Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

SYAI: Symposium AI and Data Challenges behind Emerging Self-Driving Laboratories

SYAI 1: AI and Data Challenges behind Emerging Self-Driving Laboratories

SYAI 1.3: Hauptvortrag

Donnerstag, 12. März 2026, 10:30–11:00, HSZ/AUDI

Autonomous, Data-Driven Workflows for Materials Acceleration Platforms with pyiron — •Jan Janssen and Joerg Neugebauer — Max Planck Institute for Sustainable Materials, Düsseldorf, Germany

The hierarchical nature of materials -- spanning electronic, atomistic, microstructural, and macroscopic scales -- necessitates an integrated multiscale strategy for accelerating materials discovery. Realizing such a strategy requires close coordination across chemistry, materials science, and physics, and an equally tight coupling of simulation and experiment. Materials Acceleration Platforms (MAPs) address this need by embedding both theoretical and experimental data streams into a unified active-learning loop. A central challenge is ensuring that information can be meaningfully transferred across scales. We approach this by propagating uncertainties throughout the entire multiscale workflow, enabling quantitative comparison from ab-initio predictions to experimental measurements. This uncertainty-aware integration makes it possible to reduce the number of required experiments by several orders of magnitude. The resulting hierarchy of simulations is powered by machine-learning models -- including ML interatomic potentials, Bayesian optimization, and large-language-model (LLM) agents -- each contributing to decision-making within the MAP. The pyiron workflow framework provides the digital backbone for this integration. By encoding domain expertise into modular, machine-actionable workflows, pyiron enables simulation specialists to formalize their methods in a way that is both reproducible and interoperable.

Keywords: Materials Acceleration Platform; Workflow; Multiscale Simulation; Large Language Model; Machine Learning