1: \documentclass[pre,twocolumn,showpacs,preprintnumbers,superscriptaddress,floatfix]{revtex4}

2: \usepackage{graphicx,citesort,psfrag,amsmath,amssymb}

3: \usepackage[dvips]{color}

4: 

5: \renewcommand{\vec}[1]{\pmb{#1}}

6: \newcommand{\avg}[1]{\left\langle #1\right\rangle}

7: \newcommand{\spd}[2]{\left( #1|#2\right) }

8: \newcommand{\abs}[1]{\left\lvert #1\right\rvert}

9: \newcommand{\mint}[4]{\int_{#2}^{#3}\!\!#1\,#4}

10: \newcommand{\rem}[1]{\textbf{#1}}

11: \newcommand{\kil}[1]{{\scriptsize #1}}

12: \newcommand{\Ord}[1]{{\cal O}(#1)}

13: \newcommand{\lOrd}[1]{{\cal o}(#1)}

14: \newcommand{\ord}[1]{{\text{ord}}(#1)}

15: \newcommand{\pe}{\perp}

16: \newcommand{\pa}{\parallel}

17: \newcommand{\rpe}{r_\perp}

18: \newcommand{\rpa}{r_\parallel}

19: \newcommand{\dpe}{\avg{\delta\rpe^2}}

20: \newcommand{\dpa}{\avg{\delta R_\pa^2}}

21: \newcommand{\epe}{\hat{\vec e}_\pe}

22: \newcommand{\epa}{\hat{\vec e}_\pa}

23: \newcommand{\cpe}{\chi_\pe}

24: \newcommand{\tv}{\vec {\mathfrak t}}

25: \newcommand{\nv}{\vec {\mathfrak n}}

26: \newcommand{\bv}{\vec {\mathfrak b}}

27: 

28: \newcommand{\kr}{\chi}

29: \newcommand{\ts}{\tau}

30: 

31: \newcommand{\fopt}{f^{\text{OPT}}}

32: 

33: \newcommand{\fe}{\mathfrak f \,}

34: \newcommand{\efe}{\hat{\vec e}_\fe}

35: \newcommand{\fel}{\vec{f}_{\text{el}}\,}

36: \newcommand{\gex}{\mathfrak g \,}

37: \newcommand{\gext}{\mathfrak g_0}

38: \newcommand{\gfl}{\vec{ g}_{\text{fr}}}

39: \newcommand{\gel}{\vec{ g}_{\text{el}}}

40: 

41: \newcommand{\Hext}{{\cal H}_{\text{ext}}}

42: \newcommand{\Hb}{{\cal H}_{\text{WLC}}}

43: \newcommand{\Ht}{{\cal H}_{\text{tors}}}

44: \newcommand{\Hc}{{\cal H}_{\text{comp}}}

45: \newcommand{\Htot}{{\cal H}}

46: 

47: \newcommand{\msd}{\text{MSD}}

48: \newcommand{\msdl}{\text{msd}}

49: 

50: \newcommand{\fc}{c}

51: \newcommand{\ludo}{\ell_\text{SWN}}

52: \newcommand{\lralf}{\ell_\text{EJAM}}

53: \newcommand{\lpierre}{\ell_\text{BBG}}

54: \newcommand{\lflower}{\ell_\text{Flower}}

55: \newcommand{\lp}{\ell_p}

56: \newcommand{\lpe}{\ell_\perp}

57: \newcommand{\lpa}{\ell_\|}

58: \newcommand{\tfin}{t_L^\|}

59: \newcommand{\tf}{t_\fe}

60: \renewcommand{\sf}{s_\fe}

61: \newcommand{\tot}{\tau_{\text{tot}}}

62: \newcommand{\totpe}{t_L^\pe}

63: \newcommand{\totpa}{t_L^\pa}

64: 

65: \newcommand{\eom}{equations of motion}

66: \newcommand{\leom}{longitudinal equation of motion}

67: \newcommand{\teom}{transverse equation of motion}

68: \newcommand{\rhs}{right hand side}

69: 

70: \newcommand{\crr}[1]{\color{red}{#1 }\color{black}}

71: \newcommand{\crb}[1]{\color{blue}{#1 }\color{black}}

72: 

73: \newcounter{fnt}

74: 

75: \newcommand{\pull}{\text{\emph{Pulling}}}

76: \newcommand{\epull}{\text{\emph{Towing}}}

77: \newcommand{\lelo}{\text{\emph{Release}}}

78: \newcommand{\tjump}{\text{$\lp$--\emph{Quench}}}

79: \newcommand{\push}{\text{\emph{Pushing}}}

80: 

81: 

82: \DeclareMathOperator{\sgn}{sgn}

83: \DeclareMathOperator{\erf}{erf}

84:  

85: 

86: \sloppy

87: 

88: \preprint{LMU-ASC 60/06}

89: 

90: \begin{document}

91: \bibliographystyle{apsrev}

92: 

93: 

94: \title{Tension dynamics in semiflexible polymers. \\Part~II: Scaling

95:   solutions and applications }

96: 

97: 

98: \author{Oskar Hallatschek} \email{ohallats@fas.harvard.edu}

99:  \affiliation{Lyman Laboratory of Physics, %

100:    Harvard University, Cambridge, Massachusetts 02138, USA}

101:  

102:  \author{Erwin Frey} \affiliation{Arnold Sommerfeld Center for Theoretical Physics

103:    and Center for NanoScience, %

104:    LMU M\"unchen, Theresienstr.~37, 800333 M\"unchen,

105:    Germany}

106: 

107: \author{Klaus Kroy} 

108: \affiliation{Institut f\"ur Theoretische Physik,%

109:   Universit\"at Leipzig, Augustusplatz 10/11, 04109 Leipzig, Germany}

110:      

111:      

112: 

113: \date{\today}

114: 

115: \begin{abstract}

116:   In Part~I of this contribution, a systematic coarse-grained

117:   description of the dynamics of a weakly-bending semiflexible polymer

118:   was developed. Here, we discuss analytical solutions of the

119:   established deterministic partial integro-differential equation for

120:   the spatio-temporal relaxation of the backbone tension. For

121:   prototypal experimental situations, such as the sudden application

122:   or release of a strong external pulling force, it is demonstrated

123:   that the tensile dynamics reflects the self-affine conformational

124:   fluctuation spectrum in a variety of intermediate asymptotic power

125:   laws.  Detailed and explicit analytical predictions for the tension

126:   propagation and relaxation and corresponding results for common

127:   observables, such as the end-to-end distance, are obtained.

128: \end{abstract}

129: 

130: 

131: 

132: \pacs{  87.15.He, 87.15.Aa,  87.16.Ka, 83.10.-y}

133: 

134: 

135: 

136: \maketitle

137: 

138: 

139: 

140: \section{Introduction}

141: \label{sec:intro}

142: Polymer physics has traditionally focused on very flexible polymers that admit a

143: highly coarse-grained description in terms of Gaussian chains and exhibit universal

144: physical behavior that can be explained using methods from statistical mechanics such

145: as the renormalization group and scaling arguments~\cite{degennes:79}. Further

146: research has explored new terrain that lies beyond the realm of applicability of this

147: highly successful approach either because the polymers of interest are too stiff or

148: because they are subject to extreme forces. Many of these instances have recently

149: appeared in applications involving biopolymers.  Notorious examples are the nonlinear

150: mechanical response of DNA~\cite{BustamanteBS03}, which has turned out to be pivotal

151: to protein-DNA interactions, and the problem of force transduction through the

152: cytoskeleton~\cite{JanmeyW04,YasuhiroSawada02182002,GaborForgacs06012004}, which is a

153: major mechanism by which cells explore their environment and react to external

154: mechanical stimuli.  Clearly, in neither of these situations can theorists contend

155: themselves with the convenient Gaussian chain representation, but have to resort to

156: more realistic, yet still schematic descriptions, such as the freely-jointed chain

157: (e.g.\ for single-stranded DNA) or the wormlike chain model (for double-stranded DNA,

158: F--actin, microtubules etc.)~\cite{doi-edwards:86,degennes:79}.

159: 

160: Suspicions that this might entail a substantial loss of universality and render

161: systematic analytical approaches forbiddingly complex have turned out to be

162: unfounded. The wormlike chain model provides an analytically tractable standard model

163: for many of the above mentioned new applications, in particular for calculating the

164: non-equilibrium dynamical response of stiff and semiflexible but weakly bending

165: polymers to strong external fields.  As established in Part~I of this contribution,

166: the weakly bending wormlike chain lends itself to a multiple-scale perturbation

167: theory (MSPT) based on a length scale separation between longitudinal and transverse

168: dynamic correlation lengths. In the present Part~II, we demonstrate that the

169: self-affine roughness, acquired by the weakly-bending contour in thermal equilibrium,

170: plays an analogous role as the more familiar fractal conformational correlations in

171: the case of flexible polymers~\cite{degennes:79}.  The self-similarity of the static

172: conformational fluctuations entails self-similar dynamics. It manifests itself in a

173: variety of \emph{intermediate asymptotic dynamic power laws}.  Apart from the

174: restriction to polymers with a (locally) rodlike structure, these predictions are as

175: universal as those of classical polymer physics. They are moreover derived in a

176: direct way, usually including exact amplitudes, from a controlled perturbation

177: expansion.

178: 

179: 

180: 

181: As a major result of the multiple scale theory developed in Part~I, we

182: obtained a coarse-grained reformulation of the free-draining Langevin

183: equations of motion of a weakly bending rod in the form of the

184: deterministic equation

185: \begin{equation}

186:   \label{eq:cg-eom}

187:   \partial_s^2 F(s,t)=-\zeta_\pa \avg{\Delta

188:     \overline{\varrho}}\left[F(s,\tilde t \leq t),t\right] \;.

189: \end{equation}

190: It describes the long wave-length (all time) dynamics of the

191: time-integrated tension 

192: \begin{equation}

193:   \label{eq:integrated-tension}

194:   F(s,t)\equiv\mint{dt'}{0}{t}f(s,t')

195: \end{equation} 

196: with $\zeta_{\parallel}$ being the friction coefficient for longitudinal motion and

197: $\avg{\Delta \overline{\varrho}}\left[F(s,\tilde t \leq t),t\right]$ the average

198: release of contour length stored in the transverse undulations up to time $t$.

199: Written out in terms of the transverse normal mode contributions, the latter reads

200: \begin{eqnarray}

201:     \avg{\Delta\overline{\varrho}}(t)&\equiv& \mint{\frac{dq}{\pi\lp}}{0}{\infty}

202:     \left\{\frac{1}{q^2+f_<}\left(e^{-2q^2[\kappa q^2 t+ F(t)]/\zeta_\perp}-1\right) \right.\nonumber\\

203:     &&\left.+2 q^2\mint{d\tilde t}{0}{t}e^{-2q^2\left[\kappa q^2(t-\tilde

204:           t)+F(t)-F(\tilde t)\right]/\zeta_\perp}\right\},

205:     \label{eq:change-stored-length}

206: \end{eqnarray}

207: where $\kappa$, $\lp=\kappa/(k_B T)$ and $\zeta_\perp$ are the bending stiffness,

208: persistence length and friction coefficient for transverse motion, respectively. The

209: parameter $f_<\equiv f(t<0)=$const.~allows to take into account a constant

210: prestress~\footnote{The parameter $\theta$ occurring in the dynamic force extension

211:   relation of Part~I will be set to one, $\theta=1$, throughout this paper. It

212:   describes the effect of sudden changes in persistence length, which will be

213:   discussed elsewhere~\cite{obermayer-kroy-frey-hallatschek3:tbp}.}. Upon inserting

214: this dynamical force-extension relation into Eq.~(\ref{eq:cg-eom}), we arrived at our

215: central result, the closed partial integro-differential equation (PIDE) for the

216: time-integrated tension $F(s,t)$,

217: \begin{eqnarray}

218:   \label{eq:pide}

219:     &\partial_s^2 F(s,t)= \hat \zeta

220:     \mint{\frac{dq}{\pi\lp}}{0}{\infty}

221:     \left\{\frac{1}{q^2+f_<}\left(1-e^{-2q^2[ q^2 t+ F(s,t)]}\right)

222:     \right.& \nonumber \\

223:     &\left. -2  q^2\mint{d\tilde

224:         t}{0}{t}e^{-2q^2\left[ q^2(t-\tilde t)+F(s,t)-F(s,\tilde

225:           t)\right]}\right\}\;.&

226: \end{eqnarray}

227: For convenience, we have made the following choice of units: Time and

228: tension, respectively, are rescaled according to

229: \begin{eqnarray}

230:   \label{eq:rescaling-time-tension}

231:   t&\to&\zeta_\perp t/\kappa \;, \\

232:   f&\to& \kappa f\;.  

233: \end{eqnarray}

234: This corresponds to setting $\kappa\equiv\zeta_\pe\equiv 1$ and $\hat

235: \zeta\equiv 1/2=\zeta_\pa$. As a consequence all variables represent

236: powers of length, e.g., $t$ and $f$ are a length$^4$ and a

237: length$^{-2}$, respectively. 

238: 

239: To leading order in the small contour undulations, the deterministic coarse-grained

240: tension dynamics, described by Eq.~(\ref{eq:pide}), together with the microscopic

241: transverse equation of motion, represents a valid reformulation of the constrained

242: Langevin dynamics of a weakly bending rod subject to a putative pre-stress. The PIDE

243: Eq.~(\ref{eq:pide}) is the basis not only for discussing the tension dynamics itself,

244: but also the starting point for analytical and numerical calculations of the

245: longitudinal and transverse nonlinear response of a weakly bending polymer. It is the

246: purpose of the present Part~II to treat the former case in detail, while the latter,

247: somewhat more complex case is reserved for a future

248: communication~\cite{obermayer-hallatschek2:tbp}. In Sec.~\ref{sec:driving}, we derive

249: detailed solutions to Eq.~(\ref{eq:pide}) for idealized experimental protocols

250: involving a representative selection of external fields.  The analytical scaling

251: solutions obtained for a semi-infinite polymer suddenly pulled (or released) at its

252: end, reveal the non-trivial short-time phenomenon of tension propagation

253: (Sec.~\ref{sec:tp}).  At long times, the finite contour length comes in as an

254: additional characteristic length scale, which gives rise to additional scaling

255: regimes, discussed in Sec.~\ref{term-rel-tensprop}. To make contact with experiments,

256: we finally identify the repercussions of the tension dynamics on pertinent

257: observables like the (projected) end-to-end distance (Sec.~\ref{sec:MSDII}) and

258: comment on novel experimental perspectives brought up by our analysis

259: (Sec.~\ref{sec:possible-exps}).

260: 

261: 

262: 

263: \section{Generic longitudinal driving forces}

264: \label{sec:driving}

265: In general, the tension dynamics depends on how the filament is driven externally,

266: i.e., on the boundary and initial conditions imposed on Eq.~(\ref{eq:pide}).  With

267: the definition of generic experimental force protocols, this section shall provide a

268: framework for the analysis of Eq.~(\ref{eq:pide}). We introduce the scenarios

269: {\pull}, {\epull}, and {\lelo} and report on existing investigations. Apart from

270: being directly relevant for experiments, we have chosen to consider these scenarios

271: because of two properties that render them attractive from a theoretical perspective.

272: Firstly, they correspond to \emph{sudden} changes of the environment and thus do not

273: introduce an additional delay time scale.  Secondly, since external forces are

274: assumed to act at the ends, these scenarios only change boundary conditions and leave

275: the equations of motion unchanged.  In more complicated scenarios, that involve

276: forces applied not only at the ends, these problems show up as subproblems.  For

277: instance, if a single point force is applied somewhere within the bulk of a polymer,

278: the filament can be partitioned into two sections that perceive the external force

279: only at their ends.

280: 

281: 

282: 

283: \subsubsection{{\pull}}

284: \label{sec:pull-def}

285: The polymer is supposed to be free for negative times, such that it is equilibrated

286: under zero tension at time zero, i.e., we require

287: \begin{equation}

288:   \label{eq:pulling-ic}

289:   f_<=f(s,t<0)=0 \;.

290: \end{equation}

291: Then, for positive times the polymer is pulled in longitudinal direction at both ends

292: with a constant force $\fe$~\footnote{Throughout, we denote \emph{external} forces or

293:   force fields by fraktur letters.}. The corresponding external force density

294: $\fe\left[\delta(s-L)-\delta(s)\right]$ provides the boundary conditions for the

295: tension (see, e.g., the longitudinal equation of motion in Part~I)

296: \begin{equation}

297:   \label{eq:pulling-bc}

298:   f(s=0,t>0)=\fe \;,\;  f(L,t>0)=\fe \;.

299: \end{equation}

300: 

301: 

302: 

303: {\pull} was first considered by Seifert et al.~\cite{seifert-wintz-nelson:96} (SWN).

304: They predict that a ``large'' tension spreads within a time $t$ a characteristic

305: length $\lpa(t)=\ludo(t)\equiv \lp^{1/2} (\fe t)^{1/4}$ from the ends into the bulk

306: of the filament. Their analysis neglects bending forces and thermal forces for the

307: dynamics, albeit the self-affine thermal initial conditions are used

308: (\emph{taut-string approximation}).  The contribution of Everaers et

309: al.~\cite{everaers-Maggs:99} (EJAM) shed light on the linear response to longitudinal

310: forces. Their simulations established a typical propagation length of

311: $\lpa(t)=\lralf(t)\equiv \lp^{1/2} t^{1/8}$ for weak forces, which was previously

312: predicted by Morse~\cite{morse:98II}, and made it plausible by scaling arguments.

313: Brochard--Wyart et al.~\cite{brochard-buguin-de_gennes:99} (BBG) proposed a theory

314: for tension propagation claimed to be valid on scales much larger than $\lp$

315: supposing, however, the weakly bending approximation.  A \emph{quasi-static

316:   approximation} underlies their analysis, in which the polymer is at any instant of

317: time considered to be equilibrated with the local tension. Applying their results to

318: the situation considered here, tension should propagate a distance

319: $\lpa(t)\propto\lpierre(t)\equiv \lp^{1/2} \fe^{3/4} t^{1/2}$.

320: 

321: Naturally, the scaling arguments used to predict the three different scaling regimes

322: did not address the crossover and the range of validity. Below, we show that in fact

323: only two scaling regimes exist.

324: 

325: \subsubsection{{\epull}}

326: \label{sec:epull-def}

327: 

328: Pulling of a filament can also be studied for time dependent external forces. A

329: dynamic force protocol of particular experimental relevance is given by the constant

330: velocity ensemble: The polymer is pulled by a time-dependent external force $\fe(t)$

331: at the left end such that this end moves with a \emph{constant velocity} $v$

332: ({\epull}).  Besides the growth law of the boundary layer, we wish to understand the

333: time dependence of the external force.  We will see that both quantities are

334: proportional to each other because the external force essentially has to drag a

335: polymer section of length $\lpa(t)$ through the viscous solvent with the constant

336: velocity $v$, hence $\fe(t)\simeq \hat \zeta v \lpa(t)$.  By measuring the

337: time-dependent force in a constant velocity experiment one can thus directly monitor

338: $\lpa(t)$. Possible experimental realizations are outlined in

339: Sec.~\ref{sec:possible-exps}.

340: 

341: The external force density field corresponding to {\epull} is given by $-\fe(t)

342: \delta(s)$ with an external force $\fe(t)$ determined to fulfill the requirement that

343: $\partial_t r_\pa(0,t)=v$. Recall from the equations of motion derived in Part~I,

344: that, up to terms of order $\Ord{\epsilon}$, the gradient of the tension is given by

345: the longitudinal friction (as in a rigid rod).  This implies the boundary condition

346: \begin{eqnarray}

347:   \label{eq:efpulling-bc}

348:   \partial_s f(s=0,t>0)&=&-\hat\zeta v+\Ord{\epsilon} \\

349:   f(L,t>0)&=&0\;. \nonumber

350: \end{eqnarray}

351: 

352: 

353: \subsubsection{{\lelo}}

354: \label{sec:lelo-def}

355: 

356: {\lelo} refers to the process ``inverse'' to {\pull}: the filament is supposed to be

357: equilibrated at $t=0$ under a constant pulling force,

358: \begin{equation}

359:   \label{eq:let-loose-ic}

360:   f_<=f(s,t<0)=\fe>0 \;.

