next up previous

1.1 Fluids

Made up of classical mass points or rigid bodies, interacting by pair forces (and possibly torques).

Table 1.1: Isotropic model potentials in statistical-mechanical simulation: $u=u(r)$
Hard spheres $u(r)= \infty \;\;{\rm if}  r < r_{0}$
$= 0 \;\;\; {\rm if}  r \geq r_{0} $
First approximation in many applications

u(r)= 4  \epsilon \left[ \left(\frac{r}{\sigma}\right)^{-12}
- \left(\frac{r}{\sigma}\right)^{-6} \right]

Noble gas atoms; nearly spherical molecules
Isotropic Kihara $ \textstyle{
u(r)=}$ $\textstyle{
4   \epsilon \left[ \left(
- \left(\frac{\textstyle{r-a}}{\textstyle{\sigma-a}}\right)^{-6} \right]
$ Noble gas atoms; nearly spherical molecules ( $a=0.05-0.1   \sigma$)
Harmonic $u(r)= A   \left( r-r_{0}\right)^{2}$ Intramolecular bonds, if $kT$ is small compared to the bond energy
Morse $u(r)=
A \left[ e^{\textstyle{-2b(r\!-\!r_{0})}}-
2 e^{\textstyle{-b(r\!-\!r_{0})}} \right] $ Intramolecular bonds, if $kT$ is comparable to the bond energy
Born-Huggins-Mayer $u(r)=
\frac{\textstyle{q_{1}q_{2}}}{\textstyle{4\pi \epsilon_{0}r}}
+ B e^{\te...
$ Ionic melts; $q_{i}$ are the ion charges

Table: Anisotropic model potentials in statistical-mechanical simulation: $u(12)=u(\mbox{$\bf r$}_{12},\mbox{$\bf e$}_{1},\mbox{$\bf e$}_{2})$
Hard dumbbells,
hard spherocylinders, etc.
$u(12)= \infty \;\;$ if overlap
$= 0 \;\;\;$ otherwise
First approximation to rigid molecules
Interaction site models, rigid $u(12)=$ sum of isotropic pair energies $\textstyle{u(r_{i(1),j(2)})}$, where several interaction sites $i$ and $j$ are in fixed positions on molecules $1$ and $2$, respectively Rigid molecules
Interaction site models with non-rigid bonds $u(12)=$ sum of isotropic pair energies, both intra- and intermolecular Non-rigid molecules
Kramers-type $u(12)=$ sum of isotropic pair energies, exclusively between sites on different molecules; certain intramolecular distances (bonds) and/or angles are fixed Flexible molecules, from ethane to biopolymers
Stockmayer $u(12)=$ Lennard-Jones + point dipoles First approximation to small polar molecules
Anisotropic Kihara $u(12)=
4   \epsilon \left[ \left(
- \left(
\frac{\textstyle{\rho_{12}}}{\textstyle{\sigma}}\right)^{-6} \right] $
where $\rho_{12}$ is the shortest distance between two linear rods
Rigid linear molecules
with distributed
Lennard-Jones interaction
Gay-Berne $u(12)=$ $4  \epsilon(12)
\left[ \left(
\right. $
$ - \left. \left(
\frac{\textstyle{r_{12}-\sigma(12)+\sigma_{0}}}{\textstyle{\sigma_{0}}}\right)^{-6} \right]
where $\sigma(12)$ and $\epsilon(12)$ depend on $\mbox{$\bf r$}_{12},\mbox{$\bf e$}_{1},\mbox{$\bf e$}_{2}$ and certain substance-specific shape parameters
Liquid crystal molecules of ellipsoidal shape, with smoothly distributed Lennard-Jones sites

At time $t$, combine the position vectors of the $N$ atoms into a vector $\mbox{$\bf\Gamma$}_{c}\equiv$ $\{\mbox{$\bf r$}_{1} \dots $ $\mbox{$\bf r$}_{N}\}$. The set of all possible such vectors spans the $3N$-dimensional ``configuration space'' $\Gamma_{c}$.

Let $a(\mbox{$\bf\Gamma$}_{c})$ be a property of the $N$-body system, depending on the positions of all particles (i.e. of the microstate $\mbox{$\bf\Gamma$}_{c}$). The thermodynamic average of $a$ is given by

\langle a \rangle = \int_{\Gamma_{c}} a(\mbox{$\bf\Gamma$}_{c})
 p(\mbox{$\bf\Gamma$}_{c})  d\mbox{$\bf\Gamma$}_{c}

where $p(\mbox{$\bf\Gamma$}_{c})$ is the probability density at the configuration space point $\mbox{$\bf\Gamma$}_{c}$.


Problem: the probability density $p(\mbox{$\bf\Gamma$}_{c})$ is known only up to an indetermined normalizing factor. Thus, $p_{can}(\mbox{$\bf\Gamma$}_{c})$ $\propto exp[-E(\mbox{$\bf\Gamma$}_{c})/kT]$, but the normalizing denominator $Q$ (the Configurational Partition Function) in

1 = \int p(\mbox{$\bf\Gamma$}_{c})  d\mbox{$\bf\Gamma$}_{c}...
...tyle -E(\mbox{$\bf\Gamma$}_{c})/kT}   d\mbox{$\bf\Gamma$}_{c}

is not known.

next up previous
F. J. Vesely / University of Vienna