cond-mat0607534/ropo.tex
1: \documentclass[twocolumn,aps,prl]{revtex4}
2: 
3: \usepackage{graphicx}% Include figure files
4: \begin{document}
5: 
6: 
7: \title{Light diffusion and localization in 3D nonlinear disordered media}
8: 
9: \author{C. Conti$^{1,2}$, L. Angelani$^3$,  G. Ruocco$^{2,4}$}
10: 
11: \affiliation{
12: $^1$Centro studi e ricerche ``Enrico Fermi,'' Via Panisperna 89/A,
13: I-00184, Roma, Italy   \\
14: $^2$Research center Soft INFM-CNR, c/o Universit\`a di
15: Roma ``La Sapienza,'' I-00185, Roma, Italy \\
16: $^3$Research center SMC INFM-CNR, c/o Universit\`a
17: di Roma ``La Sapienza,'' I-00185, Roma, Italy \\
18: $^4$Dipartimento di Fisica, Universit\`a di Roma ``La Sapienza,''
19: I-00185, Roma, Italy }
20: 
21: \date{\today}
22: 
23: \begin{abstract}
24: Using a 3D Finite-Difference Time-Domain parallel code, we report on the linear
25: and nonlinear propagation of light pulses in a disordered assembly of scatterers, whose
26: spatial distribution is generated by a Molecular Dynamics code;
27: refractive index dispersion is also taken into account.
28: We calculate the static and dynamical diffusion constant of light, while considering a 
29: pulsed excitation. Our results are in quantitative agreement with reported
30: experiments, also furnishing evidence of a non-exponential decay of the transmitted 
31: pulse in the linear regime and in the presence of localized modes.
32: By using an high power excitation, we numerically demonstrate the ``modulational 
33: instability random laser'':
34: at high peak input powers energy is transferred to localized states from the input pulse,
35: via third-order nonlinearity and optical parametric amplification, and this process 
36: is signed by a power-dependent  non-exponential time-decay of the transmitted pulse.
37: \end{abstract}
38: 
39: 
40: \maketitle
41: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
42: %%% INTRODUCTION
43: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
44: Electromagnetic wave localization and diffusion in random media have been recently considered by several authors.
45: Experimental studies reported on various evidences of light localization, 
46: including the existence of a non-exponential long-time tail 
47: of the transmitted intensity \cite{Storzer06, Chabanov04, Chabanov01,Chabanov03, Vellekoop05, Johnson03, Wiersma97, Reil05}, 
48: and these processes were considered in numerous theoretical and numerical papers 
49: (as e.g. in \cite{RevModPhys.71.313, Skipetrov06,Vanneste05, Cheung05, Markos05,Tseng04,Tseng05,Tseng05a}). 
50: In a purely diffusive regime, pulsed light impinging on a disordered sample is rapidly dispersed while
51: propagating; this implies that nonlinear effects are typically negligible.
52: This holds true as far as localization processes do not get involved.
53: For example, in random lasers (for a recent review see \cite{Cao05}), localized or extended modes \cite{Mujumdar04}  
54: are excited in a disordered medium by means of an optically active material;
55: in this case, the large resonant nonlinear susceptibility and the fact that long living
56: (high Q) modes are excited foster a variety of nonlinear phenomena (as e.g. in \cite{Liu03,Deych05, Angelani06}).
57: For non-resonant ultra-fast nonlinearities (as $\chi^{(3)}$ processes of electronic origin \cite{BoydBook}),
58: it is in principle possible to use high-peak intensity femtosecond laser pulses to induce nonlinear effects,
59: even in the presence of strong light diffusion.
60: However, the (numerical and theoretical) analysis is enormously complicated 
61: by the need to include various effects as disorder, nonlinearity, 
62: material dispersion and to take into account a 3D environment.
63: Even in the absence of nonlinearity, no ``ab initio'' numerical investigation 
64: of 3D Maxwell equations for femtosecond pulses in disordered media has been reported, to the best of our knowledge. 
65: Some theoretical investigations considered 
66: the role of the Kerr effect (i.e. an intensity dependent refractive index) for a monochromatic excitation \cite{Spivak00,Skipetrov00, Skipetrov01,Skipetrov03};
67: however ultra-fast $\chi^{(3)}$ optical nonlinearity may also provide other mechanisms. 
68: We specifically consider the gain
69: of non-resonant origin, i.e. the optical parametric oscillation (OPO).
70: OPO, in isotropic media in the absence of disorder, 
71: has been recently considered with reference to various devices \cite{Conti04, Foster06,Kippenberg04},  
72: while in the fiber-optics community it is known as the ``modulational instability laser'' \cite{Nakazawa89}. 
73: Here we report on the  extensive 3D+1 numerical analysis of linear diffusion and nonlinear amplification processes
74: in a random dispersive nonlinear medium. We use a Molecular Dynamics (MD) code for generating a three dimensional distribution 
75: of spherical scatterers and a Finite-Difference Time-Domain (FDTD) parallel algorithm for simulating light propagation. 
76: The MD-FDTD approach not only reproduces the diffusive regime in agreement with reported experimental results in the linear regime,
77: but also provides clear indications of the random OPO process, with analogies to random lasers.
78: 
79: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
80: %%% DESCRIPTION OF THE NUMERICAL APPROACH
81: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
82: An ``ab initio'' investigation of the nonlinear light propagation in 3D disordered media, 
83: requires a realistic distribution of a colloidal medium and of its
84: nonlinear dielectric response.
85: We model the disordered system as an assembly of poly-dispersed particles in air. 
86: The MD code generates the particle configuration;
87: specifically, we considered a 50:50 mixture of spheres with diameter ratio $1.2$, 
88: interacting by a $100-200$ Lennard Jones potential \cite{Angelani04}.
89: The considered sample is composed by $1000$ scatterers; 
90: once determined the particle positions by the MD code, each of them is taken as filled by a dispersive 
91: optically nonlinear medium, which is modeled as described below. The particle dimensions are 
92: chosen in order to simulate typical samples used in light localization and diffusion experiments,
93: as e.g. in \cite{Vellekoop05,Storzer06}.
94: We considered diameters for the biggest particles in the mixture ranging 
95: from $200$ nm to $400$ nm; correspondingly the volume filling fraction $\phi$ for
96: our samples, with volume $L^3$ ($L=4$ $\mu$m, of the order of the values considered in \cite{Johnson03}), varies in the interval $\phi\in[0.1 , 0.6]$.
97: The optical linear response of the particle medium is chosen as the best approximation of that of 
98: $TiO_2$, used in most of the reported experiments, including material dispersion.
99: The latter is modeled by a single-pole Lorentz medium, reproducing the
100:  data for $TiO_2$ reported in \cite{DeVore51}. 
101: Material absorption is also included providing an absorption coefficient
102: $\kappa_i=10$ mm$^{-1}$ at $\lambda=514$ nm.
103: The 3D Maxwell equations, solved by the parallel FDTD \cite{TafloveBook} algorithm, are written as 
104: \begin{equation}
105: \begin{array}{l}
106: \label{maxwell}
107: \nabla\times {\bf E}=-\mu_0 \partial_t {\bf H} \hspace{1cm} \nabla\times {\bf H}=\epsilon_0 \epsilon_{\infty}\partial_t {\bf E}+\partial_t {\bf P}\\
108: \partial_t^2 {\bf P} +2\gamma_0 \partial_t {\bf P} +\omega_0^2 f(P) {\bf P} =\epsilon_0 (\epsilon_s-\epsilon_{\infty}) \omega_0^2 {\bf E}
109: \end{array}
110: \end{equation}
111: The non-resonant nonlinear response is determined by 
112: the nonlinear Lorentz oscillator (see e.g. \cite{BoydBook}).
113: As $f(P)=1$,  eqs.(\ref{maxwell}) model the linear dispersive medium $Ti O_2$;
114: by appropriately choosing $f(P)$, the system displays a isotropic ultra-fast nonlinearity, as previously described in \cite{Conti04},
115: including the non-resonant $\chi^{(3)}$ susceptibility.
116: In the present case, the associated Kerr effect has a nonlinear refractive index $\Delta n= n_2 I$, with $I$ the
117: optical intensity and $n_2 \cong 10^{-18}$m$^2$W$^{-1}$.
118: The parameters in $TiO_2$ particles are given by (MKS units)  : 
119: $\epsilon_s=5.9130$, $\epsilon_\infty=2.8731$ , $\omega_0=6.6473\times 10^{15}$, $\gamma_0=8.9168\times 10^{11}$ and 
120: $f(P)=[1+(P/P_0)^2]^{3/2}$ with $P_0^{-2}=3.0$;
121: out of the particles, $P=0$ and $\epsilon_s=\epsilon_\infty=1$ (vacuum).
122: 
123: The code is parallelized using the ``MPI'' protocol,
124: and ``UPML'' absorbing boundary condition are used in the three spatial directions \cite{TafloveBook}.
125: The used discretization is $\Delta x=\Delta y=\Delta z\cong 30$ nm and $\Delta t\cong0.02$ fs.
126: The $z-$propagating input beam is taken with a Gaussian TEM$_{00}$  
127: linearly $y-$polarized profile, at the input face of the disordered medium its waist is 
128: $w_0=1\ \mu$m. The input pulse temporal profile is Gaussian, with duration $t_0=100$ fs and carrier wavelength $\lambda=520$ nm.
129: The numerical results have been verified a number of times by halving the spatio-temporal discretization. 
130: The validity of our approach has been confirmed in a variety of studies, as e.g. in \cite{Conti04,Conti04b}.
131: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
132: %%% EVIDENCE OF LIGHT DIFFUSION
133: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
134: 
135: The first numerical simulations were aimed to investigate light diffusion 
136: when the nonlinear effects can be taken as negligible (low power),
137: for a fixed sample spatial extension. We considered increasing values for the diameter 
138: of the spheres in the mixture; in the following $\sigma$ is the corresponding value of the
139: biggest particles in the considered 50:50 mixture (for the others, the diameter is $\sigma/1.2$).
140: This corresponds to increase the average density of the medium, as well as the volume packing fraction. 
141: In order to unveil the onset of diffusion, we monitored the time-dependent output intensity,
142: which is expected to develop an exponential tail in the diffusive regime.
143: This is obtained in two ways: i) we considered 
144: the electric field $E_y$ at one point in the middle of the output face of the sample
145: (see figure 1),
146: and we calculated the ``local intensity'' by squaring the field and filtering with a low 
147: pass filter to remove the optical carrier (thus mimicking a photodiode); ii) we considered the total transmission
148: from the output face of the sample as the calculated $z-$component of the Poynting vector, integrated with respect to
149: the transverse $x,y$ coordinates.
150: 
151: In figure 1 we show the results for the local intensity, while increasing the volume filling fraction $\phi$.
152: In order to compare the trailing edge of the transmitted pulses,
153: for each signal we rescaled the peak value to the unity and shifted the temporal axis to make all the
154: peaks for any $\sigma$ coincident.
155: At high packing fraction the onset of an exponential trend is clearly visible.
156: Note the residual oscillations, that are due to the fact that light
157: is collected in a specific point at the output face. 
158: %% Speckle pattern
159: A typical far-field speckle pattern calculated from our simulation is shown in the inset of the bottom panel, and 
160: obtained by storing the $E_y$ component of the electric field in the output section,
161: Fourier transforming with respect to the transverse coordinates, and averaging over time
162: (the pattern is calculated for a continuous wave excitation at $\lambda=520$ nm).
163: 
164: %% average poynting vector
165: The intensity modulation in figure 1 disappears when considering the overall transmission 
166: from the output facet, which provides an average information over all the speckles
167: and is shown in figure \ref{figpoynting}.
168: The instantaneous flux for the z-component of the Poynting vector is calculated and 
169: then filtered to remove the reactive component.
170: The appearance of an exponential tail in the transmission denotes the transition to the diffusive regime.
171: The transmitted signal is $I(t)\propto \exp[-\pi^2 D(t) t/L^2]$, with an instantaneous diffusion coefficient 
172: $D(t)\rightarrow D$ as $t\rightarrow\infty$
173: and a sample length $L$.
174: When the packing fraction increases, the tail gets longer due to the reduced diffusivity $D$.
175: The latter quantity is shown in Fig. \ref{figpoynting}(b) vs $\phi$. The results are in agreement
176: with experimentally determined values, as reported in literature (e.g. in \cite{ Johnson03, Vellekoop05,Storzer06}). 
177: Note that, as expected, $D$ decreases as the volume packing fraction increases,
178: and at high $\phi$ the diffusivity tends to an asymptotic value (for non overlapping particles).
179: The inset in Fig. \ref{figpoynting}(b) shows the calculated instantaneous diffusivity for two values of 
180: $\sigma$, which reproduces the reported trends, including some temporal oscillations \cite{Chabanov03, Storzer06}. 
181: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
182: %%% EVIDENCE OF LIGHT INDUCED LOCALIZZATION IN THE TRASMISSION
183: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
184: 
185: Next, we consider the diffusive regime for a fixed $\sigma=400$ nm ($\phi\cong0.6$), while increasing the input peak power $P_{in}$.
186: In figure \ref{fignl}(a), we show the local intensity for two input peak powers; the onset of a non-exponential tail is 
187: evident, as also shown in the inset reporting the intensity auto-correlation function.
188: In the panel (b) we show the transmitted intensity (integrated Poynting vector) for various input powers.
189: The inset shows the instantaneous diffusivity that drops at high powers and long times, when a threshold
190: value for the input peak power is reached.
191: These results foster the interpretation of the data in terms of a light-induced localization transition,
192: strongly resembling that controlled e.g.
193: by the packing fraction and experimentally observed (as e.g. in \cite{Storzer06}).
194: However, in the present case, the transition is controlled
195: by the input power, and hence it has a nonlinear origin.
196: It will be shown below that it can be explained as the excitation of
197: a ``modulational instability random laser'' (or random OPO).
198: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
199: %%% DENSITY OF STATES, SPECTRAL ANALYSIS AND INTERPRETATION
200: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
201: 
202: Indeed, to unveil the underlying  mechanism, we resorted to the spectral analysis of the
203: electric field in various points inside the random medium;
204: figure \ref{figspectrum} summarizes the typical result.
205: In panel (a) we show the calculated frequency spectra of the electric field $E_y$ in the
206: random structure, for different input peak powers.
207: The data are displayed in logarithmic scale and vertically shifted by an arbitrary amount for the
208: sake of clearness.
209: At low power the spectrum of the input pulse is reproduced, with the addition of small ripples
210: (due to the vertical scale in figure  \ref{figspectrum}a)  
211: in correspondence of the existing high-Q modes. 
212: As the power increases, the 
213: bandwidth gets larger and regions with depression and enhancement of the spectrum appear.
214: At sufficiently high power, a wide band excitation is found.
215: This behavior can be explained by recalling that in $\chi^{(3)}$ media energy can be transferred 
216: from various harmonics to others through parametric amplification.
217: We stress that the nonlinear effects are weakly pertubative, indeed we consider power level just above the
218: OPO threshold and the nonlinear index perturbation, due to the Kerr effect, is of the order of $0.001$.
219: Thus the localization process in figure \ref{fignl} is explained by  assuming that
220: energy is nonlinearly transferred to localized, long-living states. 
221: 
222: In panel (b) we show the time evolution of the spectrum (i.e. the spectrogram calculated by
223: Fourier transforming windowed parts of the temporal signal) at $P_{in}=3$ kW. 
224: In the initial interval $t<1$ ps, energy is transferred
225: to a wide band around the input pulse spectrum. However, at large times only the long-lifetime (high Q-factor) modes 
226: survive and the spectrum is characterized by several peaks in the region
227: around $\lambda\cong300$ nm ($\nu\cong 0.9\times 10^{15}$ Hz).
228: In order to sustain the fact that localized modes put into oscillation, we show in the inset
229: of figure 4(c) a snapshot of the electric field $E_y$ in the sample, taken at $P=2.5$ kW
230: and $t\cong4$ ps which clearly show the occurrence of localized excitations.
231: 
232: %----------------- FIGURE 1
233: \begin{figure}
234: \includegraphics[width=8.3cm]{figure1.eps}
235: \caption{
236: (Color online) Top panels (a-c): electric field in the middle of the output face of the sample at low 
237: input peak power ($P_{in}=1$ nW), for three different particle diameters ($\sigma=250,350,400$ nm).
238: The insets show the particle distribution in
239: the yz (top) and xz (bottom) middle section of the sample.
240: Bottom panel (d): local intensity in logarithmic scale. The inset shows 
241: the calculated speckle pattern.
242: \label{figlinear}}
243: \end{figure}
244: %----------------- FIGURE 2
245: \begin{figure}
246: \includegraphics[width=8.3cm]{figure2.eps}
247: \caption{
248: (Color online) (a) Tail of the transmitted pulse (the flux of the z-component of the Poynting vector at the output face) for various particle diameters. The inset shows the transmitted pulse in linear scale, the gray region is the input pulse.
249: (b) Calculated diffusion constant for various filling fractions. The continuous line
250: is a best-fit by an inverse power law.
251: The inset shows the dynamical diffusivity for two values of the particle diameters.
252: \label{figpoynting}}
253: \end{figure}
254: 
255: In addition, we went back to the linear regime to analyze the localized
256: modes are supported in our virtual samples. 
257: First, we repeated the pulse transmission experiment at low power $P=1$ nW and central wavelength
258: $\lambda=300$ nm ($\sigma=400$ nm) and Figure 4(c) clearly shows an non-exponential tail
259: at $t\cong 2.5$ ps, which is not present for $\lambda=521$ nm (figure 2a) also for $t>3$ ps (not shown).
260: Then, we recall that light localization
261: is typically expected in proximity of depressions of the density of states (DOS).
262: For the case of weakly disordered periodical structures, these are obtained inside the photonic band gaps \cite{John87},
263: while, for random dispersions of particles, the DOS (which
264: is numerically calculated by shining the sample by a ultra-wide band single-cycle pulse and spectrally analyzing 
265: the transmitted pulse) displays some depressions (pseudo-gaps) in proximity of the Mie resonances (see e.g. \cite{Vanneste05, Chabanov01}).
266: In our case, the distribution of states displays various depressions with the most pronounced around
267: $0.9 \times 10^{15}$ Hz, as shown in Fig. \ref{figspectrum}(d).
268: The fact that localized modes do exist in proximity of the pseudo-gaps, is also confirmed by the
269: plot of the calculated Q-factor vs frequency (bold line in Fig. \ref{figspectrum}(d)), realized by using an harmonic inversion 
270: library \cite{Mandelshtam01}.
271: Which of the localized modes are effectively excited 
272: will depend not only on the Q-factors, but also on the spatial overlaps with the pump modes and with
273: the spatial distribution of the nonlinear susceptibility (the $g$ coefficient discussed below). 
274: The localized modes are a reminiscence of the Mie resonances of the isolated spheres, which are strongly affected
275: by the presence of adjacent resonators in a disordered configuration, in analogy to what happen for periodical
276: mono-dispersed photonic crystal, investigated e.g. in \cite{Vandenbem05}.
277: 
278: %--------------------------
279: \begin{figure}
280: \includegraphics[width=8.3cm]{figure3.eps}
281: \caption{ 
282: (a) Time dynamics of the local intensity for two different input peak powers; the
283: inset shows the corresponding auto-correlation functions versus the delay $\tau$;
284: (b) transmitted signal for various input peak powers; the inset shows the calculated 
285: instantaneous diffusivity.
286: \label{fignl}}
287: \end{figure}
288: %--------------------------
289: \begin{figure}
290: \includegraphics[width=8.3cm]{figure4.eps}
291: \caption{(Color online) 
292: (a) Spectral emission for increasing peak power;
293: (b) spectrogram displaying the time evolution of the spectra for $P=3$ kW;
294: (c) transmission of a low power pulse at $\lambda=300$ nm and $\sigma=400$ nm displaying 
295: a not-exponential tails; the inset shows the field distribution for $\sigma=400$ nm 
296: at high power ($P=2.5$ kW) at $t=4$ ps in the xz middle section of the sample;
297: (d) density of states and Q-factor distribution for $\sigma=400$ nm.
298: \label{figspectrum}}
299: \end{figure}
300: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
301: %%% THEORETICAL FRAMEWORK
302: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
303: 
304: Before concluding we show that the basic features of the reported results can 
305: be reproduced by a simple theoretical model, based on the coupled mode theory in the
306: time domain (CMT) \cite{HausBook}. Indeed the small-signal parametric gain for the four-waves interaction
307: $\omega_0+\omega_0=(\omega_0+\Omega)+(\omega_0-\Omega)$, can be determined by introducing the 
308: complex amplitudes for the pump mode $a_0(t)=a(\omega_0,t)$ and for the generated modes 
309: $a_\pm(t)=a(\omega_0\pm \Omega,t)$. The complex amplitudes are normalized in a way such that $|a(\omega,t)|^2$ is
310: the energy stored in the mode at angular frequency $\omega$, and we denote with $\tau(\omega)$ the lifetime,
311:  and $Q(\omega)=\omega \tau(\omega)/2$ the corresponding Q-factor. 
312: In the un-depleted pump approximation the relevant CMT equations are written
313: in compact and obvious notation as
314: \begin{equation}
315: \displaystyle\frac{d a_\pm}{dt}=-\frac{1}{ \tau_\pm} a_\pm+ i g a_0^2 a_\mp^*\text{.}
316: \label{CMT}
317: \end{equation}
318: The coefficient $g$ is the relevant 3D spatial overlap between the mode electric field distributions and the nonlinear third-order susceptibility. 
