[molpro-user] Error in conical intersection optimization

[Boyke Schönborn]

I'm currently running CASSCF calculations on azobenzene and try to optimize a 
conical intersection structure. The input I used is:

memory,200,m
file,2,azocon_68_dz.orb
gprint,civector,basis
gdirect
geomtyp=xyz
geometry={
nosym,noorient,angstrom
   24

 C    -0.403687    -3.482982     0.427838
 C    -1.042322    -3.216027     1.643955
 C    -1.968012    -4.114800     2.167372
 C    -2.270994    -5.287977     1.507194
 C    -1.642072    -5.569209     0.305018
 C    -0.726903    -4.670854    -0.219202
 N    -0.863654    -2.051441     2.427292
 N     0.005631    -1.179979     2.164719
 C     0.000000     0.000000     1.393510
 C     0.000000     0.000000     0.000000
 C    -0.003353     1.238312    -0.641688
 C     0.017591     2.429608     0.060199 
 C     0.016579     2.411572     1.446809
 C     0.000000     1.197325     2.103638 
 H    -0.248359    -4.898202    -1.158150  
 H    -1.859429    -6.484450    -0.221906  
 H    -2.989885    -5.974233     1.924776 
 H    -2.441876    -3.866629     3.102496
 H    -0.015277     1.263432    -1.720315 
 H     0.041147     3.367182    -0.471936  
 H     0.024283     3.332131     2.007838  
 H    -0.008906     1.147670     3.178931
 H     0.278290    -2.668129     0.189421  
 H     0.000000    -0.943101    -0.544500  
}

BASIS
default=STO-3G,c=6-31g(d),
SPDF,n,6-31g(df);C;
S,n,0.0438
END

{multi
noextra
maxiter,100
closed,45  
occ,53
state,2
start,2145.2
orbital,2146.2
CPMCSCF,NACM,1.1,2.1,accu=1.0d-3,record=5100.1   
CPMCSCF,GRAD,1.1,spin=0,accu=1.0d-7,record=5101.1 
CPMCSCF,GRAD,2.1,spin=0,accu=1.0d-7,record=5102.1 
}

{Force
SAMC,5100.1             !compute coupling matrix element
CONICAL,6100.1}         !save information for optimization of conical 
intersection

{Force
SAMC,5101.1             !compute gradient for state 1
CONICAL,6100.1}         !save information for optimization of conical 
intersection

{Force
SAMC,5102.1             !compute gradient for state 2
CONICAL,6100.1}         !save information for optimization of conical 
intersection


{optg,maxit=100,startcmd=multi
}


The calculation breaks down before the first optimization step with the  
following error:

 Quadratic Steepest Descent - Minimum Search
 ? Error
 ? Error in assigning eigenvector
 ? The problem occurs in dis1

Before that error the mrogram seems not to be able to read the gradient and 
NACM information from the records properly (many 0.0000s or even NaNs).

Do you have any idea how to get the calculation running?

Thanks,
Boyke Schönborn