Dear All,
We are having issues with simple ground state DFT (PBE functional) calculations performed with the dscf module
WARNING
Symm. of e-repr. not preserved in 56953 cases
Max deviation in the symm. blocks: 826744.679996558
Obviously, the SCF cycle does not converge. This is a pretty large germanium-hydrogen system (488 electrons) with DZP-ECP basis set and a closed shell ground state at S6 symmetry. Smaller systems give similar warnings with orders of magnitude smaller deviations in "symm. blocks" but at the end they seem to be well converged and the results are meaningful. We found similar reports on the forum but no real solution was given (apart from a bug fix in an earlier version of Turbomole). The warning has obviously a lot to do with the specific point group symmetry, and certainly removing explicit S6 symmetry could help, but we would like to keep the point group order as high as we can for computational reasons.
Turbomole v. 6.3.1 was run in MPI parallel mode (8 cores) on a 64 bit machine.
Thanks,
Marton Voros