Molcas Forum

rozgonyit

Member

From: Budapest, Hungary

Registered: 2024-10-02

Posts: 1

analytical caspt2 gradient with nonzero IPEA shift

Dear Users,

I am running CASPT2 analytical gradients computations for excited states with MOLCAS. With zero IPEA shift, this always works fine. However, using the default (0.25) IPEA shift, the CASPT2 routine stops for some geometry. It computes the CASPT2 energies but then in the PROPERTY SECTION it stops with the message:
" CI iteration for non-invariant CASPT2 did not converge..."

The input for the caspt2 that I use is:

&CASPT2
SHIFT=0.0
IMAGINARY=0.200
IPEASHIFT=0.25
MAXITER=200
MULTISTATE= 9 1 2 3 4 5 6 7 8 9
OUTPUT=BRIEF
PRWF=0.1
GRDT
rlxroot = 7
CORT

&MCLR
THRESHOLD=0.000100

Any explanation and suggestion what to do in order to work with nonzero IPEA shift?
Thanks in advance!