Potential Energy Surface

A potential energy surface (PES) maps the energy of a molecular system as a function of its atomic coordinates, crucial for understanding chemical reactions and material properties. Current research heavily utilizes machine learning, particularly graph neural networks and Gaussian process regression, to build accurate and computationally efficient surrogate models of PESs, often focusing on improving extrapolation capabilities and uncertainty quantification to identify unreliable predictions. These advancements enable faster and more accurate simulations of complex systems, impacting fields like materials science, drug discovery, and catalysis by accelerating the design and optimization of new molecules and materials.