F04-Multi-paradigm Materials Design via Computation, Experiment, and Artificial Intelligence

Symposium Introduction

A single research paradigm is often insufficient to tackle complex materials problems: computations can be limited by approximations or scale; experiments are often slow and costly; and AI models require large, high‑quality datasets and may lack physical interpretability. However, integrating these approaches creates a powerful synergy. AI can guide computations toward promising regions of chemical space; computations can screen candidates before synthesis; experiments validate predictions and feed data back to improve models; and machine learning dynamically updates strategies based on new results. By combining these paradigms, researchers create a closed‑loop design cycle.
Multi‑paradigm materials design via computation, experiment, and artificial intelligence represents a transformative approach to materials discovery and optimization. It moves beyond traditional trial‑and‑error methods by synergistically combining three complementary pillars:
(1) Computational Modeling – Atomistic calculations (DFT, MD), phase‑field simulations, thermodynamics, finite‑element analysis, and other physics‑based models predict material properties and behavior from atomic to macroscopic scales.
(2) Experimental Synthesis & Characterization – High‑throughput automated synthesis and characterization, advanced microscopy, spectroscopy, in situ testing, and service‑performance evaluation provide real‑world validation, generate training data, and uncover phenomena not captured by models.
(3) Artificial Intelligence & Machine Learning – AI algorithms mine vast datasets, analyze spectra and images, accelerate simulations and experiments, suggest novel compositions/structures, and close the loop between prediction and experiment.
As computing power grows, experimental automation advances, and AI algorithms become more robust, multi‑paradigm materials design is evolving into an autonomous, agentic research paradigm—where self‑driving labs iteratively design, synthesize, and test materials with minimal human intervention, drastically compressing the timeline from concept to application. The seamless integration of all paradigms enables efficient autonomous materials design and discovery through various workflows and agents, with critical human‑in‑the‑loop oversight. In essence, this field embodies the convergence of physics‑based materials knowledge, computational design, data‑driven discovery, and experimental validation, promising a new era of rational, accelerated materials innovation.

Symposium Topics 

1.Computational simulations: QM (DFT), MD, CALPHAD, Phase field, FEM etc.
2.Experimental design & validation: high-throughput and automated materials synthesis & characterization, self-driving labs etc.
3.Artificial intelligence & machine learning: feature learning, deep learning, generative learning, active learning, reinforcement learning, transfer learning, image & spectra learning etc.
4.Autonomous agentic research: chain-of-thought, reasoning, reflection, multi-agent etc.
Structural and functional materials: alloys, ceramics, polymers, semiconductors, composites etc.

Symposium Organizers 

Organized by

Materials Genome Institute, Shanghai University

Supported by

C-MRS，European MRS，MRS-J，MRS Singapore，IUMRS

Contact with

Name: Jie Xiong

E-mail: xiongjie@shu.edu.cn

Institution: Shanghai University, China

Name: Ming Zhang

Phone:+86-18964607686

E-mail: mingzhang@shu.edu.cn

Institution: Shanghai University, China

Remark

Collaborated with the journals of Transactions of Materials Research, Journal of Materials Informatics, Journal of Structural Materials.