ECMC 2021 Engel
From Werner KRAUTH
| Revision as of 12:13, 10 May 2021 Werner (Talk | contribs) ← Previous diff |
Current revision Werner (Talk | contribs) |
||
| Line 4: | Line 4: | ||
| '''Marco Klement, <u>Michael Engel</u>''' | '''Marco Klement, <u>Michael Engel</u>''' | ||
| - | '''''Institute for Multiscale Simulation, Friedrich-Alexander University Erlangen-Nürnberg, Germany''''' | + | '''''Institute for Multiscale Simulation, Friedrich-Alexander University Erlangen-Nürnberg (Germany)''''' |
| '''Abstract''' Event-driven molecular dynamics is efficient because its Newtonian dynamics equilibrates fluctuations with the speed of sound. Monte Carlo simulation is efficient if performed with correlated position updates in event chains. Here, we combine the core concepts of both into a new algorithm involving Newtonian event chains. We implement Newtonian event-chain Monte Carlo (NEC) for spheres [1] and convex polyhedra [2] in the open source general-purpose particle simulation toolkit HOOMD-blue for serial and parallel simulation. For polyhedra, our implementation makes use of an improved computational geometry algorithm XenoSweep, which predicts sweep collision in a particularly simple way. The speed-up of NEC over Monte Carlo is between a factor of 10 for spheres or nearly spherical polyhedra and a factor of 2 for highly aspherical polyhedra. | '''Abstract''' Event-driven molecular dynamics is efficient because its Newtonian dynamics equilibrates fluctuations with the speed of sound. Monte Carlo simulation is efficient if performed with correlated position updates in event chains. Here, we combine the core concepts of both into a new algorithm involving Newtonian event chains. We implement Newtonian event-chain Monte Carlo (NEC) for spheres [1] and convex polyhedra [2] in the open source general-purpose particle simulation toolkit HOOMD-blue for serial and parallel simulation. For polyhedra, our implementation makes use of an improved computational geometry algorithm XenoSweep, which predicts sweep collision in a particularly simple way. The speed-up of NEC over Monte Carlo is between a factor of 10 for spheres or nearly spherical polyhedra and a factor of 2 for highly aspherical polyhedra. | ||
| Line 14: | Line 14: | ||
| ---- | ---- | ||
| - | '''Slides''' | + | '''Slides''' [http://www.lps.ens.fr/%7Ekrauth/images/d/d8/ECMC_2021_Engel.pdf here] |
| '''Recording''' | '''Recording''' | ||
Current revision
Newtonian Event-Chain Monte Carlo with Spheres and Polyhedra
Marco Klement, Michael Engel
Institute for Multiscale Simulation, Friedrich-Alexander University Erlangen-Nürnberg (Germany)
Abstract Event-driven molecular dynamics is efficient because its Newtonian dynamics equilibrates fluctuations with the speed of sound. Monte Carlo simulation is efficient if performed with correlated position updates in event chains. Here, we combine the core concepts of both into a new algorithm involving Newtonian event chains. We implement Newtonian event-chain Monte Carlo (NEC) for spheres [1] and convex polyhedra [2] in the open source general-purpose particle simulation toolkit HOOMD-blue for serial and parallel simulation. For polyhedra, our implementation makes use of an improved computational geometry algorithm XenoSweep, which predicts sweep collision in a particularly simple way. The speed-up of NEC over Monte Carlo is between a factor of 10 for spheres or nearly spherical polyhedra and a factor of 2 for highly aspherical polyhedra.
[1] M. Klement, M. Engel, J. Chem. Phys. 150, 174108 (2019)
[2] M, Klement, S. Lee, J.A. Anderson, M. Engel, arXiv:2104.06829 (2021)
Slides here
Recording
Further material
