ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/langevin/langevin.tex
Revision: 3317
Committed: Fri Jan 18 22:07:35 2008 UTC (16 years, 7 months ago) by xsun
Content type: application/x-tex
File size: 58027 byte(s)
Log Message:
add one ref

File Contents

# User Rev Content
1 tim 2746 %\documentclass[prb,aps,twocolumn,tabularx]{revtex4}
2     %\documentclass[aps,prb,preprint]{revtex4}
3     \documentclass[11pt]{article}
4     \usepackage{endfloat}
5     \usepackage{amsmath,bm}
6     \usepackage{amssymb}
7     \usepackage{times}
8     \usepackage{mathptmx}
9     \usepackage{setspace}
10     \usepackage{tabularx}
11     \usepackage{graphicx}
12     \usepackage{booktabs}
13     \usepackage{bibentry}
14     \usepackage{mathrsfs}
15     \usepackage[ref]{overcite}
16     \pagestyle{plain} \pagenumbering{arabic} \oddsidemargin 0.0cm
17     \evensidemargin 0.0cm \topmargin -21pt \headsep 10pt \textheight
18     9.0in \textwidth 6.5in \brokenpenalty=10000
19     \renewcommand{\baselinestretch}{1.2}
20 gezelter 3299 \renewcommand\citemid{\ } % no comma in optional referenc note
21 tim 2746
22     \begin{document}
23    
24 gezelter 3205 \title{An algorithm for performing Langevin dynamics on rigid bodies of arbitrary shape }
25 tim 2746
26 gezelter 3299 \author{Xiuquan Sun, Teng Lin and J. Daniel Gezelter\footnote{Corresponding author. \ Electronic mail:
27 tim 2746 gezelter@nd.edu} \\
28     Department of Chemistry and Biochemistry\\
29     University of Notre Dame\\
30     Notre Dame, Indiana 46556}
31    
32     \date{\today}
33    
34     \maketitle \doublespacing
35    
36     \begin{abstract}
37    
38     \end{abstract}
39    
40     \newpage
41    
42     %\narrowtext
43    
44     %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
45     % BODY OF TEXT
46     %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
47    
48     \section{Introduction}
49    
50     %applications of langevin dynamics
51 gezelter 3316 Langevin dynamics, which mimics a simple heat bath with stochastic and
52     dissipative forces, has been applied in a variety of situations as an
53     alternative to molecular dynamics with explicit solvent molecules.
54     The stochastic treatment of the solvent allows the use of simulations
55     with substantially longer time and length scales. In general, the
56     dynamic and structural properties obtained from Langevin simulations
57     agree quite well with similar properties obtained from explicit
58     solvent simulations.
59    
60     Recent examples of the usefulness of Langevin simulations include a
61     study of met-enkephalin in which Langevin simulations predicted
62     dynamical properties that were largely in agreement with explicit
63     solvent simulations.\cite{Shen2002} By applying Langevin dynamics with
64     the UNRES model, Liow and his coworkers suggest that protein folding
65     pathways can be explored within a reasonable amount of
66     time.\cite{Liwo2005}
67    
68     The stochastic nature of Langevin dynamics also enhances the sampling
69     of the system and increases the probability of crossing energy
70     barriers.\cite{Cui2003,Banerjee2004} Combining Langevin dynamics with
71     Kramers's theory, Klimov and Thirumalai identified free-energy
72     barriers by studying the viscosity dependence of the protein folding
73     rates.\cite{Klimov1997} In order to account for solvent induced
74     interactions missing from the implicit solvent model, Kaya
75     incorporated a desolvation free energy barrier into protein
76     folding/unfolding studies and discovered a higher free energy barrier
77 xsun 3317 between the native and denatured states.\cite{HuseyinKaya07012005}
78 gezelter 3316
79     Because of its stability against noise, Langevin dynamics has also
80     proven useful for studying remagnetization processes in various
81     systems.\cite{Palacios1998,Berkov2002,Denisov2003} [Check: For
82 tim 2746 instance, the oscillation power spectrum of nanoparticles from
83 gezelter 3316 Langevin dynamics has the same peak frequencies for different wave
84     vectors, which recovers the property of magnetic excitations in small
85     finite structures.\cite{Berkov2005a}]
86 tim 2746
87 gezelter 3316 In typical LD simulations, the friction and random forces on
88     individual atoms are taken from the Stokes-Einstein hydrodynamic
89     approximation,
90     \begin{eqnarray}
91     m \dot{v}(t) & = & -\nabla U(x) - \xi m v(t) + R(t) \\
92     \langle R(t) \rangle & = & 0 \\
93     \langle R(t) R(t') \rangle & = & 2 k_B T \xi m \delta(t - t')
94     \end{eqnarray}
95     where $\xi \approx 6 \pi \eta a$. Here $\eta$ is the viscosity of the
96     implicit solvent, and $a$ is the hydrodynamic radius of the atom.
97 tim 2746
98 gezelter 3316 The use of rigid substructures,\cite{???}
99     coarse-graining,\cite{Ayton,Sun,Zannoni} and ellipsoidal
100     representations of protein side chains~\cite{Schulten} has made the
101     use of the Stokes-Einstein approximation problematic. A rigid
102     substructure moves as a single unit with orientational as well as
103     translational degrees of freedom. This requires a more general
104     treatment of the hydrodynamics than the spherical approximation
105     provides. The atoms involved in a rigid or coarse-grained structure
106     should properly have solvent-mediated interactions with each
107     other. The theory of interactions {\it between} bodies moving through
108     a fluid has been developed over the past century and has been applied
109     to simulations of Brownian
110     motion.\cite{MarshallNewton,GarciaDeLaTorre} There a need to have a
111     more thorough treatment of hydrodynamics included in the library of
112     methods available for performing Langevin simulations.
113 tim 2746
114 gezelter 3316 \subsection{Rigid Body Dynamics}
115     Rigid bodies are frequently involved in the modeling of large
116     collections of particles that move as a single unit. In molecular
117     simulations, rigid bodies have been used to simplify protein-protein
118     docking,\cite{Gray2003} and lipid bilayer simulations.\cite{Sun2008}
119     Many of the water models in common use are also rigid-body
120     models,\cite{TIPs,SPC/E} although they are typically evolved using
121     constraints rather than rigid body equations of motion.
122    
123     Euler angles are a natural choice to describe the rotational
124     degrees of freedom. However, due to $1 \over \sin \theta$
125     singularities, the numerical integration of corresponding equations of
126     these motion can become inaccurate (and inefficient). Although an
127     alternative integrator using multiple sets of Euler angles can
128     overcome this problem,\cite{Barojas1973} the computational penalty and
129     the loss of angular momentum conservation remain. A singularity-free
130     representation utilizing quaternions was developed by Evans in
131     1977.\cite{Evans1977} Unfortunately, this approach uses a nonseparable
132     Hamiltonian resulting from the quaternion representation, which
133     prevented symplectic algorithms from being utilized until very
134     recently.\cite{Miller2002} Another approach is the application of
135     holonomic constraints to the atoms belonging to the rigid body. Each
136     atom moves independently under the normal forces deriving from
137     potential energy and constraint forces which are used to guarantee the
138     rigidness. However, due to their iterative nature, the SHAKE and
139     Rattle algorithms also converge very slowly when the number of
140     constraints increases.\cite{Ryckaert1977,Andersen1983}
141    
142     A breakthrough in geometric literature suggests that, in order to
143 tim 2746 develop a long-term integration scheme, one should preserve the
144 tim 2999 symplectic structure of the propagator. By introducing a conjugate
145     momentum to the rotation matrix $Q$ and re-formulating Hamiltonian's
146 gezelter 3316 equation, a symplectic integrator, RSHAKE,\cite{Kol1997} was proposed
147     to evolve the Hamiltonian system in a constraint manifold by
148     iteratively satisfying the orthogonality constraint $Q^T Q = 1$. An
149     alternative method using the quaternion representation was developed
150     by Omelyan.\cite{Omelyan1998} However, both of these methods are
151     iterative and suffer from some related inefficiencies. A symplectic
152     Lie-Poisson integrator for rigid bodies developed by Dullweber {\it et
153     al.}\cite{Dullweber1997} gets around most of the limitations mentioned
154     above and has become the basis for our Langevin integrator.
