next up previous
Next: Random Up: qchb Previous: Frozen Core/Frozen Virtual Approximation


The ACES II program [1] from the University of Florida has a few special tricks to keep in mind.

Two blank lines:
Input files need to have two blank lines at the very end, or strange errors will result.
Memory specification:
At least for the latest QTP version under AIX, the MEMORY keyword gives the number of 8-byte words, e.g., 40000000 is 40 million words or 320 MB.
Excited state symmetry:
ACES II prints the symmetry of the excitation, not of the final excited state. So, if you have an A2 state that undergoes excitation to become a B1 state, then ACES will say that the final state is B2, not B1, because the excitation process that takes you from A2 to B1 is B2 (A2 x B2 = B1).
Optimization via energies:
For certain methods (e.g., B-CCD) there are no analytic gradients available. Optimizations may still be performed in the old ACES II via METHOD=6 and in the new ACES II via GRAD_CALC=NUMERICAL.
ACES II runs out of disk space:
This happens very frequently for coupled-cluster computations. You might try ABCDTYPE=AOBASIS to get ACES II to do the four-virtual-index integrals in the AO basis so that they don't need to be stored on disk as transformed integrals.
ACES II gives me crazy energies or can't read my Z-matrix or claims 'Specified charge and multiplicity are impossible':
One potential reason for this is that ACES II requires you to give two-letter atomic symbols in all caps. So, lithium is LI not Li. If you give it Li, it will think it's a dummy atom.
ACES II gives strange analytic second derivatives:
There is an occasional bug with analytic second derivatives. To double-check your results, make sure that you have six frequencies (five for a diatomic) which are very close to zero (less than one wavenumber or so).

next up previous
Next: Random Up: qchb Previous: Frozen Core/Frozen Virtual Approximation
David Sherrill 2004-12-06