Install a model#

Install a KIM model#

The kim-api-collections-management command line tool from the kim-api makes it easy to install a model archived on the OpenKIM website. You can do:

$ kim-api-collections-management install user <model_name>

to automatically download a model from the OpenKIM website, and install it to the User Collection. Just replace <model_name> with the Extended KIM ID of the model you want to install as listed on OpenKIM. To see that the model has been successfully installed, do:

$ kim-api-collections-management list

See also

A model can be installed into a different collection other than the User Collection specified by user. You can use kim-api-collections-management to remove and reinstall models. See the kim-api documentation for more information.

Install a KLIFF-trained model#

As discussed in Train a Stillinger-Weber potential and Train a neural network potential, you can write a trained model to a KIM model that is compatible with the kim-api by:

path = "./kliff_trained_model"

Which writes to the current working directory a directory named kliff_trained_model.


The path argument is optional, and KLIFF automatically generates a path if it is None.

To install the local kliff_trained_model, do:

$ kim-api-collections-management install user kliff_trained_model

which installs the model into the User Collection of the kim-api, and, of course, you can see the installed model by:

$ kim-api-collections-management list

The installed model can then be used with simulation codes like LAMMPS, ASE, and GULP via the kim-api.