319: Looking for solutions in the form 
320: $a_\pm=A_\pm \exp(\alpha t)$, the real-valued gain $\alpha$ is given by
321: \begin{equation}
322: \alpha=-\left(\frac{1}{2\tau_+}+\frac{1}{2\tau_-}\right)+\left[ \left( 
323: \frac{1}{2\tau_+}-\frac{1}{2\tau_-} \right)^2 +|g a_0^2|^2 \right]^{1/2} \text{.}
324: \end{equation}
325: It is readily seen 
326: that a condition for having amplification ($\alpha>0$) is given by 
327: $|g a_0^2|^2> 1/(\tau_+ \tau_-)= \omega_+ \omega_+ /4 Q_+ Q_-$.
328: Assuming that for all the involved modes $\omega\cong \omega_0$ and the Q-factor is almost the same,
329: with the exception of those localized that display higher values, it is concluded that the localized modes 
330: (those with not negligible overlap $g$ with the pump modes)
331: are put into oscillations at lower pump energies (see line at $2.25$ kW in figure \ref{figspectrum}(a)). 
332: Additionally the asymmetry, at low power, in the spectrum (see \ref{figspectrum}a) can be explained by observing from Eq. (\ref{CMT}) that $|A_+/A_-|>1$
333: if $Q_+>Q_-$. At higher powers most of the modes are put into oscillation and the spectrum mimics the DOS.
334: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
335: %%% CONCLUSIONS
336: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
337: 
338: In conclusion, the considered disordered, nonlinear and dispersive samples support
339: localized modes in proximity of Mie resonances. These can be excited by high-intensity
340: pulses through optical parametric amplification.
341: The overall process can be interpreted as the excitation of a ``modulational instability random laser''.
342: The main difference with standard random lasers is
343:  given by the fact that an ultra-fast optical process provides the required gain
344: and does not need doping of the sample by an active media.
345: These results can be generalized to other nonlinear amplification processes, 
346: like the Raman effect, and can find application in all-optical storage of light,
347: ultra-fast laser-tissue interaction, soft-matter and granular systems.
348: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
349: %%% AKNOWLEDGTEMS
350: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
351: We thank S. Trillo and D. Wiersma for fruitful discussions.
352: We acknowledge support from the INFM-CINECA initiative for parallel computing.
353: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
354: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% REFERENCES 
355: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
356: 
357: %\bibliography{glassy}
358: \begin{thebibliography}{38}
359: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
360: \expandafter\ifx\csname bibnamefont\endcsname\relax
361:   \def\bibnamefont#1{#1}\fi
362: \expandafter\ifx\csname bibfnamefont\endcsname\relax
363:   \def\bibfnamefont#1{#1}\fi
364: \expandafter\ifx\csname citenamefont\endcsname\relax
365:   \def\citenamefont#1{#1}\fi
366: \expandafter\ifx\csname url\endcsname\relax
367:   \def\url#1{\texttt{#1}}\fi
368: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
369: \providecommand{\bibinfo}[2]{#2}
370: \providecommand{\eprint}[2][]{\url{#2}}
371: 
372: \bibitem[{\citenamefont{Storzer et~al.}(2006)\citenamefont{Storzer, Gross,
373:   Aegerter, and Maret}}]{Storzer06}
374: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Storzer}},
375:   \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Gross}},
376:   \bibinfo{author}{\bibfnamefont{C.~M.} \bibnamefont{Aegerter}},
377:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Maret}},
378:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{96}},
379:   \bibinfo{pages}{063904} (\bibinfo{year}{2006}).
380: 
381: \bibitem[{\citenamefont{Chabanov et~al.}(2004)\citenamefont{Chabanov, Hu, and
382:   Genack}}]{Chabanov04}
383: \bibinfo{author}{\bibfnamefont{A.~A.} \bibnamefont{Chabanov}},
384:   \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Hu}}, \bibnamefont{and}
385:   \bibinfo{author}{\bibfnamefont{A.~Z.} \bibnamefont{Genack}},
386:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{93}},
387:   \bibinfo{pages}{123901} (\bibinfo{year}{2004}).
388: 
389: \bibitem[{\citenamefont{Chabanov and Genack}(2001)}]{Chabanov01}
390: \bibinfo{author}{\bibfnamefont{A.~A.} \bibnamefont{Chabanov}} \bibnamefont{and}
391:   \bibinfo{author}{\bibfnamefont{A.~Z.} \bibnamefont{Genack}},
392:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{87}},
393:   \bibinfo{pages}{153901} (\bibinfo{year}{2001}).
394: 
395: \bibitem[{\citenamefont{Chabanov et~al.}(2003)\citenamefont{Chabanov, Zhang,
396:   and Genack}}]{Chabanov03}
397: \bibinfo{author}{\bibfnamefont{A.~A.} \bibnamefont{Chabanov}},
398:   \bibinfo{author}{\bibfnamefont{Z.~Q.} \bibnamefont{Zhang}}, \bibnamefont{and}
399:   \bibinfo{author}{\bibfnamefont{A.~Z.} \bibnamefont{Genack}},
400:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{90}},
401:   \bibinfo{pages}{203903} (\bibinfo{year}{2003}).
402: 
403: \bibitem[{\citenamefont{Vellekoop et~al.}(2005)\citenamefont{Vellekoop, Lodahl,
404:   and Lagendijk}}]{Vellekoop05}
405: \bibinfo{author}{\bibfnamefont{I.~M.} \bibnamefont{Vellekoop}},
406:   \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Lodahl}}, \bibnamefont{and}
407:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Lagendijk}},
408:   \bibinfo{journal}{\pre} \textbf{\bibinfo{volume}{71}},
409:   \bibinfo{pages}{056604} (\bibinfo{year}{2005}).
