1: %===========================================================================
2: % See the Rev Tex 4 README file
3: % It also requires running BibTeX. The commands are as follows:
4: %
5: % 1) latex apssamp.tex
6: % 2) bibtex apssamp
7: % 3) latex apssamp.tex
8: % 4) latex apssamp.tex
9: %====================================================================================
10: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
11: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
12:
13: % Some other (several out of many) possibilities
14: %\documentclass[preprint,aps]{revtex4}
15: %\documentclass[preprint,aps,draft]{revtex4}
16: %\documentclass[prb]{revtex4}% Physical Review B
17:
18: \usepackage{graphicx}% Include figure files
19: \usepackage{dcolumn}% Align table columns on decimal point
20: \usepackage{bm}% bold math
21: \usepackage{tabularx}
22: \usepackage{rotating}
23: \usepackage{graphicx}
24: \usepackage{epsfig}
25: \usepackage{amsmath,amssymb,bbm}
26: \usepackage{subfigure}
27:
28:
29: \renewcommand{\thefootnote}{\fnsymbol{footnote}}
30: %=================================================================newcommands
31:
32: %===========================================================================
33: %\nofiles
34:
35: \begin{document}
36:
37: \title{A Macromolecule in a Solvent: Adaptive Resolution Molecular
38: Dynamics Simulation}
39: \author{Matej Praprotnik}
40: \altaffiliation{On leave from the National Institute of Chemistry, Hajdrihova 19,
41: SI-1001 Ljubljana, Slovenia. Electronic Mail: praprot@cmm.ki.si}
42: \author{Luigi Delle Site}
43: \author{Kurt Kremer}
44: \affiliation{%
45: Max-Planck-Institut f\"ur Polymerforschung, Ackermannweg 10, D-55128 Mainz, Germany
46: }%
47:
48:
49:
50:
51: %====================================================================Abstract
52: \begin{abstract}
53: We report adaptive resolution molecular dynamics simulations of a
54: flexible linear polymer in solution. The solvent, i.e., a liquid
55: of tetrahedral molecules, is represented within a certain radius from the
56: polymer's center of mass with a high level of
57: detail, while a lower coarse-grained resolution is used for the
58: more distant solvent. The high resolution sphere moves with
59: the polymer and freely exchanges molecules with the low resolution
60: region through a transition regime. The solvent molecules change
61: their resolution and number of degrees of freedom on-the-fly.
62: We show that our approach correctly reproduces the static and
63: dynamic properties of the polymer chain and surrounding solvent.
64: \end{abstract}
65: %===========================================================================
66: \pacs{02.70.Ns, 61.20.Ja, 61.25.Em, 61.25.Hq}
67: % Classification Scheme.
68: %\keywords{Suggested keywords}%Use showkeys class option if keyword
69: %display desired
70: \maketitle
71: %==============================================================================================================INTRO
72: \section{Introduction}
73: The structure of polymers in solution is determined by the
74: solvent-polymer interaction. In the case of a nonpolar polymer in
75: an nonpolar solvent, one typically distinguishes three ''types''
76: of solvent, good, $\Theta$ or marginal, and poor. In the case of a
77: good solvent the solvent-solvent, solvent-polymer, and polymer-polymer
78: interactions effectively result in a situation, where the
79: chain monomers are preferably surrounded by solvent molecules. As
80: a result the chains are extended and the size scales as $\langle
81: R^2 \propto N^{2\nu}\rangle$ with N being the number of monomers and
82: $\nu \cong 0.6$ in three dimensions. For poor solvent one observes
83: just the opposite and the chains collapse into a dense globule,
84: $\langle R^2 \propto N^{2/3}\rangle$. The $\Theta$ regime is where
85: these two effects compensate and the chains behaves
86: to a first approximation as a random walk, i.e., $\langle R^2
87: \propto N\rangle$. In the limit of $N\rightarrow\infty$ the
88: $\Theta$-point is a tricritical point in the phase diagram.
89: As long as the solvent does
90: not induce special local correlations beyond an unspecific
91: attraction/repulsion and one is not studying dynamical
92: properties the collapse of polymers is usually studied with an
93: implicit solvent. The complicated local interactions are accounted
94: for by an effective interaction between the chain beads. Studies
95: of that kind have a long tradition in polymer science and the behavior
96: is now well understood. Beyond that there
97: are however many situations, where it becomes difficult or even
98: questionable, to ignore the local structure of the solvent.
99: Solvent can play an important role in the functional properties
100: of macromolecules. For example, dehydration studies of proteins
101: solvated in water demonstrated that at least a monolayer of water
102: is needed for full protein functionality\cite{Careri:1999}. The
103: influence of a macromolecule on the structure and dynamics of the
104: surrounding solvent is also an important issue. Therefore, a detailed study of interactions of a
105: macromolecule with a solvent beyond effective coupling parameters
106: is quite often required for an understanding of the
107: macromolecule's structure, dynamics, and function. To determine
108: the interactions of a solvent with a macromolecular solute
109: chemistry specific interactions on the atomic level of detail have
110: to be considered. However, the resulting solvating phenomena
111: manifest themselves at mesoscopic and macroscopic
112: scales\cite{Das:2005} and in the overall structure of the chains. Due
113: to large number of degrees of freedom (DOFs) such systems are
114: difficult to tackle using all-atom computer
115: simulations\cite{Villa:2005}. Moreover, the vast majority of the
116: simulation time is typically spent treating the solvent and not
117: the polymer or protein. A step to bridge the gap between the time
118: and length scales accessible to simulations that still retain an
119: atomistic level of detail and the solvating phenomena on
120: longer time and larger length scales, is given by hybrid
121: multiscale schemes that concurrently couple different physical
122: descriptions of the system (see, e.g., Refs.
123: \cite{Rafii:1998,Broughton:1999,Ahlrichs:1999,Malevanets:2000,
124: Csanyi:2004,Abrams:2005,Neri:2005, Fabritiis:2006}).
125:
126: Recently, we have proposed an adaptive resolution molecular
127: dynamics (MD) scheme (AdResS) that concurrently couples the
128: atomistic and mesoscopic length scales of a generic
129: solvent\cite{Praprotnik:2005:4,Praprotnik:2006,Praprotnik:2006:1}.
130: In the first application we studied a liquid of tetrahedral
131: molecules where an atomistic region was separated from the mesoscopic
132: one by a flat or a spherical boundary. The two regimes with
133: different resolutions freely exchanged molecules while maintaining
134: the thermodynamical equilibrium in the system. The spatial regions
135: of different resolutions, however, remained constant during the
136: course of the simulations. More recently this approach was
137: extended to the study of water\cite{Praprotnik:2006:2}. In the
138: present paper, we generalize our approach to the study of a
139: polymer chain in solution. The chain is surrounded by solvent with
140: ''atomistic'' resolution. When the chain moves around, the sphere
141: of atomistically resolved solvent molecules moves together with
142: the center of mass of the chain. In this way the chain is free to
143: move around, although the explicit resolution sphere is much
144: smaller than the overall simulation volume. This enables us to
145: efficiently treat solvation phenomena, because only the solvent in
146: the vicinity of a macromolecule is represented with a
147: sufficiently high level of detail to take the specific
148: interactions between the solvent and the solute into account.
149: Solvent farther away from the solute, where the high
150: resolution is no longer required, is represented on a more
151: coarse-grained level. In this work, a macromolecule is represented
152: by a generic flexible polymer chain\cite{Kremer:1990} embedded in
153: a solvent of tetrahedral molecules introduced in Refs.
154: \cite{Praprotnik:2005:4,Praprotnik:2006}. This study represents a
155: first methodological step towards adaptive resolution MD
156: simulations of systems of biological relevance, e.g., a
157: protein in water.
158:
159:
160: The paper is organized as follows: In section II the dual scale
161: model of a polymer chain in a liquid is presented. The hybrid
162: numerical scheme and computational details are given in section
163: III. The results and discussion are reported in section IV,
164: followed by a summary and outlook in section V.
165:
166:
167: \section{Multiscale Model}
168: We study a single generic bead spring polymer solvated in a
169: molecular liquid as illustrated in figure \ref{Fig.1}.
170: \begin{figure}[!ht]
171: \centering
172: \includegraphics[width=8.0cm]{fig1.eps}
173: \caption{(Color online) A schematic plot of a solvated generic
174: bead-spring polymer. The solvent is modeled on different levels of
175: detail: solvent molecules within a certain radius from the
176: polymer's center of mass are represented with a high (atomistic)
177: resolution while a lower mesoscopic resolution is used for the
178: more distant solvent. The high resolution sphere moves with
179: the polymer's center of mass. The polymer beads are represented
180: smaller than the solvent molecules for presentation
181: convenience, for details see text.}\label{Fig.1}
182: \end{figure}
183: Solvent molecules within a distance $r_0$ from the polymer's
184: center of mass are modeled with all 'atomistic' details to
185: properly describe the specific polymer-solvent interactions. For
186: the description of the solvent farther away, where the high
187: resolution is not required, we use a lower resolution. The solvent
188: molecules then, depending on their distance to the polymer's
189: center of mass, automatically adapt their resolution on-the-fly.
190:
191: The model solvent is a liquid of $n$ tetrahedral molecules as
192: introduced in Refs. \cite{Praprotnik:2005:4,Praprotnik:2006}. The
193: solvent molecules in the high resolution regime are composed of
194: four equal atoms with mass $m_0$. Their size $\sigma$ is fixed via
195: the repulsive Weeks-Chandler-Andersen potential
196: \begin{multline}
197: U_{rep}^{atom}(r_{i\alpha j\beta})=\\\left\{\begin{array}{rc}
198: 4\varepsilon\bigl[\bigl(\frac{\sigma}{r_{i\alpha
199: j\beta}}\bigr)^{12}-\bigl(\frac{\sigma}{r_{i\alpha j\beta}}\bigr)^6+\frac{1}{4}\bigr];
200: & r_{i\alpha j\beta}\le 2^{1/6}\sigma\\
201: 0; & r_{i\alpha j\beta}> 2^{1/6}\sigma
202: \end{array}\right.\label{eq.1}
203: \end{multline}
204: with the cutoff at $2^{1/6}\sigma$. $\sigma$ and $\varepsilon$ are
205: the standard Lennard Jones units for lengths and energy
206: respectively. $r_{i\alpha j\beta}$ is the distance between the
207: atom $i$ of the molecule $\alpha$ and the atom $j$ of the molecule
208: $\beta$. The neighboring atoms in a given molecule $\alpha$ are
209: connected by finite extensible nonlinear elastic (FENE) bonds
210: \begin{multline}
211: U_{bond}^{atom}(r_{i\alpha j\alpha})=\\\left\{\begin{array}{rc}
212: -\frac{1}{2}kR_0^2\ln\bigl[1-\bigl(\frac{r_{i\alpha
213: j\alpha}}{R_0}\bigl)^2\bigr]; & r_{i\alpha j\alpha}\le R_0\\
214: \infty; & r_{i\alpha j\alpha}> R_0
215: \end{array}\right.\label{eq.2}
216: \end{multline}
217: with divergence length $R_0=1.5\sigma$ and stiffness
218: $k=30\varepsilon/\sigma^2$, so that the average bond length is
219: approximately $0.97\sigma$ for $k_BT=\varepsilon$, where $T$ is the
220: temperature of the system and $k_B$ is Boltzmann's constant. For
221: the coarse-grained solvent model in the low resolution regime we
222: use one-site spherical molecules interacting via an effective pair
223: potential\cite{Praprotnik:2006}, which was derived such that the
224: statistical properties, i.e., the center of mass radial
225: distribution function and pressure, of the high resolution liquid
226: are accurately reproduced. This is also needed for the present
227: study, since the motion of the high resolution sphere should not
228: be linked to strong rearrangements in the liquid. The high and low
229: resolution freely exchange molecules through a transition regime
230: containing hybrid molecules (see figure \ref{Fig.1}), where the
231: molecules with no extra equilibration adapt their
232: resolution and change the number of DOFs
233: accordingly\cite{Praprotnik:2005:4,Praprotnik:2006,Praprotnik:2006:1}.
234:
235: The polymer is modeled as a standard bead-spring polymer
236: chain\cite{Kremer:1990}. It contains $N$ monomers, which represent
237: chemical repeat units, usually comprising several atoms. The
238: interactions between monomers (beads) are defined using Eqs.
239: (\ref{eq.1}) and (\ref{eq.2}) with the rescaled values
240: $\sigma_B=1.8\sigma$, $R_{0_B}=R_0\sigma_B/\sigma=1.5\sigma_B$,
241: and $k_B=k\sigma^2/\sigma_B^2=30\varepsilon/\sigma_B^2$, such that
242: the size of the polymer bead $\sigma_B$ is approximately the same
243: as the size of the solvent molecule\cite{Praprotnik:2006}. The
244: average bond length between beads is rescaled accordingly. The
245: bead mass is also increased $m_B=5m_0$ to make them behave more
246: like Brownian particles. Standard Lorentz-Berthelot mixing
247: rules\cite{Allen:1987} are used for the interaction between
248: monomers and the 'atoms' of the solvent molecules.
249:
250:
251:
252: \section{Numerical Scheme and Computational Details}
253: To smoothly couple the regimes of high and low level of detail of
254: the description of the solvent molecules, we apply the recently
255: introduced AdResS scheme\cite{Praprotnik:2005:4}. There the
256: molecules can freely move between the regimes, they are in
257: equilibrium with each other with no barrier in between. The
258: transition is governed by a weighting function $w(r)\in[0,1]$ that
259: interpolates the molecular interaction forces between the two
260: regimes, and assigns the identity of the solvent molecule. We
261: resort here to the weighting function defined in Ref.
262: \cite{Praprotnik:2006}:
263: \begin{eqnarray}
264: w(r)=\left\{\begin{array}{rc}
265: 1; & r_0 > r\ge 0\\
266: 0; & r\ge r_0+d\\
267: \cos^2[\frac{\pi}{2d}(r-r_0)]; & r_0+d > r\ge r_0
268: \end{array}\right.\label{eq.3}
269: \end{eqnarray}
270: where $r_0$ is the radius of the high resolution region and $d$
271: the interface region width, cf. \ref{Fig.1}. The radius $r_0$ must
272: be chosen sufficiently large so that the whole polymer always
273: stays within the high resolution solvent regime. $w(r)$ is defined
274: in such a way that $w=1$ corresponds to the high resolution, $w=0$
275: to the low resolution, and values $0<w<1$ to the transition regime,
276: respectively. This leads to intermolecular force acting between
277: centers of mass of solvent molecules $\alpha$ and $\beta$:
278: \begin{multline}
279: {\bf F}_{\alpha\beta}=w(|{\bf R}_\alpha-{\bf R}|)w(|{\bf R}_\beta-{\bf
280: R}|){\bf
281: F}_{\alpha\beta}^{ex}\\+[1-w(|{\bf R}_\alpha-{\bf R}|)w(|{\bf R}_\beta-{\bf
282: R}|)]{\bf
283: F}_{\alpha\beta}^{cg}.\label{eq.4}
284: \end{multline}
285: ${\bf F}_{\alpha \beta}$ is the total intermolecular force acting
286: between centers of mass of the solvent molecules $\alpha$ and
287: $\beta$. ${\bf F}_{\alpha\beta}^{ex}$ is the sum of all pair
288: 'atom' interactions between explicit tetrahedral 'atoms' of the
289: solvent molecule $\alpha$ and explicit tetrahedral 'atoms' of the
290: solvent molecule $\beta$, ${\bf F}_{\alpha\beta}^{cg}$ is the
291: effective pair force between the two solvent molecules, and ${\bf
292: R}_\alpha$, ${\bf R}_\beta$, and ${\bf R}$ are the centers of mass
293: of the molecules $\alpha$, $\beta$ and the polymer, respectively.
294: Note that one has to interpolate the forces and not the
295: interaction potentials in Eq. (\ref{eq.4}) if the Newton's Third
296: Law is to be satisfied~\cite{Praprotnik:2006:1}. To suppress the
297: unphysical density and pressure fluctuations emerging as artifacts
298: of the scheme given in Eq. (\ref{eq.4}) within the transition zone
299: we employ an interface pressure correction\cite{Praprotnik:2006}.
300: The latter involves a reparametrization of the effective
301: potential in the system composed of exclusively hybrid molecules with $w=1/2$.
302: Each time a solvent molecule crosses a boundary between the
303: different regimes it gains or looses on-the-fly (depending on
304: whether it leaves or enters the coarse-grained region) its
305: equilibrated rotational and vibrational DOFs while retaining its
306: linear momentum\cite{Praprotnik:2006:1, Praprotnik:2005,
307: Praprotnik:2005:1, Praprotnik:2005:2}. This change in resolution
308: requires to supply or remove ''latent heat'' and thus must be
309: employed together with a thermostat that couples locally to the
310: particle motion~\cite{Praprotnik:2005:4,Praprotnik:2006:1}. This
311: is achieved by coupling the particle motion to the Dissipative
312: Particle Dynamics (DPD)
313: thermostat\cite{Soddemann:2003}. This bears the additional
314: advantage of preserving momentum conservation and correct
315: reproduction of hydrodynamics in our $nVT$ MD simulations. Because
316: of the freely moving polymer chain and solvent molecules, the
317: above scheme requires the center of the high resolution sphere to
318: move with the polymer but slowly compared to the
319: surrounding solvent molecules, so that they at the boundary
320: between different regimes have enough time to adapt to the
321: resolution change. The validity condition for our approach thus
322: requires $D_{polymer} \ll D_{solvent}$, where $D_{polymer}$ and
323: $D_{solvent}$ the corresponding diffusion constants. This
324: condition is trivially fulfilled in polymeric solutions and thus
325: also in our simulations (see the next section).
326:
327: We conducted all MD simulations using the ESPResSo
328: package\cite{Espresso:2005}. We integrated Newtons equations of
329: motion by a standard velocity Verlet algorithm with a time step
330: $\Delta t=0.005 \tau$ and coupled the motion of the particles to a
331: DPD theromstat\cite{Soddemann:2003} with the temperature set to
332: $T=\varepsilon/k_B$. The DPD friction constant
333: $\zeta=0.5\tau^{-1}$, where
334: $\tau=(\varepsilon/m_0\sigma^2)^{-1/2}$, and the DPD cutoff radius
335: was set equal to the cuttoff radius of the effective pair
336: interaction between solvent molecules, i.e.,
337: $3.5\sigma$\cite{Praprotnik:2006}. The width of the transition
338: regime is $2.5\sigma$\cite{Praprotnik:2006}. Periodic boundary conditions and the minimum image
339: convention\cite{Allen:1987} were employed. After equilibration,
340: trajectories of $5000\tau$ were obtained, with configurations
341: stored every $5\tau$. These production runs were performed with a
342: $10^{9}\varepsilon/\sigma$ force capping to prevent possible force
343: singularities that could emerge due to overlaps with the
344: neighboring molecules when a given molecule enters the transition
345: layer from the coarse-grained side \cite{Praprotnik:2005:4}. The
346: temperature was calculated using the fractional analog of the
347: equipartition theorem:
348: \begin{equation}
349: \left<K_\alpha\right>=\frac{\alpha k_BT}2,
350: \label{eq4}
351: \end{equation}
352: where $\left<K_\alpha\right>$ is the average kinetic energy per
353: fractional quadratic DOF with the weight
354: $w(r)=\alpha$\cite{Praprotnik:2006:1}. Via Eq. (\ref{eq4}) the
355: temperature is also rigorously defined in the transition regime in
356: which the vibrational and rotational DOFs are partially 'switched
357: on/off'. The molecular number density of the solvent is
358: $\rho=0.175/\sigma^3$, which corresponds to a typical high density
359: Lennard-Jones liquid\cite{Praprotnik:2006}. We considered three
360: different system sizes with corresponding cubic box sizes:
361: $L=25.0\sigma, 30.6\sigma, 34.2\sigma$. The reduced Lennard-Jones
362: units\cite{Allen:1987} are used in the remainder of the paper.
363:
364:
365: \section{Results and Discussion}
366:
367: To validate the AdResS approach for the present polymer solvent
368: system, we carried out the analysis of the structural and dynamic
369: properties of a polymer chain in the hybrid multiscale solvent
370: compared to the corresponding fully explicit system where all
371: solvent molecules are modeled with a high level of detail, i.e.,
372: as a tetrahedral molecules.
373:
374: \subsection{Statics of the Polymer Chain and Solvent}
375: First, we focus on the explicit (\emph{ex}) systems where the
376: solvent is modeled with the high resolution all over the
377: simulation box. These results are considered as the reference to
378: check how well AdResS produces the same physics as the all-atom MD
379: simulation. The reference average thermodynamic properties of the
380: corresponding \emph{ex} systems (polymer+explicitely resolved
381: solvent) are listed in table \ref{Tab.1a}.
382: \begin{table}[ht!]
383: \centering
384: \begin{tabular}{||c|c|c|c||}
385: \hline\hline
386: $N$ & $10$ & $20$ & $30$ \\
387: $L$ & $25.0$ & $30.6$ & $34.2$\\
388: \hline
389: $<p>$ & $2.01\pm 0.04$ & $2.01\pm 0.03$ & $2.02\pm 0.01$\\
390: $<T>$ & $1.0\pm 0.01$ & $1.0\pm 0.01$ & $1.0\pm 0.01$\\
391: $<T_{polymer}>$ & $1.0\pm 0.5$ & $1.0\pm 0.3$ & $1.0\pm 0.2$\\
392: $<T_{solvent}>$ & $1.0\pm 0.01$ & $1.0\pm 0.01$ & $1.0\pm 0.01$\\
393: \hline\hline
394: \end{tabular}
395: \caption{Thermodynamic properties of the fully explicit systems (\emph{ex}) of different chain lengths
396: $N$ and box sizes $L$: average total pressure
397: $\left<p\right>$, average total temperature of the system
398: $\left<T\right>$, average temperature of the polymer
399: $\left<T_{polymer}\right>$, and average temperature of the solvent
400: $\left<T_{solvent}\right>$.} \label{Tab.1a}
401: \end{table}
402:
403: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
404:
405: The static properties of the solvent are characterized by the
406: solvent radial center of mass distribution (RDF) function depicted
407: in figures \ref{Fig.2} (a). This distribution function is within
408: the thickness of the lines the same for all systems studied,
409: including the hybrid ones. This is to be expected from our
410: previous studies and the fact, that the polymer fraction of volume
411: is very small compared to that of the solvent.
412: \begin{figure}[ht!]
413: \centering
414: \subfigure[]{\includegraphics[width=7.2cm]{fig2a.eps}}
415: \subfigure[]{\includegraphics[width=7.2cm]{fig2b.eps}}
416: \subfigure[]{\includegraphics[width=7.2cm]{fig2c.eps}}
417:
418: \caption{ (a) The solvent center of mass RDF for three different
419: polymer lengths (N=10,20,30), (b) the static structure factor of
420: the polymer in the Kratky representation, and (c) the solvent
421: density around the center of mass of the chains, which illustrates
422: the so called correlation hole.
423: }\label{Fig.2}
424: \end{figure}
425:
426:
427: The statistical properties of polymers are conveniently described
428: by a number of quantities, namely the radius of gyration
429: \begin{equation}
430: \left<R_G^2\right>=\frac{1}{N}\sum_i \left<({\bf r}_i-{\bf R})^2\right>,
431: \end{equation}
432: where ${\bf r}_i$ is the position vector of the $i$th monomer and
433: ${\bf R}=N^{-1}\sum_i{\bf r}_i$ is the polymer's center of
434: mass, the end-to-end distance
435: \begin{equation}
436: \left<R_E^2\right>=\left<({\bf r}_N-{\bf r}_1)^2\right>,
437: \end{equation}
438: and the hydrodynamic radius
439: \begin{equation}
440: \left<\frac{1}{R_H}\right>=\frac{1}{N^2}\sum_{i\ne j}\left<\frac{1}{r_{ij}}\right>,\label{eq.rh}
441: \end{equation}
442: where $r_{ij}=|{\bf r}_i-{\bf r}_j|$\cite{Praprotnik:2006:3}.
443:
444:
445: $\left<R_G^2\right>$ and $\left<R_E^2\right>$ scale as
446: \begin{equation}
447: \left<R_G^2\right>\propto\left<R_E^2\right>\propto N^{2\nu}
448: \end{equation}
449: with the number of monomers $N$ where $\nu=0.5$ in $\theta$
450: solvent and $\nu\approx 0.588$ in good solvent
451: conditions\cite{Gennes:1979,Doi:1986,Sokal:1995} with rather small
452: finite size corrections, while the hydrodynamic radius is known to
453: show significant deviations from asymptotic behavior up to very
454: long chains\cite{Dunweg:1993}.
455:
456: The single-chain static structure factor $S(q)$
457: \begin{equation}
458: S(q)=\frac{1}{N}\left<\sum_{ij}\exp(i{\bf
459: q}\cdot({\bf r}_i-{\bf r}_j))\right>
460: \end{equation}
461: probes the self similar structure within the scaling regime and
462: thus provides an accurate way to determine $\nu$. $S(q)$ scales as
463: \begin{equation}
464: S(q)\propto q^{-1/\nu}\rightarrow q^2S(q)\propto q^{2-1/\nu}\label{eq.Sq}
465: \end{equation}
466: in the regime $R_G^{-1}\ll q\ll b^{-1}$, where $b$ is the typical
467: bond length. By fitting a power law to the computed $q^2S(q)$
468: plotted in figure \ref{Fig.2} (b) we obtained the values for
469: $\nu$ reported in table \ref{Tab.1}. Table \ref{Tab.1} summarizes the values of all quantities
470: defined above, which characterize the static properties of the polymer
471: chain. The calculations were performed for $N=10,20,30$.
472: \begin{table}[ht!]
473: \centering
474: \begin{tabular}{||c|c|c|c||}
475: \hline\hline
476: $N$ & $10$ & $20$ & $30$ \\
477: $L$ & $25.0$ & $30.6$ & $34.2$ \\
478: \hline
479: $\left<\Delta r_{max}\right>$ & $4.2\pm 0.8$ & $5.7\pm 1.0$ & $8.1\pm 1.4$\\
480: $R_G=\left<R_G^2\right>^{1/2}$ & $2.7\pm 0.5$ & $3.8\pm 0.6$ & $5.0\pm 0.8$\\
481: $R_E=\left<R_E^2\right>^{1/2}$ & $6.7\pm 2.0$ & $8.6\pm 3$ & $12\pm 3$\\
482: $R_H=\left<R_H^{-1}\right>^{-1}$& $3.3\pm 0.3$ & $4.0\pm 0.3$ & $4.7\pm 0.4$\\
483: $\nu$ & $0.63$ & $0.54$ & $0.57$\\
484: $2/z$ & $0.59$ & $0.71$ & $0.67$\\
485: $\tau=R_G^2/(6D)$ & $152$ & $481$ & $1390$\\
486: \hline\hline
487: \end{tabular}
488: \caption{Summary of some polymer (embedded in the explicitely
489: resolved \emph{ex} solvent) properties: number of polymer beads $N$, size of the
490: simulation box $L$, average maximal
491: distance of a monomer from the polymer's center of mass $\left<\Delta
492: r_{max}\right>$, radius of gyration $R_G$, end-to-end distance
493: $R_E$, hydrodynamic radius $R_H$, the static exponent $\nu$, the
494: exponent $2/z$, where $z$ is the dynamic exponent, and the longest
495: relaxation time $\tau$ calculated using data from table
496: \ref{Tab.4}. The error bar for the exponents $\nu$ and $2/z$ is roughly $10\%$.}\label{Tab.1}
497: \end{table}
498:
499: Another property, which directly reveals the fractal structure of
500: the chains is the correlation hole, which is shown in figure \ref {Fig.2}
501: (c). It directly shows, to which distance from the center of mass
502: of the chains, the solvent density is perturbed by the chain
503: beads. For the later application of the hybrid scheme it is
504: important to define the explicit solvent regime large enough in
505: order to cover the correlation hole completely.
506:
507: The values of $\nu$ actually differ slightly from the asymptotical
508: value for the good solvent due to the finite chain lengths.
509: Nevertheless, the agreement improves with the increasing $N$, as
510: expected.
511:
512:
513: Let us now turn our attention to the hybrid solvent studied by MD
514: simulation using AdResS.
515:
516: To assure that the polymer is surrounded only by the explicitely
517: resolved molecules, we determine first the maximal monomer
518: distance from the polymer's center of mass, $\Delta
519: r_{max}$, as shown in figure \ref {Fig.3} for all chain lengths
520: studied.
521: \begin{figure}[!ht]
522: \centering
523: \includegraphics[width=7.2cm]{fig3.eps}
524: \caption{(Color online) Time evolution of the maximal monomer
525: distance from the polymer's center of mass, $\Delta
526: r_{max}$, for polymers with $N=10$ beads and the radius of
527: the high resolution regime $r_0=7.0$ (red line), $N=20$ and
528: $r_0=11.0$ (green line), and $N=30$ and $r_0=12.0$ (blue
529: line).}\label{Fig.3}
530: \end{figure}
531: As shown $\Delta r_{max}$ always stays within the
532: high resolution regime.
533:
534: Because for $N=30$ $\Delta r_{max}$ gets rather close
535: to $r_0$, we checked the static polymer properties for that case
536: again. In figure \ref{Fig.4} we compare the all explicit
537: simulation to the two hybrid simulation schemes (with
538: (\emph{ex-cg}$_{ic}$) and without (\emph{ex-cg})
539: the pressure correction\cite{Praprotnik:2006} in the transition regime) for the chain
540: form factor and the correlation hole. The agreement is excellent,
541: showing that the proposed scheme should at least be capable of
542: properly reproducing the conformational statistics of the embedded
543: polymer in solution.
544: \begin{figure}[!ht]
545: \centering
546: \subfigure[]{\includegraphics[width=7.2cm]{fig4a.eps}}
547: \subfigure[]{\includegraphics[width=7.2cm]{fig4b.eps}}
548: \caption{(a) The the static structure factor of the polymer with
549: $N=30$ in the Kratky representation for all three cases studied: the
550: fully explicite, the AdResS scheme with and without the
551: interface pressure correction. (b) The correlation hole for the same systems
552: as in (a).
553: }\label{Fig.4}
554: \end{figure}
555:
556:
557:
558: This is first checked by comparing the average thermodynamic
559: properties as given in table \ref{Tab.2a} for the hybrid
560: \emph{ex-cg} and \emph{ex-cg}$_{ic}$ systems (polymer+hybrid
561: solvent). While the temperatures are identical the pressure
562: correction in the interface layer reduces the pressure slightly,
563: so that the hybrid system now also there agrees quite well with the
564: all explicit simulation.
565: \begin{table}[ht!]
566: \centering
567: \begin{tabular}{||c|c|c|c||}
568: \hline\hline
569: $N$ & $10$ & $20$ & $30$ \\
570: $L$ & $25.0$ & $30.6$ & $34.2$\\
571: \hline
572: $<p>_{ex-cg}$ & $2.03\pm 0.02$ & $2.04\pm 0.01$ & $2.04\pm 0.03$\\
573: $<p>_{ex-cg_{ic}}$ & $2.01\pm 0.01$ & $2.01\pm 0.01$ & $2.01\pm 0.01$\\
574: $<T>$ & $1.0\pm 0.02$ & $1.0\pm 0.01$ & $1.0\pm 0.01$\\
575: $<T_{polymer}>$ & $1.0\pm 0.5$ & $1.0\pm 0.3$ & $1.0\pm 0.2$\\
576: $<T_{solvent}>$ & $1.0\pm 0.02$ & $1.0\pm 0.01$ & $1.0\pm 0.01$\\
577: \hline\hline
578: \end{tabular}
579: \caption{Thermodynamic properties of systems with the polymer
580: solvated in the hybrid \emph{ex-cg} solvent and the hybrid \emph{ex-cg}$_{ic}$
581: solvent: average total pressure
582: $\left<p\right>$, average total temperature of the system
583: $\left<T\right>$, average temperature of the polymer
584: $\left<T_{polymer}\right>$, and average temperature of the solvent
585: $\left<T_{solvent}\right>$. For the temperatures the results cannot be distinguished.} \label{Tab.2a}
586: \end{table}
587: The agreement with the reference values from table \ref{Tab.1a} is
588: very good. This is in line with the general static properties of
589: the polymers, which are given in \ref{Tab.2} and \ref{Tab.3}, and
590: compare very well to the data from table \ref{Tab.1}.
591: \begin{table}[ht!]
592: \centering
593: \begin{tabular}{||c|c|c|c||}
594: \hline\hline
595: $N$ & $10$ & $20$ & $30$ \\
596: $L$ & $25.0$ & $30.6$ & $34.2$ \\
597: \hline
598: $r_0$ & $7.0$ & $11.0$ & $12.0$ \\
599: $\left<\Delta r_{max}\right>$ & $4.2\pm 0.8$ & $6.6\pm 1.2$ & $7.7\pm 1.3$\\
600: $R_G=\left<R_G^2\right>^{1/2}$ & $2.7\pm 0.4$ & $4.0\pm 0.6$ & $4.6\pm 0.7$\\
601: $R_E=\left<R_E^2\right>^{1/2}$ & $6.7\pm 2.0$ & $10.4\pm 2.8$ & $10.8\pm 2.5$\\
602: $R_H=\left<R_H^{-1}\right>^{-1}$& $3.3\pm 0.3$ & $4.0\pm 0.3$ & $4.5\pm 0.4$\\
603: $\nu$ & $0.63$ & $0.58$ & $0.54$\\
604: $2/z$ & $0.56$ & $0.69$ & $0.62$\\
605: $\tau=R_G^2/(6D)$ & $122$ & $381$ & $882$\\
606: \hline\hline
607: \end{tabular}
608: \caption{Summary of some polymer (embedded in the hybrid \emph{ex-cg}
609: solvent) properties: number of polymer beads $N$, size of the
610: simulation box $L$, radius of the high resolution regime $r_0$, average maximal
611: distance of a monomer from the polymer's center of mass
612: $\left<\Delta r_{max}\right>$, radius of gyration $R_G$, end-to-end
613: distance $R_E$, hydrodynamic radius $R_H$, the static exponent
614: $\nu$, the exponent $2/z$, where $z$ is the dynamic exponent, and the
615: longest relaxation time $\tau$ calculated using data from table
616: \ref{Tab.4}. The error bar for the exponents $\nu$ and $2/z$ is roughly $10\%$.} \label{Tab.2}
617: \end{table}
618:
619: \begin{table}[ht!]
620: \centering
621: \begin{tabular}{||c|c|c|c||}
622: \hline\hline
623: $N$ & $10$ & $20$ & $30$ \\
624: $L$ & $25.0$ & $30.6$ & $34.2$ \\
625: \hline
626: $r_0$ & $7.0$ & $11.0$ & $12.0$ \\
627: $<\Delta r_{max}>$ & $4.0\pm 0.8$ & $5.9\pm 1.2$ & $8.6\pm 1.5$\\
628: $R_G=\left<R_G^2\right>^{1/2}$ & $2.7\pm 0.5$ & $3.9\pm 0.7$ & $5.2\pm 0.8$\\
629: $R_E=\left<R_E^2\right>^{1/2}$ & $6.6\pm 2.1$ & $9.4\pm 3.0$ & $13.3\pm 3.3$\\
630: $R_H=\left<R_H^{-1}\right>^{-1}$& $3.2\pm 0.3$ & $4.0\pm 0.4$ & $4.8\pm 0.4$\\
631: $\nu$ & $0.59$ & $0.55$ & $0.57$\\
632: $2/z$ & $0.69$ & $0.71$ & $0.77$\\
633: $\tau=R_G^2/(6D)$ & $152$ & $362$ & $1127$\\
634: \hline\hline
635: \end{tabular}
636: \caption{Same data as in table \ref{Tab.2}, but now for the hybrid \emph{ex-cg}$_{ic}$, where
637: interface pressure correction is applied.} \label{Tab.3}
638: \end{table}
639:
640: From the presented results we can conclude that AdResS faithfully
641: reproduces the reference statics obtained from the simulations
642: with a polymer embedded in the explicitely resolved solvent.
643:
644:
645: \subsection{Dynamics of the Polymer Chain and Solvent}
646:
647: While the conformational properties of the polymer in solution are
648: well understood and properly described by the adaptive resolution
649: approach, the situation for the dynamics is much less clear. By
650: changing the degrees of freedom not only the structure but also
651: the dynamical properties are altered, however in a way which is less
652: understood. It is also not a priori clear, whether an approach,
653: which produces a precise coarse graining for structural properties,
654: does this for dynamical properties as well. In a recent
655: study of small additive molecules to a polymer melt it was shown,
656: that while the length scaling is identical, the time scaling can
657: be different\cite{Harmandaris:2007}. In the present situation
658: the influence of the transition regime poses additional difficulties.
659:
660: In order to
661: determine the dynamical properties of the solvent and solute we
662: calculated the respective diffusion coefficients. The diffusion
663: coefficient of a species is computed from the center of mass
664: displacements using the Einstein relation
665: \begin{equation}
666: D=\frac{1}{6}\lim_{t\rightarrow\infty}\frac{\langle|{\bf
667: R}_i(t)-{\bf
668: R}_i(0)|^2\rangle}{t}=\frac{1}{6}\lim_{t\rightarrow\infty}\frac{\langle\Delta
669: R^2\rangle}{t},
670: \end{equation}
671: where ${\bf R}_i(t)$ is the center-of-mass position of the
672: molecule $i$ (which can be either a solvent or a solute molecule)
673: at time $t$ and averaging is performed over all choices of time
674: origin and, in the case of solvent, over all solvent molecules.
675:
676: Figure \ref{Fig.5} shows this for the solvent molecules' centers
677: of mass as a function of time for the different systems indicated.
678: All the curves in figure \ref{Fig.5}, except the one for the
679: coarse-grained solvent coincide. Thus the effect of the polymer on
680: the diffusivity of the solvent molecules is negligible. In other
681: words, the dilution is strong enough that the polymer effect on
682: the solvent dynamics is very small. The coarse grained solvent
683: molecules however move faster than the explicit ones. This is a
684: consequence of the reduced number of DOFs causing a time scale
685: difference in the dynamics of the coarse-grained
686: system\cite{Praprotnik:2005:4, Praprotnik:2006}. While this can be
687: very advantageous in some cases, one can also adjust $D$ by an
688: increased background friction in the DPD
689: thermostat\cite{Kremer:1990,Izvekov:2006}.
690: \begin{figure}[!ht]
691: \centering
692: \subfigure[]{\includegraphics[width=7.2cm]{fig5.eps}}
693: \caption{Log-log plot of the time dependence of the mean square
694: displacement of the solvent molecule's center of mass in the.
695: time interval [0,5000]: explicitely resolved
696: (bulk$_{ex}$) and coarse-grained (bulk$_{cg}$) solvents without solvated
697: polymer and the explicitely resolved solvent for the systems with three different
698: lengths of the solvated polymer
699: (N=10,20,30).
700: }\label{Fig.5}
701: \end{figure}
702: The diffusion coefficient $D$ of the solvent was obtained by
703: fitting a linear function to the curves depicted in figure
704: \ref{Fig.5} (a) and the obtained values are $D_{bulk_{ex}}=0.036$
705: and $D_{bulk_{cg}}=0.057$ for the explicit and coarse-grained
706: solvent, respectively. The question to ask here is, to what extend
707: does this have any influence on the dynamics of the embedded
708: polymer.
709:
710: Within the Zimm model\cite{Doi:1986} for polymer chain dynamics,
711: which is known to describe the scaling of the dynamics in dilute
712: solutions of polymers rather well and which takes into account the
713: hydrodynamic interactions, the polymer diffusion coefficient
714: scales as
715: \begin{equation}
716: D\propto N^{-\nu}\propto R_H^{-1}\propto R_G^{-1}.
717: \end{equation}
718: The longest relaxation time $\tau=R_G^2/(6D)$, i.e., the Zimm time
719: $\tau_Z \propto R_G^3 = R_G^z$, is the time the chain needs to
720: move its own size. $z=3$ is the dynamic exponent. Note that the
721: motion of inner monomers within the appropriate scaling regime
722: should be independent of $N$. For the mean square displacements of
723: the monomers a scaling analysis immediately yields for the mean
724: square displacement of a monomer $i$,
725: \begin{equation}\langle\Delta r^2\rangle=\langle({\bf r}_i(t)-{\bf
726: r}_i(0))^2\rangle \propto t^{2/z}=t^{2/3},
727: \end{equation}
728: for distances significantly larger than the bond length and
729: smaller than $\langle R^2 \rangle$, i.e., times smaller than
730: $\tau_Z$. For the center of mass of the chains a diffusive
731: behavior for the mean square displacement $\langle\Delta R^2 (t)
732: \rangle$ is always observed. Although the chains are relatively short, at
733: least for $N=30$ one expects a behavior relatively close to the
734: above mentioned idealized scheme\cite{Dunweg:1993}. Figure
735: \ref{Fig.6} shows $\langle\Delta r^2 (t) \rangle$ and
736: $\langle\Delta R^2 (t) \rangle$ for polymer chains with
737: $N=10,20,30$ embedded in the different solvent scenarios studied.
738: \begin{figure}[!ht]
739: \centering
740: \subfigure[]{\includegraphics[width=7.2cm]{fig6a.eps}}
741: \subfigure[]{\includegraphics[width=7.2cm]{fig6b.eps}}
742: \subfigure[]{\includegraphics[width=7.2cm]{fig6c.eps}}
743: \caption{Log-log plot of the time dependence of the mean square
744: displacement of a single monomer (considered are only monomers
745: near the chain's center of mass) for polymers and their centers of
746: mass with $N=10,20,30$ solvated in the \emph{ex} solvent (a), the
747: hybrid \emph{ex-cg} solvent (b) and the \emph{ex-cg}$_{ic}$
748: solvent (c) as indicated.
749: }\label{Fig.6}
750: \end{figure}
751:
752: In all cases the observed exponents for $\langle\Delta r^2 (t)
753: \rangle$ and $\langle\Delta R^2 (t) \rangle$ are close to $0.7\pm
754: 0.05$ and $1\pm 0.05$ respectively. Also the amplitudes of the
755: displacements of the inner beads are almost the same for all chain
756: lengths studied. This is in good agreement with earlier studies on
757: different generic polymer models in a explicit solvent as well as
758: studies of chains in a hybrid lattice Boltzmann
759: solvent\cite{Dunweg:1993,Ahlrichs:1999}. This is
760: to be expected since we preserve the hydrodynamic interactions by
761: employing the DPD thermostat in our simulations. This suggests
762: that our hybrid scheme is also applicable to study dynamic
763: properties of a polymer in a solution. Small deviations however
764: occur in the diffusion constant itself. The diffusion constants
765: for the polymers with $N=10,20,30$ in the hybrid \emph{ex-cg} and
766: \emph{ex-cg}$_{ic}$ solvents were obtained by fitting the straight
767: curve to the polymer's center of mass mean square displacement
768: presented in figure \ref{Fig.6}. The fit yields data listed in
769: table \ref{Tab.4}. The corresponding static and dynamic exponents
770: and the longest relaxation times are given in tables \ref{Tab.2}
771: and \ref{Tab.3}.
772:
773: \begin{table}[ht!]
774: \centering
775: \begin{tabular}{||c|c|c|c||}
776: \hline\hline
777: $N$ & $10$ & $20$ & $30$ \\
778: \hline
779: D(\emph{ex}) & $0.008$ & $0.005$ & $0.003$ \\
780: D(\emph{ex-cg}) & $0.009$ & $0.006$ & $0.0045$ \\
781: D(\emph{ex-cg}$_{ic}$) & $0.0085$ & $0.006$ & $0.0035$ \\
782: \hline\hline
783: \end{tabular}
784: \caption{Diffusion constant of the polymer chain embedded in three
785: different solvents: explicitely resolved \emph{ex}, hybrid
786: \emph{ex-cg}, and hybrid \emph{ex-cg}$_{ic}$. Though the
787: statistics of the data is rather poor we can estimate the error
788: bar roughly to $10 - 15 \%$. For comparison, the diffusion
789: coefficients of the explicit and coarse-grained solvents are
790: $D_{bulk_{ex}}=0.036$ and $D_{bulk_{cg}}=0.057$, respectively with
791: an effect of the polymers too small to determine here. Hence
792: $D_{polymer} \ll D_{solvent}$.} \label{Tab.4}
793: \end{table}
794:
795:
796: While the ratio of the diffusion constants for different chain
797: lengths roughly follow the expected scaling, even though it cannot
798: hold precisely due to the different box sizes, we here observe a
799: tendency to a weakly accelerated diffusion in the hybrid regime.
800: This is most evident for the hybrid \emph{ex-cg} case. Two different
801: aspects might play a role here. First the viscosity in the coarse
802: grained outer regime is smaller, which must have an effect on the
803: diffusion. Second, the small pressure and density fluctuations in
804: the transition regime might contribute to the effect as well.
805:
806: Although this is only a very first and incomplete test, it shows
807: that within the AdResS scheme essential aspects of the dynamical
808: properties of the embedded polymer chain are reasonably well
809: reproduced.
810:
811:
812: \section{Summary and Outlook}
813: In this paper we presented a hybrid multiscale MD simulation of a
814: generic macromolecule in a solvent using the recently proposed
815: AdResS method. The solvent surrounding the macromolecule is
816: represented with a sufficiently high level of detail so that the
817: specific interactions between the solvent and the solute are correctly
818: taken into account. The solvent farther away from the
819: macromolecule, where the high resolution is not needed, is
820: represented on a coarse-grained level. The high and low resolution
821: regimes freely exchange solvent molecules, which change
822: their resolution accordingly. To correctly simulate momentum
823: transport through the solvent, we use the DPD thermostat. The simulation results show
824: that AdResS accurately reproduces the thermodynamic and structural
825: properties of the system. The presented methodology is an
826: extension of AdResS to simulations of a solvation cavity, and
827: represents a first step towards the treatment of more realistic
828: systems such as biomacromolecules embedded in water. Work along
829: these lines is already under way\cite{Praprotnik:2006:2}.
830:
831:
832:
833: \section*{\small ACKNOWLEDGMENTS}
834: We thank Thomas Vettorel, Vagelis Harmandaris, Benedict Reynolds, and Burkhard D{\"
835: u}nweg for useful discussions. This work is supported in part by
836: the Volkswagen foundation. One of the authors (M.~P.) acknowledges
837: the financial support from the state budget by the Slovenian
838: research Agency under grant No. P1-0002.
839:
840: %============================================================================================================
841:
842: \begin{thebibliography}{10}
843:
844: \bibitem{Careri:1999}
845: G.~Careri,
846: \newblock in {\em Hydration Processes in Biology: Theoretical and Experimental
847: Approaches}, edited by M.~C. Bellisient-Funel, pages 143--155, IOS,
848: Amsterdam, 1999.
849:
850: \bibitem{Das:2005}
851: P.~Das, S.~Matysiak, and C.~Clementi,
852: \newblock Proc. Natl. Acad. Sci. U. S. A. {\bf 102}, 10141 (2005).
853:
854: \bibitem{Villa:2005}
855: E.~Villa, A.~Balaeff, and K.~Schulten,
856: \newblock Proc. Natl. Acad. Sci. U.S.A. {\bf 102}, 6783 (2005).
857:
858: \bibitem{Rafii:1998}
859: H.~Rafii-Tabar, L.~Hua, and M.~Cross,
860: \newblock J. Phys.: Condens. Matter {\bf 10}, 2375 (1998).
861:
862: \bibitem{Broughton:1999}
863: J.~Q. Broughton, F.~F. Abraham, N.~Bernstein, and E.~Kaxiras,
864: \newblock Phys. Rev. B {\bf 60}, 2391 (1999).
865:
866: \bibitem{Ahlrichs:1999}
867: P.~Ahlrichs and B.~D{\" u}nweg,
868: \newblock J. Chem. Phys. {\bf 111}, 8225 (1999).
869:
870: \bibitem{Malevanets:2000}
871: A.~Malevanets and R.~Kapral,
872: \newblock J. Chem. Phys. {\bf 112}, 7260 (2000).
873:
874: \bibitem{Csanyi:2004}
875: G.~Csanyi, T.~Albaret, M.~C. Payne, and A.~DeVita,
876: \newblock Phys. Rev. Lett. {\bf 93}, 175503 (2004).
877:
878: \bibitem{Abrams:2005}
879: C.~F. Abrams,
880: \newblock J. Chem. Phys. {\bf 123}, 234101 (2005).
881:
882: \bibitem{Neri:2005}
883: M.~Neri, C.~Anselmi, M.~Cascella, A.~Maritan, and P.~Carloni,
884: \newblock Phys. Rev. Lett. {\bf 95}, 218102 (2005).
885:
886: \bibitem{Fabritiis:2006}
887: G.~D. Fabritiis, R.~Delgado-Buscalioni, and P.~V. Coveney,
888: \newblock Phys. Rev. Lett. {\bf 97}, 134501 (2006).
889:
890: \bibitem{Praprotnik:2005:4}
891: M.~Praprotnik, L.~Delle~Site, and K.~Kremer,
892: \newblock J. Chem. Phys. {\bf 123}, 224106 (2005).
893:
894: \bibitem{Praprotnik:2006}
895: M.~Praprotnik, L.~Delle~Site, and K.~Kremer,
896: \newblock Phys. Rev. E {\bf 73}, 066701 (2006).
897:
898: \bibitem{Praprotnik:2006:1}
899: M.~Praprotnik, K.~Kremer, and L.~Delle~Site,
900: \newblock Phys. Rev. E {\bf 75}, 017701 (2007).
901:
902: \bibitem{Praprotnik:2006:2}
903: M.~Praprotnik, L.~{Delle Site}, K.~Kremer, S.~Matysiak, and C.~Clementi,
904: \newblock cond-mat/0611544 (2006).
905:
906: \bibitem{Kremer:1990}
907: K.~Kremer and G.~S. Grest,
908: \newblock J. Chem. Phys. {\bf 92}, 5057 (1990).
909:
910: \bibitem{Allen:1987}
911: M.~P. Allen and D.~J. Tildesley,
912: \newblock {\em Computer Simulation of Liquids},
913: \newblock Clarendon Press, Oxford, 1987.
914:
915: \bibitem{Praprotnik:2005}
916: D.~Jane\v{z}i\v{c}, M.~Praprotnik, and F.~Merzel,
917: \newblock J. Chem. Phys. {\bf 122}, 174101 (2005).
918:
919: \bibitem{Praprotnik:2005:1}
920: M.~Praprotnik and D.~Jane\v{z}i\v{c},
921: \newblock J. Chem. Phys. {\bf 122}, 174102 (2005).
922:
923: \bibitem{Praprotnik:2005:2}
924: M.~Praprotnik and D.~Jane\v{z}i\v{c},
925: \newblock J. Chem. Phys. {\bf 122}, 174103 (2005).
926:
927: \bibitem{Soddemann:2003}
928: T.~Soddemann, B.~D{\" u}nweg, and K.~Kremer,
929: \newblock Phys. Rev. E {\bf 68}, 046702 (2003).
930:
931: \bibitem{Espresso:2005}
932: http://www.espresso.mpg.de.
933:
934: \bibitem{Praprotnik:2006:3}
935: Note that the definition of $R_H$ given in Eq. (\ref{eq.rh}) is only correct
936: for an infinite simulation box. For a finite box we have to also consider the
937: hydrodynamic interactions with the periodic images. The value of in such a
938: way corrected $R_H$ deviates from the value obtained using the formula
939: (\ref{eq.rh})\cite{Dunweg:1993,Ahlrichs:1999}. However, since our aim in the
940: present work is only to compare the properties of the polymer chain from the
941: hybrid simulation with the corresponding ones from the all-atom simulation we
942: here do not take the finite size correction into account and use the formula
943: (\ref{eq.rh}) for the definition of $R_H$.
944:
945: \bibitem{Gennes:1979}
946: P.-G. de~Gennes,
947: \newblock {\em Scaling Concept in Polymer Physics},
948: \newblock Cornell University, Ithaca, 1979.
949:
950: \bibitem{Doi:1986}
951: M.~Doi and S.~F. Edwards,
952: \newblock {\em The Theory of Polymer Dynamics},
953: \newblock Clarendon, Oxford, 1986.
954:
955: \bibitem{Sokal:1995}
956: A.~D. Sokal,
957: \newblock in {\em Monte Carlo and Molecular Dynamics Simulation in Polymer
958: Science}, edited by K.~Binder, chapter~2, Clarendon, Oxford, 1995.
959:
960: \bibitem{Dunweg:1993}
961: B.~D{\" u}nweg and K.~Kremer,
962: \newblock J. Chem. Phys. {\bf 99}, 6983 (1993).
963:
964: \bibitem{Harmandaris:2007}
965: V.~A. Harmandaris, N.~P. Adhikari, N.~F.~A. van der Vegt, K. Kremer,
966: R. Voelkel, H. Weiss, and CheeChin Liew,
967: \newblock Macromolecules,
968: \newblock submitted.
969:
970: \bibitem{Izvekov:2006}
971: S.~Izvekov and G.~A. Voth,
972: \newblock J. Chem. Phys. {\bf 125}, 151101 (2006).
973:
974: \end{thebibliography}
975:
976: %====================================================================
977: %\bibliographystyle{physics}
978:
979: %\bibliography{bibliography}
980: %===================================================================
981:
982:
983:
984: \end{document}
985: %
986: % ****** End of file man.tex *******
987: