F - Minimum energy paths and dynamical pathways of activated processes in materials

Symposium organizers

Fabio Pietrucci (Université Pierre et Marie Curie, France) - point of contact
Tony Lelievre (École des Ponts ParisTech, France)
Antonino Marco Saitta (Université Pierre et Marie Curie, France)
Rodolphe Vuilleumier (École Normale Supérieure et UPMC, France)

Symposium description

The true potential of a new material in real life applications is determined by the availability of synthetic routes and by the behavior and stability in realistic conditions. In this respect, well-established and powerful approaches, like the nudged elastic band method, are available to map potential energy landscapes along minimum energy paths. In combination with computational models like kinetic Monte Carlo, the knowledge of minimum energy paths in turn grants access to the long-time kinetic evolution of the material. A different approach is taken by molecular dynamics and Monte Carlo simulations, which feature the additional advantage of directly accessing experimentally relevant conditions of finite temperature and pressure. Unfortunately, a straightforward reconstruction of the free energy landscape and reaction rate network of a material, albeit possible in principle, is often hampered in practice by the occurrence of barriers separating the relevant metastable states. To tackle this problem - often dubbed the rare event problem - a number of ingenious enhanced sampling strategies have been put forward, ranging from umbrella sampling to transition path sampling, from parallel tempering to metadynamics, etc. After some decades of turbulent development, the field is coming of age and the main bottlenecks and challenges get more clearly delineated. Two issues in particular are crucial, and are the focus of the symposium: the cumbersome identification of optimal reaction coordinates and the far-from-optimal robustness of the algorithms. In both cases, progress is driven by rigorous, mathematically sound contributions able to unify different problems, e.g., exploiting cross-fertilization with fields like complex network analysis and machine learning. Achieving truly wide-spectrum and fully reproducible simulations under precise control of statistical uncertainties, in a suitable multi-scale framework, would be the major breakthrough opening to a routine use of enhanced sampling approaches in the development and design of new materials.

Invited speakers~

• Mihai-Cosmin Marinica (CEA-Saclay, France)
  Energy landscape and pathways of point defects in body-centered-cubic metals

• Francesco Montalenti (University of Milano-Bicocca, Italy)
  Continuum modeling of heteroepitaxy tackling elastic and plastic relaxation

• Gerhard Stock (University of Freiburg, Germany)

• Mark Tuckerman (New York University, USA)
  Enhanced free energy based structure prediction in materials science