2004
DOI: 10.1002/nme.1234
|View full text |Cite
|
Sign up to set email alerts
|

A Green's function approach to deriving non‐reflecting boundary conditions in molecular dynamics simulations

Abstract: SUMMARYComputer simulations of atomic scale processes in solids are often associated with the issue of spurious reflection of elastic waves at the boundaries of a molecular dynamics domain. In this paper, we propose an approach to emulate non-reflecting boundary conditions in atomistic simulations of crystalline solids. Harmonic response of the outer, non-simulated, region is accurately represented by a memory function, related to the lattice dynamics Green's function. The outward wave flow is cancelled due to… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
2

Citation Types

0
92
0

Year Published

2005
2005
2013
2013

Publication Types

Select...
4
4

Relationship

3
5

Authors

Journals

citations
Cited by 99 publications
(92 citation statements)
references
References 16 publications
(27 reference statements)
0
92
0
Order By: Relevance
“…Differential equations of motion have the form (31), where the interaction force f ij (see Eq. (32)) is (38) where and f̃i j are the conservative (repulsive), viscous and random forces, respectively. These forces can be expressed as follows (see, for example, [3,10,12,21]): (39) (40) (41) where a is a material coefficient, r i and r j are the position vectors of particles "i" and "j", r ij = r i − r j , r ij = ||r i − r j ||, and ; r max is the radius of interaction domain between particles; v ij = v i − v j is the relative velocity; γ is the friction coefficient or normal damping coefficient; w γ and w̃ are the weight functions for viscous and random forces, with the relation w γ = w̃2 [8,21]; σ is the random force amplitude (for unit mass) σ = (2γk B T ) 1/2 [8,21]; ζ ij is a random number with zero mean and unit variance (chosen independently for each pair of DPD particles and time step); and Δt is a time step size used in the integration of differential equations of motion.…”
Section: Differential Equations Of Motion According To Dpd Methodsmentioning
confidence: 99%
See 2 more Smart Citations
“…Differential equations of motion have the form (31), where the interaction force f ij (see Eq. (32)) is (38) where and f̃i j are the conservative (repulsive), viscous and random forces, respectively. These forces can be expressed as follows (see, for example, [3,10,12,21]): (39) (40) (41) where a is a material coefficient, r i and r j are the position vectors of particles "i" and "j", r ij = r i − r j , r ij = ||r i − r j ||, and ; r max is the radius of interaction domain between particles; v ij = v i − v j is the relative velocity; γ is the friction coefficient or normal damping coefficient; w γ and w̃ are the weight functions for viscous and random forces, with the relation w γ = w̃2 [8,21]; σ is the random force amplitude (for unit mass) σ = (2γk B T ) 1/2 [8,21]; ζ ij is a random number with zero mean and unit variance (chosen independently for each pair of DPD particles and time step); and Δt is a time step size used in the integration of differential equations of motion.…”
Section: Differential Equations Of Motion According To Dpd Methodsmentioning
confidence: 99%
“…We implement boundary conditions, such as, for example, no-slip, Maxwellian or specular boundary conditions at the walls, or periodicity conditions at surfaces where particles are entering or leaving the flow domain [50][51][52][53][54][55]. Similar boundary conditions have been introduced in the MD, MD-FE multiscale modeling [26,38,56,57]. Boundary conditions at the common FE-DPD boundary are described in Section 3.4.…”
Section: Differential Equations Of Motion According To Dpd Methodsmentioning
confidence: 99%
See 1 more Smart Citation
“…The THK is the optimal solution for propagation in linear media and, with perfect precision arithmetic and an arbitrarily long history, it is reflectionless. For more complex, three-dimensional (3D) systems the kernel can be determined numerically [22] or analytically [91], although evaluation of the analytical expressions for typical 3D systems remains a challenge. In all cases, the kernel unfortunately has a very slow and oscillatory decay in time and therefore is very expensive computationally.…”
Section: Introductionmentioning
confidence: 99%
“…strongly coupled lattice structures, are the following: (1) energy dissipation is modelled with the viscous friction model, where the frictional force is proportional to the instantaneous atomic velocities; however, energy dissipation in lattice structures is determined by a time history of the atomic motion [5][6][7][8][9]; (2) initial conditions are sampled as independent random quantities, though there is a statistical correlation in the motion of adjacent lattice atoms.…”
Section: Introductionmentioning
confidence: 99%