Abstract Scope |
In thermodynamics, materials are considered systems. The properties of a system represent its responses to disturbances from its surroundings, observable as measurable quantities. Based on non-equilibrium thermodynamics, these responses reflect internal processes due to external disturbances and are related to derivatives of the system's energy or free energy. Consequently, accurate predictions of properties depend on the quantitative prediction of the system's free energy landscape, considering both internal and external variables. In this presentation, we will discuss the general framework of our quantitative predictive theories (DOI: 10.1088/1361-648X/ad4762), based on a multiscale entropy approach (recently termed zentropy theory) that integrates quantum mechanics and statistical mechanics. We will present their successful applications to magnetic and thermoelectric materials and melting, along with key challenges for complex systems. Additionally, we will explore potential paths forward, particularly the zentropy-enhanced neural networks (ZENNs) currently being developed by Liu’s and Hao’s groups. |