155 tim 2746
156 gezelter 3316
157     \subsection{The Hydrodynamic tensor and Brownian dynamics}
158     Combining Brownian dynamics with rigid substructures, one can study
159     slow processes in biomolecular systems. Modeling DNA as a chain of
160     beads which are subject to harmonic potentials as well as excluded
161     volume potentials, Mielke and his coworkers discovered rapid
162     superhelical stress generations from the stochastic simulation of twin
163     supercoiling DNA with response to induced torques.\cite{Mielke2004}
164     Membrane fusion is another key biological process which controls a
165     variety of physiological functions, such as release of
166     neurotransmitters \textit{etc}. A typical fusion event happens on the
167     time scale of a millisecond, which is impractical to study using
168     atomistic models with newtonian mechanics. With the help of
169     coarse-grained rigid body model and stochastic dynamics, the fusion
170     pathways were explored by Noguchi and others.\cite{Noguchi2001,Noguchi2002,Shillcock2005}
171    
172     Due to the difficulty of numerically integrating anisotropic
173     rotational motion, most of the coarse-grained rigid body models are
174     treated as spheres, cylinders, ellipsoids or other regular shapes in
175     stochastic simulations. In an effort to account for the diffusion
176     anisotropy of arbitrarily-shaped particles, Fernandes and Garc\'{i}a
177     de la Torre improved the original Brownian dynamics simulation
178     algorithm~\cite{Ermak1978,Allison1991} by incorporating a generalized
179     $6\times6$ diffusion tensor and introducing a rotational evolution
180     scheme consisting of three consecutive rotations.\cite{Fernandes2002}
181     Unfortunately, biases are introduced into the system due to the
182 tim 2999 arbitrary order of applying the noncommuting rotation
183     operators.\cite{Beard2003} Based on the observation the momentum
184 tim 2746 relaxation time is much less than the time step, one may ignore the
185 gezelter 3316 inertia in Brownian dynamics. However, the assumption of zero average
186     acceleration is not always true for cooperative motion which is common
187     in proteins. An inertial Brownian dynamics (IBD) was proposed to
188     address this issue by adding an inertial correction
189 tim 2999 term.\cite{Beard2000} As a complement to IBD which has a lower bound
190 tim 2746 in time step because of the inertial relaxation time, long-time-step
191     inertial dynamics (LTID) can be used to investigate the inertial
192     behavior of the polymer segments in low friction
193 tim 2999 regime.\cite{Beard2000} LTID can also deal with the rotational
194 tim 2746 dynamics for nonskew bodies without translation-rotation coupling by
195 gezelter 3316 separating the translation and rotation motion and taking advantage of
196     the analytical solution of hydrodynamics properties. However, typical
197     nonskew bodies like cylinders and ellipsoids are inadequate to
198     represent most complex macromolecular assemblies.
199 tim 2746
200     The goal of the present work is to develop a Langevin dynamics
201 tim 2999 algorithm for arbitrary-shaped rigid particles by integrating the
202 gezelter 3316 accurate estimation of friction tensor from hydrodynamics theory into
203     a symplectic rigid body dynamics propagator. In the sections below,
204     we review some of the theory of hydrodynamic tensors developed for
205     Brownian simulations of rigid multi-particle systems, we then present
206     our integration method for a set of generalized Langevin equations of
207     motion, and we compare the behavior of the new Langevin integrator to
208     dynamical quantities obtained via explicit solvent molecular dynamics.
209 tim 2746
210 gezelter 3316 \subsection{\label{introSection:frictionTensor}The Friction Tensor}
211     Theoretically, a complete friction kernel can be determined using the
212 tim 2999 velocity autocorrelation function. However, this approach becomes
213 gezelter 3316 impractical when the solute becomes complex. Instead, various
214     approaches based on hydrodynamics have been developed to calculate the
215     friction coefficients. In general, the friction tensor $\Xi$ is a
216     $6\times 6$ matrix given by
217     \begin{equation}
218 tim 2999 \Xi = \left( {\begin{array}{*{20}c}
219     {\Xi _{}^{tt} } & {\Xi _{}^{rt} } \\
220     {\Xi _{}^{tr} } & {\Xi _{}^{rr} } \\
221     \end{array}} \right).
222 gezelter 3316 \end{equation}
223     Here, $\Xi^{tt}$ and $\Xi^{rr}$ are $3 \times 3$ translational and
224     rotational resistance (friction) tensors respectively, while
225     $\Xi^{tr}$ is translation-rotation coupling tensor and $\Xi^{rt}$ is
226     rotation-translation coupling tensor. When a particle moves in a
227     fluid, it may experience friction force ($\mathbf{F}_f$) and torque
228     ($\mathbf{\tau}_f$) in opposition to the directions of the velocity
229     ($\mathbf{v}$) and body-fixed angular velocity ($\mathbf{\omega}$),
230     \begin{equation}
231 tim 2746 \left( \begin{array}{l}
232 gezelter 3316 \mathbf{F}_f \\
233     \mathbf{\tau}_f \\
234 tim 2746 \end{array} \right) = - \left( {\begin{array}{*{20}c}
235 gezelter 3316 \Xi ^{tt} & \Xi ^{rt} \\
236     \Xi ^{tr} & \Xi ^{rr} \\
237 tim 2746 \end{array}} \right)\left( \begin{array}{l}
238 gezelter 3316 \mathbf{v} \\
239     \mathbf{\omega} \\
240     \end{array} \right).
241     \end{equation}
242 tim 2746
243 tim 2999 \subsubsection{\label{introSection:resistanceTensorRegular}\textbf{The Resistance Tensor for Regular Shapes}}
244 gezelter 3316 For a spherical particle under ``stick'' boundary conditions, the
245     translational and rotational friction tensors can be calculated from
246     Stoke's law,
247     \begin{equation}
248     \Xi^{tt} = \left( {\begin{array}{*{20}c}
249 tim 2999 {6\pi \eta R} & 0 & 0 \\
250     0 & {6\pi \eta R} & 0 \\
251     0 & 0 & {6\pi \eta R} \\
252     \end{array}} \right)
253 gezelter 3316 \end{equation}
254 tim 2999 and
255 gezelter 3316 \begin{equation}
256 tim 2999 \Xi ^{rr} = \left( {\begin{array}{*{20}c}
257     {8\pi \eta R^3 } & 0 & 0 \\
258     0 & {8\pi \eta R^3 } & 0 \\
259     0 & 0 & {8\pi \eta R^3 } \\
260     \end{array}} \right)
261 gezelter 3316 \end{equation}
262 tim 2999 where $\eta$ is the viscosity of the solvent and $R$ is the
263     hydrodynamic radius.
264    
265     Other non-spherical shapes, such as cylinders and ellipsoids, are
266 gezelter 3316 widely used as references for developing new hydrodynamics theories,
267 tim 2999 because their properties can be calculated exactly. In 1936, Perrin
268     extended Stokes's law to general ellipsoids, also called a triaxial
269     ellipsoid, which is given in Cartesian coordinates
270 gezelter 3316 by\cite{Perrin1934,Perrin1936}
271     \begin{equation}
272     \frac{x^2 }{a^2} + \frac{y^2}{b^2} + \frac{z^2 }{c^2} = 1
273     \end{equation}
274     where the semi-axes are of lengths $a$, $b$, and $c$. Due to the
275     complexity of the elliptic integral, only uniaxial ellipsoids,
276     {\it i.e.} prolate ($ a \ge b = c$) and oblate ($ a < b = c $), can
277     be solved exactly. Introducing an elliptic integral parameter $S$ for
278     prolate ellipsoids :
279     \begin{equation}
280     S = \frac{2}{\sqrt{a^2 - b^2}} \ln \frac{a + \sqrt{a^2 - b^2}}{b},
281     \end{equation}
282 tim 2999 and oblate ellipsoids:
283 gezelter 3316 \begin{equation}
284     S = \frac{2}{\sqrt {b^2 - a^2 }} \arctan \frac{\sqrt {b^2 - a^2}}{a},
285     \end{equation}
286 tim 2999 one can write down the translational and rotational resistance
287 gezelter 3316 tensors for oblate,
288 tim 2999 \begin{eqnarray*}
289 gezelter 3316 \Xi_a^{tt} & = & 16\pi \eta \frac{a^2 - b^2}{(2a^2 - b^2 )S - 2a}. \\
290     \Xi_b^{tt} = \Xi_c^{tt} & = & 32\pi \eta \frac{a^2 - b^2 }{(2a^2 - 3b^2 )S + 2a},
291 tim 2999 \end{eqnarray*}
292 gezelter 3316 and prolate,
293 tim 2999 \begin{eqnarray*}
294 gezelter 3316 \Xi_a^{rr} & = & \frac{32\pi}{3} \eta \frac{(a^2 - b^2 )b^2}{2a - b^2 S}, \\
295     \Xi_b^{rr} = \Xi_c^{rr} & = & \frac{32\pi}{3} \eta \frac{(a^4 - b^4)}{(2a^2 - b^2 )S - 2a}
296 tim 2999 \end{eqnarray*}
297 gezelter 3316 ellipsoids. For both spherical and ellipsoidal particles, the
298     translation-rotation and rotation-translation coupling tensors are
299     zero.
300 tim 2746
301 tim 2999 \subsubsection{\label{introSection:resistanceTensorRegularArbitrary}\textbf{The Resistance Tensor for Arbitrary Shapes}}
302    
303     Unlike spherical and other simply shaped molecules, there is no
304     analytical solution for the friction tensor for arbitrarily shaped
305     rigid molecules. The ellipsoid of revolution model and general
306     triaxial ellipsoid model have been used to approximate the
307 gezelter 3316 hydrodynamic properties of rigid bodies. However, the mapping from all
308     possible ellipsoidal spaces, $r$-space, to all possible combination of
309     rotational diffusion coefficients, $D$-space, is not
310     unique.\cite{Wegener1979} Additionally, because there is intrinsic
311     coupling between translational and rotational motion of rigid bodies,
312     general ellipsoids are not always suitable for modeling arbitrarily
313     shaped rigid molecules. A number of studies have been devoted to
314 tim 2999 determining the friction tensor for irregularly shaped rigid bodies
315 gezelter 3316 using more advanced methods where the molecule of interest was modeled
316     by a combinations of spheres\cite{Carrasco1999} and the hydrodynamics
317     properties of the molecule can be calculated using the hydrodynamic
318     interaction tensor. Let us consider a rigid assembly of $N$ beads
319     immersed in a continuous medium. Due to hydrodynamic interaction, the
320     ``net'' velocity of $i$th bead, $v'_i$ is different than its
321     unperturbed velocity $v_i$,
322 tim 2746 \[
323 tim 2999 v'_i = v_i - \sum\limits_{j \ne i} {T_{ij} F_j }
324 tim 2746 \]
325 tim 2999 where $F_i$ is the frictional force, and $T_{ij}$ is the
326     hydrodynamic interaction tensor. The friction force of $i$th bead is
327     proportional to its ``net'' velocity
328 tim 2746 \begin{equation}
329 tim 2999 F_i = \zeta _i v_i - \zeta _i \sum\limits_{j \ne i} {T_{ij} F_j }.
330     \label{introEquation:tensorExpression}
331 tim 2746 \end{equation}
332 tim 2999 This equation is the basis for deriving the hydrodynamic tensor. In
333     1930, Oseen and Burgers gave a simple solution to
334     Eq.~\ref{introEquation:tensorExpression}
335 tim 2746 \begin{equation}
336 tim 2999 T_{ij} = \frac{1}{{8\pi \eta r_{ij} }}\left( {I + \frac{{R_{ij}
337     R_{ij}^T }}{{R_{ij}^2 }}} \right). \label{introEquation:oseenTensor}
338 tim 2746 \end{equation}
339 tim 2999 Here $R_{ij}$ is the distance vector between bead $i$ and bead $j$.
340     A second order expression for element of different size was
341     introduced by Rotne and Prager\cite{Rotne1969} and improved by
342     Garc\'{i}a de la Torre and Bloomfield,\cite{Torre1977}
343 tim 2746 \begin{equation}
344 tim 2999 T_{ij} = \frac{1}{{8\pi \eta R_{ij} }}\left[ {\left( {I +
345     \frac{{R_{ij} R_{ij}^T }}{{R_{ij}^2 }}} \right) + R\frac{{\sigma
346     _i^2 + \sigma _j^2 }}{{r_{ij}^2 }}\left( {\frac{I}{3} -
347     \frac{{R_{ij} R_{ij}^T }}{{R_{ij}^2 }}} \right)} \right].
348     \label{introEquation:RPTensorNonOverlapped}
349 tim 2746 \end{equation}
350 tim 2999 Both of the Eq.~\ref{introEquation:oseenTensor} and
351     Eq.~\ref{introEquation:RPTensorNonOverlapped} have an assumption
352     $R_{ij} \ge \sigma _i + \sigma _j$. An alternative expression for
353     overlapping beads with the same radius, $\sigma$, is given by
354 tim 2746 \begin{equation}
355 tim 2999 T_{ij} = \frac{1}{{6\pi \eta R_{ij} }}\left[ {\left( {1 -
356     \frac{2}{{32}}\frac{{R_{ij} }}{\sigma }} \right)I +
357     \frac{2}{{32}}\frac{{R_{ij} R_{ij}^T }}{{R_{ij} \sigma }}} \right]
358     \label{introEquation:RPTensorOverlapped}
359 tim 2746 \end{equation}
360 tim 2999 To calculate the resistance tensor at an arbitrary origin $O$, we
361     construct a $3N \times 3N$ matrix consisting of $N \times N$
362     $B_{ij}$ blocks
363     \begin{equation}
364     B = \left( {\begin{array}{*{20}c}
365     {B_{11} } & \ldots & {B_{1N} } \\
366     \vdots & \ddots & \vdots \\
367     {B_{N1} } & \cdots & {B_{NN} } \\
368     \end{array}} \right),
369     \end{equation}
370     where $B_{ij}$ is given by
371 tim 2746 \[
372 tim 2999 B_{ij} = \delta _{ij} \frac{I}{{6\pi \eta R}} + (1 - \delta _{ij}
373     )T_{ij}
374 tim 2746 \]
375 tim 2999 where $\delta _{ij}$ is the Kronecker delta function. Inverting the
376     $B$ matrix, we obtain
377 tim 2746 \[
378 tim 2999 C = B^{ - 1} = \left( {\begin{array}{*{20}c}
379     {C_{11} } & \ldots & {C_{1N} } \\
380     \vdots & \ddots & \vdots \\
381     {C_{N1} } & \cdots & {C_{NN} } \\
382     \end{array}} \right),
383 tim 2746 \]
384 tim 2999 which can be partitioned into $N \times N$ $3 \times 3$ block
385     $C_{ij}$. With the help of $C_{ij}$ and the skew matrix $U_i$
386 tim 2746 \[
387 tim 2999 U_i = \left( {\begin{array}{*{20}c}
388     0 & { - z_i } & {y_i } \\
389     {z_i } & 0 & { - x_i } \\
390     { - y_i } & {x_i } & 0 \\
391     \end{array}} \right)
392 tim 2746 \]
393 tim 2999 where $x_i$, $y_i$, $z_i$ are the components of the vector joining
394     bead $i$ and origin $O$, the elements of resistance tensor at
395     arbitrary origin $O$ can be written as
396     \begin{eqnarray}
397     \Xi _{}^{tt} & = & \sum\limits_i {\sum\limits_j {C_{ij} } } \notag , \\
398     \Xi _{}^{tr} & = & \Xi _{}^{rt} = \sum\limits_i {\sum\limits_j {U_i C_{ij} } } , \\
399 gezelter 3310 \Xi _{}^{rr} & = & - \sum\limits_i {\sum\limits_j {U_i C_{ij} } }
400     U_j + 6 \eta V {\bf I}. \notag
401 tim 2999 \label{introEquation:ResistanceTensorArbitraryOrigin}
402     \end{eqnarray}
403 gezelter 3310 The final term in the expression for $\Xi^{rr}$ is correction that
404     accounts for errors in the rotational motion of certain kinds of bead
405     models. The additive correction uses the solvent viscosity ($\eta$)
406     as well as the total volume of the beads that contribute to the
407     hydrodynamic model,
408     \begin{equation}
409     V = \frac{4 \pi}{3} \sum_{i=1}^{N} \sigma_i^3,
410     \end{equation}
411     where $\sigma_i$ is the radius of bead $i$. This correction term was
412     rigorously tested and compared with the analytical results for
413     two-sphere and ellipsoidal systems by Garcia de la Torre and
414     Rodes.\cite{Torre:1983lr}
415    
416    
417 tim 2999 The resistance tensor depends on the origin to which they refer. The
418     proper location for applying the friction force is the center of
419     resistance (or center of reaction), at which the trace of rotational
420     resistance tensor, $ \Xi ^{rr}$ reaches a minimum value.
421     Mathematically, the center of resistance is defined as an unique
422     point of the rigid body at which the translation-rotation coupling
423     tensors are symmetric,
424     \begin{equation}
425     \Xi^{tr} = \left( {\Xi^{tr} } \right)^T
426     \label{introEquation:definitionCR}
427     \end{equation}
428     From Equation \ref{introEquation:ResistanceTensorArbitraryOrigin},
429     we can easily derive that the translational resistance tensor is
430     origin independent, while the rotational resistance tensor and
431     translation-rotation coupling resistance tensor depend on the
432     origin. Given the resistance tensor at an arbitrary origin $O$, and
433     a vector ,$r_{OP}(x_{OP}, y_{OP}, z_{OP})$, from $O$ to $P$, we can
434     obtain the resistance tensor at $P$ by
435     \begin{equation}
436     \begin{array}{l}
437     \Xi _P^{tt} = \Xi _O^{tt} \\
438     \Xi _P^{tr} = \Xi _P^{rt} = \Xi _O^{tr} - U_{OP} \Xi _O^{tt} \\
439     \Xi _P^{rr} = \Xi _O^{rr} - U_{OP} \Xi _O^{tt} U_{OP} + \Xi _O^{tr} U_{OP} - U_{OP} \Xi _O^{{tr} ^{^T }} \\
440     \end{array}
441     \label{introEquation:resistanceTensorTransformation}
442     \end{equation}
443     where
444 tim 2746 \[
445 tim 2999 U_{OP} = \left( {\begin{array}{*{20}c}
446     0 & { - z_{OP} } & {y_{OP} } \\
447     {z_i } & 0 & { - x_{OP} } \\
448     { - y_{OP} } & {x_{OP} } & 0 \\
449     \end{array}} \right)
450 tim 2746 \]
451 tim 2999 Using Eq.~\ref{introEquation:definitionCR} and
452     Eq.~\ref{introEquation:resistanceTensorTransformation}, one can
453     locate the position of center of resistance,
454     \begin{eqnarray*}
455     \left( \begin{array}{l}
456     x_{OR} \\
457     y_{OR} \\
458     z_{OR} \\
459     \end{array} \right) & = &\left( {\begin{array}{*{20}c}
460     {(\Xi _O^{rr} )_{yy} + (\Xi _O^{rr} )_{zz} } & { - (\Xi _O^{rr} )_{xy} } & { - (\Xi _O^{rr} )_{xz} } \\
461     { - (\Xi _O^{rr} )_{xy} } & {(\Xi _O^{rr} )_{zz} + (\Xi _O^{rr} )_{xx} } & { - (\Xi _O^{rr} )_{yz} } \\
462     { - (\Xi _O^{rr} )_{xz} } & { - (\Xi _O^{rr} )_{yz} } & {(\Xi _O^{rr} )_{xx} + (\Xi _O^{rr} )_{yy} } \\
463     \end{array}} \right)^{ - 1} \\
464     & & \left( \begin{array}{l}
465     (\Xi _O^{tr} )_{yz} - (\Xi _O^{tr} )_{zy} \\
466     (\Xi _O^{tr} )_{zx} - (\Xi _O^{tr} )_{xz} \\
467     (\Xi _O^{tr} )_{xy} - (\Xi _O^{tr} )_{yx} \\
468     \end{array} \right) \\
469     \end{eqnarray*}
470     where $x_OR$, $y_OR$, $z_OR$ are the components of the vector
471     joining center of resistance $R$ and origin $O$.
472 tim 2746
473    
474 gezelter 3310 \section{Langevin Dynamics for Rigid Particles of Arbitrary Shape\label{LDRB}}
475 tim 2999 Consider the Langevin equations of motion in generalized coordinates
476 tim 2746 \begin{equation}
477     M_i \dot V_i (t) = F_{s,i} (t) + F_{f,i(t)} + F_{r,i} (t)
478     \label{LDGeneralizedForm}
479     \end{equation}
480     where $M_i$ is a $6\times6$ generalized diagonal mass (include mass
481     and moment of inertial) matrix and $V_i$ is a generalized velocity,
482 tim 2999 $V_i = V_i(v_i,\omega _i)$. The right side of
483     Eq.~\ref{LDGeneralizedForm} consists of three generalized forces in
484 tim 2746 lab-fixed frame, systematic force $F_{s,i}$, dissipative force
485     $F_{f,i}$ and stochastic force $F_{r,i}$. While the evolution of the
486     system in Newtownian mechanics typically refers to lab-fixed frame,
487     it is also convenient to handle the rotation of rigid body in
488     body-fixed frame. Thus the friction and random forces are calculated
489     in body-fixed frame and converted back to lab-fixed frame by:
490     \[
491     \begin{array}{l}
492 tim 2999 F_{f,i}^l (t) = Q^T F_{f,i}^b (t), \\
493     F_{r,i}^l (t) = Q^T F_{r,i}^b (t). \\
494     \end{array}
495 tim 2746 \]
496     Here, the body-fixed friction force $F_{r,i}^b$ is proportional to
497     the body-fixed velocity at center of resistance $v_{R,i}^b$ and
498 tim 2999 angular velocity $\omega _i$
499 tim 2746 \begin{equation}
500     F_{r,i}^b (t) = \left( \begin{array}{l}
501     f_{r,i}^b (t) \\
502     \tau _{r,i}^b (t) \\
503     \end{array} \right) = - \left( {\begin{array}{*{20}c}
504     {\Xi _{R,t} } & {\Xi _{R,c}^T } \\
505     {\Xi _{R,c} } & {\Xi _{R,r} } \\
506     \end{array}} \right)\left( \begin{array}{l}
507     v_{R,i}^b (t) \\
508     \omega _i (t) \\
509     \end{array} \right),
510     \end{equation}
511     while the random force $F_{r,i}^l$ is a Gaussian stochastic variable
512     with zero mean and variance
513     \begin{equation}
514     \left\langle {F_{r,i}^l (t)(F_{r,i}^l (t'))^T } \right\rangle =
515     \left\langle {F_{r,i}^b (t)(F_{r,i}^b (t'))^T } \right\rangle =
516 tim 2999 2k_B T\Xi _R \delta (t - t'). \label{randomForce}
517 tim 2746 \end{equation}
518     The equation of motion for $v_i$ can be written as
519     \begin{equation}
520     m\dot v_i (t) = f_{t,i} (t) = f_{s,i} (t) + f_{f,i}^l (t) +
521     f_{r,i}^l (t)
522     \end{equation}
523     Since the frictional force is applied at the center of resistance
524     which generally does not coincide with the center of mass, an extra
525     torque is exerted at the center of mass. Thus, the net body-fixed
526     frictional torque at the center of mass, $\tau _{n,i}^b (t)$, is
527     given by
528     \begin{equation}
529     \tau _{r,i}^b = \tau _{r,i}^b +r_{MR} \times f_{r,i}^b
530     \end{equation}
531     where $r_{MR}$ is the vector from the center of mass to the center
532 tim 2999 of the resistance. Instead of integrating the angular velocity in
533     lab-fixed frame, we consider the equation of angular momentum in
534     body-fixed frame
535 tim 2746 \begin{equation}
536 tim 2999 \dot j_i (t) = \tau _{t,i} (t) = \tau _{s,i} (t) + \tau _{f,i}^b (t)
537     + \tau _{r,i}^b(t)
538 tim 2746 \end{equation}
539     Embedding the friction terms into force and torque, one can
540     integrate the langevin equations of motion for rigid body of
541     arbitrary shape in a velocity-Verlet style 2-part algorithm, where
542     $h= \delta t$:
543    
544 tim 2999 {\tt moveA:}
545 tim 2746 \begin{align*}
546 tim 2999 {\bf v}\left(t + h / 2\right) &\leftarrow {\bf v}(t)
547     + \frac{h}{2} \left( {\bf f}(t) / m \right), \\
548     %
549     {\bf r}(t + h) &\leftarrow {\bf r}(t)
550     + h {\bf v}\left(t + h / 2 \right), \\
551     %
552     {\bf j}\left(t + h / 2 \right) &\leftarrow {\bf j}(t)
553     + \frac{h}{2} {\bf \tau}^b(t), \\
554     %
555     \mathsf{Q}(t + h) &\leftarrow \mathrm{rotate}\left( h {\bf j}
556     (t + h / 2) \cdot \overleftrightarrow{\mathsf{I}}^{-1} \right).
557 tim 2746 \end{align*}
558     In this context, the $\mathrm{rotate}$ function is the reversible
559 tim 2999 product of the three body-fixed rotations,
560 tim 2746 \begin{equation}
561     \mathrm{rotate}({\bf a}) = \mathsf{G}_x(a_x / 2) \cdot
562     \mathsf{G}_y(a_y / 2) \cdot \mathsf{G}_z(a_z) \cdot \mathsf{G}_y(a_y
563     / 2) \cdot \mathsf{G}_x(a_x /2),
564     \end{equation}
565     where each rotational propagator, $\mathsf{G}_\alpha(\theta)$,
566 tim 2999 rotates both the rotation matrix ($\mathsf{Q}$) and the body-fixed
567     angular momentum (${\bf j}$) by an angle $\theta$ around body-fixed
568     axis $\alpha$,
569 tim 2746 \begin{equation}
570     \mathsf{G}_\alpha( \theta ) = \left\{
571     \begin{array}{lcl}
572 tim 2999 \mathsf{Q}(t) & \leftarrow & \mathsf{Q}(0) \cdot \mathsf{R}_\alpha(\theta)^T, \\
573 tim 2746 {\bf j}(t) & \leftarrow & \mathsf{R}_\alpha(\theta) \cdot {\bf
574     j}(0).
575     \end{array}
576     \right.
577     \end{equation}
578     $\mathsf{R}_\alpha$ is a quadratic approximation to the single-axis
579     rotation matrix. For example, in the small-angle limit, the
580     rotation matrix around the body-fixed x-axis can be approximated as
581     \begin{equation}
582     \mathsf{R}_x(\theta) \approx \left(
583     \begin{array}{ccc}
584     1 & 0 & 0 \\
585     0 & \frac{1-\theta^2 / 4}{1 + \theta^2 / 4} & -\frac{\theta}{1+
586     \theta^2 / 4} \\
587     0 & \frac{\theta}{1+ \theta^2 / 4} & \frac{1-\theta^2 / 4}{1 +
588     \theta^2 / 4}
589     \end{array}
590     \right).
591     \end{equation}
592 tim 2999 All other rotations follow in a straightforward manner. After the
593     first part of the propagation, the forces and body-fixed torques are
594     calculated at the new positions and orientations
595 tim 2746
596 tim 2999 {\tt doForces:}
597     \begin{align*}
598     {\bf f}(t + h) &\leftarrow
599     - \left(\frac{\partial V}{\partial {\bf r}}\right)_{{\bf r}(t + h)}, \\
600     %
601     {\bf \tau}^{s}(t + h) &\leftarrow {\bf u}(t + h)
602     \times \frac{\partial V}{\partial {\bf u}}, \\
603     %
604     {\bf \tau}^{b}(t + h) &\leftarrow \mathsf{Q}(t + h)
605     \cdot {\bf \tau}^s(t + h).
606     \end{align*}
607 tim 2746 Once the forces and torques have been obtained at the new time step,
608     the velocities can be advanced to the same time value.
609    
610 tim 2999 {\tt moveB:}
611 tim 2746 \begin{align*}
612 tim 2999 {\bf v}\left(t + h \right) &\leftarrow {\bf v}\left(t + h / 2
613     \right)
614     + \frac{h}{2} \left( {\bf f}(t + h) / m \right), \\
615     %
616     {\bf j}\left(t + h \right) &\leftarrow {\bf j}\left(t + h / 2
617     \right)
618     + \frac{h}{2} {\bf \tau}^b(t + h) .
619 tim 2746 \end{align*}
620    
621 gezelter 3310 \section{Validating the Method\label{sec:validating}}
622 gezelter 3302 In order to validate our Langevin integrator for arbitrarily-shaped
623 gezelter 3305 rigid bodies, we implemented the algorithm in {\sc
624     oopse}\cite{Meineke2005} and compared the results of this algorithm
625     with the known
626 gezelter 3302 hydrodynamic limiting behavior for a few model systems, and to
627     microcanonical molecular dynamics simulations for some more
628     complicated bodies. The model systems and their analytical behavior
629     (if known) are summarized below. Parameters for the primary particles
630     comprising our model systems are given in table \ref{tab:parameters},
631     and a sketch of the arrangement of these primary particles into the
632 gezelter 3305 model rigid bodies is shown in figure \ref{fig:models}. In table
633     \ref{tab:parameters}, $d$ and $l$ are the physical dimensions of
634     ellipsoidal (Gay-Berne) particles. For spherical particles, the value
635     of the Lennard-Jones $\sigma$ parameter is the particle diameter
636     ($d$). Gay-Berne ellipsoids have an energy scaling parameter,
637     $\epsilon^s$, which describes the well depth for two identical
638     ellipsoids in a {\it side-by-side} configuration. Additionally, a
639     well depth aspect ratio, $\epsilon^r = \epsilon^e / \epsilon^s$,
640     describes the ratio between the well depths in the {\it end-to-end}
641     and side-by-side configurations. For spheres, $\epsilon^r \equiv 1$.
642     Moments of inertia are also required to describe the motion of primary
643     particles with orientational degrees of freedom.
644 gezelter 3299
645 gezelter 3302 \begin{table*}
646     \begin{minipage}{\linewidth}
647     \begin{center}
648     \caption{Parameters for the primary particles in use by the rigid body
649     models in figure \ref{fig:models}.}
650     \begin{tabular}{lrcccccccc}
651     \hline
652     & & & & & & & \multicolumn{3}c{$\overleftrightarrow{\mathsf I}$ (amu \AA$^2$)} \\
653     & & $d$ (\AA) & $l$ (\AA) & $\epsilon^s$ (kcal/mol) & $\epsilon^r$ &
654     $m$ (amu) & $I_{xx}$ & $I_{yy}$ & $I_{zz}$ \\ \hline
655 gezelter 3308 Sphere & & 6.5 & $= d$ & 0.8 & 1 & 190 & 802.75 & 802.75 & 802.75 \\
656 gezelter 3302 Ellipsoid & & 4.6 & 13.8 & 0.8 & 0.2 & 200 & 2105 & 2105 & 421 \\
657 gezelter 3308 Dumbbell &(2 identical spheres) & 6.5 & $= d$ & 0.8 & 1 & 190 & 802.75 & 802.75 & 802.75 \\
658 gezelter 3302 Banana &(3 identical ellipsoids)& 4.2 & 11.2 & 0.8 & 0.2 & 240 & 10000 & 10000 & 0 \\
659     Lipid: & Spherical Head & 6.5 & $= d$ & 0.185 & 1 & 196 & & & \\
660     & Ellipsoidal Tail & 4.6 & 13.8 & 0.8 & 0.2 & 760 & 45000 & 45000 & 9000 \\
661     Solvent & & 4.7 & $= d$ & 0.8 & 1 & 72.06 & & & \\
662     \hline
663     \end{tabular}
664     \label{tab:parameters}
665     \end{center}
666     \end{minipage}
667     \end{table*}
668    
669 gezelter 3305 \begin{figure}
670     \centering
671     \includegraphics[width=3in]{sketch}
672     \caption[Sketch of the model systems]{A sketch of the model systems
673     used in evaluating the behavior of the rigid body Langevin
674     integrator.} \label{fig:models}
675     \end{figure}
676    
677 gezelter 3302 \subsection{Simulation Methodology}
678     We performed reference microcanonical simulations with explicit
679     solvents for each of the different model system. In each case there
680     was one solute model and 1929 solvent molecules present in the
681     simulation box. All simulations were equilibrated using a
682     constant-pressure and temperature integrator with target values of 300
683     K for the temperature and 1 atm for pressure. Following this stage,
684     further equilibration and sampling was done in a microcanonical
685 gezelter 3305 ensemble. Since the model bodies are typically quite massive, we were
686 gezelter 3310 able to use a time step of 25 fs.
687    
688     The model systems studied used both Lennard-Jones spheres as well as
689     uniaxial Gay-Berne ellipoids. In its original form, the Gay-Berne
690     potential was a single site model for the interactions of rigid
691     ellipsoidal molecules.\cite{Gay81} It can be thought of as a
692     modification of the Gaussian overlap model originally described by
693     Berne and Pechukas.\cite{Berne72} The potential is constructed in the
694     familiar form of the Lennard-Jones function using
695     orientation-dependent $\sigma$ and $\epsilon$ parameters,
696     \begin{equation*}
697     V_{ij}({\mathbf{\hat u}_i}, {\mathbf{\hat u}_j}, {\mathbf{\hat
698     r}_{ij}}) = 4\epsilon ({\mathbf{\hat u}_i}, {\mathbf{\hat u}_j},
699     {\mathbf{\hat r}_{ij}})\left[\left(\frac{\sigma_0}{r_{ij}-\sigma({\mathbf{\hat u
700     }_i},
701     {\mathbf{\hat u}_j}, {\mathbf{\hat r}_{ij}})+\sigma_0}\right)^{12}
702     -\left(\frac{\sigma_0}{r_{ij}-\sigma({\mathbf{\hat u}_i}, {\mathbf{\hat u}_j},
703     {\mathbf{\hat r}_{ij}})+\sigma_0}\right)^6\right]
704     \label{eq:gb}
705     \end{equation*}
706    
707     The range $(\sigma({\bf \hat{u}}_{i},{\bf \hat{u}}_{j},{\bf
708     \hat{r}}_{ij}))$, and strength $(\epsilon({\bf \hat{u}}_{i},{\bf
709     \hat{u}}_{j},{\bf \hat{r}}_{ij}))$ parameters
710     are dependent on the relative orientations of the two ellipsoids (${\bf
711     \hat{u}}_{i},{\bf \hat{u}}_{j}$) as well as the direction of the
712     inter-ellipsoid separation (${\bf \hat{r}}_{ij}$). The shape and
713     attractiveness of each ellipsoid is governed by a relatively small set
714     of parameters: $l$ and $d$ describe the length and width of each
715     uniaxial ellipsoid, while $\epsilon^s$, which describes the well depth
716     for two identical ellipsoids in a {\it side-by-side} configuration.
717     Additionally, a well depth aspect ratio, $\epsilon^r = \epsilon^e /
718     \epsilon^s$, describes the ratio between the well depths in the {\it
719     end-to-end} and side-by-side configurations. Details of the potential
720     are given elsewhere,\cite{Luckhurst90,Golubkov06,SunGezelter08} and an
721     excellent overview of the computational methods that can be used to
722     efficiently compute forces and torques for this potential can be found
723     in Ref. \citen{Golubkov06}
724    
725     For the interaction between nonequivalent uniaxial ellipsoids (or
726     between spheres and ellipsoids), the spheres are treated as ellipsoids
727     with an aspect ratio of 1 ($d = l$) and with an well depth ratio
728     ($\epsilon^r$) of 1 ($\epsilon^e = \epsilon^s$). The form of the
729     Gay-Berne potential we are using was generalized by Cleaver {\it et
730     al.} and is appropriate for dissimilar uniaxial
731     ellipsoids.\cite{Cleaver96}
732    
733     A switching function was applied to all potentials to smoothly turn
734     off the interactions between a range of $22$ and $25$ \AA. The
735     switching function was the standard (cubic) function,
736 gezelter 3302 \begin{equation}
737     s(r) =
738     \begin{cases}
739     1 & \text{if $r \le r_{\text{sw}}$},\\
740     \frac{(r_{\text{cut}} + 2r - 3r_{\text{sw}})(r_{\text{cut}} - r)^2}
741     {(r_{\text{cut}} - r_{\text{sw}})^3}
742     & \text{if $r_{\text{sw}} < r \le r_{\text{cut}}$}, \\
743     0 & \text{if $r > r_{\text{cut}}$.}
744     \end{cases}
745     \label{eq:switchingFunc}
746     \end{equation}
747 gezelter 3310
748 gezelter 3302 To measure shear viscosities from our microcanonical simulations, we
749     used the Einstein form of the pressure correlation function,\cite{hess:209}
750     \begin{equation}
751 gezelter 3310 \eta = \lim_{t->\infty} \frac{V}{2 k_B T} \frac{d}{dt} \left\langle \left(
752     \int_{t_0}^{t_0 + t} P_{xz}(t') dt' \right)^2 \right\rangle_{t_0}.
753 gezelter 3302 \label{eq:shear}
754     \end{equation}
755     A similar form exists for the bulk viscosity
756     \begin{equation}
757 gezelter 3310 \kappa = \lim_{t->\infty} \frac{V}{2 k_B T} \frac{d}{dt} \left\langle \left(
758 gezelter 3302 \int_{t_0}^{t_0 + t}
759 gezelter 3310 \left(P\left(t'\right)-\left\langle P \right\rangle \right)dt'
760     \right)^2 \right\rangle_{t_0}.
761 gezelter 3302 \end{equation}
762     Alternatively, the shear viscosity can also be calculated using a
763     Green-Kubo formula with the off-diagonal pressure tensor correlation function,
764     \begin{equation}
765 gezelter 3310 \eta = \frac{V}{k_B T} \int_0^{\infty} \left\langle P_{xz}(t_0) P_{xz}(t_0
766     + t) \right\rangle_{t_0} dt,
767 gezelter 3302 \end{equation}
768     although this method converges extremely slowly and is not practical
769     for obtaining viscosities from molecular dynamics simulations.
770    
771     The Langevin dynamics for the different model systems were performed
772     at the same temperature as the average temperature of the
773     microcanonical simulations and with a solvent viscosity taken from
774 gezelter 3305 Eq. (\ref{eq:shear}) applied to these simulations. We used 1024
775     independent solute simulations to obtain statistics on our Langevin
776     integrator.
777 gezelter 3302
778     \subsection{Analysis}
779    
780     The quantities of interest when comparing the Langevin integrator to
781     analytic hydrodynamic equations and to molecular dynamics simulations
782     are typically translational diffusion constants and orientational
783     relaxation times. Translational diffusion constants for point
784     particles are computed easily from the long-time slope of the
785     mean-square displacement,
786     \begin{equation}
787 gezelter 3310 D = \lim_{t\rightarrow \infty} \frac{1}{6 t} \left\langle {|\left({\bf r}_{i}(t) - {\bf r}_{i}(0) \right)|}^2 \right\rangle,
788 gezelter 3302 \end{equation}
789     of the solute molecules. For models in which the translational
790 gezelter 3305 diffusion tensor (${\bf D}_{tt}$) has non-degenerate eigenvalues
791     (i.e. any non-spherically-symmetric rigid body), it is possible to
792     compute the diffusive behavior for motion parallel to each body-fixed
793     axis by projecting the displacement of the particle onto the
794     body-fixed reference frame at $t=0$. With an isotropic solvent, as we
795     have used in this study, there are differences between the three
796 gezelter 3302 diffusion constants, but these must converge to the same value at
797     longer times. Translational diffusion constants for the different
798 gezelter 3305 shaped models are shown in table \ref{tab:translation}.
799 gezelter 3302
800 gezelter 3305 In general, the three eigenvalues ($D_1, D_2, D_3$) of the rotational
801 gezelter 3302 diffusion tensor (${\bf D}_{rr}$) measure the diffusion of an object
802     {\it around} a particular body-fixed axis and {\it not} the diffusion
803     of a vector pointing along the axis. However, these eigenvalues can
804     be combined to find 5 characteristic rotational relaxation
805 gezelter 3305 times,\cite{PhysRev.119.53,Berne90}
806 gezelter 3302 \begin{eqnarray}
807 gezelter 3305 1 / \tau_1 & = & 6 D_r + 2 \Delta \\
808     1 / \tau_2 & = & 6 D_r - 2 \Delta \\
809     1 / \tau_3 & = & 3 (D_r + D_1) \\
810     1 / \tau_4 & = & 3 (D_r + D_2) \\
811     1 / \tau_5 & = & 3 (D_r + D_3)
812 gezelter 3302 \end{eqnarray}
813     where
814     \begin{equation}
815     D_r = \frac{1}{3} \left(D_1 + D_2 + D_3 \right)
816     \end{equation}
817     and
818     \begin{equation}
819 gezelter 3305 \Delta = \left( (D_1 - D_2)^2 + (D_3 - D_1 )(D_3 - D_2)\right)^{1/2}
820 gezelter 3302 \end{equation}
821 gezelter 3305 Each of these characteristic times can be used to predict the decay of
822     part of the rotational correlation function when $\ell = 2$,
823 gezelter 3302 \begin{equation}
824 gezelter 3305 C_2(t) = \frac{a^2}{N^2} e^{-t/\tau_1} + \frac{b^2}{N^2} e^{-t/\tau_2}.
825 gezelter 3302 \end{equation}
826 gezelter 3305 This is the same as the $F^2_{0,0}(t)$ correlation function that
827     appears in Ref. \citen{Berne90}. The amplitudes of the two decay
828     terms are expressed in terms of three dimensionless functions of the
829     eigenvalues: $a = \sqrt{3} (D_1 - D_2)$, $b = (2D_3 - D_1 - D_2 +
830     2\Delta)$, and $N = 2 \sqrt{\Delta b}$. Similar expressions can be
831     obtained for other angular momentum correlation
832     functions.\cite{PhysRev.119.53,Berne90} In all of the model systems we
833     studied, only one of the amplitudes of the two decay terms was
834     non-zero, so it was possible to derive a single relaxation time for
835     each of the hydrodynamic tensors. In many cases, these characteristic
836     times are averaged and reported in the literature as a single relaxation
837     time,\cite{Garcia-de-la-Torre:1997qy}
838 gezelter 3302 \begin{equation}
839 gezelter 3305 1 / \tau_0 = \frac{1}{5} \sum_{i=1}^5 \tau_{i}^{-1},
840     \end{equation}
841     although for the cases reported here, this averaging is not necessary
842     and only one of the five relaxation times is relevant.
843    
844     To test the Langevin integrator's behavior for rotational relaxation,
845     we have compared the analytical orientational relaxation times (if
846     they are known) with the general result from the diffusion tensor and
847     with the results from both the explicitly solvated molecular dynamics
848     and Langevin simulations. Relaxation times from simulations (both
849     microcanonical and Langevin), were computed using Legendre polynomial
850     correlation functions for a unit vector (${\bf u}$) fixed along one or
851     more of the body-fixed axes of the model.
852     \begin{equation}
853 gezelter 3310 C_{\ell}(t) = \left\langle P_{\ell}\left({\bf u}_{i}(t) \cdot {\bf
854     u}_{i}(0) \right) \right\rangle
855 gezelter 3302 \end{equation}
856     For simulations in the high-friction limit, orientational correlation
857     times can then be obtained from exponential fits of this function, or by
858     integrating,
859     \begin{equation}
860 gezelter 3305 \tau = \ell (\ell + 1) \int_0^{\infty} C_{\ell}(t) dt.
861 gezelter 3302 \end{equation}
862 gezelter 3305 In lower-friction solvents, the Legendre correlation functions often
863     exhibit non-exponential decay, and may not be characterized by a
864     single decay constant.
865 gezelter 3302
866     In table \ref{tab:rotation} we show the characteristic rotational
867     relaxation times (based on the diffusion tensor) for each of the model
868     systems compared with the values obtained via microcanonical and Langevin
869     simulations.
870    
871 gezelter 3305 \subsection{Spherical particles}
872 gezelter 3299 Our model system for spherical particles was a Lennard-Jones sphere of
873     diameter ($\sigma$) 6.5 \AA\ in a sea of smaller spheres ($\sigma$ =
874     4.7 \AA). The well depth ($\epsilon$) for both particles was set to
875 gezelter 3302 an arbitrary value of 0.8 kcal/mol.
876 gezelter 3299
877     The Stokes-Einstein behavior of large spherical particles in
878     hydrodynamic flows is well known, giving translational friction
879     coefficients of $6 \pi \eta R$ (stick boundary conditions) and
880 gezelter 3302 rotational friction coefficients of $8 \pi \eta R^3$. Recently,
881     Schmidt and Skinner have computed the behavior of spherical tag
882     particles in molecular dynamics simulations, and have shown that {\it
883     slip} boundary conditions ($\Xi_{tt} = 4 \pi \eta R$) may be more
884 gezelter 3299 appropriate for molecule-sized spheres embedded in a sea of spherical
885 gezelter 3310 solvent particles.\cite{Schmidt:2004fj,Schmidt:2003kx}
886 gezelter 3299
887     Our simulation results show similar behavior to the behavior observed
888 gezelter 3302 by Schmidt and Skinner. The diffusion constant obtained from our
889 gezelter 3299 microcanonical molecular dynamics simulations lies between the slip
890     and stick boundary condition results obtained via Stokes-Einstein
891     behavior. Since the Langevin integrator assumes Stokes-Einstein stick
892     boundary conditions in calculating the drag and random forces for
893     spherical particles, our Langevin routine obtains nearly quantitative
894     agreement with the hydrodynamic results for spherical particles. One
895     avenue for improvement of the method would be to compute elements of
896     $\Xi_{tt}$ assuming behavior intermediate between the two boundary
897 gezelter 3302 conditions.
898 gezelter 3299
899 gezelter 3310 In the explicit solvent simulations, both our solute and solvent
900     particles were structureless, exerting no torques upon each other.
901     Therefore, there are not rotational correlation times available for
902     this model system.
903 gezelter 3299
904 gezelter 3310 \subsection{Ellipsoids}
905     For uniaxial ellipsoids ($a > b = c$), Perrin's formulae for both
906 gezelter 3299 translational and rotational diffusion of each of the body-fixed axes
907     can be combined to give a single translational diffusion
908 gezelter 3302 constant,\cite{Berne90}
909 gezelter 3299 \begin{equation}
910     D = \frac{k_B T}{6 \pi \eta a} G(\rho),
911     \label{Dperrin}
912     \end{equation}
913     as well as a single rotational diffusion coefficient,
914     \begin{equation}
915     \Theta = \frac{3 k_B T}{16 \pi \eta a^3} \left\{ \frac{(2 - \rho^2)
916     G(\rho) - 1}{1 - \rho^4} \right\}.
917     \label{ThetaPerrin}
918     \end{equation}
919     In these expressions, $G(\rho)$ is a function of the axial ratio
920     ($\rho = b / a$), which for prolate ellipsoids, is
921     \begin{equation}
922     G(\rho) = (1- \rho^2)^{-1/2} \ln \left\{ \frac{1 + (1 -
923     \rho^2)^{1/2}}{\rho} \right\}
924     \label{GPerrin}
925     \end{equation}
926     Again, there is some uncertainty about the correct boundary conditions
927     to use for molecular-scale ellipsoids in a sea of similarly-sized
928     solvent particles. Ravichandran and Bagchi found that {\it slip}
929 gezelter 3302 boundary conditions most closely resembled the simulation
930     results,\cite{Ravichandran:1999fk} in agreement with earlier work of
931     Tang and Evans.\cite{TANG:1993lr}
932 gezelter 3299
933 gezelter 3305 Even though there are analytic resistance tensors for ellipsoids, we
934     constructed a rough-shell model using 2135 beads (each with a diameter
935 gezelter 3310 of 0.25 \AA) to approximate the shape of the model ellipsoid. We
936 gezelter 3305 compared the Langevin dynamics from both the simple ellipsoidal
937     resistance tensor and the rough shell approximation with
938     microcanonical simulations and the predictions of Perrin. As in the
939     case of our spherical model system, the Langevin integrator reproduces
940     almost exactly the behavior of the Perrin formulae (which is
941     unsurprising given that the Perrin formulae were used to derive the
942 gezelter 3299 drag and random forces applied to the ellipsoid). We obtain
943     translational diffusion constants and rotational correlation times
944     that are within a few percent of the analytic values for both the
945     exact treatment of the diffusion tensor as well as the rough-shell
946     model for the ellipsoid.
947    
948 gezelter 3308 The translational diffusion constants from the microcanonical simulations
949     agree well with the predictions of the Perrin model, although the rotational
950     correlation times are a factor of 2 shorter than expected from hydrodynamic
951     theory. One explanation for the slower rotation
952     of explicitly-solvated ellipsoids is the possibility that solute-solvent
953     collisions happen at both ends of the solute whenever the principal
954     axis of the ellipsoid is turning. In the upper portion of figure
955     \ref{fig:explanation} we sketch a physical picture of this explanation.
956     Since our Langevin integrator is providing nearly quantitative agreement with
957     the Perrin model, it also predicts orientational diffusion for ellipsoids that
958     exceed explicitly solvated correlation times by a factor of two.
959 gezelter 3299
960 gezelter 3310 \subsection{Rigid dumbbells}
961 gezelter 3302 Perhaps the only {\it composite} rigid body for which analytic
962     expressions for the hydrodynamic tensor are available is the
963     two-sphere dumbbell model. This model consists of two non-overlapping
964     spheres held by a rigid bond connecting their centers. There are
965     competing expressions for the 6x6 resistance tensor for this
966     model. Equation (\ref{introEquation:oseenTensor}) above gives the
967     original Oseen tensor, while the second order expression introduced by
968     Rotne and Prager,\cite{Rotne1969} and improved by Garc\'{i}a de la
969     Torre and Bloomfield,\cite{Torre1977} is given above as
970 gezelter 3299 Eq. (\ref{introEquation:RPTensorNonOverlapped}). In our case, we use
971     a model dumbbell in which the two spheres are identical Lennard-Jones
972     particles ($\sigma$ = 6.5 \AA\ , $\epsilon$ = 0.8 kcal / mol) held at
973 gezelter 3302 a distance of 6.532 \AA.
974 gezelter 3299
975     The theoretical values for the translational diffusion constant of the
976     dumbbell are calculated from the work of Stimson and Jeffery, who
977     studied the motion of this system in a flow parallel to the
978 gezelter 3302 inter-sphere axis,\cite{Stimson:1926qy} and Davis, who studied the
979     motion in a flow {\it perpendicular} to the inter-sphere
980     axis.\cite{Davis:1969uq} We know of no analytic solutions for the {\it
981     orientational} correlation times for this model system (other than
982 gezelter 3305 those derived from the 6 x 6 tensors mentioned above).
983 gezelter 3299
984 gezelter 3305 The bead model for this model system comprises the two large spheres
985     by themselves, while the rough shell approximation used 3368 separate
986     beads (each with a diameter of 0.25 \AA) to approximate the shape of
987     the rigid body. The hydrodynamics tensors computed from both the bead
988     and rough shell models are remarkably similar. Computing the initial
989     hydrodynamic tensor for a rough shell model can be quite expensive (in
990     this case it requires inverting a 10104 x 10104 matrix), while the
991     bead model is typically easy to compute (in this case requiring
992 gezelter 3308 inversion of a 6 x 6 matrix).
993 gezelter 3305
994 gezelter 3308 \begin{figure}
995     \centering
996 gezelter 3310 \includegraphics[width=2in]{RoughShell}
997 gezelter 3308 \caption[Model rigid bodies and their rough shell approximations]{The
998     model rigid bodies (left column) used to test this algorithm and their
999     rough-shell approximations (right-column) that were used to compute
1000     the hydrodynamic tensors. The top two models (ellipsoid and dumbbell)
1001     have analytic solutions and were used to test the rough shell
1002     approximation. The lower two models (banana and lipid) were compared
1003     with explicitly-solvated molecular dynamics simulations. }
1004     \label{fig:roughShell}
1005     \end{figure}
1006    
1007    
1008 gezelter 3305 Once the hydrodynamic tensor has been computed, there is no additional
1009     penalty for carrying out a Langevin simulation with either of the two
1010     different hydrodynamics models. Our naive expectation is that since
1011     the rigid body's surface is roughened under the various shell models,
1012     the diffusion constants will be even farther from the ``slip''
1013     boundary conditions than observed for the bead model (which uses a
1014     Stokes-Einstein model to arrive at the hydrodynamic tensor). For the
1015     dumbbell, this prediction is correct although all of the Langevin
1016     diffusion constants are within 6\% of the diffusion constant predicted
1017     from the fully solvated system.
1018    
1019 gezelter 3308 For rotational motion, Langevin integration (and the hydrodynamic tensor)
1020     yields rotational correlation times that are substantially shorter than those
1021     obtained from explicitly-solvated simulations. It is likely that this is due
1022     to the large size of the explicit solvent spheres, a feature that prevents
1023     the solvent from coming in contact with a substantial fraction of the surface
1024     area of the dumbbell. Therefore, the explicit solvent only provides drag
1025     over a substantially reduced surface area of this model, while the
1026     hydrodynamic theories utilize the entire surface area for estimating
1027     rotational diffusion. A sketch of the free volume available in the explicit
1028     solvent simulations is shown in figure \ref{fig:explanation}.
1029 gezelter 3305
1030 gezelter 3310
1031     \begin{figure}
1032     \centering
1033     \includegraphics[width=6in]{explanation}
1034     \caption[Explanations of the differences between orientational
1035     correlation times for explicitly-solvated models and hydrodynamics
1036     predictions]{Explanations of the differences between orientational
1037     correlation times for explicitly-solvated models and hydrodynamic
1038     predictions. For the ellipsoids (upper figures), rotation of the
1039     principal axis can involve correlated collisions at both sides of the
1040     solute. In the rigid dumbbell model (lower figures), the large size
1041     of the explicit solvent spheres prevents them from coming in contact
1042     with a substantial fraction of the surface area of the dumbbell.
1043     Therefore, the explicit solvent only provides drag over a
1044     substantially reduced surface area of this model, where the
1045     hydrodynamic theories utilize the entire surface area for estimating
1046     rotational diffusion.
1047     } \label{fig:explanation}
1048     \end{figure}
1049    
1050    
1051    
1052     \subsection{Composite banana-shaped molecules}
1053     Banana-shaped rigid bodies composed of three Gay-Berne ellipsoids have
1054     been used by Orlandi {\it et al.} to observe mesophases in
1055     coarse-grained models for bent-core liquid crystalline
1056     molecules.\cite{Orlandi:2006fk} We have used the same overlapping
1057 gezelter 3299 ellipsoids as a way to test the behavior of our algorithm for a
1058     structure of some interest to the materials science community,
1059     although since we are interested in capturing only the hydrodynamic
1060 gezelter 3310 behavior of this model, we have left out the dipolar interactions of
1061     the original Orlandi model.
1062 gezelter 3308
1063     A reference system composed of a single banana rigid body embedded in a
1064     sea of 1929 solvent particles was created and run under standard
1065     (microcanonical) molecular dynamics. The resulting viscosity of this
1066     mixture was 0.298 centipoise (as estimated using Eq. (\ref{eq:shear})).
1067     To calculate the hydrodynamic properties of the banana rigid body model,
1068 gezelter 3310 we created a rough shell (see Fig.~\ref{fig:roughShell}), in which
1069 gezelter 3308 the banana is represented as a ``shell'' made of 3321 identical beads
1070 gezelter 3310 (0.25 \AA\ in diameter) distributed on the surface. Applying the
1071 gezelter 3308 procedure described in Sec.~\ref{introEquation:ResistanceTensorArbitraryOrigin}, we
1072 gezelter 3310 identified the center of resistance, ${\bf r} = $(0 \AA, 0.81 \AA, 0 \AA), as
1073     well as the resistance tensor,
1074     \begin{equation*}
1075     \Xi =
1076 gezelter 3308 \left( {\begin{array}{*{20}c}
1077     0.9261 & 0 & 0&0&0.08585&0.2057\\
1078     0& 0.9270&-0.007063& 0.08585&0&0\\
1079     0&-0.007063&0.7494&0.2057&0&0\\
1080 gezelter 3310 0&0.0858&0.2057& 58.64& 0&0\\0.08585&0&0&0&48.30&3.219&\\0.2057&0&0&0&3.219&10.7373\\\end{array}} \right),
1081     \end{equation*}
1082     where the units for translational, translation-rotation coupling and
1083     rotational tensors are (kcal fs / mol \AA$^2$), (kcal fs / mol \AA\ rad),
1084     and (kcal fs / mol rad$^2$), respectively.
