Posted by Ungur Liviu on December 13, 2007 at 15:04:50:
In Reply to: Re: Max number of SO states handled by RASSI posted by Per Ake Malmqvist on October 30, 2007 at 14:51:11:
Actually there are several problems with RASSI when trying to compute a large number of states. The first one is the some of the temporary files (TDM or other...) that gets quite large and the error comes from the IO. Is it possible to increase the filesize or the number of files in which this temporary logical unit is splited in. For example the calculation only on Dy atom (TZP) including all the states (2002 SO-states for f9 configuration) needed a total intermediate space of almost 125GB. This job could only be done on a 64-bit machine. The Gd atom needed more than 180GB. If I have a Lanthanide complex, and I wish to take into account all the states, the required disc space is increasing drastically.
The second problem encountered in RASSI, was , I suppose, the diagonalization, which fails. The error in this case is " Non decreasing Von Newman sum...". I still don't understand why this error happens. If I encounter it again, I will submit a bug report.