Force Field Query¶
A trained sGDML model (see Force Field Reconstruction) is straightforward to use. The following example shows how to predict energy and forces using the pre-trained ethanol force field (Pre-trained Models) for a ethanol geometry file
Download) stored in XYZ format:
import numpy as np from sgdml.predict import GDMLPredict from sgdml.utils import io model = np.load('ethanol.npz') gdml = GDMLPredict(model) r,_ = io.read_xyz('ethanol.xyz') # 9 atoms e,f = gdml.predict(r) print r.shape # (1,27) print e.shape # (1,) print f.shape # (1,27)
We instantiate the sGDML predictor with the model file
ethanol.npz and query it using the ethanol geometry stored in
ethanol.xyz. It returns the energy and all interatomic forces for this structure.
The order of atoms in
ethanol.xyz must be the same as in the dataset from which the model
The distance unit of the query geometry (here: Ångström) must match the unit in the dataset used for model reconstruction.