iOpenShell » Q-Chem specific questions » Splitting up a frequency calculation

Splitting up a frequency calculation

Page: 1

Author Post
Member
Registered: Jun 2018
Posts: 2
Often times there are time limits on supercomputer clusters that are well below the time needed for a frequency calculation to be performed on a large molecule with a more complicated DFT method. I suspect not, but is there any way to pick up where a previous unfinished frequency calculation left off? For example, an optimization job unfinished is easy since one can use the latest optimization iteration's geometry as the starting guess for a new submission. If Q-Chem could add a feature like this, my guess is it would be very helpful to many people.
Administrator
Registered: Oct 2017
Posts: 15
So, in other words, you would like to have a restart option for a frequency job? Restart for geometry optimization is trivial, because you can start from the latest geometry from the output, but for other calculations it may lead to additional storage requirements.
Member
Registered: Jun 2018
Posts: 2
Yes just like geometry optimization, where you can always pick up where you left off. My issue is the cluster I submit to has a 48 hour limit that I cannot finish certain frequency analysis jobs under, and I am seeing if there is any way I can pick up where a timed out frequency calculation left off
Administrator
Registered: Oct 2017
Posts: 15
The manual has "MOPROP_RESTART" keyword, which should do this. I do not think that its usage would be trivial in your case, since it relies on the previous scratch files. So, you need to run the first calculation with "save" option and periodically copy the scratch files to your home directory. At restart, you will need to recreate scratch from the previous job and try this option. Test this scheme on something small and quick to see whether it works. I have never tried it.

Page: 1

iOpenShell » Q-Chem specific questions » Splitting up a frequency calculation