Hi all,
I’m trying to run some PES scan calculations but they keep OOM crashing even though I’ve set mem_total to well under the allocated memory for the job. Can someone please provide insight to this without actually seeing the input?
Does the problem manifest with a geometry optimization at the same level of theory with the same molecule?
No it doesn’t, I think it might be due to a poor SCF guess at that given step since (typically) if I restart from that point or do a geometry optimization from there, the PES is not continuous or the structure changes a decent bit. Is there a way to clear the cache and restart the SCF guess at each given point on the PES scan?
For geometry opt, GUESS_ALWAYS = TRUE will force Q-Chem to generate a new SCF guess at each new geometry, rather than using the previous MOs. (This probably works for PES scans but I don’t know that for sure.) Is that what you mean?
It seems strange to me that the SCF is causing memory problems.
I’ll give that a try to see if it works. I’m also not sure why the memory problems are arising. I’m currently trying to run on 16 cores with 29G requested in my slurm script and mem_total set to 24000, which I’ve tried throttling to 20000 but still get OOM killed.
For what it’s worth, this is happening consistently across several different PES scans I’m trying to do, it doesn’t seem localized to one single PES scan.
Can you post an input file? If it’s a correlated wave function method there may be some other memory settings that you need. I can’t imagine that this is an SCF problem.