361: \end{equation}

362: Then, at $t=0$, the external force is suddenly switched off and the filament begins to

363: relax. The ends are considered to be free for $t>0$,

364: \begin{equation}

365:   \label{eq:let-loose-bc}

366:   f(s=0,t>0)=0 \;,\; f(L,t>0)=0 \;.

367: \end{equation}

368: {\lelo} has been discussed by Brochard et al.~\cite{brochard-buguin-de_gennes:99}. According to that work

369: the characteristic size of the boundary layer, where the tension is appreciably

370: decreased from $\fe$, should be given by $\lpa(t)\propto\lpierre(t)$ (the same as for

371: {\pull}). 

372: 

373: Furthermore, Brochard et al.~predict that the tension is relaxed as soon as the

374: tension has spread over the whole filament yielding a relaxation time $\totpa$ for

375: the tension that satisfies $\lpa(\totpa)=L$. This is in conflict with what we will

376: find Sec.~\ref{term-rel-tensprop}, where we identify a novel scaling regime of

377: homogeneous tension relaxation.

378: 

379: \section{Tension propagation}

380: \label{sec:tp}

381: To unravel the physical implications of Eq.~(\ref{eq:pide}) for the scenarios

382: introduced above, we begin with the tension propagation regime $\lpa\ll L$, where the

383: total length $L$ of the polymer is irrelevant.  In this regime, it is legitimate to

384: discuss the dynamics on a (formally) semi-infinite arc length interval $[ 0,

385: \infty[$.  Problems like {\pull} and {\lelo} still depend on four independent length

386: scales ($\lp, \fe^{-1/2},s , t^{1/4}$).  Yet, it is shown in

387: Sec.~\ref{sec:scaling-forms} that, Eq.~(\ref{eq:pide}) is \emph{solved exactly} by a

388: tension profile that obeys a crossover scaling form depending on only two arguments,

389: which can be identified as a reduced time and arc length variable, respectively.  In

390: Sec.~\ref{sec:asymptotic-growth}, we then argue that for asymptotically short ($<$)

391: and long ($>$) times the scaling function reduces to a function of only one scaling

392: variable $\xi^{\gtrless}=s/\lpa^{\gtrless}(t)$. Our major results concerning tension

393: propagation, particularly our classification of tension propagation laws

394: $\lpa^{\gtrless}(t)$, are summarized in Sec.~\ref{sec:tensprop-summary}.

395: 

396: 

397: 

398: \subsection{Scaling forms}

399: \label{sec:scaling-forms}

400: For each of the generic problems introduced in Sec.~\ref{sec:driving}, we shall see

401: that the tension profile obeys certain crossover scaling forms, that cannot be

402: inferred from dimensional analysis. These scaling forms greatly simplify the further

403: analysis of the tension dynamics by reducing the number of independent parameters.

404: 

405: To solve the equation of motion, Eq.~(\ref{eq:pide}), for a given force protocol, we

406: proceed in the following way.  A scaling ansatz is postulated and shown to eliminate

407: the parameter dependence in Eq.~(\ref{eq:pide}) and the boundary conditions after a

408: suitable choice of length, time and force scales.  These crossover scales turn out to

409: separate two different regimes, a short- and long-time regime,

410: respectively~\footnote{Such a crossover was also found in Part~I for the stored

411:   length under a spatially constant tension from ordinary perturbation theory.}.

412: 

413: Although being ultimately interested in the tension profile $f(s,t)$, it is

414: convenient to first discuss the \emph{time integrated} tension $F(s,t)$, defined in

415: Eq.~(\ref{eq:integrated-tension}), because the equation of motion for the tension,

416: Eq.~(\ref{eq:pide}), is naturally formulated in terms of $F(s,t)$.  The physically

417: more intuitive quantity $f(s,t) =\partial_t F(s,t)$ is extracted afterwards by a

418: differentiation with respect to time.

419:  

420: We make the following ansatz for the time integral $F(s,t)$ of the tension

421: \begin{equation}

422:   \label{eq:sansatz}

423:   F(s,t)=\fe \tf \; \phi\left(\frac s {\sf } , \frac t {\tf} \right) \;,

424: \end{equation}

425: in terms of as yet unknown crossover time and length scales $\tf$ and $\sf$ to be

426: determined below. While a force scale $\fe$ is given explicitely in the case of

427: {\pull} and {\lelo} by the pulling/pre-stretching force, a natural force scale

428: \begin{equation}

429:   \label{eq:ext-force-towing}

430:   \fe\equiv \hat \zeta v

431:   \sf \qquad \text{(\epull)}\:

432: \end{equation}

433: for {\epull} is provided not unless $\sf$ is fixed. The combination of variables in

434: Eq.~(\ref{eq:ext-force-towing}) represents the force necessary to drag a polymer

435: section of length $\sf$ (longitudinally) through the fluid with the imposed towing

436: velocity $v$.

437: 

438: As long as the polymer is in equilibrium, $t<0$, the dimensionless scaling function

439: $\phi(\sigma,\tau)$ for the integrated tension is zero for both pulling scenarios,

440: but linearly increasing with time for {\lelo} due to the constant pre-stretching

441: force,

442: \begin{equation}

443:   \label{eq:ic-sansatz}

444:   \phi(\sigma,\tau<0)\equiv \fc \tau=\left\{{0 \;, \atop \tau \;,}

445:       \qquad {{\pull}/ {\epull}  \atop \lelo }

446:     \right.

447: \end{equation}

448: The constant $c$ entering the initial condition, Eq.~(\ref{eq:ic-sansatz}), is

449: given by $c= 0$ for {\pull}/{\epull} and $c= 1$ for {\lelo}, respectively.

450: 

451: Since we expect that the signal of a sudden change at the end of the polymer,

452: i.e.~at $\sigma=0$, takes time to propagate into the bulk of the polymer, which

453: corresponds to $\sigma\to\infty$, we look for solutions that have a time-independent

454: stored length at $\sigma\to\infty$.  According to Eq.~(\ref{eq:cg-eom}), this

455: corresponds to the boundary condition of a vanishing curvature of the tension profile

456: at infinity,

457: \begin{equation}

458:   \label{eq:bc-infty}

459:   \partial_\sigma^2\phi(\sigma\to\infty,\tau>0) =0

460:   \;.

461: \end{equation}

462: At the origin, the force, respectively, the gradient of the force are prescribed by

463: the considered experimental setup,

464: \begin{subequations}

465:   \label{eq:bc-origin}

466:   \begin{eqnarray}

467:     \label{eq:bc-origin-1}

468:     \phi(\sigma=0,\tau>0)&=&\left\{{\tau \;, \atop 0 \;,}

469:       \qquad {{\pull}  \atop \lelo } \right. \\

470:     \partial_\sigma\phi(\sigma=0,\tau>0) &=&-\tau \;, \qquad {\epull}

471:     \label{eq:bc-origin-2} 

472:   \end{eqnarray}

473: \end{subequations}

474: Inserting the scaling ansatz, Eq.~(\ref{eq:sansatz}), into Eq.~(\ref{eq:pide}) yields

475: after the variable substitutions $q\to q\sqrt{\fe}$, $t\to \tau \tf$ and $\tilde t

476: \to \tilde \tau \tf$

477: \begin{eqnarray}

478:   \label{eq:pide-inserted-1}

479:   &&\lp \frac{\fe^{3/2} \tf}{\hat \zeta \sf ^2} \partial_\sigma^2

480:   \phi(\sigma,\tau) =\\

481:   &&\mint{\frac{dq}{2\pi}}{-\infty}{\infty}

482:   \left\{\frac{1}{q^2+\fc}\left[1-e^{-2q^2\left[q^2\tau+

483:           \phi(\sigma,\tau)\right] \tf \fe^2}\right] \right. \nonumber\\

484:   &&\left. -2 q^2 \tf \fe^2\mint{d\tilde

485:       \tau}{0}{\tau}e^{-2 q^2 \left[ q^2 (\tau-\tilde

486:         \tau)+\phi(\sigma,\tau)-\tilde \tau \phi(\sigma, \tilde

487:         \tau)\right] \tf \fe^2} \right\}\;.\qquad \nonumber

488: \end{eqnarray}

489: By fixing the scales $\tf$ and $\sf $ appropriately,

490: \begin{subequations} \label{eq:scales1}

491:   \begin{eqnarray}

492:     \tf&=&\fe^{-2} \label{eq:scales1-time} \\

493:     \sf &=&\hat \zeta^{-1/2} \lp^{1/2} \fe^{-1/4}

494:     \label{eq:scales1-length} \;,

495:     \end{eqnarray}

496: \end{subequations}

497: we can eliminate the parameter dependence of

498: Eq.~(\ref{eq:pide-inserted-1}),

499: \begin{eqnarray}

500:   \label{eq:pide-nondimensional-1}

501:   &&\partial_\sigma^2 \phi(\sigma,\tau)

502:     =\mint{\frac{dq}{2\pi}}{-\infty}{\infty} \left\{\frac{1}{q^2+\fc

503:         }\left[1-e^{-2q^2\left[q^2 \tau+

504:             \phi(\sigma,\tau)\right]}\right] \right.\nonumber\\

505:     &&\qquad \left. -2 q^2 \mint{d\hat \tau}{0}{\tau}e^{-2

506:         q^2 \left[ q^2 (\tau-\hat \tau)+\phi(\sigma,\tau)-

507:           \phi(\sigma, \hat \tau)\right] } \right\}\;.

508: \end{eqnarray}

509: Note that for {\epull}, the conditions in Eqs.~(\ref{eq:scales1-time},

510: \ref{eq:scales1-length}) imply the scales

511: \begin{subequations} \label{eq:scales2}

512:   \begin{eqnarray}

513:     \tf&=&\left(v^2 \hat \zeta \lp \right)^{-4/5} 

514:     \label{eq:scales2-time} \\

515:     \sf&=&\left(v{\hat \zeta}^{3}\lp^{-2}\right)^{-1/5}

516:     \label{eq:scales2-length}  \qquad ({\epull})\;,

517:   \end{eqnarray}

518: \end{subequations} 

519: since $\fe$ depends on $\sf$ via Eq.~(\ref{eq:ext-force-towing}).

520: 

521: 

522: 

523: 

524: \subsection{Asymptotic scaling}

525: \label{sec:asymptotic-growth}

526: We have thus removed the parameter dependence of the differential equation as well as

527: the boundary and initial conditions. The remaining task is to solve

528: Eq.~(\ref{eq:pide-nondimensional-1}) for $\phi$ under the initial/boundary conditions

529: given by Eqs.~(\ref{eq:ic-sansatz},~\ref{eq:bc-infty},~\ref{eq:bc-origin}), and to

530: extract the tension

531: \begin{subequations}

532:   \label{eq:closure-1}

533:   \begin{eqnarray}

534:     f(s,t)&=&\partial_t F(s,t) \label{eq:closure-a-1} \\

535:     &=&\fe \; \varphi\left(\frac s {\sf },\frac t {\tf}\right)

536:   \end{eqnarray}

537: \end{subequations}

538: in terms of the scaling function

539: \begin{equation}

540:   \label{eq:close-scaling-form-1}

541:   \varphi(\sigma,\tau)\equiv\partial_\tau \phi(\sigma,\tau) \;.

542: \end{equation}

543: 

544: Although this remaining task of finding a solution is analytically not possible in

545: generality, we now know, at least, that it should be a function of only two

546: variables, an effective space and time variable $\sigma=s/\sf$ and $\tf$,

547: respectively. From the conditions in Eqs.~(\ref{eq:scales1-time},

548: \ref{eq:scales1-length}), it is seen that the scales $\sf $ and $\tf$ are given by a

549: combination of the characteristic force scale $\fe$ (a length$^{-2}$ in our units)

550: and the persistence length, and hence not simply a consequence of dimensional

551: analysis.  The significance of these nontrivial scales is that they mark a crossover

552: in the behavior of the tension. For it turns out that, the two-parameter scaling form

553: $\varphi(\sigma,\tau)$ collapses onto one-parameter scaling form in the limit of

554: large and small arguments, i.e.,

555: \begin{equation}

556:   \label{eq:scaling}

557:   \varphi(\sigma,\tau)\to \tau^\alpha \hat \varphi

558:   \left(\frac{\sigma}{\tau^z}\right)

559:   \;,\qquad\text{for } \tau{\ll \atop \gg}1

560: \end{equation}

561: with a positive and monotonous scaling function $\hat \varphi(\xi)$

562: that is bounded as $\xi\to \{0,\;\infty\}$ and exponents $\alpha$ and

563: $z$ depending on the problem and the limit, i.e., short or long time

564: limit. Equation (\ref{eq:scaling}) expresses the asymptotic

565: self-similarity of the tension profile: by stretching the tension

566: profile at a given time in the arclength coordinate $\sigma$ one

567: obtains the tension profile at a later time, a property inherited

568: from the self-affine conformational fluctuation spectrum of the weakly

569: bending wormlike chain.

570: 

571: Rewriting the scaling variable in Eq.~(\ref{eq:scaling}) as

572: $\sigma/\tau^z\equiv s/\lpa(t)$ identifies the tension propagation length

573: \begin{equation}

574:   \label{eq:def-lf}

575:   \lpa(t)=\sf \left( \frac{t}{\tf}  \right)^z \;.

576: \end{equation}

577: In Tab.~\ref{tab:lf-growth-laws} the actual growth laws $\lpa(t)$ are

578: tabulated depending on the problem and the asymptotic limit.

579: 

580: Before deriving these growth laws from an asymptotic analysis of

581: Eq.~(\ref{eq:pide-nondimensional-1}), let us give a simple argument as to what the

582: exponent $z$ should be on short times. As usual, we assume the crossover should occur

583: when the scaling variable $\sigma/\tau^z$ is of order one, hence

584: \begin{equation}

585:   \label{eq:crossover-requirement}

586:   \lpa^\gtrless(\tf)\simeq\sf \;.

587: \end{equation}

588: If we further assume, that on very short times the propagation length $\lpa^<(\tau)$

589: of the tension should actually be independent of the external force, we can

590: immediately infer

591: \begin{equation}

592:   \label{eq:short-time-guess}

593:   \lpa^<(t)\propto \lp^{1/2}t^{1/8} \;,

594: \end{equation}

595: which is the correct short time growth law, as will be shown in

596: Sec.~\ref{sec:short-time-asymptotics}.

597: 

598: 

599: 

600: The derivations we present in the following are consistent in the sense that we use

601: assumptions that are a posteriori legitimized by the solutions. In particular, we

602: exploit Eq.~(\ref{eq:scaling}) as a scaling ansatz in order to derive asymptotic

603: differential equations for the tension. Those equations are then shown to be indeed

604: solved by similarity solutions of the postulated type. In addition, let us assume

605: that the exponent $\alpha$ in Eq.~(\ref{eq:scaling}) is larger than $-1/2$,

606: \begin{equation}

607:   \label{eq:alpha-assumption}

608:   \alpha>-1/2 \;,

609: \end{equation}

610: i.e., the tension should increase (decrease) less rapidly than $\tau^{-1/2}$ for

611: $\tau\to 0$ ($\tau\to\infty$). The assumption is reasonable for {\pull}, because at

612: the ends, we have $\varphi(0,\tau)=1$ and therefore $\alpha=0$ in this case. It turns

613: out that Eq.~(\ref{eq:alpha-assumption}) is correct for all considered problems of

614: tension propagation except for sudden temperature changes, discussed in

615: Ref.~\cite{obermayer-kroy-frey-hallatschek3:tbp}. As a consequence, the

616: approximations that are made in the following do not apply to sudden changes in

617: persistence length, which is an indication that it is an exceptional problem. In

618: Sec.~\ref{term-rel-tensprop}, where we consider the scaling regime \emph{succeeding}

619: tension propagation, we encounter an asymptotic regime of {\lelo} characterized by an

620: exponent $\alpha=-2/3$ as another important exception of

621: Eq.~(\ref{eq:alpha-assumption}).

622: 

623: Since the central PIDE Eq.~(\ref{eq:pide-nondimensional-1}) is

624: expressed in terms of the time integrated tension $\phi(\sigma,\tau)$,

625: it is useful for the following discussion to reformulate the scaling

626: assumption, Eq.~(\ref{eq:scaling}), in terms of $\phi$,

627: \begin{equation}

628:   \label{eq:scaling-phi}

629:   \phi(\sigma,\tau)\to \tau^{\alpha+1} \hat \phi

630:   \left(\frac{\sigma}{\tau^z}\right)

631:   \;,\qquad\text{for } \tau{\ll \atop \gg}1  \;.

632: \end{equation}

633: 

634: 

635: 

636: \subsubsection{Short times ($t \ll \tf$)}

637: \label{sec:short-time-asymptotics} 

638: In case Eq.~(\ref{eq:alpha-assumption}) holds we can linearize

639: Eq.~(\ref{eq:pide-nondimensional-1}) for short times in $\phi$. This is at first

640: sight only correct for small wave numbers that satisfy $q^2\phi=\Ord{ q^2

641:   \tau^{\alpha+1}}\ll 1$. However, upon a closer inspection of the region of large

642: wave vectors $q>\tau^{-(\alpha+1)/2}$ that do not allow for a linearization, it is

643: seen that there is another term in the exponent, $q^4 \tau>\tau^{-2\alpha-1}\gg1$

644: (for $\tau\ll1$ and $\alpha>-1/2$), which renders the considered exponential

645: essentially zero.  Therefore, we can approximate Eq.~(\ref{eq:pide-nondimensional-1})

646: by

647: \begin{equation}

648:   \label{eq:pide-linearized}

649:   \begin{split}

650:     & \partial_\sigma^2 \phi(\sigma,\tau) \approx \\

651:     & \mint{\frac{dq}{2\pi}}{-\infty}{\infty} \left\{\frac{1}{q^2+\fc}

652:       \left[1-\left(1-2q^2\phi(\sigma,\tau)\right)e^{-2q^4

653:           \tau}\right] \right.\\

654:     & \left. -2 q^2 \mint{d\hat

655:         \tau}{0}{\tau}\left[1-2q^2\left(\phi(\sigma,\tau)-\phi(\sigma,\hat

656:           \tau)\right)\right]e^{-2 q^4 (\tau-\hat \tau)} \right\}\\

657:     &\qquad\qquad=\mint{\frac{dq}{2\pi}}{-\infty}{\infty}

658:     \left\{-\frac{\fc}{q^2(q^2+\fc)}\left(1-e^{-2q^4\tau}\right)

659:     \right. \\

660:     &\qquad\qquad \left. +2\phi(\sigma,\tau)\left[1-\left(\frac{\fc}

661:           {q^2+\fc}\right)e^{-2q^4\tau}\right]\right.\\

662:     &\qquad\qquad\left.- 4q^4 \mint{d\hat \tau}{0}{\tau}\phi(\sigma,\hat \tau) e^{-2

663:         q^4 (\tau-\hat \tau)} \right\}\;.

664: \end{split}

665: \end{equation}

666: Since $\tau\ll1$ we can neglect the parameter $\fc$ in the denominator of the first

667: term and set the exponential function in the second term equal to one.  Furthermore,

668: we observe that the lower bound of the time-integral can be set to $-\infty$ by

669: defining $\phi(\sigma,\tau<0)\equiv 0$. After the variable substitution $\hat \tau

670: \to \hat \tau+\tau$ we obtain

671: \begin{equation}

672:   \label{eq:pide-linearized-11}

673:   \begin{split}

674:     &\partial_\sigma^2 \phi(\sigma,\tau)

675:     =\mint{\frac{dq}{\pi}}{-\infty}{\infty} \left[\fc \frac{e^{-2q^4

676:           \tau}-1}{2q^4}+

677:        \right. \\

678:     &\left.\frac{q^2 \phi(\sigma,\tau)}{q^2+\fc} - 2q^4\mint{d\hat

679:         \tau}{-\infty}{0}\phi(\sigma,\hat \tau+\tau) e^{2 q^4 \hat

680:         \tau} \right] \;.

681:   \end{split}

682: \end{equation}

683: Now we introduce the Laplace transform of $\phi$ according to

684: \begin{subequations}

685:   \label{eq:laplace-trafo}

686:   \begin{align}

687:     \phi(\sigma,z)& \equiv \mint{d\tau}{0}{\infty}e^{-z \tau}

688:     \phi(\sigma,\tau) \\

689:     \phi(\sigma,\tau)&=\mint{\frac{dz}{2\pi i}}{c-i \infty}{c+i

690:       \infty}e^{z \tau} \phi(\sigma,z)\;,

691:   \end{align}

692: \end{subequations}

693: such that Eq.~(\ref{eq:pide-linearized-11}) in Laplace space reads

694: \begin{eqnarray}

695:   \label{eq:pide-linearized-lt-1}

696:     &&\partial_\sigma^2 \phi(\sigma,z) =

697:     \nonumber\\

698:     &&\mint{\frac{dq}{\pi}}{-\infty}{\infty}\left[

699:        -\frac{\fc}{2q^4}\left(\frac1 z-\frac1{z+2q^4}\right)+

700:       \frac{q^2 \phi(\sigma,z)}{q^2+\fc}\right.\nonumber\\

701:     &&\left.-\mint{d\tau}{0}{\infty}e^{-

702:         z \tau}\mint{d \hat\tau}{-\infty}{0}\phi(\sigma,\hat \tau+

703:       \tau)(2

704:       q^4)e^{2q^4 \hat \tau}\right] \nonumber\\

705:     &&= \mint{\frac{

706:         dq}{\pi}}{-\infty}{\infty}\left[-\frac{\fc}{z(z+2q^4)}+

707:      \frac{z \phi(\sigma,z)}{2

708:           q^4+z}-\frac{\fc \phi(\sigma,z)}{q^2+\fc}\right] \nonumber\\

709:     &&=2^{-3/4}\left(z^{1/4}\phi-\fc z^{-7/4}\right)-\fc^{1/2}\phi

710:     \;.

711: \end{eqnarray}

712: In the short time limit $z\gg 1$, this reduces to  

713: \begin{equation}

714:   \label{eq:linear-pide-laplace}

715:   \partial_\sigma^2 \phi(\sigma,z)=2^{-3/4}z^{1/4}\phi\;.

716: \end{equation}

717: Choosing the decaying solution we find for the boundary conditions in

718: Eqs.~(\ref{eq:bc-infty},~\ref{eq:bc-origin-1}) of {\pull}

719: \begin{equation}

720:   \label{eq:phi-small-t}

721:   \phi=\phi_P(\sigma,z)\equiv z^{-2} e^{-\sigma (z/8)^{1/8}} \qquad

722: (\pull) \;.

723: \end{equation}

724: Since Eq.~(\ref{eq:pide-linearized-lt-1}) is a linear differential equation, we can

725: express the solutions for the boundary conditions of {\epull} and {\lelo} in terms of

726: $\phi_P$,

727: \begin{subequations}

728:   \label{eq:letloose-from-pulling}

729:   \begin{align}

730:     \phi&=z^{-2}-\phi_P(\sigma,\tau) \qquad (\lelo) \\

731:     \phi&=\mint{d\hat \sigma}{\sigma}{\infty}\phi_P(\hat

732:     \sigma,\tau)\qquad (\epull) \;.

733:   \end{align}

734: \end{subequations}

735: Ultimately, we are interested in the tension 

736: \begin{equation}

737:   \label{eq:tension-from-phi}

738:   \varphi_P(\sigma,\tau)=\partial_\tau

739:   \phi_P(\sigma,\tau) \;,

740: \end{equation}

741: which is given by the inverse Laplace transform of $z

742: \phi_P(\sigma,z)$,

743: \begin{eqnarray}

744:   \varphi_P(\sigma,\tau)&=&\hat\varphi_P\left(\frac{\sigma}{\tau^{1/8}}\right)   \label{eq:lt-inverse-a}

745:   \\

746:   \hat\varphi_P(\xi)&=&\mint{\frac{dz}{2\pi i z}}{-i \infty+\epsilon}{i\infty+\epsilon}e^{-\xi

747:       (z/8)^{1/8}+z }\;. \label{eq:lt-inverse-b}

748: \end{eqnarray}

749: After deforming the contour of integration such that it encloses the branch cut at the

750: negative real axis, the integral in Eq.~(\ref{eq:lt-inverse-b}) becomes

751: \begin{eqnarray}

752:   \label{eq:small-time-result}

753:   \hat \varphi_P(\xi)&=&\mint{\frac{dx}{\pi}}{0}{\infty}\frac8x \sin\left[\xi x\sin{\frac \pi 8} 

754:   \right] e^{-\xi x \cos{\frac \pi 8}}\nonumber\\

755:   & &\times\left( 1-e^{-8x^8}  \right) \;,

756: \end{eqnarray}

757: which is to our knowledge not tabulated, but can be easily evaluated numerically, see

758: Fig.~\ref{fig:small-t-profile}.  Upon using the known Laplace transform,

759: \begin{equation}

760:   \label{eq:power-law-back}

761:   \frac{\Gamma(\nu)}{z^\nu}=\mint{d\tau}{0}{\infty}e^{-z \tau} \tau^{\nu-1}\;,\quad

762:     \Re \nu>0\;,

763: \end{equation}

764: and Taylor expanding the integrand of Eq.~(\ref{eq:lt-inverse-b}) one obtains an

765: expansion of $\varphi_P(\xi)$ that is particular useful for small $\xi$,

766: \begin{equation}

767:   \label{eq:small-sigma-exp}

768:    \hat \varphi_P(\xi)=\sum_{{n=0\atop n/8\notin\mathbb{N}}}^{\infty}

769:    \frac{(-2^{-3/8}\xi)^n}{n!\Gamma(1-n/8)}\;.

770: \end{equation}

771: With an absolute error less than one percent the scaling function is approximated by

772: an exponential,

773: \begin{equation}

774:   \label{eq:approx-lr}

775:   \hat \varphi_P(\xi)\approx \exp\left(-\frac{\xi}{2^{3/8}\Gamma(7/8)}

776:   \right) \;,

777: \end{equation}

778: where the pre-factor of $\xi$ in the exponent is the initial slope

779: $\partial_\xi \hat\varphi\vert_{\xi=0}$ of the scaling function.

780: \begin{figure}

781:   \centerline{\includegraphics[width=\columnwidth]

782:   {small-time-profile-psfraged.eps}}

783:   \caption{The scaling function $\hat \varphi_P(\xi)$ describing the shape

784:     of the tension profile of {\pull} on short times.}\label{fig:small-t-profile}

785: \end{figure}

786: 

787: The tension profiles of the problems under consideration are all

788: related to the scaling function $\hat \varphi_P(\xi)$,

789: \begin{subequations}

790:   \label{eq:letloose-from-pulling-2}

791:   \begin{align}

792:     f(s,t)&=\fe \hat\varphi_P\left(\frac s {\lpa(t)}\right)\qquad (\pull) \\

793:     f(s,t)&=\fe \left[1-\hat\varphi_P\left(\frac s

794:         {\lpa(t)}\right)\right]

795:     \qquad (\lelo) \\

796:     f(s,t)&=\hat \zeta v \lpa(t)\mint{d\xi}{s/\lpa(t)}{\infty}

797:     \hat\varphi_P(\xi)\qquad (\epull) \label{eq:efp-t<1} \;, 

798:   \end{align}

799: \end{subequations}

800: where the boundary layer at time $t$ has the typical size

801: \begin{equation}

802:   \label{eq:lambda-short-time}

803:   \lpa(t)=\hat \zeta^{-1/2} \lp^{1/2} t^{1/8}\propto\lralf(t) \;.

804: \end{equation}

805: The scaling was anticipated in Eq.~(\ref{eq:short-time-guess}) and in an ad hoc

806: scaling argument presented in Part~I. The result for {\epull},

807: Eq.~(\ref{eq:efp-t<1}), shows that the force at the grafted end, at $\xi=0$, scales

808: like $\hat \zeta v \lpa(t)$. This scaling will be shown to hold also in the long-time

809: limit and can be understood in the sense that the graft has to balance only the drag

810: arising within the boundary layer, since the bulk of the filament is not moving

811: longitudinally. Thus, measuring the force at the grafted end, we can monitor the

812: spreading of the tension. This gives special experimental relevance to the {\epull}

813: scenario.  In particular, we predict

814: \begin{equation}

815:   \label{eq:force-at-graft}

816:   f(0,t)=\hat \zeta v \lpa(t)2^{3/8}/\Gamma(9/8)

817: \end{equation}

818: for the force at the grafted end on short times, $\tau\ll1$. The

819: pre-factor in Eq.~(\ref{eq:force-at-graft}) has been found by

820: evaluating the remaining integral in Eq.~(\ref{eq:efp-t<1}), which

821: yields a Taylor series very similar to Eq.~(\ref{eq:small-sigma-exp}),

822: \begin{equation}

823:   \label{eq:efp-t<1-full}

824:   \mint{d\hat\xi}{\xi}{\infty}

825:     \hat\varphi_P(\hat \xi)=2^{3/8}\sum_{{n=-1\atop 

826: n/8-1\notin\mathbb{N}}}^{\infty}

827:    \frac{(-2^{-3/8}\xi)^{n+1}}{(n+1)!\Gamma\left(1-n/8\right)}\;.

828: \end{equation}

829: 

830: 

831: \subsubsection{Long times ($t \gg \tf$)}

832: \label{sec:long-time-asymptotics}

833: The present subsection deals with the dynamics of the tension on a semi-infinite

834: filament on asymptotically long times. We identify and interprete the terms

835: dominating the stored length release in this limit. Neglecting subdominant terms in

836: the continuity equation, Eq.~(\ref{eq:cg-eom}), results in differential equations

837: that can be solved by similarity solutions.

838: 

839: The right hand side of the nondimensionalized PIDE,

840: Eq.~(\ref{eq:pide-nondimensional-1}), represents the negative change of stored length

841: in adapted units. It can be written as the sum of two terms, $A$ and $B$, where

842: \begin{subequations}

843:   \begin{align}

844:     \label{eq:A-def}

845:     A&\equiv\mint{\frac{dq}{2\pi}}{-\infty}{+\infty}\frac 1 {q^2+\fc

846:     }\left[1-e^{-2q^2\left[q^2 \tau+

847:           \phi(\sigma,\tau)\right]}\right]\;,\\

848:     \label{eq:B}

849:     B&\equiv-\mint{\frac{dq}{2\pi}}{-\infty}{\infty} 

850:     2 q^2 \mint{d\hat \tau}{0}{\tau}e^{-2 q^2 \left[ q^2

851:         (\tau-\hat \tau)+\phi(\sigma,\tau)- \phi(\sigma, \hat

852:         \tau)\right] }\;.

853:   \end{align}

854: \end{subequations}

855: We already pointed out in Part~I that the term $A$ can be interpreted as the

856: ``deterministic relaxation'' of stored length (for the fictitious situation ``T=0'',

857: i.e.~no thermal noise, for $t>0$). The term $B$ describes the increase in stored

858: length due to the thermal kicks and is strictly positive. We analyze both terms

859: separately.

860: 

861: For pre-stretched initial conditions ($\fc=1$) the long-time limit

862: of $A$ follows from setting the  exponential to zero,

863: \begin{equation}

864:   \label{eq:A-let-loose}

865:   A\stackrel{\tau\gg1}{\to}\mint{\frac{dq}{2\pi}}{-\infty}{+\infty}

866:   \frac1{q^2+1}=\frac 1 2\;, \quad \text{for }\fc=1 \;.

867: \end{equation}

868: The term $A$ for $\tau\to \infty$ is nothing but the initially stored length, which

869: is completely relaxed after a purely deterministic relaxation. 

870: 

871: The same reasoning cannot be applied in cases of a tension-free

872: initial state (both pulling problems): setting the exponential to zero

873: in Eq.~(\ref{eq:A-def}) for $\fc=0$ yields an infrared divergence. The

874: exponential has to be retained to render the integrand finite at small

875: wave numbers. For the dominant small wave numbers one can, however,

876: neglect the term $2 q^4 \tau$ in the exponent because it is small

877: compared to the term $2q^2\Phi$, so that we arrive at the asymptotic

878: expression

879: \begin{equation}

880:   \label{eq:a-pulling}

881:   \begin{split}

882:     A\stackrel{\tau\gg1}{\to}&\mint{\frac{dq}{2\pi}}{-\infty}{+\infty}

883:     \frac1{q^2}

884:     \left[1-e^{-2q^2\phi(\sigma,\tau)}\right]\\

885:     &=\sqrt{\frac2\pi} \sqrt{\phi(\sigma,\tau)}\;,\quad \text{for } \fc=0 \;.

886:   \end{split}

887: \end{equation}

888: I.e. the ``deterministic relaxation'' is dominated by the tension term

889: and bending can be neglected (as heuristically assumed by Seifert et

890: al.~\cite{seifert-wintz-nelson:96}). A more formal justification for

891: Eqs.~(\ref{eq:A-let-loose},~\ref{eq:a-pulling}) is given in

892: App.~\ref{sec:A}.

893: 

894: 

895: The term $B$, describing the stored length generated by the thermal

896: kicks, takes for asymptotically large $\tau\gg1$ the form

897: \begin{equation}

898:   \label{eq:B-large-times}

899:   B\stackrel{\tau\gg1}{\to}-\mint{\frac{dq}{2\pi}}{-\infty}{\infty}

900:   \frac 1 {q^2+\partial_\tau \phi(\sigma,\tau)}

901:   =-\frac1{2\sqrt{\partial_\tau\phi(\sigma,\tau)}}

902: \end{equation}

903: independent of the initial conditions. This is shown in App.~\ref{sec:B} upon using

904: the scaling assumptions in Eqs.~(\ref{eq:scaling-phi},~\ref{eq:alpha-assumption}).

905: The result, Eq.~(\ref{eq:B-large-times}), should not come as a surprise. It simply

906: represents the (negative) stored length of a stiff polymer equilibrated at the

907: (rescaled) tension $\partial_\tau \phi$. For a constant force, i.e., $\partial_\tau

908: \phi=$const., it is obvious that the stored length should saturate for long times at

909: the corresponding equilibrium value.  But also if the tension is varying slowly

910: enough in time ($\alpha>-1/2$) the ``noise-generated'' stored length can be

911: considered as quasi-statically equilibrated with the tension.

912: 

913: Finally, we combine the ``relaxed stored length'' expressed in $A$ and

914: the ``noise-generated stored length'' $B$.

915: 

916: 

917: 

918: \paragraph{Pulling and Towing:} For $\fc=0$ and $\alpha>-1/2$ the term $A\propto

919: \sqrt{\phi}\propto\tau^{\alpha/2+1/2}$ is much larger than $B\propto

920: (\partial_\tau\phi)^{-1/2}=\Ord{\tau^{-\alpha/2}}$. I.e. the effects of noise can be

921: neglected on long times (as presumed by Seifert et

922: al.~\cite{seifert-wintz-nelson:96}). In this limit, the thermal noise is merely

923: relevant in preparing the initial state. The relaxation after force application for

924: $\tau\gg 1$ is purely mechanical, like for a pulled string that is initially prepared

925: with some contour roughness (\emph{taut-string approximation}). If we replace the

926: right-hand-side of Eq.~(\ref{eq:pide-nondimensional-1}) by the asymptotic form of

927: $A$, Eq.~(\ref{eq:a-pulling}), we obtain the partial differential equation

928: \begin{equation}

929:   \label{eq:pulling-large-time-pde}

930:   \partial_\sigma^2\phi(\sigma,\tau)=\sqrt{\frac2\pi} 

931:   \sqrt{\phi(\sigma,\tau)}

932: \end{equation}

933: for the dynamics of the integrated tension $\phi$.  

934: 

935: Eq.~(\ref{eq:pulling-large-time-pde}) represents a Newtonian equation of motion for a

936: particle moving in a conservative force field $\propto\sqrt{\phi}$ and can be

937: integrated straightforwardly. In the case of {\pull} the solution, which satisfies

938: the boundary conditions in Eqs.~(\ref{eq:bc-infty},~\ref{eq:bc-origin-1}) of {\pull} (in

939: particular $\phi(0,\tau)=\tau$), is given by

940: \begin{equation}

941:   \label{eq:pulling-large-time-phi}

942:   \phi(\sigma,\tau)=\tau\left[\sigma/(72\pi\tau)^{1/4}

943:         -1\right]^4

944: \end{equation}

945: for $\sigma<(72\pi \tau)^{1/4}$ and $\phi=0$ otherwise. The dimensionless tension

946: $\varphi$ is derived from $\phi$ via differentiation, see

947: Eq.~(\ref{eq:close-scaling-form-1}), and obeys a scaling form,

948: \begin{equation}

949:   \label{eq:pulling-large-time-varphi}  

950:   f(s,t)=\fe \varphi(s,t)=\fe \hat\varphi\left(\frac s {\lpa(t)}\right)\;,

951: \end{equation}

952: with a typical boundary layer size proportional to $\ludo(t)$,

953: \begin{equation}

954:   \label{eq:lambda-large-time}

955:   \lpa(t)=t^{1/4}\hat 

956:   \zeta^{-1/2}\lp^{1/2}\fe^{1/4} 

957: \end{equation}

958: and a scaling function $\hat \varphi$ given by

959: \begin{equation}

960:   \label{eq:scalingfct-large-times}

961:   \hat\varphi(\xi)=\left[1-\xi/(72\pi)^{1/4}\right]^3 

962: \end{equation}

963: for $ \xi<(72 \pi)^{1/4}$ and $\hat \varphi=0$ otherwise. {\epull} starts with the same initial conditions ($\fc=0$) but with

964: different boundary conditions, Eqs.~(\ref{eq:bc-infty},~\ref{eq:bc-origin-2}).

965: Again, we have to solve Eq.~(\ref{eq:pulling-large-time-pde}), but now

966: under the boundary condition $\partial_\sigma \phi(0,\tau)=-\tau$. The

967: solution is

968: \begin{equation}

969:   \label{eq:phi-electroforesis}

970:   \phi(\sigma,\tau)=\tau^{4/3}\left( \frac{9\pi}{32}  \right)^{1/3}\left[

971:       \frac{\sigma}{(18\pi\tau)^{1/3}}-1  \right]^4  

972: \end{equation}

973: for $\sigma<(18\pi\tau)^{1/3}$ and $\phi=0$ otherwise. This implies a tension profile

974: of

975: \begin{equation}

976:   \label{eq:tension-orig-electroforesis}

977:   f(s,t)=\hat \zeta v \lpa(t)\; \hat\varphi\left(\frac s 

978:     {\lpa(t)}\right)\;,

979: \end{equation}

980: with the typical boundary layer size

981: \begin{equation}

982:   \label{eq:lambda-ef-large-time}

983:   \lpa(t)=    t^{1/3}\lp^{2/3}(v/\hat \zeta)^{1/3}  \;,

984: \end{equation}

985: and the scaling function $\hat \varphi$ given by

986: \begin{equation}

987:   \label{eq:scalingfct-towing-large-times}

988:   \hat\varphi(\xi)=\left(\frac{2\pi}{3}

989:       \right)^{1/3}\left[1-\xi/(18\pi)^{1/3}\right]^3 

990: \end{equation}

991: for $ \xi<(18 \pi)^{1/3}$ and $\hat\varphi=0$ otherwise. As on short times, the

992: absolute value of the reduced tension at the left end is proportional to the size of

993: the boundary layer. Its precise value is predicted to be

994: \begin{equation}

995:   \label{eq:left-end-tension-electroforese}

996:   f(0,t)=\left(2\pi/3  \right)^{1/3}\hat \zeta v \lpa(t)\propto t^{1/3} \;,

997: \end{equation}

998: and should be directly accessible to single molecule experiments.

999: 

1000: 

1001: \paragraph{Release:}

1002: The stored length release is asymptotically given by

1003: \begin{equation}

1004:   \label{eq:letting-loose-rho-release-large-t}

1005:   -\avg{\Delta\overline{\varrho}}\propto A+B\stackrel{\tau\gg1}{\sim} \frac 1 2- 

1006:   \frac1{2\sqrt{\partial_\tau \phi(\sigma,\tau)}}\;.

1007: \end{equation}

1008: This expression for the change in stored length can also be directly obtained if one

1009: assumes that the filament was at any time equilibrated with the current tension

1010: $\varphi=\partial_\tau \phi$ (\emph{quasi-static approximation}).  Our derivations

1011: show, that this assumption, used by Brochard et

1012: al.~\cite{brochard-buguin-de_gennes:99}, is only valid in the long time limit for the

1013: scenario of {\lelo}. The dynamics of the integrated tension is then described by

1014: \begin{equation}

1015:   \label{eq:pde-letting-loose-long-t}

1016:   \partial_\sigma^2\phi=\frac 1 2-\frac1{2\sqrt{\partial_\tau 

1017:       \phi(\sigma,\tau)}}\;,

1018: \end{equation}

1019: or, in terms of the tension $\varphi=\partial_\tau \phi(\sigma,\tau)$,

1020: \begin{equation}

1021:   \label{eq:pde-letting-loose-long-t-2}

1022:   \partial_\sigma^2\varphi=\frac1 4 \varphi^{-3/2}\partial_\tau\varphi\;.

1023: \end{equation}

1024: The solution satisfying the correct boundary conditions,

1025: Eqs.~(\ref{eq:bc-infty},~\ref{eq:bc-origin-1}), is given by a scaling form

1026: \begin{equation}

1027:   \label{eq:scform-letting-loose-long-time}

1028:   \varphi(s,t)=\hat \varphi\left(\frac s {\lpa(t)}\right) \;,

1029: \end{equation}

1030: with the typical boundary layer size now growing like

1031: \begin{equation}

1032:   \label{eq:lambda-letloose-large-t}

1033:   \lpa(t)=t^{1/2}\hat\zeta^{-1/2}\lp^{1/2}\fe^{3/4} \propto \lpierre(t)

1034: \end{equation}

1035: and a scaling function $\hat\varphi(\xi)$ satisfying the ordinary differential

1036: equation

1037: \begin{equation}

1038:   \label{eq:scfct-ode}

1039:   \partial_\xi^2\hat\varphi=-\frac1 8 \xi   

1040:   {\hat\varphi}^{-3/2}\partial_\xi\hat\varphi \;.

1041: \end{equation}

1042: The scaling function depicted in Fig.~\ref{fig:degennes-scfct} was already obtained

1043: numerically in Ref.~\cite{brochard-buguin-de_gennes:99}. The slope at the origin is

1044:   $\partial_\xi \hat \varphi\vert_{\xi=0}\approx0.6193$. 

1045: 

1046: \begin{figure}

1047:   \includegraphics[width=\columnwidth]{degennes-corrected-psfraged.eps}

1048:   \caption{The scaling function $\hat \varphi(\xi)$ for 

1049:     {\lelo} obtained by a numerical solution of

1050:     Eq.~(\ref{eq:scfct-ode}).}\label{fig:degennes-scfct}

1051: \end{figure}

1052: 

1053: 

1054: 

1055: \subsection{Tension propagation (summary)}

1056: \label{sec:tensprop-summary}

1057: 

1058: This section summarizes the picture of tension propagation that emerges from the

1059: above solutions of the dynamical equation for the tension, Eq.~(\ref{eq:pide}), for

1060: sudden changes in boundary conditions.

1061: 

1062: For the considered problems {\pull}, {\epull} and {\lelo}, we have shown that the

1063: tension profile in units of $\fe$ has to obey a crossover scaling form

1064: $\varphi(\sigma,\tau)$ depending on a reduced time variable $\tau\equiv t/t_\fe$ and

1065: a reduced arc length variable $\sigma\equiv s/s_\fe$. The scaling function $\varphi$

1066: describes how sudden changes of the tension at the ends spread into the bulk of the

1067: polymer.  Its crossover structure and the expressions for $t_\fe$ and

1068: $s_\fe\approx\lpa(t_\fe)$ are consistent with our heuristic discussion of {\pull} in

1069: Part~I. In the limits $\tau \ll1$ and $\tau\gg1$ the function $\varphi(\sigma,\tau)$

1070: assumes a simple (one-variable) scaling form 

1071: \begin{equation}

1072:   \label{eq:tension-scform}

1073:   \varphi\sim \tau^{\alpha_{\gtrless}}

1074: \hat\varphi^{\gtrless} [\sigma/\tau^{z_{\gtrless}}] \;.

1075: \end{equation}

1076: The notation $\gtrless$

1077: indicates that the asymptotic form of $\hat \varphi$, $\alpha$, and $z$ will

1078: generally not only depend on the kind of external perturbation applied, but will also

1079: differ for times $t\gtrless t_\fe$.  Rewriting $\sigma/\tau^z\equiv s/\lpa$

1080: identifies the tension propagation length $\lpa\equiv s_\fe\,\tau^z$.

1081: 

1082: For $t\ll t_\fe$ Eq.~(\ref{eq:pide}) could be linearized in $f$ and the scaling

1083: function $\hat \varphi^<$ was obtained analytically. Whereas $\hat \varphi^<$ depends

1084: on the considered force protocol, the corresponding exponent $z_<=1/8$ is independent

1085: of the boundary conditions. As established by our heuristic discussion of {\pull},

1086: this is due to the relaxation of modes with Euler forces $\lpe^{-2}\gg \fe$ much

1087: larger than the external force, for which the equilibrium mode spectrum is hardly

1088: perturbed by the external force. The self-affinity of the equilibrium mode spectrum

1089: translates into a self-similar relaxation dynamics. The dynamic exponent $z$ for the

1090: growth of the boundary layer could already be anticipated from requiring

1091: $\hat\varphi^<$ to become $\fe-$independent as in linear response, see

1092: Eq.~(\ref{eq:short-time-guess}). The short-time dynamics for strong external force is

1093: thus closely related to the linear response. Note, however, that the limit $\fe\to0$

1094: is problematic, as it does not interchange with $\epsilon\to0$.  Our identification

1095: of arc length averages with (local) ensemble averages in Part~I breaks down for

1096: $\fe<(\zeta/\lp)^{1/4}t^{-7/16}$, where fluctuations in the tension become comparable

1097: to its average value.  In fact, extending Eq.~(\ref{eq:pide}) to linear response

1098: amounts to an uncontrolled factorization approximation $\avg{f r_\pe^2}\to

1099: \avg{f}\avg{r_\pe^2}$.  Even in the stiff limit the linear longitudinal dynamic

1100: response remains an open problem. The limit where fluctuations in the tension become

1101: important and its consequences will be detailed in Sec.~\ref{sec:tens-fluct}.

1102: 

1103: For $t\gg t_\fe$ the dynamics becomes nonlinear in the external force and starts to

1104: depend on the kind of external perturbation and on how precisely it is applied to the

1105: polymer.  Previously predicted power laws were recovered from Eq.~(\ref{eq:pide}) by

1106: employing different approximations to its right hand side.  In the \emph{taut-string

1107:   approximation} of Ref.~\cite{seifert-wintz-nelson:96}, one neglects for $t>0$

1108: bending and thermal forces against the tension, i.e., one drops the $q^4-$term in the

1109: relaxation time $\tau_q=q^4+f q^2$ of a mode with wave number $q$ and sets

1110: $\lp\to\infty$ for positive times (i.e.~$\theta=0$).  The complementary

1111: \emph{quasi-static approximation} of Ref.~\cite{brochard-buguin-de_gennes:99} amounts

1112: to the omission of memory effects, i.e., to the assumption of instantaneous

1113: equilibration of tension and stored length (as it would be the case for vanishing

1114: \emph{transverse} friction, $\zeta_\pe\to 0$). In cases, where either of these

1115: approximations applies, a power-law dispersion relation combines with a self-affine

1116: mode spectrum to produce self-similar tension dynamics.  Our analysis of

1117: Eq.~(\ref{eq:pide}) showed that either of these approximations becomes rigorous in

1118: the intermediate asymptotic regime defined by $t\gg t_\fe$, $\lpa\ll L$. The

1119: quasi-equilibrium approximation applies to {\lelo} and the taut-string approximation

1120: to {\pull}. We could rule out the applicability of the \emph{taut-string

1121:   approximation} for {\lelo} and of the \emph{quasi-static approximation} for

1122: {\pull}~\cite{brochard-buguin-de_gennes:99} and {\epull}. The ``pure'' scenarios of

1123: self-similar dynamics are summarized in Tab.~\ref{tab:lf-growth-laws} and

1124: Fig.~\ref{fig:lf}.

1125: 

1126: 

1127: 

1128: \begin{table}

1129:   \caption{Asymptotic growth laws for the dynamic 

1130:     size $\lpa(t)$ of the tension boundary

1131:     layer.}

1132:   \label{tab:lf-growth-laws}

1133:   \begin{center}

1134:     \begin{tabular}{c|c|c}

1135:       Problem&$t\ll \tf$&$ \tf\ll t \ll t_L^\pa$\\ \hline

1136:       {\pull}&${\displaystyle \sqrt{ \lp /\hat \zeta }\; 

1137:         t^{1/8}}$&${\displaystyle

1138:         \sqrt{ \lp /\hat \zeta }\; \fe^{1/4}t^{1/4}}$\\

1139:       {\epull}&${\displaystyle \sqrt{\lp/ \hat \zeta  }\; 

1140:         t^{1/8}}$&${\displaystyle

1141:         (\lp^2 v/\hat \zeta)^{1/3}t^{1/3}}$\\

1142:       {\lelo}&${\displaystyle \sqrt{ \lp/ \hat \zeta  }\;t^{1/8}}$

1143:       &${\displaystyle

1144:         \sqrt{ \lp / \hat \zeta  }\;\fe^{3/4}t^{1/2}}$\\

1145:     \end{tabular}

1146:   \end{center}

1147: \end{table}

1148: 

1149: \begin{figure}[t]

1150:   \includegraphics[width=\columnwidth,height=.5\columnwidth

1151:   ]{growth-laws-pre-psfraged.eps}

1152:   \caption{Schematic of the tension propagation laws $\lpa(t)\propto t^z$ on a

1153:     double-logarithmic scale.  At $t_\fe=\fe^{-2}$ they crossover from a universal

1154:     short-time regime to (problem-specific) tension-dominated intermediate

1155:     asymptotics, except for weak forces, $\fe < \lp^2/L^4$. The propagation ends when

1156:     $\lpa(t)\approx L$.}

1157:   \label{fig:lf}

1158: \end{figure}

1159: 

1160: 

1161: 

1162: \section{Terminal stress relaxation} 

1163: \label{term-rel-tensprop}

1164: Up to now, we have considered the growth of the boundary layer in a

1165: stiff polymer that has a (formally) semi-infinite arc length

1166: parameter space, $s\in [0,\infty [$, which is an idealization. However,

1167: the foregoing discussion obviously applies to a polymer of

1168: \emph{finite} length $L$ for sufficiently short times: As long as the

1169: size of the boundary layer is much smaller than the total length $L$

1170: the presence of a second end is irrelevant to the boundary layer

1171: at the first end.  The time where the boundary layers span the whole

1172: polymer marks the crossover to a new behavior. For definiteness, we

1173: define the crossover time $\totpa$ by

1174: \begin{equation}

1175:   \label{tension-rel-time-tp}

1176:   \lpa(\totpa)\equiv L \;.

1177: \end{equation}

1178: What happens for $t>\totpa$? The straightforward way to answer this question is to

1179: solve for the intermediate asymptotics of the PIDE, Eq.~(\ref{eq:pide}), for a polymer

1180: of finite length. The finiteness of $L$ amounts to replacing the boundary condition

1181: $\partial_s^2 f(s\to\infty)=0$ by the correct problem-specific one, i.e., by

1182: \begin{eqnarray}

1183:   \label{eq:bc-tension-rend}

1184:   f(L,t>0)&=&\fe\qquad {\pull} \nonumber\\

1185:   f(L,t>0)&=&0 \qquad {\lelo}, {\epull}.\nonumber

1186: \end{eqnarray}

1187: One could now proceed as in Sec.~\ref{sec:asymptotic-growth} by identifying proper

1188: scaling forms and extracting their asymptotic behavior. As compared to the

1189: semi-infinite polymer limit, this procedure is more complicated for a finite polymer

1190: because of the additional scaling variable $\lpe(t)/L$.  Therefore, we prefer to take

1191: the following short-cut, which consists in two steps.

1192: 

1193: \subsubsection{Trivial tension profiles for $t\gg t_\star$.}  In the heuristic

1194: analysis of Part~I, we found that ordinary perturbation theory (OPT) should become

1195: valid for times larger than some crossover time $t_\star$. With the exception of

1196: {\epull}, which we discuss separately below, the introduced scenarios treat both ends

1197: equally, such that the tension profile in OPT has a trivial time and arc length

1198: dependence, namely

1199: \begin{equation}

1200:   \label{eq:OPT-tension}

1201:   \fopt(s,t)=f(0,t)=\text{const.}\;,

1202: \end{equation} 

1203: up to small terms of order $\Ord{\epsilon}$. For constant tension we

1204: can easily extract the longitudinal dynamics from

1205: Eq.~(\ref{eq:change-stored-length}). The corresponding

1206: predictions for the evolution of the end-to-end distance will be

1207: discussed in Sec.~\ref{sec:tension-relaxed}.

1208: 

1209: 

1210: \subsubsection{Possibility of homogeneous tension relaxation for $\totpa \ll t \ll

1211:   t_\star$.}  Up to now, we have argued that the tension propagates for $t\ll \totpa$

1212: and is constant in time and space for $t\gg t_\star$. There remains the question

1213: whether there is a non-trivial regime of homogeneous tension relaxation in the

1214: time- interval $[\totpa;t_\star]$. Using the systematic approach outlined in

1215: App.~\ref{sec:tstar} to determine $t_\star$, we actually find for most of the

1216: problems that $t_\star\simeq \totpa$, i.e., there is no scaling regime between

1217: tension propagation and the stationary tension profiles dictated by OPT.  The

1218: {\lelo}-scenario, however, provides an important exception, as it allows for a time

1219: scale separation $\totpa\ll t_{\star}$, as we demonstrate explicitely below.  For

1220: intermediate times the tension relaxation is shown to exhibit a novel behavior with

1221: an almost parabolic tension profile and an amplitude that slowly decays in time

1222: according to a power law.

1223: 

1224: \subsection{{\lelo} for large pre-stretching force}

1225: \label{sec:homogeneous-relaxation}

1226: Let us first determine the time $t_\star$, at which OPT becomes valid, for {\lelo}.

1227: In the OPT regime, the tension should be so small, that we can calculate the change in

1228: stored length accurately by means of Eq.~(\ref{eq:change-stored-length}) (with the

1229: pre-stretched initial conditions of {\lelo}, i.e.~$f_<=\fe$) under the assumption of

1230: a vanishing tension, $\fopt=0$,

1231: \begin{subequations}

1232:   \label{eq:stole-lelo-OPT}

1233:   \begin{eqnarray}

1234:     &&\avg{\Delta \overline{\varrho}}(t)=\mint{\frac{dq}{\pi\lp}}{0}{\infty}

1235:     \left\{\frac{1}{q^2+\fe}\left(e^{-2q^4 t}-1\right)\right.\nonumber\\

1236:     &&\qquad\left.+2

1237:       q^2\mint{d\tilde t}{0}{t}e^{-2q^4 (t-\tilde

1238:         t)}\right\} \label{eq:stole-lelo-OPT-1}\\

1239:     &&=\mint{\frac{dq}{\pi\lp}}{0}{\infty} \left\{\left(

1240:         \frac{1}{q^2}-\frac{1}{q^2+\fe} \right)\left(1-e^{-2q^4

1241:           t}\right) \right\} \label{eq:stole-lelo-OPT-2}\\

1242:     &&=\frac{t^{1/4}}{\lp}\mint{\frac{dq}{\pi}}{0}{\infty}\\

1243:     &&\qquad    \times\left\{\left( \frac{1}{q^2}-\frac{1}{q^2+\sqrt{t/\tf}}

1244:       \right)\left(1-e^{-2q^4} \right)

1245:     \right\}\qquad\label{eq:stole-lelo-OPT-3}  \\

1246:     &&\stackrel{t\gg\tf}{\sim}\frac{2^{3/4}} {\Gamma(1/4)}

1247:     \frac{t^{1/4}}{\lp}\;.\label{eq:stole-lelo-OPT-4}

1248:   \end{eqnarray}

1249: \end{subequations}

1250: Here, we have substituted $q\to q t^{-1/4}$ and replaced $\fe$ by $\tf^{-1/2}$ in

1251: order to obtain Eq.~(\ref{eq:stole-lelo-OPT-3}). In the final step, we took the

1252: long-time limit $t\gg\tf$.  Let us now determine the order of magnitude of the

1253: variation $\delta f$ of the tension due to the longitudinal friction. Estimating

1254: $\delta f\simeq \hat \zeta \avg{\Delta\overline\varrho}/(t L^2)$ from Eq.~(\ref{eq:cg-eom}), we

1255: obtain

1256: \begin{equation}

1257:   \label{eq:delta-f-lelo}

1258:   \delta f\approx  \hat \zeta  \frac{L^2}{\lp}t^{-3/4}\;.

1259: \end{equation}

1260: As in the heuristic discussion of Part~I, we find a diverging tension in the limit

1261: $t\to 0$ for the OPT result, so that the OPT result can only be valid after the

1262: effect of $\delta f$ on the evolution of $\avg{\Delta\overline{\varrho}}(t)$ can be

1263: neglected.  For the particular case of {\lelo}, this time can be determined as

1264: follows. As discussed in Part~I, the stress-free dynamics is at the time $t$

1265: characterized by relaxation of modes with wave length $\lpe(t)\simeq t^{1/4}$. When

1266: $\delta f$ is larger than the critical Euler Buckling force $\lpe(t)^{-2}$

1267: corresponding to the length $\lpe(t)$ we expect that the tension cannot be neglected

1268: for the evolution of $\avg{\Delta\overline{\varrho}}$.  Hence, the above result

1269: $\avg{\Delta\overline{\varrho}}\propto L t^{1/4}/\lp$, obtained from OPT, can only be

1270: valid if

1271: \begin{equation}

1272:   \label{eq:val-cond-lelo}

1273:   \lpe(t)^2\delta f\simeq \frac{L^2}{\lp}t^{-1/4}\ll1 \;,

1274: \end{equation}

1275: i.e., for long enough times

1276: \begin{equation}

1277:   \label{eq:val-cond-lelo-2}

1278:   t\gg t_\star=L^8/\lp^4 \;.

1279: \end{equation}

1280: The time $t_\star(L)$ is obviously not identical with the time

1281: \begin{equation}

1282:   \label{eq:tf-2}

1283:   \totpa=L^2 \lp^{-1} \hat\zeta \fe^{-3/2} \qquad (t\gg\tf)

1284: \end{equation}

1285: it takes for the boundary layer to spread over the filament. To compare them, we

1286: first notice that $\totpa\gg\tf=\fe^{-2}$ implies that the polymer must have been

1287: pre-stretched by a large enough force,

1288: \begin{equation}

1289:   \label{eq:large-force}

1290:   \fe\gg \lp^2/L^4=\epsilon^{-2} L^{-2} \;,

1291: \end{equation}

1292: larger than the pre-stretching force at least necessary to enter

1293: the propagation regime $\tf\ll t\ll t_L^\pa$. By using the estimate

1294: in Eq.~(\ref{eq:large-force}) we can compare $t_\star$ and $\totpa$,

1295: \begin{equation}

1296:   \label{eq:tstar-vs-tf}

1297:   t_\star(L)= \frac{L^2}{\lp}\left(\frac{L^2}{\lp}\right)^3

1298:   \gg\frac{L^2}{\lp} \fe^{-3/2} \simeq \totpa \;.

1299: \end{equation}

1300: It is seen that the time window $\totpa\ll t\ll t_\star$ grows with the

1301: pre-stretching force, which means, in particular, that it describes the limit of an

1302: initially straight polymer.

1303: 

1304: To determine the physics of this novel regime, we have to solve the equation of

1305: motion for the tension on the finite arc length interval $[0;L]$ using the

1306: approximations developed in Sec.~\ref{sec:asymptotic-growth}.  There, we found that

1307: in the limit $t\gg\tf$ the tension profile of {\lelo} is described by

1308: Eq.~(\ref{eq:pde-letting-loose-long-t-2})

1309: \begin{equation}

1310:   \label{eq:degennes-eq}

1311:   \partial_\sigma^2\varphi=\frac1 4 \varphi^{-3/2}\partial_\tau\varphi\;.

1312: \end{equation}

1313: The right-hand-side represents the time-derivative of the stored

1314: length in the \emph{quasi-static approximation}. Going back to

1315: variables $f$, $s$ and $t$, Eq.~(\ref{eq:degennes-eq}) takes the form

1316: \begin{equation}

1317:   \label{eq:degennes-eq-2}

1318:   \partial_s^2 f=\frac1 4 \frac{\hat \zeta}{\lp}f^{-3/2}\partial_t f \;.

1319: \end{equation}

1320: For the following, we assume that this \emph{quasi-static

1321: approximation} is not only valid in the tension propagation regime for

1322: $\tf\ll t \ll\totpa$, but also for longer times until the OPT

1323: regime begins, $\tf\ll t \ll t_\star$. This is justified a posteriori

1324: in App.~\ref{sec:lelo-app}, where it is shown that the change in

1325: stored length for the solution $f(s,t)$ of

1326: Eq.~(\ref{eq:degennes-eq-2}) can indeed be calculated quasi-statically

1327: for times $t \ll t_\star$. We solve Eq.~(\ref{eq:degennes-eq-2}) for

1328: the boundary conditions

1329: \begin{equation}

1330:   \label{eq:bc}

1331:   f(s=0,t)=0 \;,\;f(s=L,t)=0

1332: \end{equation}

1333: and the initial conditions

1334: \begin{equation}

1335:   \label{eq:ini-cond}

1336:   f(s,0)=\fe  \;,\qquad \text{ for }0<s<L \;.

1337: \end{equation}

1338: In Sec.~\ref{sec:long-time-asymptotics}, we solved the differential equation

1339: Eq.~(\ref{eq:degennes-eq}) with the scaling ansatz $f=\fe \hat \varphi[s/\lpa(t)]$

1340: numerically for the correct initial condition, but ignored the boundary conditions of

1341: at one end by sending $L\to\infty$. In contrast, we now look for a simple solution

1342: that obeys the tension boundary conditions exactly, at the expense of a possible

1343: mismatch with the initial conditions.  To this end, we make the product ansatz

1344: \begin{equation}

1345:   \label{product-ansatz-2}

1346:   f(s,t)=g(t) h(s) \;.

1347: \end{equation}

1348: Separation of variables yields

1349: \begin{equation}

1350:   \label{eq:sep-var}

1351:   \frac{\hat \zeta}{4\lp}g^{-5/2}\partial_t g=C=\sqrt{h}h'' \;.

1352: \end{equation}

1353: Choosing $C=-1/(6L^2)$, for convenience, we find the long time asymptotics

1354: \begin{equation}

1355:   \label{eq:g-eq}

1356:   g(t)\sim\left(\frac{\hat \zeta L^2}{\lp t}\right)^{2/3}\;.

1357: \end{equation}

1358: The spatial part obeys a scaling form $h(s)=\hat h(s/L)$, where $\hat

1359: h$ satisfies the equation

1360: \begin{equation}

1361:   \label{eq:h-eq}

1362:   \partial_\xi^2 \hat h(\xi)=-\frac 1 6 \hat h^{-1/2}  

1363: \end{equation}

1364: with the boundary condition

1365: \begin{equation}

1366:   \label{eq:bc-hat-h}

1367:   \hat h(0)=\hat h(1)=0 \;.

1368: \end{equation}

1369: The analytical solution can be found in a standard way. The main characteristics are

1370: the slope at $\xi=0$,

1371: \begin{equation}

1372:   \label{eq:h-ini-slope}

1373:   \partial_\xi h(\xi)\vert_{\xi=0}=12^{-1/3}\approx 0.4368 \;. 

1374: \end{equation}

1375: and the maximum value of $h$,

1376: \begin{equation}

1377:   \label{eq:hmax}

1378:   h(1/2)=\frac 1 {16} \left( \frac 3 2  \right)^{2/3}\approx

1379:   0.0819 \;.

1380: \end{equation}

1381: 

1382: 

1383: \begin{figure}

1384:   \includegraphics[width=\columnwidth]{degennes3d-psfraged.eps}

1385:   \label{fig:lelo-3d}

1386:   \caption{The initial ($t\ll t_\star$) time evolution of {\lelo}. A

1387:     regime of (slow) tension relaxation begins after the sudden change

1388:     in boundary condition has propagated through the filament.}

1389: \end{figure}

1390: 

1391: 

1392: Up to now we have investigated the quasi-static approximation in the tension

1393: propagation regime ($t\ll \tfin$) and in the regime of tension relaxation ($t\gg

1394: \tfin$) separately. In order to illustrate the crossover, we have also solved the

1395: corresponding PIDE, Eq.~(\ref{eq:degennes-eq-2}), numerically. The result shown in

1396: Fig.~\ref{fig:lelo-3d} unveils the transient nature of the tension propagation

1397: regime.

1398: 

1399: 

1400: 

1401: \section{Nonlinear response of the projected length}

1402: \label{sec:MSDII}

1403: After having discussed the rich tension dynamics of stiff polymers in detail, we wish

1404: to derive its consequences for pertinent observables in order to make contact with

1405: experiments.  Although the tension may in some situations be monitored directly, see

1406: Sec.~\ref{sec:possible-exps}, a more conventional observable is the longitudinal

1407: extension $R_\pa(t)$ of the polymer, which is defined to be the end-to-end distance

1408: projected onto the longitudinal axis. Tension dynamics strongly affects the nonlinear

1409: response of the projected length, which shall be detailed in the following for the

1410: force protocols {\pull}, {\epull} and {\lelo}.

1411: 

1412: The average temporal change in the projected length $R_\pa(t)$ is directly related to

1413: the stored length release,

1414: \begin{equation}

1415:   \label{eq:end-end-length}

1416:   \avg{\Delta R_\pa} (t)= -\mint{ds}{0}{L}\avg{\Delta\varrho}(s,t)+o(\epsilon)\;,

1417: \end{equation}

1418: which was already noted in Part~I. As long as modes with wave length on the order of

1419: the total length are irrelevant (i.e.~both ends of the polymer are not correlated),

1420: \begin{equation}

1421:   \label{eq:ends-uncorrelated}

1422:   \lpe(t)\ll L\;,

1423: \end{equation}

1424: half-space solutions for $\avg{\Delta\varrho}$ may be used to evaluate

1425: Eq.~(\ref{eq:end-end-length}). 

1426: 

1427: Recall from Part~I, that $\avg{\Delta\varrho}$ can be decomposed into a bulk

1428: contribution $\avg{\Delta \overline{\varrho}}$ and a term influenced by the boundary

1429: conditions, which vanishes under a spatial average.  We show in

1430: Sec.~\ref{sec:microscale} that in the cases of hinged and clamped ends, as opposed to

1431: free ends, the contribution of the boundary term to the integral in

1432: Eq.~(\ref{eq:end-end-length}) is subdominant in the limit $t\to 0$.

1433: Nevertheless, these boundary effects represent important corrections that should be

1434: taken into account in any experimental situation (i.e.~with finite $t$) of {\pull}

1435: and {\epull}.

1436: 

1437: At first, however, let us consider the \emph{bulk} contribution to the end-to-end

1438: distance,

1439: \begin{equation}

1440:   \label{eq:end-end-bulk}

1441:   \avg{\Delta \overline{R}_\pa}(t)\equiv\mint{ds}{0}{L}

1442:   \avg{\Delta\overline{\varrho}}\left[F(s,\tilde t

1443:     \leq t),t\right] \;,

1444: \end{equation}

1445: which is universal in the sense that it does not depend on the boundary conditions of

1446: $\vec r_\pe$. In Eq.~(\ref{eq:end-end-bulk}), and in the following, we neglect contributions

1447: of higher order than $\Delta\varrho=\Ord{\epsilon}$ to the projected end-to-end

1448: distance.   According to Eqs.~(\ref{eq:end-end-bulk},~\ref{eq:cg-eom}),

1449: the bulk change of the end-to-end distance is given by the arclength integral of

1450: the curvature of the time integrated tension $F(s,t)$,

1451: \begin{subequations}

1452:   \begin{align}

1453:     \label{eq:EED-from-tension}

1454:     \avg{\Delta \overline{R}_\pa}(t)&=\hat \zeta^{-1} \mint{ds}{0}{L}F''(s,t)\\

1455:     \label{eq:EED-from-tension-slopes}

1456:     &=\hat \zeta^{-1} \left[  F'(L,t)-F'(0,t) \right]\;.

1457:   \end{align}

1458: \end{subequations}

1459: 

1460: The predictions are presented separately for the regime of tension propagation ($t\ll

1461: \totpa$) and the regime of OPT ($t\gg t_\star$), where the tension profiles are flat

1462: and time-independent up to sub-leading terms. For weak forces, an explicit expression

1463: for times $t<t_\star$ is given in Sec.~\ref{sec:small-forces-crossover}, which

1464: captures the crossover from tension propagation to the tension saturated OPT regime.

1465: As discussed in Sec.~\ref{sec:homogeneous-relaxation}, {\lelo} turns out to also have

1466: an additional time-window $\totpa< t < t_\star$, whose consequences for the

1467: end-to-end distance is described in

1468: Sec.~\ref{sec:homogeneous-relaxation-e-e-distance}.  Since {\epull} is the only

1469: problem, in which both ends do not behave in the same way, we discuss this

1470: ``asymmetrical'' problem separately in Sec.~\ref{sec:e-e-epull}.

1471: 

1472: \subsection{Tension propagation regime ($t\ll \totpa$)}

1473: \label{sec:unrelaxed-tension}

1474: In the time domain of tension propagation, where the boundary layers are growing in

1475: from both ends ($t\ll \totpa$), we may use the tension profiles for the

1476: semi-infinite (pseudo-) polymer from Sec.~\ref{sec:scaling-forms} (labeled by

1477: $F_\infty$ here) to approximate Eq.~(\ref{eq:EED-from-tension-slopes}) by

1478: \begin{equation}

1479:   \label{eq:EED-from-tension-2}

1480:   \avg{\Delta \overline{R}_\pa}(t)\stackrel{t\ll \totpa}{\sim}-2\hat

1481:   \zeta^{-1} F_\infty'(0,t)

1482: \end{equation}

1483: for scenarios with \emph{two} equally treated ends.  Now, if $t$ falls into a regime

1484: where the tension exhibits scaling,

1485: \begin{equation}

1486:   \label{eq:F-scaling}

1487:   F_\infty(s,t)\propto t^{\alpha+1} \hat F_\infty\left(\frac s {t^z} \right) \;,

1488: \end{equation}

1489: we immediately obtain from Eq.~(\ref{eq:EED-from-tension-2}) the

1490: power-law 

1491: \begin{equation}

1492:   \label{eq:ee-growth-law-principle}

1493:   \avg{\Delta \overline{R}_\pa}(t)\propto - t^{\alpha+1-z} 2 \partial_{\xi} \hat

1494:   F_\infty(\xi=0)

1495: \end{equation}

1496: for the growth of the end-to-end distance. The pre-factors can be calculated for all

1497: cases, because the scaling functions are known. In this way we obtain the list

1498: Tab.~\ref{tab:ww-growth-laws} of growth laws.

1499: 

1500: \begin{table}

1501:   \caption{Universal bulk contribution $\avg{\Delta\overline{R}_\pa}(t)$ to the dynamic change

1502:     of the end-to-end distance  in the limit

1503:     $t\ll \totpa\ll t_L^\pe$.}

1504:   \label{tab:ww-growth-laws}

1505:   \begin{center}

1506:     \begin{tabular}{c|c|c}

1507:       Problem&$t\ll \tf$&$t\gg \tf$\\ \hline

1508:       {\pull}&${\displaystyle \frac{2^{5/8}}{\Gamma(15/8)}

1509:         \frac{\fe}{\sqrt{ \hat \zeta \lp }}t^{7/8}}$&${\displaystyle

1510:         \frac{4}{\sqrt{3}}

1511:         \left(\frac{2}{\pi}\right)^{1/4}\frac{\fe^{3/4}}{\sqrt{ \hat

1512:             \zeta \lp }}t^{3/4}}$\\

1513:       {\lelo}&$-{\displaystyle \frac{2^{5/8}}{\Gamma(15/8)}

1514:         \frac{\fe}{\sqrt{ \hat \zeta \lp }}t^{7/8}}$&${\displaystyle

1515:         - 2.477\frac{\fe^{1/4}}{\sqrt{ \hat   

1516:             \zeta \lp }}t^{1/2}}$\\

1517:     \end{tabular}

1518:   \end{center}

1519: \end{table}

1520: 

1521: 

1522: \subsection{OPT regime ($t\gg t_{\star}$)}

1523: \label{sec:tension-relaxed}

1524: When calculating the release of stored length in OPT the tension profile

1525: $\fopt=f(s=0)=\text{const.}$ is assumed to be stationary and flat, so that

1526: \begin{eqnarray}

1527:   \label{eq:OPT-Rpa}

1528:   &\avg{\Delta \overline{R}_\pa}(t)=-L \avg{\Delta \overline{\varrho}}

1529:   \left[ \fopt,t  \right]\;,\nonumber &\\

1530:    &\qquad\qquad\qquad \text{for }t\gg t_\star \;.&

1531: \end{eqnarray}

1532: In the case of {\lelo}, the quantity $\avg{\Delta \overline{\varrho}}(\fopt=0,t)$ has

1533: been explicitely calculated in Eq.~(\ref{eq:stole-lelo-OPT}).  For {\pull},

1534: $\avg{\Delta \overline{\varrho}}(\fopt=\fe,t)$ can be evaluated from

1535: Eq.~(\ref{eq:change-stored-length}) in a similar straightforward manner, because the

1536: tension is spatially constant. The corresponding growth laws are summarized in

1537: Tab.~\ref{tab:ww-growth-laws-2}.

1538: 

1539: Compared to the tension propagation regime, the growth laws in the OPT regime are

1540: slowed down, see Tabs.~\ref{tab:ww-growth-laws},~\ref{tab:ww-growth-laws-2}.

1541: Actually, for all cases except {\lelo} the corresponding growth laws obey 

1542: \[ \avg{\Delta \overline{R}_\pa}^{\text{OPT}}\propto \avg{ \Delta

1543:   \overline{R}_\pa}^{\text{MSPT}} t^{-z}\;.\] This can be understood in terms of the

1544: scaling arguments used Part~I.  There, we took tension propagation heuristically into

1545: account by assuming that the stored length release, as given by OPT, is restricted to

1546: the boundary layer of size $\lpa(t)$.  Then, one has 

1547: \begin{equation}

1548:   \label{eq:dr-estimate1}

1549:   \avg{\Delta \overline{R}_\pa}^{\text{OPT}}\simeq -\lpa(t) \avg{\Delta

1550:     \overline{\varrho}}(\fopt,t) \qquad \text{for }t\ll\totpa

1551: \end{equation}

1552: as compared to 

1553: \begin{equation}

1554:   \label{eq:dr-estimate1}

1555:   \avg{\Delta \overline{R}_\pa}^{\text{MSPT}}\simeq - L

1556: \avg{\Delta\overline{\varrho}}(\fopt,t) \qquad \text{for }t\gg t_\star \;.

1557: \end{equation}

1558: This conforms with the heuristic rule noticed above,

1559: \begin{equation}

1560:   \label{eq:OPT-MSPT}

1561:   \avg{\Delta \overline{R}_\pa}^{\text{OPT}}\approx

1562:   \frac{L}{\lpa(t)}

1563: \avg{\Delta \overline{R}_\pa}^{\text{MSPT}}

1564:   \sim t^{\alpha+1-2z}\;.

1565: \end{equation}

1566: 

1567: 

1568: 

1569: \begin{table}

1570:   \caption{Universal bulk contribution $\avg{\Delta\overline{R}_\pa}(t)$ to the dynamic change

1571:     of the end-to-end distance in the limit

1572:     $t_\star\ll t \ll t_L^\pe$.}

1573:   \label{tab:ww-growth-laws-2}

1574:   \begin{center}

1575:     \begin{tabular}{c|c|c}

1576:       Problem&$t\ll \tf$&$t\gg \tf$\\ \hline

1577:       {\pull}&${\displaystyle

1578:         \frac{L

1579:           \fe}{\lp\Gamma(7/4)}\left(\frac{t}{2}\right)^{3/4}}$~\cite{granek:97,gittes-mackintosh:98_pub}

1580:       &${\displaystyle

1581:         \frac{L}{\lp}\sqrt{\frac{2\fe t}{\pi}}}$\\

1582:       {\lelo}&$-{\displaystyle

1583:         \frac{L \fe}{\lp\Gamma(7/4)}\left(\frac{t}{2}\right)^{3/4}}

1584:       $ &${\displaystyle \frac{-2^{3/4}}

1585:         {\Gamma(1/4)} \frac{L}{\lp}t^{1/4}}$\\

1586:     \end{tabular}

1587:   \end{center}

1588: \end{table}

1589: 

1590: 

1591: \subsection{{\lelo} in the limit $\totpa\ll t\ll t_{\star}$}

1592: \label{sec:homogeneous-relaxation-e-e-distance}

1593: The intuitive rule in Eq.~(\ref{eq:OPT-MSPT}) fails for {\lelo} in the limit $t\gg\tf$,

1594: which indicates that this is an exceptional scenario.  As discussed in

1595: Sec.~\ref{sec:homogeneous-relaxation}, $t_\star=L^8/\lp^4$ cannot be identified with

1596: $\totpa=L^2 \lp^{-1} \hat\zeta \fe^{-3/2}$ in this case. There exists a time window

1597: $\totpa\ll t \ll t_\star$ that expands in the limit of large forces, $\fe\gg

1598: \lp^2/L^4$.  We have shown that the tension exhibits homogeneous relaxation in this

1599: novel regime.  With the slope of the tension at the ends

1600: \begin{equation}

1601:   \label{eq:lelo-tension-slopes}

1602:   \partial_s f\vert_{s=\left\{  {0 \atop L}\right\}}=\pm \frac 1 {16} \left( \frac 3 2

1603:   \right)^{2/3} \frac{1}{L}\left(\frac{\hat \zeta L^2}{\lp t}\right)^{2/3}

1604: \end{equation}

1605: the growth law 

1606: \begin{equation}

1607:   \label{eq:homogenous-lelo-3}

1608:   \avg{\Delta \overline{R}_\pa}(t)\simeq - 18^{1/3} \left(\frac{L t}{\hat \zeta

1609:       \lp^2}\right)^{1/3} 

1610:   \qquad (\lelo)  

1611: \end{equation}

1612: follows from Eq.~(\ref{eq:EED-from-tension-slopes}). We expect the growth law

1613: $\avg{\Delta \overline{R}_\pa}\propto t^{1/3}$ during homogeneous tension relaxation to hold even

1614: for chains with $L\gg\ell_p$. The example of retracting DNA will be discussed as an

1615: experimental outlook in Sec.~\ref{sec:possible-exps}. The exponent $1/3$ coincides

1616: with that obtained by an adiabatic application of the stationary force-extension

1617: relation~\cite{bustamante-marko-siggia-smith:94} to a

1618: ``frictionless''~\cite{bhobot-wiggins-granek:04} polymer with attached beads at its

1619: ends~\cite{Hallatschek:F:K::94:p077804:2005}.

1620: 

1621: For times $t\gg t_\star=L^8/\lp^4$ the growth law in Eq.~(\ref{eq:homogenous-lelo-3})

1622: crosses over to the one noted in Tab.~\ref{tab:ww-growth-laws-2}. Interestingly, both

1623: growth laws appearing for $t\gg \totpa$ are independent of the initial tension $\fe$.

1624: In both cases, the initial conditions are completely ``forgotten'' once the tension

1625: has propagated through the whole polymer. An overview over the time scales separating

1626: the diverse regimes for {\lelo} (as compared to {\pull}) will be given in

1627: Sec.~\ref{sec:lelo-vs-pull}.

1628: 

1629: 

1630: \subsection{{\pull} and {\lelo} for small forces}

1631: \label{sec:small-forces-crossover}

1632: Provided the external force is smaller than the critical Euler buckling force of the

1633: polymer, $\fe< L^{-2}$, the crossover time $\tf$ exceeds the terminal relaxation time

1634: $\totpe$, hence the linearized PIDE, Eq.~(\ref{eq:pide-linearized-lt-1}), applies

1635: throughout the contour relaxation. The linearity allows to solve the PIDE for a

1636: polymer of finite length, and we obtain an analytic description of the crossover

1637: between the asymptotic power laws $\Delta\overline{R}_\pa(t)\propto t^{7/8}$ for

1638: $t\ll t_L^\pa$ and $\Delta\overline{R}_\pa(t)\propto t^{3/4}$ for $t_\star\ll t \ll

1639: t_L^\pe$ (cf.~Tabs.~\ref{tab:ww-growth-laws},~\ref{tab:ww-growth-laws-2}). To this

1640: end, let us first reinstall original units (which are better adapted for the

1641: present purpose) into the linearized PIDE, Eq.~(\ref{eq:pide-linearized-lt-1}),

1642: \begin{equation}

1643:   \label{eq:PIDE-linear-units}

1644:   \partial_s^2 F(s,z)=\frac{z^{1/4} \hat \zeta}{2^{3/4} \lp} F(s,z)\;,

1645: \end{equation}

1646: where $F(s,z)$ is the Laplace transformation of the integrated tension,

1647: \begin{equation}

1648:   \label{eq:Laplace trafo}

1649:   F(s,z)\equiv \mint{dt}{0}{\infty}e^{-z t}F(s,t)\;.

1650: \end{equation}

1651: For the boundary conditions of {\pull}, Eqs.~(\ref{eq:bc-infty},

1652: \ref{eq:bc-origin-1}), the solution to Eq.~(\ref{eq:PIDE-linear-units}) is given by

1653: \begin{equation}

1654:   \label{eq:weak-pulling-finite-polymer}

1655:   F(s,z)=\fe z^{-2}\frac{\cosh\left[(\hat \zeta/\lp)^{1/2} (z/8)^{1/8} (s-L/2)

1656:     \right]} {\cosh\left[(\hat \zeta/\lp)^{1/2} (z/8)^{1/8} L/2  \right]}

1657: \end{equation}

1658: This implies a growth of the end-to-end distance of

1659: \begin{equation}

1660:   \label{eq:eed-weak-forces-finitep-laplace}

1661:   \begin{split}

1662:     \avg{\Delta \overline R_\pa}(z)&=-2\hat \zeta^{-1} \partial_s F|_{s=0} \\

1663:     &=\frac{2^{5/8}\fe}{z^{15/8} (\hat \zeta\lp)^{1/2}}\tanh\left[ \left( \frac{\hat

1664:           \zeta}{\lp} \right)^{1/2}\left( \frac{z}{8} \right)^{1/8} \frac{L}{2}

1665:     \right]\;.

1666:   \end{split}

1667: \end{equation}

1668: The Laplace back-transform of Eq.~(\ref{eq:eed-weak-forces-finite-polymer}) takes

1669: the form 

1670: \begin{equation}

1671:   \label{eq:eed-weak-forces-finite-polymer}

1672:   \begin{split}

1673:   \avg{\Delta \overline R_\pa}(t)&=\mint{\frac{dz}{2\pi

1674:       i}}{-i\infty+\epsilon}{i\infty+\epsilon} e^{z t}   \avg{\Delta \overline

1675:     R_\pa}(t)\\

1676:   &=\frac{\fe \hat \zeta^3 L^7}{\lp^4} \Upsilon\left[ \left( \frac{t}{L^4}  \right)

1677:     \left( \frac{\lp}{\hat \zeta L}  \right)^4 \right]

1678:   \end{split}

1679: \end{equation}

1680: where $\Upsilon(\tau)$ is a scaling function, given by

1681: \begin{equation}

1682:   \label{eq:Upsilon}

1683:   \begin{split}

1684:     \Upsilon(\tau)&= \frac{2^{5/8}}{\pi}\mint{ds}{0}{\infty}\frac{e^{-x

1685:         \tau}-1}{x^{15/8}}\\

1686:     &\times \Im\left[ e^{-i\frac 78 \pi} \tanh\left( 2^{-11/8}x^{1/8}e^{i

1687:           \pi/8} \right) \right]

1688:   \end{split}

1689: \end{equation}

1690: We have depicted $\Upsilon(\tau)$ in a double logarithmic plot in

1691: Fig.~\ref{fig:pulling-finite-polymer-weak-forces} together with the corresponding

1692: asymptotics from Tabs.~\ref{tab:ww-growth-laws},~\ref{tab:ww-growth-laws-2}. Note

1693: that the time $\tau_c\approx 6.55 \times 10^{-4}$ where the asymptotic lines cross is

1694: a good indication for the crossover occurring in the exact solution. Assuming that

1695: this also holds quite generally, it is possible to obtain estimations of when the

1696: crossover between linear and nonlinear regimes should occur in the natural time units

1697: $\tf$. To this end, one simple equates the corresponding asymptotic power laws

1698: (including the exact pre-factors) for the growth of the end-to-end distance.  As for

1699: the present case with $\tau_c\approx 6.55 \times 10^{-4}$, these crossover times are

1700: typically not of order one in natural units because of the numerical proximity of the

1701: exponents of the asymptotic power laws, which are $7/8$ and $3/4$ in the present

1702: case. In a given experimental situation, one should therefore check carefully which

1703: regime is expected by comparing experimental time scales with these unusual crossover

1704: times.

1705: \begin{figure}

1706:   \includegraphics[width=\columnwidth]{pulling-finite-polymer-psfraged.eps}

1707:   \caption{The scaling function $\Upsilon(\tau)$ in a double logarithmic plot. The

1708:     crossover of the scaling behavior is close to the time $\tau_c =6.55\times

1709:     10^{-4}$ where the asymptotic short-(blue) and longtime (red) asymptotics

1710:     cross.}

1711:   \label{fig:pulling-finite-polymer-weak-forces}

1712: \end{figure}

1713: 

1714: 

1715:  

1716: \subsection{FDT and linear response}

1717: \label{sec:tens-fluct}

1718: According to the fluctuation-dissipation theorem

1719: (FDT~\cite{kubo83bookI,kubo91bookII}), the fluctuations in the end-to-end distance

1720: should be related to the linear response of the polymer. It is tempting to interpret

1721: the linear short-time regime, where $\avg{\Delta \overline{R}}\simeq \fe t^{7/8}(\hat

1722: \zeta \lp)^{-1/2}$, as linear response in the usual sense, in which case the

1723: fluctuations of the end-to-end distance should, according to the FDT, scale as

1724: $\avg{\Delta \overline{R}^2}\simeq t^{7/8}{\hat \zeta}^{-1/2} \lp^{-3/2}$ in

1725: equilibrium.  However, fluctuations of this strength are inconsistent with a

1726: deterministic tension dynamics, because they generate friction forces over a scale

1727: $\lpa(t)\simeq t^{7/8}(\lp/\hat \zeta)^{1/2}$ and thus imply tension fluctuations of

1728: magnitude

1729: \begin{equation}

1730:   \label{eq:tens-fluct}

1731:  \delta f\simeq \hat \zeta  \lpa(t) \avg{\Delta  \overline{R}}/t\simeq

1732:  (\zeta/\lp)^{1/4} t^{-7/16}

1733: \end{equation}

1734: Tension fluctuations exceed the applied force in magnitude at any given time for

1735: small enough external forces. Hence, in the limit $\fe\to 0$ while $\epsilon\ll1$ is

1736: fixed, the tension cannot be considered as a deterministic quantity. Recall however,

1737: that our MSPT analysis in Part~I was based on the limit $\epsilon\to 0$ while $\fe$

1738: is fixed, and only in this limit the self-averaging argument of Part~I applies.

1739: Extending our results to the usual linear response limit corresponds to the

1740: uncontrolled approximation $f \rpe'^2\to f\avg{\rpe'^2}$. To analyze this limit more

1741: carefully, one has to solve the stochastic PIDE that obtained in Part~I before the

1742: self-averaging argument was employed.  However, since the $7/8-$scaling of the

1743: fluctuations has already been confirmed in simulations~\cite{everaers-Maggs:99}, we

1744: expect that such a more rigorous analysis would yield the same scaling but a

1745: pre-factor different from the one of the deterministic short-time law

1746: (Tab.~\ref{tab:ww-growth-laws}).

1747: 

1748: 

1749: \subsection{{\lelo} versus {\pull} (overview)}

1750: \label{sec:lelo-vs-pull}

1751: The diverse regimes and their range of validity are summarized in

1752: Fig.~\ref{fig:lelo-vs-pull} for the {\pull} and the complementary

1753: {\lelo} problem. It depicts the crossover time scales as a function of

1754: the externally applied tension $\fe$. The line $\tf\simeq \fe^{-2}$

1755: separates ``linear'' from ``nonlinear'' behavior. The symmetry of the

1756: graph for $t<\tf$ with respect to the $\fe=0$-axis indicates that the

1757: scenario of {\pull} can indeed be considered as the inverse scenario

1758: of {\lelo} for weak forces $f\ll f_c$, or, more generally, on short

1759: times. This symmetry is lost in the nonlinear regime. The growing

1760: importance of uniform tension relaxation for {\lelo} with increasing

1761: initial tension $\fe$ becomes particularly apparent on the log-scale

1762: of the figure. Due to the particular choice of units ($t/L^4$ and

1763: $f/L^{-2}$) the regimes of nonlinear growth of the boundary layers

1764: appear relatively narrow in Fig.~\ref{fig:lelo-vs-pull}. Which of the

1765: various regimes will pre-dominantly be observed in measurements

1766: actually depends strongly on the ratio $L/\lp$ and on the

1767: experimentally accessible time scales.

1768: \begin{figure}

1769:   \includegraphics[width=\columnwidth]

1770:   {lelo-vs-pulling-sregion-psfraged.eps}

1771:   \caption{Characteristic times (logarithmic scale) for {\pull} and {\lelo} against

1772:     the applied external force $\fe$ (linear scale). The time $t_\star$ (stars)

1773:     separates regions where ordinary perturbation theory (OPT) applies (dark shaded)

1774:     from regions (light shaded) of linear (hatched) and nonlinear tension propagation

1775:     and from homogeneous tension relaxation (white). Whereas longitudinal friction is

1776:     negligible for $t>t_\star$, it limits the dynamics for $t<t_\star$. The innermost

1777:     funnel-shaped region indicates the regime where the tension fluctuations are

1778:     important (defined by $f<\delta f$ with $\delta f$ given by

1779:     Eq.~(\ref{eq:tens-fluct})).}

1780:    \label{fig:lelo-vs-pull}

1781: \end{figure}

1782: 

1783: 

1784: \subsection{\epull}

1785: \label{sec:e-e-epull} {\epull} was excluded from the preceeding discussion because it

1786: is the only ``asymmetric'' problem, as both ends of the polymer behave differently.

1787: In the tension propagation regime ($t\ll\totpa$), the left end is constrained to move

1788: with constant velocity, while the right end experiences no driving force.  Hence, we

1789: have

1790: \begin{equation}

1791:   \label{eq:epull-growth-law}

1792:   \avg{\Delta \overline{R}_\pa}(t)=v t \;,\qquad \text{for } t\ll\totpa\;.

1793: \end{equation}

1794: After the boundary layer of non-zero tension has reached the free end, $t\gg \totpa$,

1795: the right end starts to move. Then, the tension profile becomes linear as for a

1796: straight rod dragged through a viscous solvent. The further contour relaxation is up

1797: to pre-factors identical to the {\pull} problem for $t\gg \totpa\approx t_\star$.

1798: 

1799: \subsection{\push}

1800: \label{sec:pushing}

1801: 

1802: For completeness, we mention the scenario of a filament being compressed by external

1803: longitudinal forces. This scenario has some subtleties.  {\push} increases the stored

1804: length exponentially for $t\gg \tf$ by virtue of the Euler Buckling instability and

1805: generates a situation where the weakly bending approximation is not valid anymore.

1806: Then, hairpins are generated~\cite{ranjith:02} and for $t\gg t_f$ the rigidly

1807: oriented driving forces \emph{pull} on those hair pins. Our theory is only applicable

1808: at short times. For $t\ll t_f$ the response of the system is linear in the driving

1809: force, irrespective of the sign.

1810: 

1811: 

1812: \subsection{Boundary effects}

1813: \label{sec:microscale}

1814: Up to now, we only discussed the change $\avg{\Delta \overline{R}_\pa}(t)$ in

1815: projected end-to-end distance corresponding to the change

1816: $\avg{\Delta\overline{\varrho}}$ of the stored length in the bulk. For a hinged,

1817: respectively, clamped semi-infinite polymer, the as yet missing boundary

1818: contribution $X^{h/c}(t)\equiv\avg{\Delta R^{h/c}_\pa}-\avg{\Delta\overline{R}_\pa}$

1819: is given by

1820: \begin{eqnarray}

1821:   \label{eq:bc-contri-1}

1822:    && X^{h/c}(t)=\pm2\mint{ds}{0}{\infty}\mint{\frac{dq}{\pi\lp}}{0}{\infty}

1823:   \left\{\frac{e^{-2q^2[q^2 t+ F(s,t)]}-1}{q^2+f_<}\right. \nonumber\\ 

1824:     &&+\left. 2 

1825:     q^2\mint{d\tilde t}{0}{t}e^{-2q^2\left[q^2(t-\tilde

1826:         t)+F(s,t)-F(s,\tilde t)\right]}\right\}\cos(2 q s) \;.\nonumber\\

1827: \end{eqnarray}

1828: As discussed in Part~I, the boundary dependent term

1829: of the stored length decays on a length scale of $\Ord{1}$ due to the cosine factor.

1830: Since the tension decays on a much larger length scale of order

1831: $\Ord{\epsilon^{-1/2}}$, it is permissible to use the (integrated) tension at $s=0$

1832: to evaluate the arclength integral in Eq.~(\ref{eq:bc-contri-1}).

1833:  

1834: Upon replacing $F(s,t)\to F(0,t)$ and using 

1835: \[\mint{ds}{0}{\infty}\cos(2 q s)=(\pi/2)\delta(q)\;,\] the integral in Eq.~(\ref{eq:bc-contri-1}) can be evaluated,

1836: \begin{equation}

1837:   \label{eq:Rpa-boundary}

1838:   X^{h/c}(t)=\pm\lim_{q\to 0}\frac 2 {\lp} \frac{q^2(q^2 t +F(0,t))}{q^2+f_<}\;,

1839: \end{equation}

1840: which vanishes, unless

1841: \begin{equation}

1842:   \label{eq:Rpa-boundary2}

1843:   f_<=0 \quad \Rightarrow \quad X^{h/c}(t)=\pm\frac {2F(0,t)} {\lp}\;.

1844: \end{equation}

1845: On the semi-infinite arclength interval, these boundary contributions are therefore

1846: nonzero only for {\pull} and {\epull}, and are summarized in

1847: Tab.~\ref{tab:tab:bc-effects}. Note that, the boundary effects of clamped (hinged)

1848: ends tend to reduce (increase) the longitudinal response of the polymer in comparison

1849: to the bulk response.  This may be explained as follows. Close to a clamped end, a

1850: polymer is more stretched out than in the bulk because $\vec r_\pe'^2$ is constraint

1851: to approach zero at the end. As a consequence, the end portion of the polymer is less

1852: able to store or release excess length. For hinged ends, the boundary conditions act

1853: just in the reverse direction.

1854: 

1855: \begin{table}

1856:   \caption{Boundary contribution to the change $\avg{\Delta R^{h/c}_\pa}(t)$ of the end-to-end distance for

1857:     hinged, respectively, clamped ends in the limit $t_L^\pe \ll t$.} 

1858:   \label{tab:tab:bc-effects}

1859:   \begin{tabular}{c|c}

1860:     Problem&$\avg{\Delta R^{h/c}_\pa}(t)-\avg{\Delta \overline{R}_\pa}(t)$\\ \hline

1861:     {\pull}&$\pm 2t \fe/\lp$\\

1862:     {\epull}&$\pm 2t \fe/\lp$\\

1863:     {\lelo}&$\left\{ {\pm 2t \fe/\lp\text{, for }f_<\ll L^{-2}  \atop 

1864:         0 \text{, for }f_<\gg L^{-2}}

1865:     \right.$ \\

1866:   \end{tabular}

1867: \end{table}

1868: 

1869: 

1870: The strict vanishing of the boundary term for any finite $f_<$ and the discontinuity

1871: at $f_<=0$ is a consequence of the assumed infinite half-space. For a polymer of

1872: finite length, it turns out that the boundary term approaches zero for

1873: pre-stretching forces larger than the critical Euler-Buckling force, $f_<\gg

1874: f_c\equiv L^{-2}$. This can be seen by studying the finite integral

1875: $\mint{ds}{0}{L}(\avg{\Delta \varrho^{h/c}}(s,t)-\avg{\Delta \overline{\varrho}}(s,t))$. For

1876: forces $f_<\ll f_c$, the integrand saturates at a plateau of magnitude

1877: $\Ord{f_<^{3/2}t/\lp }$ for $\lpe(t)<s<f_<^{-1/2}$ before it finally decays to zero.

1878: Within the semi-infinite integral, the integral over this long plateau cancels the

1879: contribution stemming from $s<\lpe(t)$, as required by Eq.~(\ref{eq:Rpa-boundary}).

1880: However, for $L\ll f_<^{-1/2}$ the contribution from the plateau may be neglected. As

1881: a consequence, the value of the integral is for $f_<\ll f_c$ given by $\pm 2f_<

1882: t/\lp$.  This asymptotic behavior is important for {\lelo}, as noted in

1883: Tab.~\ref{tab:tab:bc-effects}.  

1884: 

1885: 

1886: 

1887: 

1888: 

1889: 

1890: Upon comparing Tab.~\ref{tab:tab:bc-effects} with Tabs.~\ref{tab:ww-growth-laws},

1891: \ref{tab:ww-growth-laws-2}, one may think that boundary effects are always

1892: subdominant in the short time limit.  However, our calculations were specialized to

1893: hinged or clamped boundary conditions.  In many experimental situations, one has to

1894: deal with \emph{free} boundary conditions.  Somewhat tedious but rigorously, these

1895: boundary conditions can be taken into account by means of the correct susceptibility,

1896: which can only be given in terms of an integral. Here, we discuss effects related to

1897: free boundary conditions on a heuristic basis and show that they generate a

1898: \emph{dominant} contribution to the change in the end-to-end distance for {\pull},

1899: which is proportional to $t^{3/4}$.

1900: 

1901: 

1902: 

1903: 

1904: 

1905: The argument is based on the observation, that external forces that act in the

1906: longitudinal direction ($\pa$) while the polymer is free, automatically introduce

1907: small (of order $\Ord{\epsilon^{1/2}}$) transverse forces at the ends.  This follows

1908: from the force balance equation of a semiflexible polymer (cf.~Part~I)

1909: \begin{equation}

1910:   \label{eq:force-balance}

1911:    \kappa\, \vec{r}''' +\fel =  f\, \vec{r}' \;,

1912: \end{equation}

1913: where $\fel(s)$ is the elastic force acting at arc length $s$. At the ends, where

1914: $\fel$points in longitudinal direction to cancel the external pulling force, one obtains

1915: from the projection of Eq.~(\ref{eq:force-balance}) onto the transverse axis

1916: \begin{equation}

1917:   \label{eq:perp-endforces}

1918:   \vec r_\pe'''=f \vec r_\pe'=\fe \vec r_\pe'+\Ord{\epsilon} \qquad \text{(at the ends)} \;.

1919: \end{equation}

1920: Thus, $\vec r_\pe'''$ is typically non-zero at the ends because the slope $\vec

1921: r'=\Ord{\epsilon^{1/2}}$ fluctuates.  This, however, corresponds to a transverse

1922: force at the end, which results in a transverse deformation. The corresponding bulge

1923: of contour is only visible on the micro-scale because it spreads with the transverse

1924: correlation length $\lpe(t)\simeq t^{1/4}$.  Nevertheless this deformation may

1925: dominate the growth of the end-to-end distance, as is demonstrated for {\pull}: With

1926: the transverse bulk susceptibility scaling as $\overline{\chi}(t)\simeq t^{-1/4}$ we

1927: estimate the magnitude of transverse deformation $\Delta r_\pe(s=0,t)\equiv r_\pe

1928: (0,t)-r_\pe(0,0)$ induced by a transverse force of magnitude

1929: $\ord{\fe\epsilon^{1/2}}$ by 

1930: \begin{equation}

1931:   \label{eq:tdispl-estimate}

1932:   \ord{\avg{\abs{\Delta r_\pe(0,t)}}}\simeq \fe

1933:   \epsilon^{1/2} t \, \overline{\chi}(t)\simeq \epsilon^{1/2} t^{3/4}\fe \;.

1934: \end{equation}

1935: The displacement of the end couples to the projected length because of the mismatch

1936: of the end-tangent with the $\pa$-axis by a small angle of typical magnitude

1937: $\avg{\abs{r_\pe'(s=0)}}\simeq \ord{\epsilon^{1/2}}$. The expected growth of

1938: end-to-end distance due to this effect is therefore estimated by

1939: \begin{equation}

1940:   \label{eq:micro-fluctuations}

1941:   \begin{split}

1942:     &\avg{\Delta R^{h/c}_\pa}(t)-\avg{\Delta

1943:       \overline{R}_\pa}(t)\simeq\\&

1944:     \avg{\abs{r_\pe'(s=0)}} \times\avg{\abs{\Delta

1945:       r_\pe(0,t)}}\simeq\epsilon t^{3/4}

1946:     \fe=\frac{L}{\lp}t^{3/4}\fe  \;.

1947:   \end{split}

1948: \end{equation}

1949: This dominates over the growth law of the bulk, which scales like

1950: $t^{7/8}/\sqrt{\lp}$ on short times. The only way to avoid the outlined effect is to

1951: apply the external force strictly tangentially to the end-tangents, which is however

1952: somewhat unrealistic.  The same problem will experimentally arise in the short time

1953: limit of {\lelo} if the pre-stretching force was not applied strictly tangentially.

1954: However, the long-time limits are unaffected by this subtlety because the bulk

1955: contributions dominate over contributions from the end.

1956: 

1957: To our knowledge, these end-effects have so far masked the subdominant

1958: $t^{7/8}$-contribution in experiments that monitored the time-dependent end-to-end

1959: distance (we note that in Ref.~\cite{legoff-amblard-furst:02} the $7/8$-scaling is

1960: inferred from a corresponding scaling of the measured shear modulus of an active

1961: gel).  As we outline in Sec.~\ref{sec:smfs}, force spectroscopy, on the contrary, may

1962: allow to measure the tension dynamics, which is itself truly \emph{independent} of

1963: the boundary conditions imposed on the transverse displacements.

1964: 

1965: 

1966: 

1967: 

1968: 

1969: 

1970: 

1971: \section{Suggestions for experiments}

1972: \label{sec:possible-exps}

1973: While many experiments have been done concerning flexible polymers in external force

1974: fields~\cite{perkins-smith-chu:97,quake-babcock-chu:97}, the available measurements

1975: on driven stiff or pre-stretched polymers is not sufficient to verify our

1976: predictions.  Most of these experiments have monitored the transverse and

1977: longitudinal response on intermediate times where OPT is valid

1978: (e.g.~\cite{legoff-hallatschek-frey:02}). Investigations concerning the longitudinal

1979: short-time dynamics are

1980: scarce~\cite{bhobot-wiggins-granek:04,legoff-amblard-furst:02,maier-seifert-raedler:02}.

1981: In the following, we propose several assays that might be able to fill this gap.

1982: 

1983: To facilitate the application of our predictions, we reintroduce the parameters

1984: $\kappa$ and $\zeta$ for the following.

1985: 

1986: \subsection{Strongly stretched DNA}

1987: \label{sec:dna-exp}

1988: 

1989: The experimental verification of most of our results requires stiff

1990: polymers with a total length much smaller than their persistence

1991: length.  A remarkable exception is {\lelo} in the regime of

1992: homogeneous tension relaxation, which was discussed in

1993: Sec.~\ref{sec:homogeneous-relaxation}. For polymers with $L\gg\lp$,

1994: like a typical DNA molecule, this novel regime appears if the

1995: pre-stretching force obeys

1996: \begin{equation}

1997:   \label{eq:lelo-exp-large-forces}

1998:   \fe\gg \fe_{\lp}\equiv \kappa \lp^{-2}=\frac{k_B T}{\lp} \;, 

1999: \end{equation}

2000: where $\fe_{\lp}$ is the Euler buckling force corresponding to the

2001: buckling length $\lp$. For forces much larger than the cross-over

2002: force $\fe_{\lp}$, the polymer may be considered as weakly bending.

2003: For those long, but strongly stretched polymers our analysis of

2004: {\lelo} predicts the following.

2005: 

2006: After the stretching force has been released the tension first

2007: propagates through the filament. This

2008: regime~\cite{brochard-buguin-de_gennes:99,Hallatschek:F:K::94:p077804:2005}

2009: ends at the time

2010: \begin{equation}

2011:   \label{eq:lelo-exp-crossover}

2012:   \totpa=\frac{\zeta_\pa L^2 \lp}{k_b T}\left( \frac{\fe}{\fe_{\lp}}

2013:   \right)^{-3/2}\equiv \tau_{R} \left( \frac{\fe}{\fe_{\lp}}  \right)^{-3/2}

2014: \end{equation}

2015: when the tension has to propagated through the filament. Here, $\tau_{R}\equiv

2016: \zeta_\pa L^2 \lp /(k_B T)$ scales like the longest relaxation time of a Rouse chain

2017: with segment length $\lp$. The characteristic time $\totpa\ll \tau_{R}$ marks the

2018: cross-over to a regime where the tension profile is roughly parabolic and slowly

2019: decays according to the power law $f(t)\propto t^{-2/3}$. This is associated with the

2020: projected length $R_\pa$ growing like

2021: \begin{equation}

2022:   \label{eq:lelo-exp-growth-law}

2023:   \avg{\Delta R_\pa(t)}\equiv \avg{R_\pa(t)-R_\pa(0)}=

2024:   18^{1/3} L \left( \frac{t}{\tau_R}

2025:   \right)^{1/3} \;. 

2026: \end{equation}

2027: 

2028: The above analysis strictly holds for the portion of the polymer that

2029: stays weakly bending. This is not the case at the boundaries, for

2030: which we refer to existing theories. According to the stem-flower

2031: model of Refs.~\cite{brochard:93,brochard:95,manneville:96} the

2032: boundaries will develop in time $t$ a ``flower'' of arc length

2033: \begin{equation}

2034:   \label{eq:flower-length}

2035:   \lflower(t)\simeq L \left( \frac{t}{\tau_{R}}  \right)^{1/2}  \;,

2036: \end{equation}

2037: thereby reducing the end-to-end distance by an amount of the same order of

2038: magnitude as $\lflower$ itself. It is seen, that for $t\ll \tau_R$ the shrinkage of

2039: the end-to-end distance due to the flower is much smaller than that due to the

2040: weakly bending part of the polymer (the stem), Eq.~(\ref{eq:lelo-exp-growth-law}).

2041: 

2042: Thus, the evolution of the end-to-end distance should be described

2043: by Eq.~(\ref{eq:lelo-exp-growth-law}) even for flexible polymers if

2044: the pre-stretching force is large enough.  Since DNA can be stretched

2045: by very large forces without un-zipping or destroying the covalent

2046: bonds, we think that the scaling $\avg{\Delta R(t)}\propto t^{1/3}$

2047: should be  visible in a {\lelo}-experiment with DNA. The relevant

2048: quantities in an experiment with $\lambda$-phage DNA (as in

2049: Ref.~\cite{maier-seifert-raedler:02}) in aqueous solution would be

2050: \begin{subequations}

2051:   \label{eq:lelo-exp-numbers}

2052:   \begin{align}

2053:     \lp&\approx 50\text{nm} \nonumber \\

2054:     L&\approx 20 \mu\text{m}  \nonumber\\

2055:     a&\approx 2 \mu\text{m}  \qquad \text{(thickness)}\nonumber\\

2056:     \zeta_\pe&\approx 4 \pi \eta/\log(L/a)\approx 1.3 \times 10^{-3}

2057:     \text{Pa s}  \nonumber\\

2058:     \fe_{\lp}&\approx 0.08 \text{pN}  \nonumber\\

2059:     \tau_R&\approx 7 \text{s} \;.  \nonumber

2060:   \end{align}

2061: \end{subequations}

2062: Though it might require extreme conditions to reach the asymptotic limit in

2063: Eq.~(\ref{eq:lelo-exp-growth-law}), any experiment with finite pre-stretching forces

2064: larger than $f_c$ should be suitable for deciding the question whether or not the

2065: stem dominates the retraction.  One would then compare the data with a numerical

2066: solution~\cite{obermayer-kroy-frey-hallatschek:tbp} of the governing equation,

2067: Eq.~(\ref{eq:cg-eom}), which describes the stress relaxation in stiff worm-like

2068: chains.

2069: 

2070: 

2071: 

2072: \subsection{Single molecule force spectroscopy}

2073: \label{sec:smfs}

2074: 

2075: \subsubsection{{\epull}}

2076: 

2077: As we already mentioned Sec.~\ref{sec:tensprop-summary}, {\epull} offers the possibility

2078: to measure the tension propagation by a force measurement.  The experimental idea is

2079: illustrated in Fig.~\ref{fig:tweezer} (a). The polymer's left end is suddenly pulled,

2080: e.g., by an optical tweezer whose focus moves with constant velocity. As a

2081: consequence the pulled end follows the laser beam with (almost~\footnote{Due to the

2082:   finite stiffness of the harmonic Laser potential, the trapped end will not be moved

2083:   with \emph{exactly} the same velocity as the trap during a transient, in which the

2084:   end approaches its steady state location within the Laser potential.})  constant

2085: velocity. By measuring the deflection of the trapped end from the center of the beam

2086: one can, in principle, extract the pulling force $\fe(t)$ and thus the size

2087: $\lpa(t)\simeq \fe(t)/(\hat \zeta v)$ of the boundary layer.  However, since the

2088: laser beam is moving, it might represent some problems to dynamically extract the

2089: deflection.

2090: 

2091: \begin{figure}

2092:   \centerline{\includegraphics[width=.7\columnwidth]

2093:     {tweezer1-psfraged.eps}}

2094:   \caption{Two possible realizations of {\epull}. In (a) the polymer's left end is

2095:     being dragged through the solvent with constant velocity $\vec v$, whereas in (b)

2096:     the optical tweezer is immobile while the solvent flows with constant velocity

2097:     $\vec v$. In both experiments, the length $\lpa(t)$ of the boundary layer is

2098:     derived from the pulling force of the tweezer, which can be inferred from the

2099:     displacement of the end within the focus of the tweezer.}\label{fig:tweezer}

2100: \end{figure}

2101: 

2102: 

2103: A solution to the latter problem is suggested by the following Gedanken experiment.

2104: Consider the above realization of {\epull} in the coordinate frame co-moving with

2105: the left tip, as illustrated in Fig.~\ref{fig:tweezer}(b). Then, it seems as if the

2106: bulk of the polymer was dragged by a homogeneous force field to the right while the

2107: left end is held fixed by the optical tweezer, just as if the optical tweezer was

2108: spatially fixed while the solvent was homogeneously flowing to the right. In fact,

2109: from the polymer's perspective there is no difference between both experiments. Thus,

2110: we propose to graft one end of a polymer by a tweezer or by the cantilever of an

2111: atomic force microscope.  Then, a homogeneous force field (electric field or fluid

2112: flow) is suddenly turned on that pulls the bulk of the polymer to the right.  The

2113: deflection of the tip gives the pulling force and hence the length of the boundary

2114: layer.

2115: 

2116: 

2117: \subsubsection{Onset of the nonlinear regime}

2118: 

2119: One would like to estimate typical time sales for the diverse regimes

2120: of tension propagation and relaxation. As we have seen above, most of

2121: the time-scales crucially depend on the total length of the polymer,

2122: e.g., $\lpa(t)\propto L^8$ in the linear regime. This high tunability

2123: is of experimental advantage because one can adjust the setup to the

2124: observable time scales. On the other hand, it forbids to give

2125: \emph{typical} time-scales for those quantities.

2126: 

2127: 

2128: \begin{table}

2129:   \caption{Threshold values $I_c=\sqrt{\kappa \zeta_\pe}$ for diverse biopolymers.}

2130:   \label{tab:threshold-ic}

2131:   \begin{center}

2132:     \begin{tabular}{c|c}

2133:       Polymer&$I_c$\\ \hline

2134:       microtubuli & $1.6 \times 10^2$ pN\,ms$^{1/2}$ \\ 

2135:       F--actin & $10$ pN\,ms$^{1/2}$\\

2136:       Intermediate fil.s & $4$ pN\,ms$^{1/2}$ \\

2137:       DNA & $0.9$ pN\, ms$^{1/2}$  \\

2138:     \end{tabular}

2139:   \end{center}

2140: \end{table}

2141: 

2142: 

2143: However, there are characteristic quantities that do not depend on the

2144: length of the polymer. The probably most interesting one is the

2145: threshold to the nonlinear regime: when the product of driving force

2146: and the square root of the applied time exceeds a certain value $I_c$

2147: the polymer response becomes nonlinear,

2148: \begin{equation}

2149:   \label{eq:nonlinear regime}

2150:   \fe \sqrt{t} \gg I_c\equiv\sqrt{\kappa \zeta_\pe}=\sqrt{k_B T \lp \zeta_\pe}

2151:     \;.

2152: \end{equation}

2153: With persistence lengths of $\lp=7$mm, $17\mu$m, $2\mu$m, $50$nm for

2154: microtubuli\cite{mickey-howard:95,FrancescoPampaloni07052006},

2155: F-actin~\cite{legoff-hallatschek-frey:02}, intermediate filaments and

2156: DNA~\cite{bustamante-marko-siggia-smith:94}, respectively, and corresponding friction

2157: coefficients $\zeta_\pe\approx 4 \pi \eta /\ln(\lp/a)$ (transverse friction

2158: coefficient per length of a rod of length $\lp$), we have evaluated $I_c$ for some

2159: common biopolymers, see Tab.~\ref{tab:threshold-ic}.  Those values can be used to

2160: decide whether the response of a given biopolymer under a ``typical'' time-dependent

2161: external longitudinal force is predominantly nonlinear or linear.  During a power

2162: stroke, for instance, the molecular motor myosin exerts a force of about $5$pN on an

2163: actin filament during a time of roughly $1$ms (cycle time of the power stroke).

2164: Hence, the impulse of about $0.5 I_c$ is somewhat smaller than $I_c$ for actin, so

2165: that the actin response should be linear.

2166:  

2167: 

2168: 

2169: 

2170: 

2171: \section{Summary}

2172: \label{cha:Summary}

2173: 

2174: In this paper, we have studied the tension dynamics of a weakly bending semiflexible

2175: polymer in a viscous fluid theoretically. Starting from the coarse-grained equation

2176: of motion for the tension, Eq.~(\ref{eq:pide}), we elaborated the non-linear

2177: longitudinal dynamic response to various external perturbations (mechanical

2178: excitations, hydrodynamic flows, electrical fields \dots) that can be represented as

2179: sudden changes of boundary conditions.  For the various scenarios we identified

2180: two-parameter scaling forms, that capture the crossover from linear to nonlinear

2181: tension dynamics.  In the limit of large and small arguments, where the equilibrium

2182: structure of the polymer is self-affine, they were shown to reduce to one-parameter

2183: scaling forms, which could be calculated analytically in most cases.  The growth law

2184: $\lpa(t)\sim t^z$ of the tension profiles could be inferred from the scaling

2185: variables of the respective scenarios. This enabled us to develop a unified theory of

2186: tension propagation. Not only does it contain all cases (correctly) studied in the

2187: literature so far.  It also identifies their ranges of validity and provides new

2188: predictions.  The recovered known results and our new predictions are summarized in

2189: Figs.~\ref{fig:lf},~\ref{fig:lelo-vs-pull} and

2190: Tabs.~\ref{tab:lf-growth-laws},~\ref{tab:ww-growth-laws} and

2191: \ref{tab:ww-growth-laws-2}.  

2192: 

2193: Various dynamic regimes should be well realizable for certain biopolymers.  A novel

2194: regime of homogeneous tension relaxation is a particularly remarkable result from the

2195: experimental point of view (Sec.~\ref{sec:dna-exp}).  In contrast to previous

2196: expectations, this new regime is predicted to dominate the relaxation of strongly

2197: stretched DNA. Moreover, it is an intriguing question, whether the tension

2198: propagation laws $\ell_\|(t)$ govern mechanical signal transduction through the

2199: cytoskeleton~\cite{shankar-pasquali-morse:2002,GardelVCBW03}. We expect that the

2200: force spectroscopical methods, proposed in Sec.~\ref{sec:smfs}, might be helpful to

2201: answer these questions.

2202: 

2203: Inclusion of hydrodynamic interactions merely produce logarithmic corrections but

2204: would give rise to more interesting effects for polymerized membranes to which our

2205: discussion could be generalized with otherwise little change. Other natural

2206: generalizations including the transverse nonlinear response of

2207: polymers~\cite{obermayer-hallatschek2:tbp}, quenches in the persistence

2208: length~\cite{obermayer-kroy-frey-hallatschek3:tbp} and more complex force

2209: protocols~\cite{obermayer-hallatschek-frey-kroy:tbp} are currently also under

2210: investigation.

2211: 

2212: 

2213: 

2214: \section{Acknowledgments}

2215: \label{sec:ack}

2216: It is a pleasure to acknowledge helpful conversations with Benedikt Obermayer, who,

2217: in addition, corrected several pre-factors.  This research was supported by the

2218: German Academic Exchange Service (DAAD) through a fellowship within the

2219: Postdoc-Program (OH) and by the Deutsche Forschungsgemeinschaft through grant no.~Ha

2220: 5163/1 (OH) and SFB 486 (EF).

2221: 

2222: 

2223: \appendix

2224: 

2225: 

2226: \begin{table}[b]

2227:   \caption{Some important notations}

2228:   \label{tab:common-notation}

2229:   \begin{ruledtabular}

2230:     \begin{tabular}{c|l}

2231:       Symbol(s) & General Meaning \\ \hline

2232:       $L$ & total length of the worm-like chain \\

2233:       $\kappa$ & bending stiffness \\

2234:       $\epsilon$  & small parameter, defined such that $\vec r_\pe'^2=\Ord{\epsilon}$    \\

2235:       $\vec r_\pe(s,t)$ & transverse displacement; $\vec r_\pe=\Ord{\epsilon^{1/2}}$ \\

2236:       $r_\pa(s,t)$ & longitudinal displacement; $\vec r_\pa=\Ord{\epsilon}$ \\

2237:       $\varrho(s,t)$ & stored-length density; $\varrho=\vec

2238:       r_\pe'^2/2+\Ord{\epsilon^2}=\Ord{\epsilon}$  \\

2239:       $R$ & end-to-end distance \\

2240:       $R_\pa$ & end-to-end vector projected onto the long.~axis

2241:       \\

2242:       $\simeq$ & equal up to numerical factors of

2243:       order $1$ \\

2244:       $\propto$ & proportional to \\

2245:       $\sim$ & asymptotically equal \\

2246:       $\lp$ &  persistence length \\

2247:       $k_B T$ & thermal energy \\

2248:       $f(s,t)$ & line tension \\

2249:       $\lpe(t)$ &  equilibration scale for transverse bending modes

2250:       \\ 

2251:       $\lpa(t)$ & scale  of tension variations \\

2252:       $\vec \fe$ & external force \\

2253:       $\vec \xi(s,t)$ & thermal force per arc length

2254:     \end{tabular}

2255:   \end{ruledtabular}

2256: \end{table}

2257: 

2258:   

2259: \section{Crossover scaling (details)}

2260: \label{sec:cscaling-det}

2261: 

2262: 

2263: \subsection{Deterministic relaxation on long times ($t \gg \tf$)}

2264: \label{sec:A}

2265: We have for $\fc=0$

2266: \begin{equation}

2267:   \label{eq:A-step1}

2268:   \begin{split}

2269:     &A-\mint{\frac{dq}{2\pi}}{-\infty}{\infty} \frac 1

2270:     {q^2}\left(1-e^{-2q^2 \phi}\right) = \\

2271:     &\sqrt{\phi}\mint{\frac{dq}{2\pi}}{-\infty}{\infty} \frac 1

2272:     {q^2}\left(e^{-2q^2}-e^{-2q^2(q^2\tau\phi^{-2}+1)}\right) \\

2273:     &\stackrel{\tau\gg1}{\to}0 \;,\qquad \text{if }\phi^{-2}=

2274:     o(\tau^{-1}) \;,

2275:   \end{split}

2276: \end{equation}

2277: and for $\fc=1$

2278: \begin{equation}

2279:   \label{eq:A-step2}

2280:   \begin{split}

2281:     A-\mint{\frac{dq}{2\pi}}{-\infty}{\infty}

2282:     \frac 1 {q^2+1}&=-\mint{\frac{dq}{2\pi}}{-\infty}{\infty}

2283:     \frac 1 {q^2+1} e^{-2q^2(q^2\tau+\phi)}\\

2284:     &\stackrel{\tau\gg1}{\to}0

2285:     \;,\qquad \text{if }\phi^{-2}= o(\tau^{-1}) \;.

2286:   \end{split}

2287: \end{equation}

2288: As indicated, both expressions, Eqs.~(\ref{eq:A-step1},~\ref{eq:A-step2}), go to zero

2289: for large $\tau$ if $\phi^{-2}= o(\tau^{-1})$. The latter, however, follows from the

2290: assumptions stated in the main text, namely that the tension satisfies the scaling

2291: form Eq.~(\ref{eq:scaling-phi}) with the requirement in Eq.~(\ref{eq:alpha-assumption}).

2292: N.b.~it turns out that $\phi=\Ord{\tau}$ for {\pull} and {\lelo} and

2293: $\phi=\Ord{\tau^{4/3}}$ for \epull.

2294: 

2295: 

2296: 

2297: \subsection{Thermal excitation on long times ($t \gg \tf$)}

2298: \label{sec:B}

2299: We want to show that it is justified to calculate (the negative of) the thermally

2300: generated stored length represented by the term $B$, Eq.~(\ref{eq:B}), on long times

2301: \emph{quasi-statically} for the semi-infinite polymer.  To this end, we first

2302: insert the scaling ansatz, Eq.~(\ref{eq:scaling-phi}), for $\phi(\sigma,\tau)$,

2303: \begin{equation}

2304:   \label{eq:B-app-1}

2305:   \begin{split}

2306:     B&=-4 \mint{\frac{dq}{2\pi}}{\Lambda^{-1}}{\infty} q^2 \mint{d\hat

2307:       \tau}{0}{\tau}\\

2308:     &\times e^{-2 q^2 \left[ q^2 (\tau-\hat \tau)+\tau^{\alpha+1}\left(\hat

2309:           \phi(\xi)- (\hat \tau/ \tau)^{\alpha+1}\hat\phi[\xi (\tau/\hat \tau)^{z}

2310:           ]\right)\right] } \;,

2311:   \end{split}

2312: \end{equation}

2313: where we introduced the scaling variable $\xi\equiv

2314: \sigma/\tau^{z}$. Then we substitute $\hat \tau\to x \tau$ and

2315: $q\to q \tau^{-1/4}$,

2316: \begin{equation}

2317:   \label{eq:B-app-2}

2318:   \begin{split}

2319:   B&=-4 \tau^{1/4} \mint{\frac{dq}{2\pi}}{\Lambda^{-1}\tau^{1/4}}{\infty} q^2

2320:   \mint{dx}{0}{1} \\

2321:   &\times e^{-2 q^2 \left[ q^2

2322:         (1-x)+\tau^{\alpha+1/2}\left(\hat \phi(\xi)- x^{\alpha+1} \hat\phi(\xi

2323:           x^{-z})\right)\right] } \;.

2324:   \end{split}

2325: \end{equation}

2326: Note that for $\alpha>-1/2$ (as assumed in Eq.~(\ref{eq:alpha-assumption})) the

2327: factor $\tau^{\alpha+1/2}$ in the exponent diverges in the long-time limit. Hence,

2328: for any given wave number the $x$-integral will be dominated by $x$ close to $1$ for

2329: large enough $\tau\gg1$. This allows us to linearize the exponent in $1-x$ when

2330: performing the $x$-integral for this given wave number. In contrast, for a given time

2331: $\tau$ and $\hat \phi(\xi)=\Ord{1}$, the exponent can be linearized only for large

2332: enough wave numbers, $q\gg q_\star(\tau)\equiv \tau^{-\alpha/2-1/4}$, for which the

2333: factor $q^2 \tau^{\alpha+1/2}\gg1$ in the exponent of Eq.~(\ref{eq:B-app-2}) is much

2334: larger than $1$.

2335: 

2336: Since the integral over $q$ runs over all $q$-vector, we have also to

2337: care about the small wave numbers, for which the exponent cannot be

2338: linearized in $1-x$. To this end, we split the $q$-integral at a wave

2339: vector $K$ satisfying

2340: \begin{equation}

2341:   \label{eq:K}

2342:   q_\star^{1/3}\gg K\gg q_\star\equiv\tau^{-\alpha/2-1/4} \;,

2343: \end{equation}

2344: which can be found in the limit $\tau\gg1$ under the premise of

2345: Eq.~(\ref{eq:alpha-assumption}), $\alpha>-1/2$.  The first inequality

2346: in Eq.~(\ref{eq:K}) is required for reasons that become clear later

2347: on. For the upper part $B_>$ of the integral we can linearize the

2348: exponent in $1-x$,

2349: \begin{equation}

2350:   \label{eq:B-app-upper}

2351:   \begin{split}

2352:     B_>&\equiv-4 \tau^{1/4} \mint{\frac{dq}{2\pi}}{K}{\infty} q^2

2353:     \mint{dx}{0}{1}\\& \times  e^{-2 q^2 \left[ q^2

2354:         (1-x)+\tau^{\alpha+1/2} \left(\hat \phi(\xi)-x^{\alpha+1} \hat\phi(\xi

2355:           x^{-z})\right)\right] }\\

2356:     &\sim-4 \tau^{1/4} \mint{\frac{dq}{2\pi}}{K}{\infty} q^2

2357:     \mint{dx}{0}{1} \\& \times e^{-2 q^2 (1-x)\left[ q^2

2358:         +\tau^{1/2}\partial_\tau \left(\tau^{\alpha+1}  \hat

2359:           \phi(\sigma/\tau^z)\right)\right] } \\

2360:     &=-2 \tau^{1/4} \mint{\frac{dq}{2\pi}}{K}{\infty}  \frac{1-e^{-2 q^2 \left[ q^2

2361:         +\tau^{1/2}\partial_\tau \left(\tau^{\alpha+1}  \hat

2362:           \phi(\sigma/\tau^z)\right)\right] }}{q^2

2363:         +\tau^{1/2}\partial_\tau \left(\tau^{\alpha+1}  \hat

2364:           \phi(\sigma/\tau^z)\right)} 

2365:   \end{split} 

2366: \end{equation}

2367: where we eliminated the scaling variable $\xi=\sigma/\tau^z$, again.

2368: Using the second inequality in Eq.~(\ref{eq:K}) it is seen that we can

2369: drop the exponential for the ``interesting'' regime

2370: $\hat\phi(\sigma/\tau^z)=\Ord{1}$, where the tension has an

2371: appreciable value. Inserting back

2372: $\phi=\tau^{\alpha+1}\hat\phi(\sigma,\tau)$ we obtain

2373: \begin{equation}

2374:   \label{eq:B-app-upper-2}

2375:   \begin{split}

2376:     B_> &\sim -2 \tau^{1/4} \mint{\frac{dq}{2\pi}}{K}{\infty}

2377:     \frac{1}{q^2

2378:       +\tau^{1/2}\partial_\tau \phi(\sigma,\tau)} \\

2379:     & = \frac{2}{\sqrt{\partial_\tau \phi(\sigma,\tau)}}

2380:     \mint{\frac{dq}{2\pi}}{K/\left(\tau^{1/4}\sqrt{\partial_\tau

2381:       \phi(\sigma,\tau)}\right)}{\infty} \frac{1}{q^2 +1 } \\

2382:     &\sim - \frac{1}{2\sqrt{\partial_\tau \phi(\sigma,\tau)}}\;.

2383:   \end{split}

2384: \end{equation}

2385: To obtain the last asymptotics, we have approximated the lower bound

2386: of the integral by zero. This can be justified by the first inequality

2387: in Eq.~(\ref{eq:K}),

2388: \[\frac{K}{\tau^{1/4}\sqrt{\partial_\tau \phi(\sigma,\tau)}}= K\,

2389: \Ord{\tau^{-\frac 1 4-\frac \alpha 2}=q_\star} \ll q_\star^{4/3}\ll 1 \;.\] 

2390: 

2391: The remaining lower part $B_<$ of the $q$-integral in

2392: Eq.~(\ref{eq:B-app-2}) is estimated to be small as compared to $B_>$,

2393: \begin{equation}

2394:   \label{eq:B-app-lower}

2395:   \begin{split}

2396:     B_<&\equiv-4 \tau^{1/4} \mint{\frac{dq}{2\pi}}{\Lambda^{-1}t^{1/4}}{K} q^2

2397:     \mint{dx}{0}{1}  \\& \times e^{-2 q^2 \left[ q^2

2398:         (1-x)+\tau^{\alpha+1/2} \left(\hat \phi(\xi)-x^{\alpha+1} \hat\phi(\xi

2399:           x^{-z})\right)\right] }\\

2400:     &< -4 \tau^{1/4} \mint{\frac{dq}{2\pi}}{0}{K} q^2

2401:     \mint{dx}{0}{1} \\

2402:     &= -\frac{4}{3} \tau^{1/4} K^3 /(2 \pi)\\ 

2403:     &\ll -\frac{4}{3} \tau^{-\alpha/2}  /(2 \pi) \\

2404:     &\propto B_> \;,

2405:   \end{split}

2406: \end{equation}

2407: where the first inequality in Eq.~(\ref{eq:K}) has been applied.  Therefore, $B$ is

2408: asymptotically given by Eq.~(\ref{eq:B-app-upper-2}).

2409: 

2410: 

2411: \section{Relaxation of a completely stretched polymer}

2412: \label{sec:lelo-app}

2413: 

2414: In this section, we consider more closely the intermediate asymptotics

2415: \begin{equation}

2416:   \label{eq:homogeneous-lelo}

2417:   f\propto \left(\frac{\hat \zeta L^2}{l_p t}\right)^{2/3} \;, 

2418: \end{equation}

2419: that is approached in the limit

2420: \begin{equation}

2421:   \label{eq:homo-lelo-time-frame}

2422:   \totpa=\frac{\hat \zeta L^2}{l_p {f_<}^{3/2}}\ll t \ll

2423:   t_\star=\frac{\hat \zeta^4 L^8}{l_p^4}\;

2424: \end{equation}

2425: of {\lelo}, which we analyzed in terms of a quasi-static approximation in

2426: Sec.~\ref{sec:tension-relaxed}. The purpose of this section is to justify the

2427: quasi-static assumption in the limit of $f_<\to\infty$ (i.e.~$\tf\to0$) where we

2428: start with a completely stretched polymer and all stored length is generated by the

2429: action of stochastic forces.

2430: 

2431: To this end, we show that, in the limit $t\to 0$, the change in stored length

2432: $\avg{\Delta\overline \varrho}$ given by Eq.~(\ref{eq:change-stored-length}) for the

2433: force history given by Eq.~(\ref{eq:homogeneous-lelo}) asymptotically approaches the

2434: value one obtains from the quasi-static calculation,

2435: \begin{eqnarray}

2436:   &\avg{\Delta\overline

2437:     \varrho}(t)=\mint{\frac{dq}{\pi\lp}}{0}{\infty}

2438:   \left\{\frac{1}{q^2+f_<}\left(e^{-2q^2[q^2 t+ F(t)]}-1\right)

2439:   \right.& \nonumber\\

2440:   &\left.+2 q^2\mint{d\tilde t}{0}{t}e^{-2q^2\left[q^2(t-\tilde

2441:         t)+F(t)-F(\tilde t)\right]}\right\}&\nonumber\\

2442:   &\stackrel{f_<\to\infty}{=}\mint{\frac{dq}{\pi\lp}}{0}{\infty} 2

2443:   q^2\mint{d\tilde t}{0}{t}e^{-2q^2\left[q^2(t-\tilde

2444:       t)+F(t)-F(\tilde t)\right]} & \label{eq:to-show-1} \\

2445:   &\stackrel{!}{\sim} \mint{\frac{dq}{\pi\lp}}{0}{\infty}

2446:   \frac{1}{q^2+f(t)}=\frac{1}{2l_p}\left[f(t)\right]^{-1/2}\;,\qquad\text{for } t

2447:   \to 0\;.&\nonumber

2448: \end{eqnarray}

2449: This will comprise an a posteriori justification of the quasi-static

2450: assumption that entered in the derivation of the right-hand-side of

2451: Eq.~(\ref{eq:degennes-eq-2}).

2452: 

2453: The argument closely follows App.~\ref{sec:B}. Inserting the force history

2454: \begin{equation}

2455:   \label{eq:F-1/3}

2456:   F(t)=\mint{d\hat t}{0}{t}f(\hat t)= C t^{\alpha+1}

2457: \end{equation}

2458: with

2459: \begin{equation}

2460:   \label{eq:prefactor}

2461:   \alpha=-2/3\stackrel{!}{<}-1/2 \qquad \text{and}\qquad C\approx \left(\frac{\hat \zeta L^2}{l_p}\right)^{2/3}

2462: \end{equation}

2463: into Eq.~(\ref{eq:to-show-1}) and changing variables $\hat t \to x t$

2464: and $q\to q t^{-1/4}$ yields

2465: \begin{eqnarray}

2466:   \label{eq:1/3-noise-rescaled}

2467:   &&\avg{\Delta\overline

2468:       \varrho}(t)=2t^{1/4}\mint{\frac{dq}{\pi\lp}}{\Lambda^{-1}t^{1/4}}{\infty} 

2469:     q^2  \nonumber \\

2470:     &&\qquad\times \mint{dx}{0}{1}e^{-2q^2\left[q^2(1-x)+C t^{\alpha+1/2}\left(1-x^{\alpha+1}\right)\right]}

2471: \end{eqnarray}

2472: As in Sec.~\ref{sec:B} an approximation to the integral can be found in

2473: the limit

2474: \begin{equation}

2475:   \label{eq:central-limit}

2476:   C t^{\alpha+1/2}\gg1 \;.

2477: \end{equation}

2478: We split the $q$-integral at $K$ satisfying

2479: \begin{equation}

2480:   \label{eq:K2}

2481:   q_\star^{1/3}\gg K\gg q_\star\equiv (C t^{\alpha+1/2})^{-1/2} \;,

2482: \end{equation}

2483: which can be found in the limit $t\ll 1$ ($\Rightarrow

2484: q_\star\ll1$) because $\alpha<-1/2$. The upper part of

2485: the integral

2486: \begin{equation}

2487:   \label{eq:rho>}

2488:   \avg{\Delta\overline

2489:       \varrho^>}(t)\equiv \left(\dots\right)\mint{dq}{K}{\infty}\left(\dots\right)

2490: \end{equation}

2491: is dominated by values of $x$ close to $1$ and we can linearize the

2492: exponent in $1-x$,

2493: \begin{equation}

2494:   \label{eq:rho>-2}

2495:   \begin{split}

2496:     \avg{\Delta\overline{

2497:         \varrho}^>}(t)&=2t^{1/4}\mint{\frac{dq}{\pi}}{K}{\infty}q^2\mint{dx}{0}{1}

2498:     \\&\times e^{-2q^2 (1-x) \left[q^2+(\alpha+1)C t^{\alpha+1/2} \right]} \\

2499:     &= \tau^{1/4}\mint{\frac{dq}{\pi}}{K}{\infty} \frac{1-e^{-2q^2 (1-x)

2500:         \left[q^2+(\alpha+1)C t^{\alpha+1/2}

2501:         \right]}} {q^2+(\alpha+1)C t^{\alpha+1/2}} \\

2502:     &\sim \tau^{1/4}\mint{\frac{dq}{\pi}}{K}{\infty}

2503:     \frac{1}{q^2+(\alpha+1)C t^{\alpha+1/2}} \\

2504:     &=\frac{1}{l_p\sqrt{(\alpha+1)C t^\alpha}}\mint{\frac{dq}{1

2505:         \pi}}{K/\sqrt{(\alpha+1)C t^{\alpha+1/2}}}{\infty}\frac{1}{q^2+1} \\

2506:     &\sim \frac{1}{2 l_p \sqrt{f(t)}}

2507: \end{split}

2508: \end{equation}

2509: where the asymptotics follow from both inequalities in

2510: Eqs.~(\ref{eq:central-limit},~\ref{eq:K2}). The lower part $\avg{\Delta\overline{

2511:     \varrho}^<}$ of the integral is estimated to be subdominant as compared to

2512: $\avg{\Delta\overline{ \varrho}^>}$,

2513: \begin{equation}

2514:   \label{eq:rho<}

2515:   \begin{split}

2516:     \avg{\Delta\overline{

2517:       \varrho}^<}(t)&=(\dots)\mint{dq}{0}{K}(\dots) \\

2518:     &<\frac{2 t^{1/4}}{l_p}\mint{\frac{dq}{\pi}}{0}{K} q^2

2519:     \mint{dx}{0}{1} \\

2520:     &=\frac{t^{1/4}K^3}{3\pi l_p}\\

2521:     &\ll\frac{C^{-1/2}t^{-\alpha/2}}{3\pi l_p} \\

2522:     &\propto \avg{\Delta\overline{

2523:       \varrho}^>}(t) \;,

2524:   \end{split}

2525: \end{equation}

2526: where the first inequality in Eq.~(\ref{eq:K2}) has been applied. Therefore

2527: $\avg{\Delta\overline{ \varrho}}(t)$ is asymptotically given by

2528: Eq.~(\ref{eq:rho>-2}).

2529: 

2530: Finally, we want to emphasize the central condition for the validity

2531: of the quasi-static approximation,

2532: \begin{equation}

2533:   \label{eq:central-limit-2}

2534:   C t^{\alpha+1/2}\approx\left(\frac{\hat \zeta L^2}{l_p}\right)^{2/3}t^{-1/6}\gg 1\;,

2535: \end{equation}

2536: or $t\ll t_\star$ with $t_\star=(L^2/\lp)^4$ as given by

2537: Eq.~(\ref{eq:val-cond-lelo-2}).

2538: 

2539: 

2540: \section{Defining $t_\star$}

2541: \label{sec:tstar}

2542: 

2543: As anticipated Part~I, there is a problem-specific time $t_\star$ limiting the

2544: short-time validity of OPT.  Physically, the crossover at $t_\star$ can be understood

2545: as follows. For $t\gg t_\star$ the ``speed'' of the structural relaxation is

2546: determined solely by the relaxation times of the bending modes, which are related to

2547: the transverse friction while the longitudinal friction is irrelevant. In contrast,

2548: for $t\ll t_\star$ the longitudinal friction substantially limits the speed of the

2549: relaxation. This suggests to estimate the time $t_\star$ as follows.  From the

2550: continuity equation, Eq.~(\ref{eq:cg-eom}), derived via the multiple-scale

2551: perturbation theory (MSPT), we can estimate the order of magnitude of the correction

2552: $\delta f(s,t)=f(s,t)-f^{\text{OPT}}$ to the flat tension profile,

2553: Eq.~(\ref{eq:OPT-tension}), by

2554: \begin{equation}

2555:   \label{eq:f-correction}

2556:   \delta f \approx \hat \zeta \partial_t \avg{\Delta \overline{\varrho}}\left(

2557:     f^{\text{OPT}},t  \right) L^2 \;.

2558: \end{equation}

2559: OPT can only be applicable if the correction $\delta f$ has negligible

2560: effect on the evolution of the stored length,

2561: \begin{equation}

2562:   \label{eq:OPT-validity}

2563:   \abs{1-\frac{\avg{\Delta\overline{\varrho}}\left(

2564:     f^{\text{OPT}}+\delta f,t  \right)}{\avg{\Delta\overline{\varrho}}\left(

2565:     f^{\text{OPT}},t  \right)}}\ll 1 \;,\qquad \text{for } t\gg t_\star\;.

2566: \end{equation}

2567: The time for which the left hand side of Eq.~(\ref{eq:OPT-validity}) becomes of order

2568: may thus be identified with the time $t_\star$ before which OPT is not valid.

2569: 

2570: 

2571: 

2572: 

2573: 

2574: 

2575: 

2576: \bibliographystyle{apsrev}

2577: 

2578: \bibliography{bibis/tdpre06,bibis/elastic-rod-dynamics,bibis/journals,bibis/sfpdynamics-tcited,bibis/unpub,bibis/actin-viscoelastic,bibis/semiflexibleA04,bibis/sf,bibis/klaussf,bibis/mysf,bibis/sfnet,bibis/unpub,bibis/mysfnet}

2579: 

2580: \end{document}