[molpro-user] MCSCF convergence in OPTG

Hrusak Jan hrusak at kav.cas.cz
Tue Aug 6 10:44:59 BST 2013


Hi all,
is there a general strategy how to avoid convergence problems in a MCSCF geometry optimization? While doing a  scan (with rather small geometry changes between the steps) I see after several successful optimization runs the following program termination...

BFGS update of hessian
Geometry optimization using default procedure for command MULTI
Inactive variables:    OX HRK TW11 ACO REQ NUL
                        CXO
Active variables:      HCO HC1 HCH1 HC2 HCH2
Geometry written to block  1 of record 700
Making model hessian for Z-matrix coordinates
Quadratic Steepest Descent - Minimum Search
Optimization point 1

Variable                      Last           Current        Next           Gradient       Hessian

E(MULTI001) / Hartree       0.00000000  -114.61632254     0.00000000
HCO / DEGREE                0.00000000   218.24848639   219.99920570    -0.00011188     0.00005270
HC1 / ANGSTROM              0.00000000     1.09944819     1.08058056     0.03145839     1.66732133
HCH1 / DEGREE               0.00000000    98.00807296    99.06812000    -0.00004758     0.00006127
HC2 / ANGSTROM              0.00000000     1.17378547     1.17231681     0.00438589     2.73744328
HCH2 / DEGREE               0.00000000    36.69990284    38.48723128    -0.00041947     0.00020013
Convergence:                0.00000000  (line search)     0.05939719     0.03014397  (total)
? Error
? No convergence
? The problem occurs in Multi

ERROR EXIT
CURRENT STACK:      MAIN

Kindly
Jan

________________________
RNDr. Jan H R U S A K, CSc.
Academy of Sciences of the Czech Republic
Narodni 3, 117 20 Prague 1, Czech Republic
tel. +420 221 403 267,
Institute +420 266 053 436
fax +420 221 403 476,
mobil +420 724 073 180
hrusak at kav.cas.cz<mailto:hrusak at kav.cas.cz>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20130806/9bd6b3e3/attachment.html>


More information about the Molpro-user mailing list