Go back to publications

TransRheo: A Physics-Informed Multi-Task Numeric Transformer for Polymer Nanocomposite Rheology

March 2026 Bily Cheng, collaborators Preparing for submission
#AI for Science #Materials #Transformer #Scientific ML

Overview

This project focuses on data-driven modeling for polymer nanocomposite rheology and was developed in collaboration with researchers across multiple universities.

Research Context

The broader effort combines:

  • scientific machine learning for material property prediction,
  • high-dimensional variable modeling,
  • agent-based tooling to support materials R&D workflows.

My Role

I contributed to the algorithmic and experimental side of the project while independently working on the LLM Agent direction for material-research workflows.

Current Status

The manuscript is being prepared for submission.