Optimizing transition interface sampling simulations

Author(s)
Ernesto Emmanuel Borrero, Marcus Weinwurm, Christoph Dellago
Abstract

We demonstrate that a recently proposed adaptive optimization algorithm for forward flux sampling simulations [E. E. Borrero and F. A. Escobedo, J. Chem. Phys. 129, 024115 (2008)] can be easily applied within the framework of transition interface sampling. This optimization algorithm systematically identifies the kinetic bottlenecks along the order parameter used to partition phase space via interfaces and improves the statistical accuracy of the reaction rate constant estimate. In different versions of the algorithm, the number or the placement of the interfaces (or both) are varied in order to allocate the numerical effort in a balanced way. The algorithm is demonstrated for a simple two-dimensional model and for the dipole flip transition of icelike structures inside carbon nanotubes. For these test systems, the optimization yielded an efficiency increase by a factor of 2-15.

Organisation(s)
Computational and Soft Matter Physics
Journal
Journal of Chemical Physics
Volume
134
No. of pages
9
ISSN
0021-9606
DOI
https://doi.org/10.1063/1.3601919
Publication date
2011
Peer reviewed
Yes
Austrian Fields of Science 2012
103036 Theoretical physics, 103009 Solid state physics, 104022 Theoretical chemistry, 103029 Statistical physics
Portal url
https://ucrisportal.univie.ac.at/en/publications/339de7d2-5a2f-42e1-bea3-a7499336bb72