[molpro-user] Infinity and beyond in CPHF!!!

Mon Sep 7 19:41:33 BST 2009

Dear Molpro users
 We are trying to run some geometry optimisations on several molecular 
complexes using DF-LMP2. The calculations were done on a Itanium 
computer with the 2006.1 version of Molpro. We got the following message 
in the CPHF section:

Starting Coupled-perturbed Hartree-Fock, convergence threshold= 1.0D-06, 
integral threshold= 1.0D-11

 ITER   MIC  DIIS      VAR         VARC       CPU
 *  1     0     1 Infinity    Infinity        0.99*
   2     0     2 NaN         NaN             1.71

Below are extracts of the .log file and the .out file
We have tried many different options and none of them seem to work.
Any help will be kindly appreciated.

--------------------out file--------------------

 ***,furan
 memory,20,m

 gdirect
 gprint

!,cpu=2
 gthresh,energy=1.d-8
 basis=vdz

 geomtyp=xyz
 geometry={nosym;ANGSTROM;
 4
 GeomXYZ
 O                  0.01811300    2.60092900    0.00000000
 H                  0.16830700    1.61247400    0.00000000
 H                 -0.94718800    2.73973200    0.00000000
 Br                 0.01811300   -5.71884700    0.00000000
 }
 set,charge=-1

 basis={
 default,vdz
 set,mp2fit
 default,vdz/mp2fit
 set,jkfit
 default,vdz/jkfit
 }

 {df-hf,basis=jkfit,locfit_scf=0;accu,14}
 {df-lmp2,DOMONLY=1,interact=1,locfit=1,save=6000.2a}

 geomtyp=xyz
 geometry={nosym;ANGSTROM;
 4
 GeomXYZ
 O                  0.01811300    2.60092900    0.00000000
 H                  0.16830700    1.61247400    0.00000000
 H                 -0.94718800    2.73973200    0.00000000
 Br                 0.01811300   -0.71884700    0.00000000
 }
 set,charge=-1

 basis={
 default,vdz
 set,mp2fit
 default,vdz/mp2fit
 set,jkfit
 default,vdz/jkfit
 }

 {df-hf,basis=jkfit,locfit_scf=0;accu,14}
 {df-lmp2,interact=1,locfit=1,start=6000.2,save=6000.2}
 optg,gradient=1.d-5
 show,energy,dmx,dmy,dmz
........
PROGRAM * OPT (Geometry optimization)     Authors: F. Eckert and H.-J. 
Werner

 *** Long output written to logfile /home/uvi/qf/jhr/Prueba1.log ***

 Geometry optimization using default procedure for command DF-LMP2

 Geometry written to block  1 of record 700

 Making model hessian for cartesian coordinates

 Quadratic Steepest Descent - Minimum Search

 Optimization point  1

 Variable                      Last           Current        
Next           Gradient       Hessian
 E(DF-LMP2) / Hartree        0.00000000 -2648.12958513     0.00000000
 GX1 / ANGSTROM              0.00000000     0.01811300     0.02531587    
-0.00360123     0.00000000
 GY1 / ANGSTROM              0.00000000     2.60092900     2.61296980    
-0.00251705     0.00000000
 GZ1 / ANGSTROM              0.00000000     0.00000000     
0.00000000     0.00000000     0.00000000
 GX2 / ANGSTROM              0.00000000     0.16830700     
0.16223504     0.02638191     0.00000000
 GY2 / ANGSTROM              0.00000000     1.61247400     1.63466704    
-0.02606942     0.00000000
 GZ2 / ANGSTROM              0.00000000     0.00000000     
0.00000000     0.00000000     0.00000000
 GX3 / ANGSTROM              0.00000000    -0.94718800    -0.93633472    
-0.02321478     0.07743510
 GY3 / ANGSTROM              0.00000000     2.73973200     
2.69419191     0.02799422     0.41544290
 GZ3 / ANGSTROM              0.00000000     0.00000000     
0.00000000     0.00000000     1.44502414
 GX4 / ANGSTROM              0.00000000     0.01811300     
0.00612881     0.00043410     4.45576059
 GY4 / ANGSTROM              0.00000000    -0.71884700    
-0.70754075     0.00059225     5.24589755
 GZ4 / ANGSTROM              0.00000000     0.00000000     
0.00000000     0.00000000     5.52586962
 Convergence:                0.00000000  (line search)     
0.10672198     0.02758870  (total)
 ? Error
 ? failure in dsyev
 ? The problem occurs in diag2

 ERROR EXIT
 CURRENT STACK:      MAIN

-----------------------------log 
file----------------------------------------------------------------------------------------------------------
 !LMP2 STATE 1.1 ENERGY             -2648.131234809772
 Time for making Fock matrix (overall):      1.37 sec

 Solving coupled perturbed localization equations

 Number of orbital parameters:                 28
 Norm of right-hand side:                  0.6854D-03
 Gradient of localization functional:      0.7576D-09

 Solving CPL equations. Requested accuracy=  0.10D-05

 ITER      GRAD        VAR
   1    0.91D-03    0.75D-03
   2    0.13D-03    0.20D-04
   3    0.32D-05    0.29D-05
   4    0.51D-06    0.78D-07

 Time for CPL C-matrix:                      0.00 sec
 Time for CPL solve:                         0.00 sec
 Time for CPL (total):                       0.00 sec

 CPHF trial vector read from record  4102.2

 Local density fitting used:    Locfit=1, Rdomaux= 5.00, Invert=1

 Thresholds used in cphf: THRAO= 1.0D-11, THRMO= 1.0D-11, THROV= 
1.0D-12, THRSW= 1.0D-10, THRPROD= 1.0D-11

 Starting Coupled-perturbed Hartree-Fock, convergence threshold= 
1.0D-06, integral threshold= 1.0D-11

 ITER   MIC  DIIS      VAR         VARC       CPU
 *  1     0     1 Infinity    Infinity        0.99*
   2     0     2 NaN         NaN             1.71

Regards,
-- 

_Carlos Manuel Estévez Valcárcel
_Departamento de Química Física
Facultade de Química
Universidade de Vigo
36310 Vigo

Tlf: +34 986 813813
Fax: +34 986 812321
E-mail: cestevez at uvigo.es
<mailto:cestevez at uvigo.es>Web: http://webs.uvigo.es/cestevez