[molpro-user] Failure in large CASPT2(23,19) run

[Vadim V. Kislov]

Dear All, 
We have encountered a memory problem during of CASPT2(23,15) calculation of C2H3O2 species. We are using a dual Itanum 2 1.5GHz workstation with 8Gb of memory and 8 Gb of swap space.

Input looks like

***,C2H3O2 TS 9
memory,1000,m;
file,2,c2h3o2_9.wfu;
geometry={angstrom;
 c,                   0   ,-0.01897, -0.0666 ,  0.4391
 c,                   0  , -1.36099, -0.18654, -0.11713
 h,                   0  ,  0.05533, -0.03293,  1.51564
 h,                   0  , -2.2241 ,  0.09429,  0.42556
 h,                   0 ,  -1.45278, -0.35638, -1.17026
 o,                   0,    0.50073,  1.0631 , -0.21201
 o,                   0,    0.97344, -0.84162, -0.16122}
basis=vtz
int
!rhf;occ,16;closed,15;wf,31,1,1;
!mcscf;occ,19;closed,4;wf,31,1,1;orbprint,10;
rs2;occ,19;closed,4;core,4;wf,31,1,1;

In the end I have the following output:

1PROGRAM * RS2 (Multireference RS Perturbation Theory)     Authors: H.-J. Werner
 (1993), P. Celani (1998)


 Convergence thresholds:  THRVAR = 1.00D-08  THRDEN = 1.00D-06

 Number of optimized states:  1  Roots:   1
 Number of reference states:  1  Roots:   1

 Reference symmetry:               1   Doublet
 Maximum shell inside CICON        7

 Maximum number of shells reduced from  7 to  6
 Maximum number of spin couplings:    132

 Reference space:    71955 conf   305760 CSFs
 N elec internal:    71955 conf   305760 CSFs
 N-1 el internal:   157950 conf  1337700 CSFs
 N-2 el internal:   306735 conf  5245240 CSFs
Number of electrons in valence space:                     23
 Maximum number of open shell orbitals in reference space:  7
 Maximum number of open shell orbitals in internal spaces: 11


 Number of core orbitals:           4 (   4 )
 Number of active  orbitals:       15 (  15 )
 Number of external orbitals:     143 ( 143 )

 Molecular orbitals read from record     2140.2  Type=MCSCF/NATURAL (state 1.1)

 Integral transformation finished. Total CPU:  16.20 sec, npass=  1  Memory used
:   5.77 MW

 Number of p-space configurations:  49

 Reference wavefunction optimized for reference space (refopt=1)

 State     Reference Energy
   1        -227.20596745
 Number of blocks in overlap matrix:     2   Smallest eigenvalue:  0.11D-01
 Number of N-2 electron functions:     225
 Number of N-1 electron functions: 1337700

 Number of internal configurations:               305760
 Number of singly external configurations:     191291100
 Number of doubly external configurations:       2301585
 Total number of contracted configurations:    193898445
 Total number of uncontracted configurations:53707038840

 Weight factors for SA-density in H0:    1.000000

 FIMAX=  0.29D+00  FXMAX=  0.11D-04  DIAG=  F  F  NOREF=1  NOINT=0  IHPPD=2

 Nuclear energy:                     112.31304506
 Core energy:                       -233.39134100
 Zeroth-order valence energy:        -17.12532307
 Zeroth-order total energy:         -138.20361901
 First-order energy:                 -89.00234844

 There are only all-active singles.
 Contracted Singles code is turned off.


 Using uncontracted singles

 noint is 2

 Diagonal Coupling coefficients finished.               Storage: 1477293 words,
CPU-Time:      1.48 seconds.
 Energy denominators for pairs finished in 0 passes.    Storage: 1411722 words,
CPU-time:      0.00 seconds.

  ITER. STATE  ROOT      NORM       CORR.ENERGY   TOTAL ENERGY   ENERGY CHANGE
     DEN1      VAR(S)    VAR(P)      TIME
    1     1     1     1.06866120     0.00000000  -227.20596745     0.00000000
 -0.65884541  0.69D-01  0.11D+00  9509.93


 Disastrous lack of store, CIGPS



I was able to run CASPT2 calculations with smaller active space (9,9) where the number of configurations just sligthly smaller. 


I will gladly appreciate any suggestions how to override this problem (if possible). Also I would like to know what causes this problem (lack of disk space, memory or system swap). 

Thanks.  


Vadim V.Kislov
Postdoctoral Research Associate,
Department of Chemistry and Biochemistry,
Florida International University
11200 SW 8th Street, Miami, FL 33199
e-mail: kvv_bne at mail.ru
tel: 305-348-1945

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20050525/44181288/attachment.html>