[molpro-user] error in CASPT2 calculation

[Chen Hui]

Accidently, I found a message from Prof. Werner in 2004:

There is one unresolved problem with the atlas blas/lapack (i4 version): 
the dsyev diagonalization routine crashes on opteron (only for large 
matrices). This problem can be avoided by adding "olddiag2" to the 
FTCFLAGS in CONFIG: FTCFLAGS="blas1 blas2 blas3 lapack olddiag2" After 
doing this, delete src/util.o and remake the program.

I use opteron system with i8 and 64bit compilation of molpro2006. So is 
the above message related to my problem?

Chen Hui wrote:
> Hi, Molpro-Users,
> I encountered an error when I was using molpro2006 to do CASPT2 
> calculation. The output is:
>
> ******
> ...
> Reference wavefunction optimized for reference space (refopt=1)
>
> State     Reference Energy
>   1       -2843.65168572
> ** ACML error: on entry to DSYEV  parameter number  8 had an illegal 
> value
> ******
>
>
> This error message appeared after I increased the memory assignment, 
> before the output is:
>
>
>
> ******
> ...
> Reference wavefunction optimized for reference space (refopt=1)
>
> State     Reference Energy
>   1       -2843.66569756
>
> Number of blocks in overlap matrix:     2   Smallest eigenvalue:  
> 0.10D-01
> Number of N-2 electron functions:   10201
> Number of N-1 electron functions:  339768
>
> Number of internal configurations:               169884
> Number of singly external configurations:     101930400
> Number of doubly external configurations:     459060150
> Total number of contracted configurations:    561160434
> Total number of uncontracted configurations:27472536234
>
> Weight factors for SA-density in H0:    1.000000
>
> Using g3 option
>
> FIMAX=  0.10D+00  FXMAX=  0.53D-03  DIAG=  F  F  NOREF=1  NOINT=0  
> IHPPD=2
>
> Nuclear energy:                    4853.30173939
> Core energy:                      -4951.98885294
> Zeroth-order valence energy:       -141.24447341
> Zeroth-order total energy:         -239.93158695
> First-order energy:               -2603.73411061
>
>
> Using contracted singles
>
> Number of contracted N-1 electron functions:     120595
> Number of contracted internal configurations:       144
>
>
> Diagonal Coupling coefficients finished.               
> Storage:56512028 words, CPU-Time:   1492.84 seconds.
> Energy denominators for pairs finished in 0 passes.    Storage: 
> 7579466 words, CPU-time:      0.02 seconds.
>
> A level shift of 1.00 is applied.
>
>  ITER. STATE  ROOT      NORM       CORR.ENERGY   TOTAL ENERGY   ENERGY 
> CHANGE       DEN1      VAR(S)    VAR(P)      TIME
>    1     1     1     1.06194494    -0.06194494 -2843.72764250    
> -0.06194494    -3.10762919  0.62D-01  0.73D+00 26953.35
> insufficient memory available - require             1381725450  have
>             612843167
> the request was for real words
>
> ERROR EXIT
> CURRENT STACK:      CIPRO  MAIN
> ******
>
>
> Is this problem due to the ACML library used? Could I use the molpro's 
> own optimised lib to solve this?
>
>
>


-- 
Dr. Chen Hui
Department of Organic Chemistry
Hebrew University of Jerusalem
The Hebrew University, Givat Ram Campus
Jerusalem, 91904,Israel
Email: chenh at yfaat.ch.huji.ac.il
phone: 02 6585069
fax:   02 6584033