Endless rescaling after optimization is done
Hello developers,
First of all thank you very much for sharing this code, so far i have had no problems to speak of but now I have run into a strange issue and was wondering if you could assist me.
I currently train the MTP for a two component system (AlN) on 600 2x2x2 supercell configurations with random atomic displacements and cell deformations and sometimes, after the BFGS optimizations is done, the final rescaling part just won't finish... I have only had this now for 600 training configurations and not with less.
Even more strange is that it only happens for a specific number of BFGS steps. I wrote a small python program, that runs a number of steps via maxiter and then saves and benchmarks the current MTP, so that i can see how everything evolves. When i change the number of steps from currently 10 per iteration of the python program to anything else, it doesn't get stuck at the rescaling part.
So this seem to have something to do with a specific combinations of parameters or something because if i run even 1 more/less BFGS step the parameters are slightly different and the rescaling works...
Any help would be greatly appreciated!
best regards Simon Fernbach
Used files: