Differences
This shows you the differences between two versions of the page.
| Next revision | Previous revision Last revision Both sides next revision | ||
|
user-meso:main [2014/09/18 14:47] ytang created |
user-meso:main [2015/03/17 02:45] ytang |
||
|---|---|---|---|
| Line 1: | Line 1: | ||
| - | ''//<sub>USER</sub>//MESO'' is a GPU-accelerated package for mesoscopic modelling using **LAMMPS** \cite{plimpton1995lammps}. Currently supported methods include Dissipative Particle Dynamics (DPD) and Smoothed Particle Hydrodynamics (SPH). The software is written in a combination of C++ and CUDA C++ language. It achieves tremendous speedup over the CPU code, while at the same time provide good scaling performance. | + | ====== USER-MESO ====== |
| - | As a LAMMPS package, its usage pattern follows that of the trunk code. It is shipped with a complete collection of atom styles, fix styles, pair styles, bonded interaction styles and a new integrator style. | ||
| - | \section{Dissipative Particle Dynamics} | + | ''//<sub>USER</sub>//MESO'' is a GPU-accelerated package of **LAMMPS** \cite{plimpton1995lammps} for mesoscopic modelling using Dissipative Particle Dynamics and Smoothed Particle Hydrodyanmics. The code achieves tremendous speedup over the CPU code, and also exhibits good scalability. |
| - | Dissipative Particle Dynamics (DPD) method is a stochastic particle simulation method particularly useful for the model of soft matter systems such as polymer melt and solution. Detailed description of the implementation of the DPD functionality within \USERMESO can be found in Ref.~\cite{tang2013gpudpd}. | + | Existing LAMMPS users can follow the [[user-meso:quickstart|quick start guide]] to obtain and set up USER-MESO. |
| - | \subsection{The DPD formulation} | + | As a LAMMPS package, its usage pattern follows that of the trunk code. It is shipped with a complete collection of atom styles, fix styles, pair styles, bonded interaction styles and a new integrator style. |
| - | \label{sec:dpd_formulation} | + | |
| - | In DPD, the force $\textbf{f}$ acting on each particle consists of three pairwise additive parts: a conservative one, a dissipative one and a random one, \textit{i.e.} | ||
| - | \begin{align} | ||
| - | \textbf{f}_i = \sum_{i\neq j} \textbf{F}_{ij} = \sum_{i\neq j} ( \textbf{F}_{ij}^C + \textbf{F}_{ij}^D + \textbf{F}_{ij}^R ) | ||
| - | \end{align} | ||
| - | given by | ||
| - | \begin{align} | ||
| - | \textbf{F}_{ij}^C &= a_{ij} w_C(\textbf{r}_{ij}) \textbf{e}_{ij} \\ | ||
| - | \textbf{F}_{ij}^D &= -\gamma_{ij} w_D(\textbf{r}_{ij}) (\textbf{e}_{ij} \cdot \textbf{v}_{ij}) \textbf{e}_{ij} \\ | ||
| - | \textbf{F}_{ij}^R &= \sigma_{ij} w_R(\textbf{r}_{ij}) \xi_{ij} {\delta t}^{-\frac{1}{2}} \textbf{e}_{ij} | ||
| - | \end{align} | ||
| - | if $ |\textbf{r}_{ij}| \leq r_c $, where $r_c$ is the cutoff distance, and | ||
| - | \begin{align} | ||
| - | \textbf{F}_{ij}^C = \textbf{F}_{ij}^D &= \textbf{F}_{ij}^R = 0,\;\;\; |\textbf{r}_{ij}| > r_c. | ||
| - | \end{align} | ||
| - | Between the two weight functions $w_D$ (dissipative) and $w_R$ (random), one of them can be chosen arbitrarily, while the other is then fixed as dictated by the fluctuation-dissipation theorem \cite{espanol1995dpd}. | ||
| - | \begin{align} | ||
| - | w_D(\textbf{r}_{ij}) = w_R^2(\textbf{r}_{ij}) | ||
| - | \end{align} | ||
| - | In \USERMESO, $w_R$ (random) is assumed to be a power of $w_C$ (conservative): | ||
| - | \begin{align} | ||
| - | w_R(\textbf{r}_{ij}) = w_C^s(\textbf{r}_{ij}) = (1-\frac{|\textbf{r}_{ij}|}{r_c})^s | ||
| - | \end{align} | ||
| - | The choice of exponent $s$ is somewhat arbitrary and can be conveniently tuned for reproducing dynamical properties such as viscosity and diffusivity. The coefficients $\sigma_{ij}$ and $\gamma_{ij}$ are also related to each other by | ||
| - | \begin{align} | ||
| - | \sigma_{ij}^2 = 2 \gamma_{ij} k_B T | ||
| - | \end{align} | ||
| - | as also dictated by the fluctuation-dissipation theorem. | ||
| \subsection{Quick Start Guide} | \subsection{Quick Start Guide} | ||
