[molpro-user] MCSCF and save,ref= card

[Peter REINHARDT]

Dear MOLPRO users and maintainers,
while trying to calculate MCSCF wavefunctions with 26 electrons in 18 
active orbitals I encounter the following strange behaviour of MOLPRO:

when asking for a simple MCSCF with 

! MCSCF septet with many open shells, symmetry B1u
!
multi;
closed, 7,2,4,0,5,2,2,0;
wf,70,5,6;

all is going well with about 250 seconds/iteration. The job is finished 
within half an hour and 4 iterations in D2H.

However, when asking for 
! MCSCF septet with many open shells, symmetry B1u
!
multi;
closed, 7,2,4,0,5,2,2,0;
save,ref=2150;
wf,70,5,6;

for saving the configurations on a permanent file for a restart,
the job takes about 15 hours (!) for the first iteration, doing what?
There are these 'new' lines in the output 

 Maximum number of shells:             5
 Maximum number of spin couplings:   637

 Storage for configuration lists12277771, coupling coefficients7418564, address lists*******

and final configurations (and total energies) are different:

with save,ref=2150 (after 7 iterations):

 CI vector
 ---------
 22+0 22 2+ 2+ 22+0 2 2+ +      0.9533840
 22+0 22 2+ 2+ 22+2 2 0+ +     -0.0748168
 22++ 22 -+ 2+ 22++ 2 -+ +     -0.0571668

 TOTAL ENERGIES                     -2912.42558696

without save,ref=2150:

 CI vector
 ---------
 2220 2+ 2+ 2+ 2220 + 2+ +      0.9471155
 2220 2+ 2+ 2+ 2222 + 0+ +     -0.0733663
 2220 2+ 2+ 2+ 2022 + 2+ +     -0.0533393

 TOTAL ENERGIES                     -2912.43052852

Cpu times are (on a 64bit Opteron 2.4 GHz)


 PROGRAMS   *       MULTI     MULTI   RHF-SCF       INT
 CPU TIMES  *      2149.72      6.71      0.32      2.60


 PROGRAMS   *       MULTI     MULTI   RHF-SCF       INT
 CPU TIMES  *     50168.32      6.28      0.34      2.60


Any hint to this behaviour/problem(?) is highly welcome!

Sincerely yours,
   Peter Reinhardt

--
      v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v
     >  Peter Reinhardt              |                             <
     >  Lab. Chimie Theorique        |                             <
     >  Universite Paris VI          |                             <
     >  4 place Jussieu, case 137    |  Tel.: +33(0)1 44 27 9657   <
     >  F -- 75252 Paris, France     |  Fax.:               4117   <
     >          email: Peter.Reinhardt at lct.jussieu.fr              <
     >          Web: http://www.lct.jussieu.fr/reinhardt           <
      ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^