Transition properties

Ronen Shai sronen at
Sun Dec 1 22:50:17 GMT 2002

I've encountered various problems with computing transition properties
(of the same symmetry) and outputing them on a cube. I would be very
thankful for any help.

1) To output a transition density on a cube I first did MULTI with a few
states, then CI;NOEXC;DM,RECORD to save the transition matrix. I then used
this record as input to a CUBE command and selected the desired transition
density. So far it's OK. The problem is this:
I believe that computed dipole moments with the cube density should be
the same as the transition dipole moments reported in the log file. Please
correct me if I'm wrong in this. When I do this, I find that the
values  are NOT the same. I did  128x128x128 cube and tried  various BRAGG
settings between 1 to 5.

2) I'd like to do a bigger active space in MULTI. I have no problem with
the bigger MULTI calculation.  But when I do CI;NOEXC I get the ERROR
MESSAGE: TOO MANY SPIN COUPLINGS. I  have more then 1100 or so which is the limit. This
in spite of the NOEXC.  In fact I'm not interested in the CI at all. I only
use it to output the transition matrix with DM. Is there any way around this?

3) In fact I'm going to calculate an effective "transition electric
potential"  from the transition density. So I thought an alternative  way
to do it is to MATROP with the operator for an electric potential. But
MATROP gives an ERROR message for LOAD,name,OPER,POT,x,y,z.

4) I believe that I should be able to check a few points of
the  cube in (1) by using MATROP to compute the TRACE of the DENSITY multiplied by  a
DELTA(x,y,z) Operator. But again I get an ERROR message for LOAD,name,OPER,DELT,x,y,z.

5) I used MATROP to calculate the trace of the transition matrix and I got
0.03; I believe a transition matrix trace should give zero for orthogonal
states(correct me if I'm wrong). If so isn't this a bit too large numerical
deviation from zero?

Many thanks in advance,
Shai Ronen

School of Chemistry
Tel Aviv University

More information about the Molpro-user mailing list