[ase-users] Unable to restart from a previous minima hopping run

[Rasmus Karlsson]

Dear Andrew,
Thanks for your reply. I'm sorry for not responding earlier.

Together with the administrators of the cluster I'm using, we've been
trying to trace down the origin of the problem. As far as I can tell, the ASE code works as it should (it always works if I just start
it locally on the login node using gpaw-python minima_hopping.py, but it
fails, especially if running on higher numbers of cores (nodes?), when the
job is run using the job scheduler (SLURM)). However, we haven't quite
been able to understand what causes the issue. For now, it seems the job
works (mostly) when running it with e.g. 48 cores, but it crashes
more often when using e.g. 128 cores. As far as I can tell, the problem
is somehow related either to the cluster or to the account rights of my
user on the cluster, and not related to the ASE code
itself. Furthermore, all jobs I need to run right now are running
successfully when using 48 cores instead of 128, so I'm able to
make progress for now.

Thanks,
Rasmus

Peterson, Andrew writes:

> The restart has worked quite robustly on our system. There have been a few
> updates to the minima hopping module since the version you are using, so you
> may want to download the latest ASE. 


-- 
Rasmus Karlsson, PhD student
Applied Electrochemistry
School of Chemical Science and Engineering KTH Royal Institute of Technology
SE-100 44 Stockholm
Sweden