Generative design of polymeric materials
Аннотация
Polymeric materials play an extremely important role in many key fields. However, polymer research still remains constrained by vast, sparsely sampled design spaces and strong coupling among chemistry, processing, structure, and performance. Generative design is shifting this paradigm from prediction and screening of predefined libraries toward target-driven creation of candidates under property, synthesis, and application constraints. This article discusses the emerging framework of generative polymer design, focusing on polymer representations, data infrastructure, mainstream generative models, candidate screening, and closed-loop validation. We emphasize that credible generative design requires complementary assessment of structural validity, synthesis and processing feasibility, uncertainty-aware property evaluation, physics-based simulation, and experimental validation, while computational or experimental feedback can guide subsequent design decisions toward closed-loop discovery. Finally, we discuss the remaining challenges and future directions for generative polymer design. It is our hope that this article may serve as a useful reference and, in some small way, help stimulate further discussion in AI for polymer materials research. Graphical abstract