The field of hierarchical modeling and optimization is witnessing significant developments, with a focus on addressing complex systems and high-dimensional problems. Researchers are exploring innovative approaches to capture hierarchical relationships, conditional dependencies, and heterogeneous structures in data. The integration of surrogate models, graph theory, and deep generative models is enabling more efficient and adaptable optimization methods. Notably, the application of variational autoencoders and flow-based models is improving the scalability and performance of Bayesian inference and constrained black-box optimization. Furthermore, the interpretation of generative adversarial networks as probabilistic generative models is leading to new regularization strategies and optimization techniques. Overall, these advancements are poised to impact various fields, including complex system design, subsurface energy storage, and multiphase flow behavior. Noteworthy papers include:

• Hierarchical Modeling and Architecture Optimization, which introduces a unified framework for modeling and optimizing hierarchical domains.
• Posterior Inference in Latent Space for Scalable Constrained Black-box Optimization, which proposes a novel framework for constrained optimization using flow-based models and posterior inference.