ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/group/trunk/tengDissertation/Langevin.tex
Revision: 2889
Committed: Mon Jun 26 13:42:53 2006 UTC (18 years ago) by tim
Content type: application/x-tex
File size: 27267 byte(s)
Log Message:
Change the case caption of tables

File Contents

# Content
1
2 \chapter{\label{chapt:langevin}LANGEVIN DYNAMICS FOR RIGID BODIES OF ARBITRARY SHAPE}
3
4 \section{Introduction}
5
6 %applications of langevin dynamics
7 As an excellent alternative to newtonian dynamics, Langevin
8 dynamics, which mimics a simple heat bath with stochastic and
9 dissipative forces, has been applied in a variety of studies. The
10 stochastic treatment of the solvent enables us to carry out
11 substantially longer time simulation. Implicit solvent Langevin
12 dynamics simulation of met-enkephalin not only outperforms explicit
13 solvent simulation on computation efficiency, but also agrees very
14 well with explicit solvent simulation on dynamics
15 properties\cite{Shen2002}. Recently, applying Langevin dynamics with
16 UNRES model, Liow and his coworkers suggest that protein folding
17 pathways can be possibly exploited within a reasonable amount of
18 time\cite{Liwo2005}. The stochastic nature of the Langevin dynamics
19 also enhances the sampling of the system and increases the
20 probability of crossing energy barrier\cite{Banerjee2004, Cui2003}.
21 Combining Langevin dynamics with Kramers's theory, Klimov and
22 Thirumalai identified the free-energy barrier by studying the
23 viscosity dependence of the protein folding rates\cite{Klimov1997}.
24 In order to account for solvent induced interactions missing from
25 implicit solvent model, Kaya incorporated desolvation free energy
26 barrier into implicit coarse-grained solvent model in protein
27 folding/unfolding study and discovered a higher free energy barrier
28 between the native and denatured states. Because of its stability
29 against noise, Langevin dynamics is very suitable for studying
30 remagnetization processes in various
31 systems\cite{Palacios1998,Berkov2002,Denisov2003}. For instance, the
32 oscillation power spectrum of nanoparticles from Langevin dynamics
33 simulation has the same peak frequencies for different wave
34 vectors,which recovers the property of magnetic excitations in small
35 finite structures\cite{Berkov2005a}. In an attempt to reduce the
36 computational cost of simulation, multiple time stepping (MTS)
37 methods have been introduced and have been of great interest to
38 macromolecule and protein community\cite{Tuckerman1992}. Relying on
39 the observation that forces between distant atoms generally
40 demonstrate slower fluctuations than forces between close atoms, MTS
41 method are generally implemented by evaluating the slowly
42 fluctuating forces less frequently than the fast ones.
43 Unfortunately, nonlinear instability resulting from increasing
44 timestep in MTS simulation have became a critical obstruction
45 preventing the long time simulation. Due to the coupling to the heat
46 bath, Langevin dynamics has been shown to be able to damp out the
47 resonance artifact more efficiently\cite{Sandu1999}.
48
49 %review langevin/browninan dynamics for arbitrarily shaped rigid body
50 Combining Langevin or Brownian dynamics with rigid body dynamics,
51 one can study the slow processes in biomolecular systems. Modeling
52 the DNA as a chain of rigid spheres beads, which subject to harmonic
53 potentials as well as excluded volume potentials, Mielke and his
54 coworkers discover rapid superhelical stress generations from the
55 stochastic simulation of twin supercoiling DNA with response to
56 induced torques\cite{Mielke2004}. Membrane fusion is another key
57 biological process which controls a variety of physiological
58 functions, such as release of neurotransmitters \textit{etc}. A
59 typical fusion event happens on the time scale of millisecond, which
60 is impracticable to study using all atomistic model with newtonian
61 mechanics. With the help of coarse-grained rigid body model and
62 stochastic dynamics, the fusion pathways were exploited by many
63 researchers\cite{Noguchi2001,Noguchi2002,Shillcock2005}. Due to the
64 difficulty of numerical integration of anisotropy rotation, most of
65 the rigid body models are simply modeled by sphere, cylinder,
