCF (that is, _complete_ rather than _restricted_ active space SCF) calculations, _remove_ fors=0 and IEXCIT key. You need _all_ excitations within the given active space. The default is fors=1 (aka CAS).

With these keys, you'll be surprised how fast CASSCF can be.

[ Previous ] [ Next ] [ Index ]           Fri Apr 24 '15 3:48am
[ Reply ] [ Edit ] [ Delete ]           This message read 504 times