dscf - problem with 6d basis set?

[kulhanek]

Hi all,

I have faced following problem with dscf from Turbomole 6. I have downloaded 6-31G* basis set from EMSL and used it for a single point calculation on HF level of theory. Everything was fine until I have switched to 6d using bp keyword from define. Then the calculation stopped with an error message after scf was converged:

    anisotropy=     408.681831

   2161 is greater than mxcao=(ndi4*ndi19)=   2160 - use higher ndi19 !


 MODTRACE: no modules on stack

 <makcao> sao/cao dimension overflow


rundimensions in control files are following:

5d case:
$rundimensions
   dim(fock,dens)=1846680
   natoms=180
   nshell=840
   nbf(CAO)=1920
   nbf(AO)=1800
   dim(trafo[SAO<-->AO/CAO])=2160
   rhfshells=1

6d case:
$rundimensions
   dim(fock,dens)=1846680
   natoms=180
   nshell=840
   nbf(CAO)=1920
   nbf(AO)=1920
   dim(trafo[SAO<-->AO/CAO])=2520
   rhfshells=1

So it seems to me that dscf for some reason does not read correct value of dim(trafo[SAO<-->AO/CAO]) from control file. It rather calculates the value as it would be 5d case. The same problem occurs in version 5.10. When we used version 5.6 some time ago everything worked fine.

Any suggestion what can be wrong?

Thanks in advance,

          Petr

PS: I would like to mention that it is not restarted calculation. Every run was prepared from scratch using the same version of define as used dscf.

   

[uwe]

Hi,

the calculation itself is correct, just the final property section (population analysis, 3D data for visualization, etc.) is not possible with Cartesian basis functions, and hence dscf stops at this point. This will be fixed in Turbomole 6.1.

6d/10f ($pople CAO instead of the default) is meanwhile almost only needed to compare with (very) old results from some old versions of other quantum chemistry programs, and there usually only the absolute energy is important. The energy that comes out of your dscf job is correct, so it can be used.

Regards,

Uwe