The field of multiscale modeling and simulation is witnessing significant advancements, driven by the need to efficiently and accurately capture complex phenomena across multiple length and time scales. Researchers are explorings innovative approaches to learn stochastic multiscale models directly from observational data, leveraging techniques such as amortized variational inference and physics-based multiscale modeling. Additionally, new finite element methods and domain decomposition techniques are being developed to improve the simulation of particulate flows and other complex systems. These developments have the potential to revolutionize fields such as fluid dynamics, materials science, and biomedicine. Noteworthy papers include:

• One that proposes a learnable and differentiable finite volume solver for accelerated simulation of fluid flows, demonstrating state-of-the-art performance and superior generalizability.
• Another that introduces a new multimesh finite element method for direct numerical simulation of incompressible particulate flows, showing significant gains in accuracy and robustness.