1085 gezelter 3299
1086 gezelter 3308 The Langevin rigid-body integrator (and the hydrodynamic diffusion tensor)
1087     are essentially quantitative for translational diffusion of this model.
1088     Orientational correlation times under the Langevin rigid-body integrator
1089     are within 11\% of the values obtained from explicit solvent, but these
1090     models also exhibit some solvent inaccessible surface area in the
1091     explicitly-solvated case.
1092    
1093 gezelter 3310 \subsection{Composite sphero-ellipsoids}
1094 gezelter 3299 Spherical heads perched on the ends of Gay-Berne ellipsoids have been
1095 xsun 3312 used recently as models for lipid
1096     molecules.\cite{SunGezelter08,Ayton01}
1097 gezelter 3310 MORE DETAILS
1098 xsun 3298
1099 xsun 3312 A reference system composed of a single lipid rigid body embedded in a
1100     sea of 1929 solvent particles was created and run under standard
1101     (microcanonical) molecular dynamics. The resulting viscosity of this
1102     mixture was 0.349 centipoise (as estimated using
1103     Eq. (\ref{eq:shear})). To calculate the hydrodynamic properties of
1104     the lipid rigid body model, we created a rough shell (see
1105     Fig.~\ref{fig:roughShell}), in which the lipid is represented as a
1106     ``shell'' made of 3550 identical beads (0.25 \AA\ in diameter)
1107     distributed on the surface. Applying the procedure described in
1108     Sec.~\ref{introEquation:ResistanceTensorArbitraryOrigin}, we
1109     identified the center of resistance, ${\bf r} = $(0 \AA, 0 \AA, 1.46
1110     \AA).
1111 gezelter 3310
1112 gezelter 3315
1113 gezelter 3310 \subsection{Summary}
1114 xsun 3298 According to our simulations, the langevin dynamics is a reliable
1115     theory to apply to replace the explicit solvents, especially for the
1116     translation properties. For large molecules, the rotation properties
1117     are also mimiced reasonablly well.
1118    
1119 gezelter 3315 \begin{figure}
1120     \centering
1121     \includegraphics[width=\linewidth]{graph}
1122     \caption[Mean squared displacements and orientational
1123     correlation functions for each of the model rigid bodies.]{The
1124     mean-squared displacements ($\langle r^2(t) \rangle$) and
1125     orientational correlation functions ($C_2(t)$) for each of the model
1126     rigid bodies studied. The circles are the results for microcanonical
1127     simulations with explicit solvent molecules, while the other data sets
1128     are results for Langevin dynamics using the different hydrodynamic
1129     tensor approximations. The Perrin model for the ellipsoids is
1130     considered the ``exact'' hydrodynamic behavior (this can also be said
1131     for the translational motion of the dumbbell operating under the bead
1132     model). In most cases, the various hydrodynamics models reproduce
1133     each other quantitatively.}
1134     \label{fig:results}
1135     \end{figure}
1136    
1137 xsun 3298 \begin{table*}
1138     \begin{minipage}{\linewidth}
1139     \begin{center}
1140 gezelter 3305 \caption{Translational diffusion constants (D) for the model systems
1141     calculated using microcanonical simulations (with explicit solvent),
1142     theoretical predictions, and Langevin simulations (with implicit solvent).
1143     Analytical solutions for the exactly-solved hydrodynamics models are
1144     from Refs. \citen{Einstein05} (sphere), \citen{Perrin1934} and \citen{Perrin1936}
1145     (ellipsoid), \citen{Stimson:1926qy} and \citen{Davis:1969uq}
1146     (dumbbell). The other model systems have no known analytic solution.
1147     All diffusion constants are reported in units of $10^{-3}$ cm$^2$ / ps (=
1148     $10^{-4}$ \AA$^2$ / fs). }
1149     \begin{tabular}{lccccccc}
1150 xsun 3298 \hline
1151 gezelter 3305 & \multicolumn{2}c{microcanonical simulation} & & \multicolumn{3}c{Theoretical} & Langevin \\
1152     \cline{2-3} \cline{5-7}
1153     model & $\eta$ (centipoise) & D & & Analytical & method & Hydrodynamics & simulation \\
1154 xsun 3298 \hline
1155 xsun 3312 sphere & 0.279 & 3.06 & & 2.42 & exact & 2.42 & 2.33 \\
1156 gezelter 3305 ellipsoid & 0.255 & 2.44 & & 2.34 & exact & 2.34 & 2.37 \\
1157     & 0.255 & 2.44 & & 2.34 & rough shell & 2.36 & 2.28 \\
1158 xsun 3312 dumbbell & 0.308 & 2.06 & & 1.64 & bead model & 1.65 & 1.62 \\
1159     & 0.308 & 2.06 & & 1.64 & rough shell & 1.59 & 1.62 \\
1160 gezelter 3305 banana & 0.298 & 1.53 & & & rough shell & 1.56 & 1.55 \\
1161     lipid & 0.349 & 0.96 & & & rough shell & 1.33 & 1.32 \\
1162 xsun 3298 \end{tabular}
1163     \label{tab:translation}
1164     \end{center}
1165     \end{minipage}
1166     \end{table*}
1167    
1168     \begin{table*}
1169     \begin{minipage}{\linewidth}
1170     \begin{center}
1171 gezelter 3305 \caption{Orientational relaxation times ($\tau$) for the model systems using
1172     microcanonical simulation (with explicit solvent), theoretical
1173     predictions, and Langevin simulations (with implicit solvent). All
1174     relaxation times are for the rotational correlation function with
1175     $\ell = 2$ and are reported in units of ps. The ellipsoidal model has
1176     an exact solution for the orientational correlation time due to
1177     Perrin, but the other model systems have no known analytic solution.}
1178     \begin{tabular}{lccccccc}
1179 xsun 3298 \hline
1180 gezelter 3305 & \multicolumn{2}c{microcanonical simulation} & & \multicolumn{3}c{Theoretical} & Langevin \\
1181     \cline{2-3} \cline{5-7}
1182     model & $\eta$ (centipoise) & $\tau$ & & Perrin & method & Hydrodynamic & simulation \\
1183 xsun 3298 \hline
1184 xsun 3312 sphere & 0.279 & & & 9.69 & exact & 9.69 & 9.64 \\
1185 gezelter 3305 ellipsoid & 0.255 & 46.7 & & 22.0 & exact & 22.0 & 22.2 \\
1186     & 0.255 & 46.7 & & 22.0 & rough shell & 22.6 & 22.2 \\
1187 xsun 3312 dumbbell & 0.308 & 14.1 & & & bead model & 50.0 & 50.1 \\
1188     & 0.308 & 14.1 & & & rough shell & 41.5 & 41.3 \\
1189 gezelter 3305 banana & 0.298 & 63.8 & & & rough shell & 70.9 & 70.9 \\
1190     lipid & 0.349 & 78.0 & & & rough shell & 76.9 & 77.9 \\
1191     \hline
1192 xsun 3298 \end{tabular}
1193     \label{tab:rotation}
1194     \end{center}
1195     \end{minipage}
1196     \end{table*}
1197    
1198 gezelter 3310 \section{Application: A rigid-body lipid bilayer}
1199    
1200     The Langevin dynamics integrator was applied to study the formation of
1201     corrugated structures emerging from simulations of the coarse grained
1202     lipid molecular models presented above. The initial configuration is
1203 xsun 3298 taken from our molecular dynamics studies on lipid bilayers with
1204 gezelter 3310 lennard-Jones sphere solvents. The solvent molecules were excluded
1205     from the system, and the experimental value for the viscosity of water
1206     at 20C ($\eta = 1.00$ cp) was used to mimic the hydrodynamic effects
1207     of the solvent. The absence of explicit solvent molecules and the
1208     stability of the integrator allowed us to take timesteps of 50 fs. A
1209     total simulation run time of 100 ns was sampled.
1210     Fig. \ref{fig:bilayer} shows the configuration of the system after 100
1211     ns, and the ripple structure remains stable during the entire
1212     trajectory. Compared with using explicit bead-model solvent
1213     molecules, the efficiency of the simulation has increased by an order
1214 xsun 3298 of magnitude.
1215    
1216 gezelter 3310 \begin{figure}
1217     \centering
1218     \includegraphics[width=\linewidth]{bilayer}
1219     \caption[Snapshot of a bilayer of rigid-body models for lipids]{A
1220     snapshot of a bilayer composed of rigid-body models for lipid
1221     molecules evolving using the Langevin integrator described in this
1222     work.} \label{fig:bilayer}
1223     \end{figure}
1224    
1225 tim 2746 \section{Conclusions}
1226    
1227 tim 2999 We have presented a new Langevin algorithm by incorporating the
1228     hydrodynamics properties of arbitrary shaped molecules into an
1229 gezelter 3308 advanced symplectic integration scheme. Further studies in systems
1230     involving banana shaped molecules illustrated that the dynamic
1231     properties could be preserved by using this new algorithm as an
1232     implicit solvent model.
1233 tim 2999
1234    
1235 tim 2746 \section{Acknowledgments}
1236     Support for this project was provided by the National Science
1237     Foundation under grant CHE-0134881. T.L. also acknowledges the
1238     financial support from center of applied mathematics at University
1239     of Notre Dame.
1240     \newpage
1241    
1242 gezelter 3305 \bibliographystyle{jcp}
1243 tim 2746 \bibliography{langevin}
1244    
1245     \end{document}