LAMMPS_for_beginners - A brief survey of the LAMMPS MD...

Info iconThis preview shows pages 1–8. Sign up to view the full content.

View Full Document Right Arrow Icon
A brief survey of the LAMMPS MD code: intro, case studies, and future development Paul Crozier, Steve Plimpton, Aidan Thompson, Mike Brown February 24, 2010 LAMMPS Users’ Workshop CSRI Building, Albuquerque, NM Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE-AC04-94AL85000.
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
MD: molecular dynamics F = ma Classical dynamics Rapidly grown in popularity and use in research Computationally intensive, especially computation of nonbonded interactions Uses force fields: mathematical models of interatomic interactions A brief introduction to MD
Background image of page 2
MD uses empirical force fields Particles interact via empirical potentials analytic equations, fast to compute coefficients fit to expt or quantum calcs Potential energy = Φ = f(x) Force = -Grad Φ Pair-wise forces Van der Waals (dipole-dipole) Coulombic (charge-charge) Many-body forces EAM, Tersoff, bond-order, ReaxFF Molecular forces springs, torsions, dihedrals, . .. Long-range Coulombic forces Ewald, particle-mesh methods, FFTs
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
MD in the Middle Quantum mechanics electronic degrees of freedom, chemical reactions Schrodinger equation, wave functions sub-femtosecond timestep, 1000s of atoms, O(N 3 ) Atomistic models molecular dynamics (MD), Monte Carlo (MC) point particles, empirical forces, Newton's equations femtosecond timestep, millions of atoms, O(N) Mesoscale to Continuum finite elements or finite difference on grids coarse-grain particles: DPD, PeriDynamics, . .. PDEs, Navier-Stokes, stress-strain microseconds seconds, microns meters, O(N 3/2 ) Distance Time Å m 10 -15 s years QM MD MESO FEA Design
Background image of page 4
Algorithmic Issues in MD Speed parallel implementation Accuracy long-range Coulombics Time scale slow versus fast degrees of freedom Length scale coarse-graining
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Classical MD in Parallel MD is inherently parallel forces on each atom can be computed simultaneously X and V can be updated simultaneously Most MD codes are parallel via distributed-memory message-passing paradigm (MPI) Computation scales as N = number of atoms ideally would scale as N/P in parallel Can distribute: atoms communication = scales as N forces communication = scales as N/sqrt(P) space communication = scales as N/P or (N/P) 2/3
Background image of page 6
Parallelism via Spatial-Decomposition Physical domain divided into 3d boxes, one per processor Each proc computes forces on atoms in its box using info from nearby procs Atoms "carry along" molecular topology as they migrate to new procs Communication via nearest-neighbor 6-way stencil Optimal scaling for MD: N/P so long as load-balanced Computation scales as N/P Communication scales sub-linear as (N/P) 2/3 (for large problems) Memory scales as N/P
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 8
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 12/07/2010 for the course MAT 500 taught by Professor Unknown during the Spring '10 term at University of Illinois at Urbana–Champaign.

Page1 / 41

LAMMPS_for_beginners - A brief survey of the LAMMPS MD...

This preview shows document pages 1 - 8. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online