ParAMS, the parameter optimization toolkit

Parametrized models such as ReaxFF and Machine Learning Potentials can model large, realistic, systems with high efficency. However, the accuracy of such models varies.

ParAMS lets you is refit the parameters of such models to increase the accuracy, or enable the study of new systems by creating new parameters.

ParAMS helps you to create and manage your training data, run high-dimensional parameter fits in smart and efficient ways, and evaluate the fitness of your new model.

ParAMS features a graphical user interface and Python library to make parametrization projects easy, with extensive documentation and many hands-on tutorials to get you started.

Create your own Machine Learning Potentials:

• Fine-tune the M3GNet universal potential with only little training data
• Automatically generate and train your model with active learning

Create your own ReaxFF reactive force fields and DFTB models:

• Sensitivity analysis of parameter space: Which parameters should be optimized and which not?
• Apply recommended ReaxFF parameter constraints during the optimization
• Run multiple optimizers in parallel
• Automatically turn off and restart optimizers not performing well

• ParAMS is part of Advanced workflows and tools
• You also need to license the compute engine you want to train: ReaxFF and/or DFTB and/or ML potentials
• For reference data, we recommend to use ADF, BAND, or Quantum ESPRESSO. You may also user other training data, for example from other software or experimental data.
• See also Pricing and Licensing