Documentation

PyAMFF must be run with a config.ini file with the options specified in the documentation. Each section header is denoted by square brackets, and is followed by the key/value pairs. For example, to set the run_type key of the [Main] section to the value trainFF, your config.ini file would include the lines:

[Main]
run_type = trainFF

PyAMFF also requires a fpParas.dat file to specifiy the values for fingerprint calculations. The fpParas.dat should be in the following format:

#type  centralElement  neighborElement  eta   Rs
 G1    H               H                0.05  0.0
 G1    H               Pd               0.05  0.0
 G1    Pd              Pd               0.05  0.0
#type  centralElement  neighborElement1  neighborElement2  eta  zeta  gamma  theta
 G2    Pd              Pd                Pd                0.02 100.0 1.0      0.5

Lastly, your images must be in a trajectory file names train.traj.

There are specific options for each method, and a set of general options which are shared between methods.

Flags

Main: General settings.

Parralelization: Settings for parallelization of calculations.

Fingerprints: Settings for fingerprints.

MachineLearningModel: Choose which machine learning model to train.

LossFunction: the loss functions available.

Optimizer: Settings for optimization.

Debug: Debug Settings.