[molpro-user] rs2 excited state geometry optimizations

[Hans-Joachim Werner]

Dear Jayashree,
I looked at your problem and found several problems:

1.) The main error in your input is that you cannot optimize two  
states at the same time.
You must specify on the state directive in the rs2 input which state  
you want to compute
and optimize (e.g., state,1,1 means the ground state and state,1,2 the  
second state).

2.) There is a conical intersection between the two states, and  
therefore you cannot optimize
the excited state using the standard method (the gradient does not  
become zero at the conical intersection,
and therefore it will not converge). Conical intersections can only be  
optimized using casscf, not
using rs2. See section 40.2.19 of the manual for details.

3.) You should not have uhf in the procedure, and the start directive  
in the casscf input should be removed.
With this directive you force the casscf program to use the UHF  
orbitals, and it happens that it uses
then the uno-cas occupations rather than the ones you have specified.  
This is a bug that has been
fixed in the developers version already. The fix will be backported to  
2008.1 after more extensive testing.
In general it is safer not to use a procedure in geometry optimizations.

4.) The CPMCSCF directives are not needed for rs2 geometry optimizations
(they are only needed for sa-casscf gradients).

5.) If one tries a ground state calculation (using a corrected input),  
the calculation still fails
because it finds at some stage that the casscf and rs2 reference  
coefficients slightly differ. This
is due to the nearly degenerate states which causes convergence of the  
Davidson procedure to be slow.
The problem is avoided by setting option,nstati=2 in the rs2 input. I  
added a more meaningful
error message in case that this problem occurs.

I attach inputs that work with version 2008.1 for
(i) ground state rs2 optimization using a procedure
(ii) ground state rs2 optimization without a procedure
(iii) excited state conical intersection search using casscf

Best regards
Joachim Werner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test1_proc.inp
Type: application/octet-stream
Size: 820 bytes
Desc: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20090303/4f239235/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test1_proc.out
Type: application/octet-stream
Size: 37492 bytes
Desc: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20090303/4f239235/attachment-0001.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test1.inp
Type: application/octet-stream
Size: 766 bytes
Desc: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20090303/4f239235/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test1.out
Type: application/octet-stream
Size: 37428 bytes
Desc: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20090303/4f239235/attachment-0003.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: testmc.inp
Type: application/octet-stream
Size: 1336 bytes
Desc: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20090303/4f239235/attachment-0004.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: testmc.out
Type: application/octet-stream
Size: 45586 bytes
Desc: not available
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20090303/4f239235/attachment-0005.obj>
-------------- next part --------------

-- 
Prof. Hans-Joachim Werner
Institute for Theoretical Chemistry
University of Stuttgart
Pfaffenwaldring 55
D-70569 Stuttgart, Germany
Tel.: (0049) 711 / 685 64400
Fax.: (0049) 711 / 685 64442
e-mail: werner at theochem.uni-stuttgart.de