Development of Machine-Learning Interatomic Potential to Study Organic Materials under Dynamics Compression

We present the development of machine-learning interatomic potential based on Chebyshev polynomials to study organic crystal under dynamic compression. Our potential accurately predicts the structure properties and chemistry of organic material for a wide range of thermodynamic conditions. While training only on a specific system, our model also shows good transferability to other materials that are outside the training data. We discuss the application of our interatomic potential to understand the complicated chemistry of organic material under shockwave. Our methods provide a way to efficiently achieve highly accurate simulations that can greatly facilitate experimental design and interpretation.