Welcome to a tutorial on atomate2 in connection with universal ML potentials#
This tutorial will illustrate some of the basic features of atomate2. This tutorial has been written by J. George and A. Naik from BAM.
It is build based on workflows that can be executed with universal ML potentials such as M3GNet and CHGNet. Please be aware that they are not very accurate for these tasks.
If you want accurate results, you could for example train our own ML potential for a specific composition. We currently provide a GAP interface but any other ML potential that is connected to ASE should be interfaced with a few lines of code.