66 ellipsoid or other regular shapes in stochastic simulations. In an
67 effort to account for the diffusion anisotropy of the arbitrary
68 particles, Fernandes and de la Torre improved the original Brownian
69 dynamics simulation algorithm\cite{Ermak1978,Allison1991} by
70 incorporating a generalized $6\times6$ diffusion tensor and
71 introducing a simple rotation evolution scheme consisting of three
72 consecutive rotations\cite{Fernandes2002}. Unfortunately, unexpected
73 error and bias are introduced into the system due to the arbitrary
74 order of applying the noncommuting rotation
75 operators\cite{Beard2003}. Based on the observation the momentum
76 relaxation time is much less than the time step, one may ignore the
77 inertia in Brownian dynamics. However, assumption of the zero
78 average acceleration is not always true for cooperative motion which
79 is common in protein motion. An inertial Brownian dynamics (IBD) was
80 proposed to address this issue by adding an inertial correction
81 term\cite{Beard2000}. As a complement to IBD which has a lower bound
82 in time step because of the inertial relaxation time, long-time-step
83 inertial dynamics (LTID) can be used to investigate the inertial
84 behavior of the polymer segments in low friction
85 regime\cite{Beard2000}. LTID can also deal with the rotational
86 dynamics for nonskew bodies without translation-rotation coupling by
87 separating the translation and rotation motion and taking advantage
88 of the analytical solution of hydrodynamics properties. However,
89 typical nonskew bodies like cylinder and ellipsoid are inadequate to
90 represent most complex macromolecule assemblies. These intricate
91 molecules have been represented by a set of beads and their
92 hydrodynamics properties can be calculated using variant
93 hydrodynamic interaction tensors.
94
95 The goal of the present work is to develop a Langevin dynamics
96 algorithm for arbitrary rigid particles by integrating the accurate
97 estimation of friction tensor from hydrodynamics theory into the
98 sophisticated rigid body dynamics.
99
100 \section{Computational Methods{\label{methodSec}}}
101
102 \subsection{\label{introSection:frictionTensor}Friction Tensor}
103 Theoretically, the friction kernel can be determined using the
104 velocity autocorrelation function. However, this approach become
105 impractical when the system become more and more complicate.
106 Instead, various approaches based on hydrodynamics have been
107 developed to calculate the friction coefficients. In general,
108 friction tensor $\Xi$ is a $6\times 6$ matrix given by
109 \[
110 \Xi = \left( {\begin{array}{*{20}c}
111 {\Xi _{}^{tt} } & {\Xi _{}^{rt} } \\
112 {\Xi _{}^{tr} } & {\Xi _{}^{rr} } \\
113 \end{array}} \right).
114 \]
115 Here, $ {\Xi^{tt} }$ and $ {\Xi^{rr} }$ are translational friction
116 tensor and rotational resistance (friction) tensor respectively,
117 while ${\Xi^{tr} }$ is translation-rotation coupling tensor and $
118 {\Xi^{rt} }$ is rotation-translation coupling tensor. When a
119 particle moves in a fluid, it may experience friction force or
120 torque along the opposite direction of the velocity or angular
121 velocity,
122 \[
123 \left( \begin{array}{l}
124 F_R \\
125 \tau _R \\
126 \end{array} \right) = - \left( {\begin{array}{*{20}c}
127 {\Xi ^{tt} } & {\Xi ^{rt} } \\
128 {\Xi ^{tr} } & {\Xi ^{rr} } \\
129 \end{array}} \right)\left( \begin{array}{l}
130 v \\
131 w \\
132 \end{array} \right)
133 \]
134 where $F_r$ is the friction force and $\tau _R$ is the friction
135 toque.
136
137 \subsubsection{\label{introSection:resistanceTensorRegular}\textbf{The Resistance Tensor for Regular Shape}}
138
139 For a spherical particle with slip boundary conditions, the
140 translational and rotational friction constant can be calculated
141 from Stoke's law,
142 \[
143 \Xi ^{tt} = \left( {\begin{array}{*{20}c}
144 {6\pi \eta R} & 0 & 0 \\
145 0 & {6\pi \eta R} & 0 \\
146 0 & 0 & {6\pi \eta R} \\
147 \end{array}} \right)
148 \]
149 and
150 \[
151 \Xi ^{rr} = \left( {\begin{array}{*{20}c}
152 {8\pi \eta R^3 } & 0 & 0 \\
153 0 & {8\pi \eta R^3 } & 0 \\
154 0 & 0 & {8\pi \eta R^3 } \\
155 \end{array}} \right)
156 \]
157 where $\eta$ is the viscosity of the solvent and $R$ is the
158 hydrodynamics radius.
159
160 Other non-spherical shape, such as cylinder and ellipsoid
161 \textit{etc}, are widely used as reference for developing new
162 hydrodynamics theory, because their properties can be calculated
163 exactly. In 1936, Perrin extended Stokes's law to general
164 ellipsoids, also called a triaxial ellipsoid, which is given in
165 Cartesian coordinates by\cite{Perrin1934, Perrin1936}
166 \[
167 \frac{{x^2 }}{{a^2 }} + \frac{{y^2 }}{{b^2 }} + \frac{{z^2 }}{{c^2
168 }} = 1
169 \]
170 where the semi-axes are of lengths $a$, $b$, and $c$. Unfortunately,
171 due to the complexity of the elliptic integral, only the ellipsoid
172 with the restriction of two axes having to be equal, \textit{i.e.}
173 prolate($ a \ge b = c$) and oblate ($ a < b = c $), can be solved
174 exactly. Introducing an elliptic integral parameter $S$ for prolate
175 :
176 \[
177 S = \frac{2}{{\sqrt {a^2 - b^2 } }}\ln \frac{{a + \sqrt {a^2 - b^2
178 } }}{b},
179 \]
180 and oblate :
181 \[
182 S = \frac{2}{{\sqrt {b^2 - a^2 } }}arctg\frac{{\sqrt {b^2 - a^2 }
183 }}{a}
184 \],
185 one can write down the translational and rotational resistance
186 tensors
187 \[
188 \begin{array}{l}
189 \Xi _a^{tt} = 16\pi \eta \frac{{a^2 - b^2 }}{{(2a^2 - b^2 )S - 2a}}. \\
190 \Xi _b^{tt} = \Xi _c^{tt} = 32\pi \eta \frac{{a^2 - b^2 }}{{(2a^2 - 3b^2 )S +
191 2a}},
192 \end{array}
193 \]
194 and
195 \[
196 \begin{array}{l}
197 \Xi _a^{rr} = \frac{{32\pi }}{3}\eta \frac{{(a^2 - b^2 )b^2 }}{{2a - b^2 S}}, \\
198 \Xi _b^{rr} = \Xi _c^{rr} = \frac{{32\pi }}{3}\eta \frac{{(a^4 - b^4 )}}{{(2a^2 - b^2 )S - 2a}} \\
199 \end{array}.
200 \]
201
202 \subsubsection{\label{introSection:resistanceTensorRegularArbitrary}\textbf{The Resistance Tensor for Arbitrary Shapes}}
203
204 Unlike spherical and other simply shaped molecules, there is no
205 analytical solution for the friction tensor for arbitrarily shaped
206 rigid molecules. The ellipsoid of revolution model and general
207 triaxial ellipsoid model have been used to approximate the
208 hydrodynamic properties of rigid bodies. However, since the mapping
209 from all possible ellipsoidal spaces, $r$-space, to all possible
210 combination of rotational diffusion coefficients, $D$-space, is not
211 unique\cite{Wegener1979} as well as the intrinsic coupling between
212 translational and rotational motion of rigid body, general
213 ellipsoids are not always suitable for modeling arbitrarily shaped
214 rigid molecule. A number of studies have been devoted to determine
215 the friction tensor for irregularly shaped rigid bodies using more
216 advanced method where the molecule of interest was modeled by
217 combinations of spheres(beads)\cite{Carrasco1999} and the
218 hydrodynamics properties of the molecule can be calculated using the
219 hydrodynamic interaction tensor. Let us consider a rigid assembly of
220 $N$ beads immersed in a continuous medium. Due to hydrodynamics
221 interaction, the ``net'' velocity of $i$th bead, $v'_i$ is different
222 than its unperturbed velocity $v_i$,
223 \[
224 v'_i = v_i - \sum\limits_{j \ne i} {T_{ij} F_j }
225 \]
