Tools

Parametrization

Scm white molecules 1
Create custom models or fine-tune existing ones for your research

Atomistic modeling can help reduce time and wasteful experiments by suggesting the best materials out of millions of candidates. Interatomic potentials are fundamental to perform atomistic simulations beyond quantum mechanics. However, the parameters might be unavailable or unsuitable for your application. With ParAMS, you can create your own models to describe novel materials and molecules.

Params gui

ParAMS will support every step of your parametrization journey: From the generation of references to the handling of data (sorting, weighting, refining…) to the actual fitting of parameters and evaluation of the parametrization result. Choose from several established models such as the reactive force field ReaxFF, the tight-binding codes DFTB and GFN-xTB, machine learning potentials or use your own. The built-in sensitivity analysis aids the choice of active parameters in highly-parametrized models such as ReaxFF and DFTB.

  • Import reference data from AMS, VASP, QE, Gaussian, or experiments
  • Use training and validation sets to prevent overfitting
  • Explore sensitivity of parameters for ReaxFF and DFTB
  • Submit multiple optimizations in parallel, interactively stop and restart
  • Submit jobs to remote machine
  • Visualize dataset and interactively display results
  • Select and inspect outliers

SCM Product Manager Dr. Matti Hellström about the sensitivity analysis in ParAMS.


Comparing existing ReaxFF parametersets with ParAMS.

Not sure what modeling tools you need for you research project?