Molpro: conical intersection search

[IM Grant, Science 99]

Hi,

On the subject of conical intersections, I wonder if anyone can help me 
with my search..
I'm just beginning to get to grips with MOLPRO, and I'm trying to locate a 
CI in HCO, but the job crashes with this:

 ? Error
 ? Record not found
 ? The problem occurs in readm

 ERROR EXIT

..at the first FORCE command, because the previous CPMCSCF calculation 
failed to converge:

 ITER. MIC  NCI  NEG     ENERGY(VAR)     ENERGY(PROJ)   ENERGY CHANGE 
GRAD(0)  GRAD(ORB)   GRAD(CI)     STEP       TIME

   1   37   30    0    -113.25491259    -113.25428587    0.00062672 
0.00462640 0.00000010 0.00076643  0.60D-02     11.77

 *** NO CONVERGENCE ***

This happens at varying stages of the do loop, depending on the basis set 
(mainly vdz, vtz), although when I tried it with an sto-3g basis set, the 
job didn't crash, but failed to converge overall before reaching the 50th 
iteration (which is the maximum possible) in the do loop.

If anyone has any pointers, it would be greatfully appreciated..

Thanks,

Ian Grant.


--
Here is the input file:

***,HCO - find Conical Intersection, vtz (extended) basis set, geommetry 
opt for minimum

memory,10,m
file,2,/home/ig8572/docs/conical_intersections/hco/hco_ci_vtz_extended.wfu,
new

basis={
s,c,vtz,,10;
p,c,vtz,,5;
d,c,1.091,0.318;
f,c, 0.761;
s,o,vtz,,10;
p,o,vtz,,5;
d,o,2.314,0.645;
f,o,1.428;
s,h,vtz,,5;
p,h,1.407,0.388;
d,h,1.057;
}


r=2.029833300
rr=1.743194464
tet=86.51022682
Geometry={c;
          o,c,r;
          h,o,rr,c,tet}

                                                     ! Fix the rnn distance 
at the experimentally determined result

thresh,orbital=1.d-5,civec=1.d-5

int;
if(orbital.eq.0) then
rhf;wf,15,1,1;save,2130.2;maxit,70;
endif

maxstep=35						! CI optimisation loop iterations

do i = 1,maxstep

text,--------- Minimize energy difference, step $i of $maxstep ---------
multi;occ,9,2;closed,2,0;
config,csf;
wf,15,1,1;state,3; 
! 23 elecrons, symmetry 1, doblet
weight,0.95,0.95,0.05;
natorb,2140.2;
cpmcscf,grad,1.1,2.1,accu=1.d-7,record=5101.1;
cpmcscf,nacm,1.1,2.1,accu=1.d-7,record=5102.1;
cpmcscf,grad,2.1,accu=1.d-7,record=5100.1;
! Gradient evaluation using CPMCSCF results
! The order here is not important
Force
SAMC,5100.1
CONICAL,6100.1 ! Always put your record on file 1
Force
SAMC,5101.1
CONICAL,6100.1 ! Always put your record on file 1
Force
SAMC,5102.1
CONICAL,6100.1 ! Always put your record on file 1
opt
STEP,0.15,0.2,2
COORD,bmat
UPDATE,bfgs=1
if(optconv.lt.1.d-5) exit
enddo

---


----------------------
IM Grant, Science 98
Ian.Grant.99 at bristol.ac.uk