226 where $F_i$ is the frictional force, and $T_{ij}$ is the
227 hydrodynamic interaction tensor. The friction force of $i$th bead is
228 proportional to its ``net'' velocity
229 \begin{equation}
230 F_i = \zeta _i v_i - \zeta _i \sum\limits_{j \ne i} {T_{ij} F_j }.
231 \label{introEquation:tensorExpression}
232 \end{equation}
233 This equation is the basis for deriving the hydrodynamic tensor. In
234 1930, Oseen and Burgers gave a simple solution to Equation
235 \ref{introEquation:tensorExpression}
236 \begin{equation}
237 T_{ij} = \frac{1}{{8\pi \eta r_{ij} }}\left( {I + \frac{{R_{ij}
238 R_{ij}^T }}{{R_{ij}^2 }}} \right). \label{introEquation:oseenTensor}
239 \end{equation}
240 Here $R_{ij}$ is the distance vector between bead $i$ and bead $j$.
241 A second order expression for element of different size was
242 introduced by Rotne and Prager\cite{Rotne1969} and improved by
243 Garc\'{i}a de la Torre and Bloomfield\cite{Torre1977},
244 \begin{equation}
245 T_{ij} = \frac{1}{{8\pi \eta R_{ij} }}\left[ {\left( {I +
246 \frac{{R_{ij} R_{ij}^T }}{{R_{ij}^2 }}} \right) + R\frac{{\sigma
247 _i^2 + \sigma _j^2 }}{{r_{ij}^2 }}\left( {\frac{I}{3} -
248 \frac{{R_{ij} R_{ij}^T }}{{R_{ij}^2 }}} \right)} \right].
249 \label{introEquation:RPTensorNonOverlapped}
250 \end{equation}
251 Both of the Equation \ref{introEquation:oseenTensor} and Equation
252 \ref{introEquation:RPTensorNonOverlapped} have an assumption $R_{ij}
253 \ge \sigma _i + \sigma _j$. An alternative expression for
254 overlapping beads with the same radius, $\sigma$, is given by
255 \begin{equation}
256 T_{ij} = \frac{1}{{6\pi \eta R_{ij} }}\left[ {\left( {1 -
257 \frac{2}{{32}}\frac{{R_{ij} }}{\sigma }} \right)I +
258 \frac{2}{{32}}\frac{{R_{ij} R_{ij}^T }}{{R_{ij} \sigma }}} \right]
259 \label{introEquation:RPTensorOverlapped}
260 \end{equation}
261
262 To calculate the resistance tensor at an arbitrary origin $O$, we
263 construct a $3N \times 3N$ matrix consisting of $N \times N$
264 $B_{ij}$ blocks
265 \begin{equation}
266 B = \left( {\begin{array}{*{20}c}
267 {B_{11} } & \ldots & {B_{1N} } \\
268 \vdots & \ddots & \vdots \\
269 {B_{N1} } & \cdots & {B_{NN} } \\
270 \end{array}} \right),
271 \end{equation}
272 where $B_{ij}$ is given by
273 \[
274 B_{ij} = \delta _{ij} \frac{I}{{6\pi \eta R}} + (1 - \delta _{ij}
275 )T_{ij}
276 \]
277 where $\delta _{ij}$ is Kronecker delta function. Inverting the $B$
278 matrix, we obtain
279
280 \[
281 C = B^{ - 1} = \left( {\begin{array}{*{20}c}
282 {C_{11} } & \ldots & {C_{1N} } \\
283 \vdots & \ddots & \vdots \\
284 {C_{N1} } & \cdots & {C_{NN} } \\
285 \end{array}} \right)
286 \]
287 , which can be partitioned into $N \times N$ $3 \times 3$ block
288 $C_{ij}$. With the help of $C_{ij}$ and skew matrix $U_i$
289 \[
290 U_i = \left( {\begin{array}{*{20}c}
291 0 & { - z_i } & {y_i } \\
292 {z_i } & 0 & { - x_i } \\
293 { - y_i } & {x_i } & 0 \\
294 \end{array}} \right)
295 \]
296 where $x_i$, $y_i$, $z_i$ are the components of the vector joining
297 bead $i$ and origin $O$. Hence, the elements of resistance tensor at
298 arbitrary origin $O$ can be written as
299 \begin{equation}
300 \begin{array}{l}
301 \Xi _{}^{tt} = \sum\limits_i {\sum\limits_j {C_{ij} } } , \\
302 \Xi _{}^{tr} = \Xi _{}^{rt} = \sum\limits_i {\sum\limits_j {U_i C_{ij} } } , \\
303 \Xi _{}^{rr} = - \sum\limits_i {\sum\limits_j {U_i C_{ij} } } U_j \\
304 \end{array}
305 \label{introEquation:ResistanceTensorArbitraryOrigin}
306 \end{equation}
307
308 The resistance tensor depends on the origin to which they refer. The
309 proper location for applying friction force is the center of
310 resistance (or center of reaction), at which the trace of rotational
311 resistance tensor, $ \Xi ^{rr}$ reaches a minimum value.
312 Mathematically, the center of resistance is defined as an unique
313 point of the rigid body at which the translation-rotation coupling
314 tensor are symmetric,
315 \begin{equation}
316 \Xi^{tr} = \left( {\Xi^{tr} } \right)^T
317 \label{introEquation:definitionCR}
318 \end{equation}
319 From Equation \ref{introEquation:ResistanceTensorArbitraryOrigin},
320 we can easily find out that the translational resistance tensor is
321 origin independent, while the rotational resistance tensor and
322 translation-rotation coupling resistance tensor depend on the
323 origin. Given the resistance tensor at an arbitrary origin $O$, and
324 a vector ,$r_{OP}(x_{OP}, y_{OP}, z_{OP})$, from $O$ to $P$, we can
325 obtain the resistance tensor at $P$ by
326 \begin{equation}
327 \begin{array}{l}
328 \Xi _P^{tt} = \Xi _O^{tt} \\
329 \Xi _P^{tr} = \Xi _P^{rt} = \Xi _O^{tr} - U_{OP} \Xi _O^{tt} \\
330 \Xi _P^{rr} = \Xi _O^{rr} - U_{OP} \Xi _O^{tt} U_{OP} + \Xi _O^{tr} U_{OP} - U_{OP} \Xi _O^{{tr} ^{^T }} \\
331 \end{array}
332 \label{introEquation:resistanceTensorTransformation}
333 \end{equation}
334 where
335 \[
336 U_{OP} = \left( {\begin{array}{*{20}c}
337 0 & { - z_{OP} } & {y_{OP} } \\
338 {z_i } & 0 & { - x_{OP} } \\
339 { - y_{OP} } & {x_{OP} } & 0 \\
340 \end{array}} \right)
341 \]
342 Using Equations \ref{introEquation:definitionCR} and
343 \ref{introEquation:resistanceTensorTransformation}, one can locate
344 the position of center of resistance,
345 \begin{eqnarray*}
346 \left( \begin{array}{l}
347 x_{OR} \\
348 y_{OR} \\
349 z_{OR} \\
350 \end{array} \right) & = &\left( {\begin{array}{*{20}c}
351 {(\Xi _O^{rr} )_{yy} + (\Xi _O^{rr} )_{zz} } & { - (\Xi _O^{rr} )_{xy} } & { - (\Xi _O^{rr} )_{xz} } \\
352 { - (\Xi _O^{rr} )_{xy} } & {(\Xi _O^{rr} )_{zz} + (\Xi _O^{rr} )_{xx} } & { - (\Xi _O^{rr} )_{yz} } \\
353 { - (\Xi _O^{rr} )_{xz} } & { - (\Xi _O^{rr} )_{yz} } & {(\Xi _O^{rr} )_{xx} + (\Xi _O^{rr} )_{yy} } \\
354 \end{array}} \right)^{ - 1} \\
355 & & \left( \begin{array}{l}
356 (\Xi _O^{tr} )_{yz} - (\Xi _O^{tr} )_{zy} \\
357 (\Xi _O^{tr} )_{zx} - (\Xi _O^{tr} )_{xz} \\
358 (\Xi _O^{tr} )_{xy} - (\Xi _O^{tr} )_{yx} \\
359 \end{array} \right) \\
360 \end{eqnarray*}
361
362 where $x_OR$, $y_OR$, $z_OR$ are the components of the vector
363 joining center of resistance $R$ and origin $O$.
364
365 \subsection{Langevin Dynamics for Rigid Particles of Arbitrary Shape\label{LDRB}}
366
367 Consider a Langevin equation of motions in generalized coordinates
368 \begin{equation}
369 M_i \dot V_i (t) = F_{s,i} (t) + F_{f,i(t)} + F_{r,i} (t)
370 \label{LDGeneralizedForm}
371 \end{equation}
372 where $M_i$ is a $6\times6$ generalized diagonal mass (include mass
373 and moment of inertial) matrix and $V_i$ is a generalized velocity,
374 $V_i = V_i(v_i,\omega _i)$. The right side of Eq.~
375 (\ref{LDGeneralizedForm}) consists of three generalized forces in
376 lab-fixed frame, systematic force $F_{s,i}$, dissipative force
377 $F_{f,i}$ and stochastic force $F_{r,i}$. While the evolution of the
378 system in Newtownian mechanics typically refers to lab-fixed frame,
379 it is also convenient to handle the rotation of rigid body in
380 body-fixed frame. Thus the friction and random forces are calculated
381 in body-fixed frame and converted back to lab-fixed frame by:
382 \[
383 \begin{array}{l}
384 F_{f,i}^l (t) = A^T F_{f,i}^b (t), \\
385 F_{r,i}^l (t) = A^T F_{r,i}^b (t) \\
386 \end{array}.
387 \]
388 Here, the body-fixed friction force $F_{r,i}^b$ is proportional to
389 the body-fixed velocity at center of resistance $v_{R,i}^b$ and
390 angular velocity $\omega _i$,
391 \begin{equation}
392 F_{r,i}^b (t) = \left( \begin{array}{l}
393 f_{r,i}^b (t) \\
394 \tau _{r,i}^b (t) \\
395 \end{array} \right) = - \left( {\begin{array}{*{20}c}
396 {\Xi _{R,t} } & {\Xi _{R,c}^T } \\
397 {\Xi _{R,c} } & {\Xi _{R,r} } \\
398 \end{array}} \right)\left( \begin{array}{l}
399 v_{R,i}^b (t) \\
400 \omega _i (t) \\
401 \end{array} \right),
402 \end{equation}
403 while the random force $F_{r,i}^l$ is a Gaussian stochastic variable
404 with zero mean and variance
405 \begin{equation}
406 \left\langle {F_{r,i}^l (t)(F_{r,i}^l (t'))^T } \right\rangle =
407 \left\langle {F_{r,i}^b (t)(F_{r,i}^b (t'))^T } \right\rangle =
408 2k_B T\Xi _R \delta (t - t'). \label{randomForce}
409 \end{equation}
410
411 The equation of motion for $v_i$ can be written as
412 \begin{equation}
413 m\dot v_i (t) = f_{t,i} (t) = f_{s,i} (t) + f_{f,i}^l (t) +
414 f_{r,i}^l (t)
415 \end{equation}
416 Since the frictional force is applied at the center of resistance
417 which generally does not coincide with the center of mass, an extra
418 torque is exerted at the center of mass. Thus, the net body-fixed
419 frictional torque at the center of mass, $\tau _{n,i}^b (t)$, is
420 given by
421 \begin{equation}
422 \tau _{r,i}^b = \tau _{r,i}^b +r_{MR} \times f_{r,i}^b
423 \end{equation}
424 where $r_{MR}$ is the vector from the center of mass to the center
425 of the resistance. Instead of integrating angular velocity in
426 lab-fixed frame, we consider the equation of motion of angular
427 momentum in body-fixed frame
428 \begin{equation}
429 \dot j_i (t) = \tau _{t,i} (t) = \tau _{s,i} (t) + \tau _{f,i}^b (t)
430 + \tau _{r,i}^b(t)
431 \end{equation}
432
433 Embedding the friction terms into force and torque, one can
434 integrate the langevin equations of motion for rigid body of
435 arbitrary shape in a velocity-Verlet style 2-part algorithm, where
436 $h= \delta t$:
437
438 {\tt moveA:}
439 \begin{align*}
440 {\bf v}\left(t + h / 2\right) &\leftarrow {\bf v}(t)
441 + \frac{h}{2} \left( {\bf f}(t) / m \right), \\
442 %
443 {\bf r}(t + h) &\leftarrow {\bf r}(t)
444 + h {\bf v}\left(t + h / 2 \right), \\
445 %
446 {\bf j}\left(t + h / 2 \right) &\leftarrow {\bf j}(t)
447 + \frac{h}{2} {\bf \tau}^b(t), \\
448 %
449 \mathsf{A}(t + h) &\leftarrow \mathrm{rotate}\left( h {\bf j}
450 (t + h / 2) \cdot \overleftrightarrow{\mathsf{I}}^{-1} \right).
451 \end{align*}
452
453 In this context, the $\mathrm{rotate}$ function is the reversible
454 product of the three body-fixed rotations,
455 \begin{equation}
456 \mathrm{rotate}({\bf a}) = \mathsf{G}_x(a_x / 2) \cdot
457 \mathsf{G}_y(a_y / 2) \cdot \mathsf{G}_z(a_z) \cdot \mathsf{G}_y(a_y
458 / 2) \cdot \mathsf{G}_x(a_x /2),
459 \end{equation}
460 where each rotational propagator, $\mathsf{G}_\alpha(\theta)$,
461 rotates both the rotation matrix ($\mathsf{A}$) and the body-fixed
462 angular momentum (${\bf j}$) by an angle $\theta$ around body-fixed
463 axis $\alpha$,
464 \begin{equation}
465 \mathsf{G}_\alpha( \theta ) = \left\{
466 \begin{array}{lcl}
467 \mathsf{A}(t) & \leftarrow & \mathsf{A}(0) \cdot \mathsf{R}_\alpha(\theta)^T, \\
468 {\bf j}(t) & \leftarrow & \mathsf{R}_\alpha(\theta) \cdot {\bf
469 j}(0).
470 \end{array}
471 \right.
472 \end{equation}
473 $\mathsf{R}_\alpha$ is a quadratic approximation to the single-axis
474 rotation matrix. For example, in the small-angle limit, the
475 rotation matrix around the body-fixed x-axis can be approximated as
476 \begin{equation}
477 \mathsf{R}_x(\theta) \approx \left(
478 \begin{array}{ccc}
479 1 & 0 & 0 \\
480 0 & \frac{1-\theta^2 / 4}{1 + \theta^2 / 4} & -\frac{\theta}{1+
481 \theta^2 / 4} \\
482 0 & \frac{\theta}{1+ \theta^2 / 4} & \frac{1-\theta^2 / 4}{1 +
483 \theta^2 / 4}
484 \end{array}
485 \right).
486 \end{equation}
487 All other rotations follow in a straightforward manner.
488
489 After the first part of the propagation, the forces and body-fixed
490 torques are calculated at the new positions and orientations
491
492 {\tt doForces:}
493 \begin{align*}
494 {\bf f}(t + h) &\leftarrow
495 - \left(\frac{\partial V}{\partial {\bf r}}\right)_{{\bf r}(t + h)}, \\
496 %
497 {\bf \tau}^{s}(t + h) &\leftarrow {\bf u}(t + h)
498 \times \frac{\partial V}{\partial {\bf u}}, \\
499 %
500 {\bf \tau}^{b}(t + h) &\leftarrow \mathsf{A}(t + h)
501 \cdot {\bf \tau}^s(t + h).
502 \end{align*}
503
504 {\sc oopse} automatically updates ${\bf u}$ when the rotation matrix
505 $\mathsf{A}$ is calculated in {\tt moveA}. Once the forces and
506 torques have been obtained at the new time step, the velocities can
507 be advanced to the same time value.
508
509 {\tt moveB:}
510 \begin{align*}
511 {\bf v}\left(t + h \right) &\leftarrow {\bf v}\left(t + h / 2
512 \right)
513 + \frac{h}{2} \left( {\bf f}(t + h) / m \right), \\
514 %
515 {\bf j}\left(t + h \right) &\leftarrow {\bf j}\left(t + h / 2
516 \right)
517 + \frac{h}{2} {\bf \tau}^b(t + h) .
518 \end{align*}
519
520 \section{Results and Discussion}
521
522 The Langevin algorithm described in previous section has been
523 implemented in {\sc oopse}\cite{Meineke2005} and applied to the
524 studies of kinetics and thermodynamic properties in several systems.
525
526 \subsection{Temperature Control}
527
528 As shown in Eq.~\ref{randomForce}, random collisions associated with
529 the solvent's thermal motions is controlled by the external
530 temperature. The capability to maintain the temperature of the whole
531 system was usually used to measure the stability and efficiency of
532 the algorithm. In order to verify the stability of this new
533 algorithm, a series of simulations are performed on system
534 consisiting of 256 SSD water molecules with different viscosities.
535 Initial configuration for the simulations is taken from a 1ns NVT
536 simulation with a cubic box of 19.7166~\AA. All simulation are
537 carried out with cutoff radius of 9~\AA and 2 fs time step for 1 ns
538 with reference temperature at 300~K. Average temperature as a
539 function of $\eta$ is shown in Table \ref{langevin:viscosity} where
540 the temperatures range from 303.04~K to 300.47~K for $\eta = 0.01 -
541 1$ poise. The better temperature control at higher viscosity can be
542 explained by the finite size effect and relative slow relaxation
543 rate at lower viscosity regime.
544 \begin{table}
545 \caption{AVERAGE TEMPERATURES FROM LANGEVIN DYNAMICS SIMULATIONS OF
546 SSD WATER MOLECULES WITH REFERENCE TEMPERATURE AT 300~K.}
547 \label{langevin:viscosity}
548 \begin{center}
549 \begin{tabular}{|l|l|l|}
550 \hline
551 $\eta$ & $\text{T}_{\text{avg}}$ & $\text{T}_{\text{rms}}$ \\
552 1 & 300.47 & 10.99 \\
553 0.1 & 301.19 & 11.136 \\
554 0.01 & 303.04 & 11.796 \\
555 \hline
556 \end{tabular}
557 \end{center}
558 \end{table}
559
560 Another set of calculation were performed to study the efficiency of
561 temperature control using different temperature coupling schemes.
562 The starting configuration is cooled to 173~K and evolved using NVE,
563 NVT, and Langevin dynamic with time step of 2 fs.
564 Fig.~\ref{langevin:temperature} shows the heating curve obtained as
565 the systems reach equilibrium. The orange curve in
566 Fig.~\ref{langevin:temperature} represents the simulation using
567 Nos\'e-Hoover temperature scaling scheme with thermostat of 5 ps
568 which gives reasonable tight coupling, while the blue one from
569 Langevin dynamics with viscosity of 0.1 poise demonstrates a faster
570 scaling to the desire temperature. In extremely lower friction
571 regime (when $ \eta \approx 0$), Langevin dynamics becomes normal
572 NVE (see green curve in Fig.~\ref{langevin:temperature}) which loses
573 the temperature control ability.
574
575 \begin{figure}
576 \centering
577 \includegraphics[width=\linewidth]{temperature.eps}
578 \caption[Plot of Temperature Fluctuation Versus Time]{Plot of
579 temperature fluctuation versus time.} \label{langevin:temperature}
580 \end{figure}
581
582 \subsection{Langevin Dynamics of Banana Shaped Molecule}
583
584 In order to verify that Langevin dynamics can mimic the dynamics of
585 the systems absent of explicit solvents, we carried out two sets of
586 simulations and compare their dynamic properties.
587
588 Fig.~\ref{langevin:twoBanana} shows a snapshot of the simulation
589 made of 256 pentane molecules and two banana shaped molecules at
590 273~K. It has an equivalent implicit solvent system containing only
591 two banana shaped molecules with viscosity of 0.289 center poise. To
592 calculate the hydrodynamic properties of the banana shaped molecule,
593 we create a rough shell model (see Fig.~\ref{langevin:roughShell}),
594 in which the banana shaped molecule is represented as a ``shell''
595 made of 2266 small identical beads with size of 0.3 $\AA$ on the
596 surface. Applying the procedure described in
597 Sec.~\ref{introEquation:ResistanceTensorArbitraryOrigin}, we
598 identified the center of resistance at $(0, 0.7482, -0.1988)$, as
599 well as the resistance tensor,
600 \[
601 \left( {\begin{array}{*{20}c}
602 0.9261 & 1.310e-14 & -7.292e-15&5.067e-14&0.08585&0.2057\\
603 3.968e-14& 0.9270&-0.007063& 0.08585&6.764e-14&4.846e-14\\
604 -6.561e-16&-0.007063&0.7494&0.2057&4.846e-14&1.5036e-14\\
605 5.067e-14&0.0858&0.2057& 58.64& 8.563e-13&-8.5736\\
606 0.08585&6.764e-14&4.846e-14&1.555e-12&48.30&3.219&\\
607 0.2057&4.846e-14&1.5036e-14&-3.904e-13&3.219&10.7373\\
608 \end{array}} \right).
609 \]
610
611
612
613 \begin{figure}
614 \centering
615 \includegraphics[width=\linewidth]{roughShell.eps}
616 \caption[Rough shell model for banana shaped molecule]{Rough shell
617 model for banana shaped molecule.} \label{langevin:roughShell}
618 \end{figure}
619
620 \begin{figure}
621 \centering
622 \includegraphics[width=\linewidth]{twoBanana.eps}
623 \caption[Snapshot from Simulation of Two Banana Shaped Molecules and
624 256 Pentane Molecules]{Snapshot from simulation of two Banana shaped
625 molecules and 256 pentane molecules.} \label{langevin:twoBanana}
626 \end{figure}
627
628 \begin{figure}
629 \centering
630 \includegraphics[width=\linewidth]{vacf.eps}
631 \caption[Plots of Velocity Auto-correlation functions]{Velocity
632 Auto-correlation function of NVE (blue) and Langevin dynamics
633 (red).} \label{langevin:twoBanana}
634 \end{figure}
635
636 \begin{figure}
637 \centering
638 \includegraphics[width=\linewidth]{uacf.eps}
639 \caption[Snapshot from Simulation of Two Banana Shaped Molecules and
640 256 Pentane Molecules]{Snapshot from simulation of two Banana shaped
641 molecules and 256 pentane molecules.} \label{langevin:twoBanana}
642 \end{figure}
643
644 \section{Conclusions}