| Abstract Scope |
The behavior of materials at the atomic scale influences their overall properties, including strength, and failure mechanisms. Understanding these behaviors is essential for complex materials, such as polymer matrix composites, which are widely used in structural and functional applications. These composites present particular challenges due to their multi-phase nature and history dependent nonlinear constitutive behaviors. Molecular dynamics (MD) offers valuable microscopic insights into the mechanisms and timing of material failures. However, the high computational cost of MD simulations often limits their practical use. To overcome these limitations, we develop a physics-informed surrogate modeling framework under variable loading conditions. The approach is carefully designed to model each constituent and its interface enabling insights into stress responses and failures in each phase. Numerical results show that our model can predict the constitutive behavior of materials under various loading conditions in 3D space, capturing stress and failure states with high fidelity and accuracy. |