Scientific Calendar Event



Starts 18 Oct 2022 14:00
Ends 18 Oct 2022 15:30
Central European Time
Hybrid Seminar
Oppenheimer + Zoom
Zhi Li
(ICTP)
 
 
Abstract:
 
In this tutorial, I will briefly introduce the theoretical background of the DeePMD package and its practical use. Prior knowledge of machine learning is not required. At the end of this tutorial, you will obtain a working knowledge of how to construct a deep-learning potential with the DeePMD package. The tutorial is divided into two parts. The first part is dedicated to the theoretical background of DeePMD, while the second explains the key parameters in the manual. In DeePMD, fortunately, a high-quality potential does not depend too much on the neural network structure, like the number of neurons in each hidden layer. The tedious theoretical part will also be interspersed with several practical examples from my research to make it more accessible. If time allows, I will show how to use the DeePMD C++ interface to compute the vibrational spectrum with lattice dynamics for a system with 16800 atoms.