CatLearn
CatLearn copied to clipboard
MLMIN bug with ASE 3.19
There is a major bug in MLMIN when used with ASE 3.19. It behaves like previous NEB bug (if im not mistaken) The bug is: the iteration of geo_opt using GPR in line 283 doesn't works well
I have compared them when used with ASE 3.17 and 3.19 the one that optimized quickly is done with ASE 3.17
Some context: the GPR MDMIN part of the catlearn doesn't work if used with ase 3.19
Several root cause: the ml_opt.converged
method or mml_opt.run
method
So i found out that ase 3.19 has different MDMIN.run
function than ase 3.17 in the optimizer class
https://gitlab.com/ase/ase/-/blob/3.19.0/ase/optimize/optimize.py#L267 https://gitlab.com/ase/ase/-/blob/3.17.0/ase/optimize/optimize.py#L193
But the MDMIN.converged
are the same
I have tinkered on it and fixed it in my computer
Thank you for posting this issue and #91 and #93 I hope we can get this reviewed soon. As for MLMin, there is an alternative built natively into ASE called GPMin (https://wiki.fysik.dtu.dk/ase/ase/optimize.html?highlight=gp#ase.optimize.GPMin). In the meantime, I recommend anyone here to try GPMin.