PC GAMESS/Firefly-related discussion club

Learn how to ask questions correctly

Alex Granovsky
gran@classic.chem.msu.su

>1) Input file - for example for NbCln complexes (n - number of Cl-ligands) - works when the number of ligands (n) is equal 1 or 2. At addition of the third Cl-anion the program cannot read parameters of ECP for it.

Note you must specify ECPs for all, not just symmetry-unique atoms.
The order of ECPs in $ECP group should match that of atoms
after applying symmetry operations to the symmetry-unique ones;
so be careful to look at the atom numbering scheme (found e.g.,
in output of exetyp=check run).

>How to enter ECP parameters in a case when the complex contains more than two identical atoms?

This is actually well-documented in manual. In brief, when you enter
ECP in $ECP goup, you give it a name. This name can be used to specify ECP
for other atoms of the same type, e.g., for Na₃ system:

 $ECP
NA-ECP GEN     10     2
 5      ----- d potential     -----
    -10.0000000        1    175.5502590        
    -47.4902024        2     35.0516791        
    -17.2283007        2      7.9060270        
     -6.0637782        2      2.3365719        
     -0.7299393        2      0.7799867        
 5      ----- s-d potential     -----
      3.0000000        0    243.3605846        
     36.2847626        1     41.5764759        
     72.9304880        2     13.2649167        
     23.8401151        2      3.6797165        
      6.0123861        2      0.9764209        
 6      ----- p-d potential     -----
      5.0000000        0   1257.2650682        
    117.4495683        1    189.6248810        
    423.3986704        2     54.5247759        
    109.3247297        2     13.7449955        
     31.3701656        2      3.6813579        
      7.1241813        2      0.9461106 
NA-ECP
NA-ECP
 $END

> 2) With LANL2DZ the virial ratio in out-file is much greater than 2. Is it normal for ECPs?

Yes, this is normal.

>3) When I use LANL2DZ parameters for NbFn (1 or 2 F-anions), after a first step of SCF calculation (density converged!) the further calculations always stops because of too big value of RMS gradient. In out file there is the following text:
>OPTIMIZATION ABORTED. GRADIENT OUT OF RANGE. MAXIMUM ALLOWED FORCE (FMAXT) = 10.000
>Perhaps, it is necessary to take another potential? If it so, what you would recommend to use in case when for the central atom of a complex the LANL2DZ ECP is used?

This is most likely the sequence of some hidden bugs in your input file.

> And one more: the FMAXT value should not be changed?

It cannot be changed.

Regards,
Alex