Physics-Informed Neural Networks in Polymers: A Review.

Area of Science:

• Polymer Science and Engineering
• Computational Materials Science
• Artificial Intelligence in Chemistry

Background:

• Polymer systems exhibit complex multi-scale behavior, posing significant challenges for traditional modeling and simulation techniques.
• Balancing accuracy and computational efficiency remains a key hurdle in bridging atomistic and macroscopic scales for polymer systems.

Purpose of the Study:

• To review the development and application of physics-informed neural networks (PINNs) in polymer science.
• To summarize recent advances, methodologies, benefits, and limitations of PINNs for polymer modeling.
• To identify future research directions for advanced polymer simulations using PINNs.

Main Methods:

• Review of existing literature on PINNs applied to polymer systems.
• Analysis of methodologies integrating data-driven learning with physical laws.
• Evaluation of PINNs for property prediction, structural design, and process optimization.

Main Results:

• PINNs show promise in overcoming limitations of traditional methods by combining data-driven insights with physical principles.
• Applications span polymer property prediction, structural design, and process optimization.
• Key methodologies and benefits are identified, alongside current limitations.

Conclusions:

• PINNs represent a powerful emerging tool for advanced polymer modeling and simulation.
• Further research is needed to address current challenges and fully exploit PINNs' potential in polymer science.
• PINNs offer a pathway to more accurate and efficient simulations across multiple scales.