Tutorials

autoplex tutorials written by Christina Ertural (christina.ertural@bam.de).

The user is advised to have a general familiarity with the following software packages and tools:

• VASP

• the machine-learned interatomic potential (MLIP) framework that is aimed to be used

• the Materials Project framework

• atomate2

• ase

• basic Python knowledge

The tutorials are aimed to demonstrate the usage of autoplex to generate MLIPs and benchmark it to DFT results. By the end of these tutorials, you should be able to:

• use autoplex out of the box with default settings

• customize the workflow settings

• use any of the submodule workflows to build your own workflows

Tutorial table of content

• Before you start using autoplex
• Out-of-the-box workflow
  • General workflow
  • Now start the workflow
  • Workflow management
  • Output and results
• Generating reference data
  • Single-atom displaced supercell structures
  • Rattled supercell structures
  • Adjust supercell settings
  • VASP settings
• Fitting potentials
  • General settings
  • GAP
  • ACE
  • Nequip
  • M3GNet
  • MACE
  • Example script for autoplex workflow using GAP to fit and benchmark a Si database
  • Running a MLIP fit only
• Benchmark
  • General settings
  • Error metrics
  • Run a benchmark with a pre-existing DFT calculation and GAP potential