1: \begin{abstract}

2: Generalized Langevin dynamics (GLD) arise in the modeling of a number of 

3: systems, ranging from structured fluids that exhibit a viscoelastic mechanical 

4: response, to biological systems, and other media that exhibit anomalous 

5: diffusive phenomena.  Molecular dynamics (MD) simulations that include GLD in 

6: conjunction with external and/or pairwise forces require the development of 

7: numerical integrators that are efficient, stable, and have known convergence 

8: properties. In this article, we derive a family of extended variable 

9: integrators for the Generalized Langevin equation (GLE) with a positive Prony 

10: series memory kernel.  Using stability and error analysis, we identify a 

11: superlative choice of parameters and implement the corresponding numerical 

12: algorithm in the LAMMPS MD software package. 

13: Salient features of the algorithm include exact conservation of the first and 

14: second moments of the equilibrium velocity distribution in some important cases, 

15: stable behavior in the limit of conventional Langevin dynamics, and the use of a convolution-free formalism 

16: that obviates the need for explicit storage of the time history of particle 

17: velocities.  Capability is demonstrated with respect to accuracy in numerous

18: canonical examples, stability in certain limits, and an exemplary application 

19: in which the effect of a harmonic confining potential is mapped onto a memory

20: kernel. \\ 

21: \vspace{1in} 

22: 

23: {\footnotesize \it \noindent Copyright 2013 American Institute of Physics. This article may be downloaded for personal use only. Any other use requires prior permission of the author and the American Institute of Physics.}

24: \end{abstract}

25: