The use of machine learning in chemistry is on the rise for the prediction of chemical properties. In these applications, the input feature representation or descriptor is an important factor that affects the accuracy as well as the extent of the explored chemical space. We recently developed a universal periodic table tensor descriptor that combines features from Behler–Parrinello’s symmetry functions and a periodic table representation. Using our descriptor and a convolutional neural network model, we achieved a low mean absolute error for the prediction of the atomization energy of organic molecules in the QM9 data set and the formation energy of materials from Materials Project data set, respectively. We also modified a neural network potential (NNP) to simultaneously predict energy, band edges, and partial density of states of Cu2O. Our NNP can reproduce the density functional theory (DFT) potential energy surface and properties at a fraction of the computational cost. Furthermore, we show that the standard deviation of the energies predicted by the ensemble of training snapshots can be used to estimate the uncertainty in the predictions. This allows us to switch from the NNP to DFT on-the-fly during molecular dynamics simulations to evaluate the forces when the uncertainty is high.
This event will be recorded and uploaded on ICIQ’s YouTube Channel.
If you would like to attend this talk, please, register here!