BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IMX Colloquium - Modelling Materials Failure Processes at the Atom
 istic and Electronic Structure Scales with Scientific Machine Learning
DTSTART:20251124T131500
DTEND:20251124T141500
DTSTAMP:20261022T131822Z
UID:842b7b9e8e23e5f3a5ca69fa3b54f92f0a90a828a28523a989e64103
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. James Kermode\, University of Warwick\, UK\nI will descr
 ibe recent\, rapid progress in the development and application of machine 
 learning interatomic potentials (MLIPs) to 'chemomechanical' problems in s
 tructural materials that simultaneously require accurate local chemistry a
 nd long-range stress fields - e.g. fracture and plasticity. Examples will 
 include concurrent coupling of quantum mechanics and MLIPs  [1] or even t
 he use of MLIPs standalone [2] to describe plasticity in tungsten. Combini
 ng two potentials with different accuracy/cost tradeoff choices in differe
 nt parts of a large system is also possible [3]. The recent arrival of fou
 ndation MLIP models that leverage large datasets and deep learning have pr
 oduced models capable of describing much of the periodic table with reason
 able accuracy. I will critically assess the applicability of the MACE MP0 
 and MPA models [4] to chemomechanical problems\, and present results from 
 fine-tuning them to improve their (already reasonable) out-of-the-box desc
 ription of these systems [5]. Finally\, I will discuss the importance of r
 obust uncertainty estimates when using these surrogate models and report r
 ecent efforts in this direction [6].\n \n[1] P. Grigorev\, A. M. Goryaeva
 \, M.-C. Marinica\, J. R. Kermode\, and T. D. Swinburne\,\nCalculation of 
 Dislocation Binding to Helium-Vacancy Defects in Tungsten Using Hybrid Ab 
 Initio-Machine Learning Methods\, Acta Mater. 247 118734 (2023) [arXiv:211
 1.11262]\n[2] M. Nutter\, J. R. Kermode and A. P. Bartók\, Kink-helium in
 teractions in tungsten: Opposing effects of assisted nucleation and hinder
 ed migration arXiv:2406.08368 (2024)
[3] F. Birks\, T. D. Swinburne and 
 J. R. Kermode\, Efficient and Accurate Spatial Mixing of Machine Learned I
 nteratomic Potentials for Materials Science\, arXiv:2502.19081 (2025)
[4
 ] I. Batatia et al.\, A Foundation Model for Atomistic Materials Chemistr
 y arXiv:2401.00096 (2024)
[5] P. Grigorev\, F. Birks\, T. D. Swinburne a
 nd J.R. Kermode\, In Prep (2025)
[6] I. R. Best\, T. J. Sullivan and J. 
 R. Kermode\, Uncertainty Quantification in Atomistic Simulations of Silico
 n Using Interatomic Potentials\, J. Chem. Phys. 161\, 064112 (2024)\n\nBio
 : Prof. James Kermode\, Materials Modelling in the School of Engineering
  at the University of Warwick\, where he is also currently serving as Res
 earch Cluster Leader for the Predictive Modelling research cluster. He a
 lso directs the EPSRC Centre for Doctoral Training in Modelling of Hetero
 geneous Systems (HetSys) and the Warwick Centre for Predictive Modelling
  (WCPM) university research centre. He develops multiscale materials mod
 elling algorithms and the software that implements them\, with a partic
 ular focus on machine learning and data-driven approaches\, and on qu
 antifying the uncertainty in the output of electronic structure and at
 omistic models. He is also active in applying parameter-free modelling tec
 hniques to make quantitative predictions of "chemomechanical" materials fa
 ilure processes where stress and chemistry are tightly coupled\, e.g. near
  the tip of a propagating crack\, where local bond-breaking chemistry is d
 riven by long-range stress fields.
LOCATION:MXF 1 https://plan.epfl.ch/?room==MXF%201
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR