LM3 Lab, University of Michigan

Thomas D Swinburne, PhD
Assistant Professor, Mechanical Engineering
Visiting researcher, CNRS Physique
CV  /  Google Scholar  /  GitHub
tswin-at-umich-dot-edu

LM3: Learning Multiscale Material Models
Metallic alloys are essential for the modern world as they uniquely combine strength, ductility and reusability. These properties result from a complex hierarchy of plasticity mechanisms which are only partially understood.

Experiments are typically indirect, destructive and expensive, meaning data is sparse. With fatigue alone costing 2-3% of US GDP, predictive models which help diagnose and mitigate component failure are urgently needed.

Our research pushes the frontier of multi-scale modelling for metals, connecting atomic dynamics, microstructure and mechanical properties. We draw broadly from applied math, physics, mechanics and machine learning.

We are hiring in Ann Arbor and Paris!

Current research topics
Projects balance theory, method development and applications, depending on your interest. See team members/alumni and open positions.

• Application focus:
  • Slip transfer in complex dislocation networks
  • High-dpa irradiation damage for nuclear fusion
  • Exploring phase diagrams and kinetics
  • Mechanistic inverse design of dilute alloys
• Theory/method development focus:
  • Model-form uncertainty quantification (UQ) for ML surrogates
  • Data-driven forecasting of long-timescale simulation behaviour
  • End-to-end differentiable simulations for UQ and design
  • Efficient fine-tuning of universal potentials (e.g. MACE)

Selected recent work
See Google Scholar for an up-to-date list.

• Score matching free energies (Nat. Comm. 2025 / code)
  First differentiable free energy method. Extends NPJ 2025 for minima
• POPS model-form UQ for deterministic models (ML:S&T 2025 / code)
  POPS solves Bayes’ ignorance of model-form UQ. For MLIPs: NPJ 2025
• Entropy of dislocation glide (Nat. Comm. 2025 / PAFI / Rodney Group)
  Why ML potentials are required to accurately model bcc plastic flow
  
• MACE foundation model for ML potentials (Csyani Group / arXiv 2024)
  Section A.13. Dislocations are a tough extrapolation test for UMLIPs.
  
• Coarse-graining & forecasting atomic simulations ( PRL 2023)
  ML feature vectors used for MLIPs can also forecast complex futures
• Embedding ab initio in MLIPs for solute studies (Acta Mat. 2023 )
  QM/ML for dislocation-solute interactions. Applied to W alloys
  
• A15 defects in fcc (Nat. Comm. 2023 / TAMMBER / Marinica Group)
  Irradiation defects can grow as three-dimensional Laves phase clusters

Recent & Upcoming Invited Conferences/Seminars
SIAM UQ, Minneapolis, USA, 03/26
Dislocations 2025, Miami, USA, 11/25
CECAM: UQ from DFT to ML, EPFL, Lausanne, 11/25
CoMPASs: Math and Comp. Mat. Sci., ICMS, Edinburgh, 11/25
USACM Nanomechanics 2025, Urbana-Champaign, USA, 09/25
Speaker & Senior Scholar, IPAM, UCLA, 9/25-12/25
E-MRS Meeting, Prague, 09/25 (unavailable)
Fritz Haber Institute of the Max Planck Society, Berlin, 04/25
APS March Meeting, Anaheim, 03/25 (unavailable)
Condensed Matter Physics Seminar, École Polytechnique, 03/25
CECAM Promises of the multiscale, EPFL, Lausanne, 03/25
Keynote, MecaNano symposium, online, 02/25
AppliedML 2025, EPFL, Lausanne, 02/25