São Paulo, April 9, 2026
The Department of Materials Physics and Mechanics (DFMT/IFUSP) congratulates Prof. Caetano Rodrigues Miranda for the article "Bayesian Optimization for High-Dimensional Coarse-Grained Model Parameterization: A Case Study on Pebax Polymer", selected for the front cover of the journal Journal of Chemical Theory and Computation (Vol. 22, No. 5, p. 2358-2368, 2026).
Coarse-grained (CG) models are widely used in materials simulations due to their scalability. Traditionally parameterized by hybrid top-down/bottom-up strategies, they limit joint optimization of parameters. This study applies Bayesian optimization with tree-structured Parzen estimator (TPE) to a 41-parameter CG model of the Pebax-1657 copolymer, optimizing density, radius of gyration, and glass transition temperature. The approach converges faster than traditional methods, accurately reproducing atomistic properties.
Authors: Carlos A. Martins Jr., Daniela A. Damasceno, Keat Yung Hue, Caetano Rodrigues Miranda*, Erich A. Müller, and Rodrigo A. Vargas-Hernández.
Comment by Prof. Caetano Miranda:
Our recent work, 'Bayesian Optimization for High-Dimensional Coarse-Grained Model Parameterization: A Case Study on Pebax Polymer', presents an important advance in using artificial intelligence to accelerate simulations of complex materials. Coarse-grained models, which simplify the molecular description to enable simulations on larger time and size scales, rely on a delicate parameter adjustment process. In this study, we demonstrate that Bayesian optimization, using the tree-structured Parzen estimator (TPE) method, can be successfully applied even to high-dimensional models with dozens of parameters. The strategy showed faster convergence and more consistent results than traditional approaches. As a case study, we parameterized a coarse-grained model of the Pebax-1657 copolymer, of interest for membrane technologies in greenhouse gas separation applications. As a highlight, the scientific artwork was selected for the front cover of the Journal of Chemical Theory and Computation.
Congratulations to Prof. Miranda and the team for the international recognition!
Department of Materials Physics and Mechanics (DFMT/IFUSP)