410: 
411: \bibitem[{\citenamefont{Johnson et~al.}(2003)\citenamefont{Johnson, Imhof,
412:   Bret, Rivas, and Lagendijk}}]{Johnson03}
413: \bibinfo{author}{\bibfnamefont{P.~M.} \bibnamefont{Johnson}},
414:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Imhof}},
415:   \bibinfo{author}{\bibfnamefont{B.~P.~J.} \bibnamefont{Bret}},
416:   \bibinfo{author}{\bibfnamefont{J.~G.} \bibnamefont{Rivas}}, \bibnamefont{and}
417:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Lagendijk}},
418:   \bibinfo{journal}{\pre} \textbf{\bibinfo{volume}{68}},
419:   \bibinfo{pages}{016604} (\bibinfo{year}{2003}).
420: 
421: \bibitem[{\citenamefont{Wiersma et~al.}(1997)\citenamefont{Wiersma, Bartolini,
422:   Lagendijk, and Righini}}]{Wiersma97}
423: \bibinfo{author}{\bibfnamefont{D.~S.} \bibnamefont{Wiersma}},
424:   \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Bartolini}},
425:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Lagendijk}},
426:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Righini}},
427:   \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{390}},
428:   \bibinfo{pages}{671} (\bibinfo{year}{1997}).
429: 
430: \bibitem[{\citenamefont{Reil and Thomas}(2005)}]{Reil05}
431: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Reil}} \bibnamefont{and}
432:   \bibinfo{author}{\bibfnamefont{J.~E.} \bibnamefont{Thomas}},
433:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{95}},
434:   \bibinfo{pages}{143903} (\bibinfo{year}{2005}).
435: 
436: \bibitem[{\citenamefont{van Rossum and
437:   Nieuwenhuizen}(1999)}]{RevModPhys.71.313}
438: \bibinfo{author}{\bibfnamefont{M.~C.~W.} \bibnamefont{van Rossum}}
439:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{T.~M.}
440:   \bibnamefont{Nieuwenhuizen}}, \bibinfo{journal}{Rev. Mod. Phys.}
441:   \textbf{\bibinfo{volume}{71}}, \bibinfo{pages}{313} (\bibinfo{year}{1999}).
442: 
443: \bibitem[{\citenamefont{Skipetrov and van Tiggelen}(2006)}]{Skipetrov06}
444: \bibinfo{author}{\bibfnamefont{S.~E.} \bibnamefont{Skipetrov}}
445:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{B.~A.} \bibnamefont{van
446:   Tiggelen}}, \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{96}},
447:   \bibinfo{pages}{043902} (\bibinfo{year}{2006}).
448: 
449: \bibitem[{\citenamefont{Vanneste and Sebbah}(2005)}]{Vanneste05}
450: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Vanneste}} \bibnamefont{and}
451:   \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Sebbah}},
452:   \bibinfo{journal}{\pre} \textbf{\bibinfo{volume}{71}},
453:   \bibinfo{pages}{026612} (\bibinfo{year}{2005}).
454: 
455: \bibitem[{\citenamefont{Cheung and Zhang}(2005)}]{Cheung05}
456: \bibinfo{author}{\bibfnamefont{S.~K.} \bibnamefont{Cheung}} \bibnamefont{and}
457:   \bibinfo{author}{\bibfnamefont{Z.~Q.} \bibnamefont{Zhang}},
458:   \bibinfo{journal}{\prb} \textbf{\bibinfo{volume}{72}},
459:   \bibinfo{pages}{235102} (\bibinfo{year}{2005}).
460: 
461: \bibitem[{\citenamefont{Markos and Soukoulis}(2005)}]{Markos05}
462: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Markos}} \bibnamefont{and}
463:   \bibinfo{author}{\bibfnamefont{C.~M.} \bibnamefont{Soukoulis}},
464:   \bibinfo{journal}{\prb} \textbf{\bibinfo{volume}{71}},
465:   \bibinfo{pages}{054201} (\bibinfo{year}{2005}).
466: 
467: \bibitem[{\citenamefont{Tseng et~al.}(2004)\citenamefont{Tseng, Greene,
468:   Taflove, Maitland, Backman, V., and Walsh~Jr.}}]{Tseng04}
469: \bibinfo{author}{\bibfnamefont{S.~H.} \bibnamefont{Tseng}},
470:   \bibinfo{author}{\bibfnamefont{J.~H.} \bibnamefont{Greene}},
471:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Taflove}},
472:   \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Maitland}},
473:   \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Backman}},
474:   \bibinfo{author}{\bibnamefont{V.}}, \bibnamefont{and}
475:   \bibinfo{author}{\bibfnamefont{J.~T.} \bibnamefont{Walsh~Jr.}},
476:   \bibinfo{journal}{\ol} \textbf{\bibinfo{volume}{29}}, \bibinfo{pages}{1393}
477:   (\bibinfo{year}{2004}).
478: 
479: \bibitem[{\citenamefont{Tseng et~al.}(2005{\natexlab{a}})\citenamefont{Tseng,
480:   Kim, Taflove, Maitland, Backman, and Walsh~Jr.}}]{Tseng05}
481: \bibinfo{author}{\bibfnamefont{S.~H.} \bibnamefont{Tseng}},
482:   \bibinfo{author}{\bibfnamefont{Y.~L.} \bibnamefont{Kim}},
483:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Taflove}},
484:   \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Maitland}},
485:   \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Backman}}, \bibnamefont{and}
486:   \bibinfo{author}{\bibfnamefont{J.~T.} \bibnamefont{Walsh~Jr.}},
487:   \bibinfo{journal}{Opt. Express} \textbf{\bibinfo{volume}{13}},
488:   \bibinfo{pages}{3666} (\bibinfo{year}{2005}{\natexlab{a}}).
489: 
490: \bibitem[{\citenamefont{Tseng et~al.}(2005{\natexlab{b}})\citenamefont{Tseng,
491:   Taflove, Maitland, Backman, and Walsh~Jr.}}]{Tseng05a}
492: \bibinfo{author}{\bibfnamefont{S.~H.} \bibnamefont{Tseng}},
493:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Taflove}},
494:   \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Maitland}},
495:   \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{Backman}}, \bibnamefont{and}
496:   \bibinfo{author}{\bibfnamefont{J.~T.} \bibnamefont{Walsh~Jr.}},
497:   \bibinfo{journal}{Opt. Express} \textbf{\bibinfo{volume}{13}},
498:   \bibinfo{pages}{6127} (\bibinfo{year}{2005}{\natexlab{b}}).
499: 
500: \bibitem[{\citenamefont{Cao}(2005)}]{Cao05}
501: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Cao}}, \bibinfo{journal}{J.
502:   Phys. A. : Math. Gen.} \textbf{\bibinfo{volume}{38}}, \bibinfo{pages}{10497}
503:   (\bibinfo{year}{2005}).
504: 
505: \bibitem[{\citenamefont{Mujumdar et~al.}(2004)\citenamefont{Mujumdar, Ricci,
506:   Torre, and Wiersma}}]{Mujumdar04}
507: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Mujumdar}},
508:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Ricci}},
509:   \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Torre}}, \bibnamefont{and}
510:   \bibinfo{author}{\bibfnamefont{D.~S.} \bibnamefont{Wiersma}},
511:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{93}},
512:   \bibinfo{pages}{053903} (\bibinfo{year}{2004}).
513: 
514: \bibitem[{\citenamefont{Liu et~al.}(2003)\citenamefont{Liu, Yamilov, Ling, Xu,
515:   and Cao}}]{Liu03}
516: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Liu}},
517:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Yamilov}},
518:   \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Ling}},
519:   \bibinfo{author}{\bibfnamefont{J.~Y.} \bibnamefont{Xu}}, \bibnamefont{and}
520:   \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Cao}},
521:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{91}},
522:   \bibinfo{pages}{063903} (\bibinfo{year}{2003}).
523: 
524: \bibitem[{\citenamefont{Deych}(2005)}]{Deych05}
525: \bibinfo{author}{\bibfnamefont{L.~I.} \bibnamefont{Deych}},
526:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{95}},
527:   \bibinfo{pages}{043902} (\bibinfo{year}{2005}).
528: 
529: \bibitem[{\citenamefont{Angelani et~al.}(2006)\citenamefont{Angelani, Conti,
530:   Ruocco, and Zamponi}}]{Angelani06}
531: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Angelani}},
532:   \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Conti}},
533:   \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Ruocco}}, \bibnamefont{and}
534:   \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Zamponi}},
535:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{96}},
536:   \bibinfo{pages}{065702} (\bibinfo{year}{2006}).
537: 
538: \bibitem[{\citenamefont{Boyd}(2002)}]{BoydBook}
539: \bibinfo{author}{\bibfnamefont{R.~W.} \bibnamefont{Boyd}},
540:   \emph{\bibinfo{title}{Nonlinear Optics}} (\bibinfo{publisher}{Academic
541:   Press}, \bibinfo{address}{New York}, \bibinfo{year}{2002}),
542:   \bibinfo{edition}{2nd} ed.
543: 
544: \bibitem[{\citenamefont{Spivak and Zyuzin}(2000)}]{Spivak00}
545: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Spivak}} \bibnamefont{and}
546:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Zyuzin}},
547:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{84}}, \bibinfo{pages}{1970}
548:   (\bibinfo{year}{2000}).
549: 
550: \bibitem[{\citenamefont{Skipetrov and Maynard}(2000)}]{Skipetrov00}
551: \bibinfo{author}{\bibfnamefont{S.~E.} \bibnamefont{Skipetrov}}
552:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Maynard}},
553:   \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{85}}, \bibinfo{pages}{736}
554:   (\bibinfo{year}{2000}).
555: 
556: \bibitem[{\citenamefont{Skipetrov}(2001)}]{Skipetrov01}
557: \bibinfo{author}{\bibfnamefont{S.~E.} \bibnamefont{Skipetrov}},
558:   \bibinfo{journal}{\pre} \textbf{\bibinfo{volume}{63}},
559:   \bibinfo{pages}{056614} (\bibinfo{year}{2001}).
560: 
561: \bibitem[{\citenamefont{Skipetrov}(2003)}]{Skipetrov03}
562: \bibinfo{author}{\bibfnamefont{S.~E.} \bibnamefont{Skipetrov}},
563:   \bibinfo{journal}{\pre} \textbf{\bibinfo{volume}{67}},
564:   \bibinfo{pages}{016601} (\bibinfo{year}{2003}).
565: 
566: \bibitem[{\citenamefont{Conti et~al.}(2004{\natexlab{a}})\citenamefont{Conti,
567:   Di~Falco, and Assanto}}]{Conti04}
568: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Conti}},
569:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Di~Falco}}, \bibnamefont{and}
570:   \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Assanto}},
571:   \bibinfo{journal}{Opt. Express} \textbf{\bibinfo{volume}{12}},
572:   \bibinfo{pages}{823} (\bibinfo{year}{2004}{\natexlab{a}}),
573:   \bibinfo{note}{arXiv:physics/0403013}.
574: 
575: \bibitem[{\citenamefont{Foster et~al.}(2006)\citenamefont{Foster, Turner,
576:   Sharping, Schmidt, Lipson, and Gaeta}}]{Foster06}
577: \bibinfo{author}{\bibfnamefont{M.~A.} \bibnamefont{Foster}},
578:   \bibinfo{author}{\bibfnamefont{A.~C.} \bibnamefont{Turner}},
579:   \bibinfo{author}{\bibfnamefont{J.~E.} \bibnamefont{Sharping}},
580:   \bibinfo{author}{\bibfnamefont{B.~S.} \bibnamefont{Schmidt}},
581:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Lipson}}, \bibnamefont{and}
582:   \bibinfo{author}{\bibfnamefont{A.~L.} \bibnamefont{Gaeta}},
583:   \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{441}},
584:   \bibinfo{pages}{960} (\bibinfo{year}{2006}).
585: 
586: \bibitem[{\citenamefont{Kippenberg et~al.}(2004)\citenamefont{Kippenberg,
587:   Spillane, and Vahala}}]{Kippenberg04}
588: \bibinfo{author}{\bibfnamefont{T.~J.} \bibnamefont{Kippenberg}},
589:   \bibinfo{author}{\bibfnamefont{S.~M.} \bibnamefont{Spillane}},
590:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{K.~J.}
591:   \bibnamefont{Vahala}}, \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{93}},
592:   \bibinfo{pages}{083904} (\bibinfo{year}{2004}).
593: 
594: \bibitem[{\citenamefont{Nakazawa et~al.}(1989)\citenamefont{Nakazawa, Suzuki,
595:   and Haus}}]{Nakazawa89}
596: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Nakazawa}},
597:   \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Suzuki}}, \bibnamefont{and}
598:   \bibinfo{author}{\bibfnamefont{H.~A.} \bibnamefont{Haus}},
599:   \bibinfo{journal}{{IEEE} J. Quantum Electron.} \textbf{\bibinfo{volume}{25}},
600:   \bibinfo{pages}{2036} (\bibinfo{year}{1989}).
601: 
602: \bibitem[{\citenamefont{Angelani et~al.}(2005)\citenamefont{Angelani, Foffi,
603:   Sciortino, and Tartaglia}}]{Angelani04}
604: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Angelani}},
605:   \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Foffi}},
606:   \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Sciortino}},
607:   \bibnamefont{and}
608:   \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Tartaglia}},
609:   \bibinfo{journal}{J. Phys. : Condens. Matter} \textbf{\bibinfo{volume}{17}},
610:   \bibinfo{pages}{L113} (\bibinfo{year}{2005}), \bibinfo{note}{we used the data
611:   at temperature T=0.26 in the units of this reference}.
612: 
613: \bibitem[{\citenamefont{DeVore}(1951)}]{DeVore51}
614: \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{DeVore}},
615:   \bibinfo{journal}{J. Opt. Soc. Am.} \textbf{\bibinfo{volume}{41}},
616:   \bibinfo{pages}{416} (\bibinfo{year}{1951}).
617: 
618: \bibitem[{\citenamefont{Taflove and Hagness}(2000)}]{TafloveBook}
619: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Taflove}} \bibnamefont{and}
620:   \bibinfo{author}{\bibfnamefont{S.~C.} \bibnamefont{Hagness}},
621:   \emph{\bibinfo{title}{Computational Electrodynamics: the finite-difference
622:   time-domain method}} (\bibinfo{publisher}{Artech House},
623:   \bibinfo{year}{2000}), \bibinfo{edition}{3rd} ed.
624: 
625: \bibitem[{\citenamefont{Conti et~al.}(2004{\natexlab{b}})\citenamefont{Conti,
626:   Di~Falco, and Assanto}}]{Conti04b}
627: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Conti}},
628:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Di~Falco}}, \bibnamefont{and}
629:   \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Assanto}},
630:   \bibinfo{journal}{\ol} \textbf{\bibinfo{volume}{29}}, \bibinfo{pages}{2902}
631:   (\bibinfo{year}{2004}{\natexlab{b}}).
632: 
633: \bibitem[{\citenamefont{John}(1987)}]{John87}
634: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{John}}, \bibinfo{journal}{\prl}
635:   \textbf{\bibinfo{volume}{58}}, \bibinfo{pages}{2486} (\bibinfo{year}{1987}).
636: 
637: \bibitem[{\citenamefont{Mandelshtam}(2001)}]{Mandelshtam01}
638: \bibinfo{author}{\bibfnamefont{V.~A.} \bibnamefont{Mandelshtam}},
639:   \bibinfo{journal}{Progress in NMR Spectroscopy}
640:   \textbf{\bibinfo{volume}{38}}, \bibinfo{pages}{159} (\bibinfo{year}{2001}),
641:   \bibinfo{note}{we used the Harminv library by Steven G. Johnson (MIT).}
642: 
643: \bibitem[{\citenamefont{Vandenbem and Vigneron}(2005)}]{Vandenbem05}
644: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Vandenbem}} \bibnamefont{and}
645:   \bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Vigneron}},
646:   \bibinfo{journal}{\josaa} \textbf{\bibinfo{volume}{22}},
647:   \bibinfo{pages}{1042} (\bibinfo{year}{2005}).
648: 
649: \bibitem[{\citenamefont{Haus}(1984)}]{HausBook}
650: \bibinfo{author}{\bibfnamefont{H.~A.} \bibnamefont{Haus}},
651:   \emph{\bibinfo{title}{Waves and Fields in Optoelectronics}}
652:   (\bibinfo{publisher}{Prentice-Hall}, \bibinfo{address}{Englewood Cliffs, N.
653:   J.}, \bibinfo{year}{1984}).
654: 
655: \end{thebibliography}
656: 
657: 
658: \end{document}
659: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
660: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
661: