[molpro-user] Inconsistent memory?

zhendong zhao zzhao at olemiss.edu
Mon Jun 29 19:24:05 BST 2009


Dear Molpro Users,

I am trying to do MCSCF calculations. The below input works using 1
core, but it does not work using 2 or more cores.

***,Input file generated by gabedit;
Memory,50m;
Gprint,basis;				! Print basis information 
Gprint,orbital;				! Print orbitals in SCF
and MCSCF 

geomtyp=xyz
geometry={ANGSTROM;
10 ! number of atoms 
GeomXYZ
C,    0.00000000,  0.00000000,  0.00000000
C,   -1.24354248, -0.40869640,  0.00709616
C,   -2.39121354,  0.61767383,  0.03895941
C,   -3.63475602,  0.20897743,  0.04605559
H,   -4.44706865,  0.93543429,  0.06860813
H,   -3.85793712, -0.85779958,  0.02941201
H,   -2.16803244,  1.68445084,  0.05560303
H,   -1.46672358, -1.47547341, -0.00954746
H,    0.81231264, -0.72645686, -0.02255256
H,    0.22318110,  1.06677701,  0.01664362
}
cartesian;
basis=6-31G(d,p)
{rhf
save,2101.2
PRINT,2
}
put,molden,rhf.molden;
 
LOCALI,BOYS;
ORBITAL,2101.2;
OCC,15;
CORE,4;
GROUP,1.1,-15.1;
GROUP,16.1,-90.1;
NOORDER;
put,molden,localize.molden;orbital,2101.2
 
merge;
orbital,2101.2,local(boys);
move,1.1,90.1;
rotate,7.1,11.1,90;
rotate,8.1,12.1,90;
rotate,16.1,45.1,90;
rotate,17.1,58.1,90;
rotate,18.1,63.1,90;
rotate,19.1,81.1,90;
rotate,20.1,82.1,90;
print,2;
save,4102.2;
put,molden,alter.molden;orbital,4102.2
{multi;
occ,20;
closed,10;
wf,30,1;
start,4102.2;
maxit,64;
}
put,molden,multi.molden;
optg;
put,molden,opt.molden;

When the job runs in parallel, it will gives the error message:

 Number of orbital rotations:    1500     ( 100 Core/Active  700
 Core/Virtual   0 Active/Active  700 Active/Virtual) Total number of
 variables:     65004


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

 USED MEMORY IN cislow:         1952647   1977757
 FREE MEMORY IN cislow:        48047253  48022143
 ? Error
 ?  Inconsistent memory
 ? The problem occurs in check_address

 GLOBAL ERROR fehler on processor   0


What is wrong with my input, or is this a bug?

Thank you,

ZZ



More information about the Molpro-user mailing list