1: %\documentclass[preprint,aps,showpacs]{revtex4}
2: %\documentclass[prl,aps,twocolumn,amssymb,showpacs]{revtex4}
3: \documentclass[aps,floatfix,nofootinbib,showpacs,espfig,twocolumn,amssymb]{revtex4}
4: \usepackage{graphicx}
5: \usepackage{latexsym}
6: \begin{document}
7: %\draft
8: %
9: \title{UNIVERSAL BEHAVIOR OF THE COEFFICIENTS OF THE CONTINUOUS EQUATION IN COMPETITIVE GROWTH MODELS}
10: \author{D. Muraca$^1$, L. A. Braunstein$^{1,2}$ and R. C. Buceta$^1$}
11:
12: \affiliation{$^1$Departamento de F\'{\i}sica, Facultad de Ciencias Exactas y
13: Naturales\\ Universidad Nacional de Mar del Plata\\ Funes 3350,
14: $7600$ Mar del Plata, Argentina\\
15: $^2$Center for Polymer Studies and Department of Physics\\
16: Boston University, Boston, MA 02215, USA}
17: \pacs{81.15.Aa, 05.40.-a, 05.10.Gg}
18:
19: \begin{abstract}
20: The competitive growth models involving only one
21: kind of particles (CGM), are a mixture of two processes one with probability $p$ and the
22: other with probability $1-p$. The $p-$dependance produce crossovers between two different regimes.
23: We demonstrate that the coefficients of the continuous equation, describing their universality
24: classes, are quadratic in $p$ (or $1-p$). We show that the origin of such dependance is the
25: existence of two different average time rates. Thus, the quadratic $p-$dependance is a universal
26: behavior of all the CGM. We derive analytically the continuous equations for two CGM, in $1+1$
27: dimensions, from the microscopic rules using a regularization procedure. We propose
28: generalized scalings that reproduce the scaling behavior in each
29: regime. In order to verify the analytic results and the scalings,
30: we perform numerical integrations of the derived analytical equations.
31: The results are in excellent agreement with those of the
32: microscopic CGM presented here and with the proposed scalings.
33: \end{abstract}
34:
35: \maketitle
36:
37: Evolving growing interfaces or surfaces can be found in many
38: physical, chemical and biological processes. For example, in film growth either
39: by vapour deposition or chemical deposition \cite{family,barabasi}, bacterial growth \cite{albanobact} and
40: propagation of forest fire \cite{clar}.
41: The resulting interface has a rough surface that is characterized
42: through scaling of the interfacial width $W$ defined as
43: \begin{math}
44: W(L,t)=\left \{ \langle[h_i- \langle h_i \rangle]^2\rangle^{1/2} \right\}
45: \end{math},
46: where $h_i$ is the height at the position $i$, $\langle h_i \rangle = \sum_{i=1}^{L^d} h_i$ is the
47: spatial average, $L$ is the linear size, $d$ is the spatial dimension and $\left\{ \right\}$ denote configuration averages.
48: The general scaling relation \cite{family} for these growing interfaces that evolves through a
49: single model can be summarized in the form $W(L,t) \sim L^\alpha f(t/L^z)$,
50: where the scaling function $f(u)$ behaves as $f(u) \sim u^\beta$
51: ($\beta= z/\alpha$), for $u \ll 1$ and $f(u) \sim$ const for $u
52: \gg 1$. The exponent $\alpha$ describes the asymptotic behavior where the width saturates due to finite sizes
53: effects, while the exponent $\beta$ represent the early time
54: regime where finite-size effects are weak. The crossover time
55: between the two regimes is $t_s=L^{z}$.
56:
57: The study of growth models involving one kind of particles in competitive processes (CGM) has received little attention, in spite of the fact that they are more realistic describing the growing in real materials, where usually there exist a competition between different growing processes. As an example, in a colony of bacteria growing on a substrate, a new bacteria can born near to another and stay there, move into another place looking for food or died. This ``bacteria'' can be thought as a particle undergoing either a deposition/evaporation process or deposition/surface relaxation.
58:
59: The processes involved in the CGM could have different characteristic average time rate. Recently Shapir et al. \cite{shapir} reported experimental results of surface roughening during cyclical electrodeposition dissolution of silver. Horowitz et al. \cite{albano} introduced a competitive growth model between random deposition with surface relaxation (RDSR) with probability $p$ and random deposition (RD) with probability $1-p$, called RDSR/RD. The authors proposed that the scaling behavior is characteristic of an Edward Wilkinson (EW) equation, where the coefficient associated to the surface tension $\nu$ depends on $p$. The dependance of $\nu$ on $p$ governs the transition from RDRS to RD. Using a dynamic scaling ansatz for the interface width
60: $W$ they found that the results are consistent provide that
61: $\nu \propto p^2$. Also Pellegrini and Jullien \cite{pellegrini1} have introduced CGM between ballistic deposition (BD) with probability $p$ and RDSR with probability $1-p$, called BD/RDSR. For this model Chame and Aar$\tilde{a}$o Ries \cite{reis}
62: presented a more carefully analysis in $1+1$-$d$ and showed that
63: there exist a slow crossover from an EW to a
64: Kardar-Parisi-Zhang (KPZ) for any $p >0$. They also found that the
65: parameter $p$ is connected to the coefficient $\lambda$ of the
66: nonlinear term of the KPZ equation by $\lambda \sim p^{\gamma}$,
67: with $\gamma=2.1$.
68:
69: In this letter, we show that the origin of such dependance is the
70: existence of two different average time rates. Thus, the quadratic $p-$dependance is an universal
71: feature of all the CGM. To our knowledge this is the first
72: time that the $p$-dependance on the coefficient of the continuous
73: equations is obtained analytical from the microscopic dynamics.
74:
75: In order to test our hypothesis, we derive the analytical continuous equations for the local height for the RDSR/RD and BD/RDSR models. The procedure chosen here is based on regularization and
76: coarse-graining of the discrete Langevin equations obtained from a
77: Kramers-Moyal expansion of the master equation \cite{VK,Vvedensky,lidia}.
78:
79: Letīs introduce first the general treatment of
80: this problem. Let us denote by $h_i(t)$ the height of the $i$-th
81: generic site at time $t$. The set $\{h_i,\,i=1,\dots ,L\}$ defines
82: the interface. Here we distinguish between two competitive
83: processes: A with probability p and average time of deposition $\tau_A$,
84: and B with probability 1-p and average time of deposition $\tau_B$.
85: In deposition processes with $p=1$ the average rate of deposition
86: is given by $\tau_0^{-1}=\left\{dh_i/dt\right\}_{(p=1)}$. If the
87: process is made with probability $p$ the average rate of deposition
88: is given by $\tau_A^{-1}=\left\{dh_i/dt\right\}_{(p)}=p\,\left\{dh_i/dt\right\}_{(p=1)}$.
89: The same hold for a process with probability $(1-p)$.
90: Thus, the particles are deposited at an average rate
91: \begin{equation}\label{Eq.tau}
92: \tau_A=\frac{\tau_0}{p}\;,\hspace{5ex}\
93: \tau_B =\frac{\tau_0}{1-p}\;.
94: \end{equation}
95: In the average time of each process, the height in
96: the site $i$ increases by
97: \begin{eqnarray} \label{Eq.rd}
98: h_i(t + \tau_A)&=& h_i(t) + a_{\perp}\,p\;R_i^A\;,\nonumber\\
99: h_i(t + \tau_B)&=& h_i(t) + a_{\perp}\,(1-p)\;R_i^B\;,
100: \end{eqnarray}
101: where $R_i^A$ and $R_i^B$ are the growing rules for
102: processes $A$ and $B$ respectively and $a_{\perp}$ is the vertical
103: lattice spacing. Expanding $h_i(t + \tau_A)$
104: and $h_i(t + \tau_B)$ to second order in Taylor series around
105: $\tau_A$ and $\tau_B$, we obtain
106: \begin{equation} \label{Eq.dt}
107: h_i(t + \tau_J)- h_i(t) \approx\frac{dh_i}{dt} \, \tau_J\;,
108: \end{equation}
109: for the process $J=A,B$. Thus, the evolution equation for the height (in the site $i$) for
110: this CGM is given by
111: \begin{equation}\label{Eq.3}
112: \frac{d h_i}{d t} =K_i^{(1,A)} + K_i^{(1,B)}+\eta_i(t)\;,
113: \end{equation}
114: where the first moments of the transition rate for each
115: process \cite{foot} are
116: \begin{eqnarray}\label{Eq.k}
117: K_i^{(1,A)}&=&\frac{a_{\perp}}{\tau_A}\,p\;R_i^A\;,\nonumber \\
118: K_i^{(1,B)}&=&\frac{a_{\perp}}{\tau_B}\,(1-p)\;R_i^B\;,
119: \end{eqnarray}
120: and the Gaussian thermal noise $\eta_i(t)$ has zero mean and covariance
121: \begin{equation}\label{Eq.4}
122: \left\{\eta_i(t) \eta_j(t')\right\} = a_{\perp}
123: \left(K^{(1,A)}_i + K^{(1,B)}_i\right)\delta_{ij}\,\delta(t-t')\;.
124: \end{equation}
125:
126: In order to test our analytical result, we use two models. The
127: first model RDSR/RD considers a mixture \cite{albano} of RDSR
128: (process A) with probability $p$ and RD (process B) with
129: probability $1-p$. Letīs introduce the growth rule for each
130: process for the first model. In the RD growth model one chose a
131: column of a lattice, at random, among $L$ and a particle is
132: launched until it reaches the top of the selected column. The RDSR
133: is a variant of the RD: a particle is released from a random
134: position but when it reaches the top of the selected column is
135: allowed to relax to the nearest neighbor (nn) column if their
136: height are lower that the selected one. If the height of both of
137: the nn are lower than the selected one the relaxation takes place
138: with equal probability to one of them. For RD, $W(L,t)$ does not
139: depend on $L$, this means that the width $W$ does not saturate due
140: to the lack of lateral correlations. Thus, in this model: $W(t)
141: \sim t^{\beta_{RD}}$. Moreover, the RDSR model generates lateral
142: correlations, therefore one has $\beta_{RDSR}= 1/4$ and
143: $\alpha_{RDSR}= 1/2$. The first moment of the transition rate for
144: these processes are
145: \begin{eqnarray}\label{Eq.k1}
146: K_i^{(1,A)}&=& \frac{a_{\perp}}{\tau_A} \,p \left(
147: \omega_i^{(2)}+\omega_{i+1}^{(3)}+\omega_{i-1}^{(4)} \right)\;,\nonumber \\
148: K_i^{(1,B)}&=& \frac{a_{\perp}}{\tau_B} \,(1-p)\;\omega^{(1)}_i\;,
149: \end{eqnarray}
150: where the rules for both processes can we written as
151: \begin{eqnarray}\label{Eq.rules1}
152: \omega^{(1)}_i&=&1\;,\nonumber\\
153: \omega^{(2)}_i&=&\Theta(H^{i+1}_i)\;\Theta(H^{i-1}_i)\;,\\
154: \omega^{(3)}_i&=& \left\{\textstyle\frac{1}{2}\left[1-\Theta(H^{i+1}_{i})\right]+
155: \Theta(H^{i+1}_{i})\right\}\left[1-\Theta(H^{i-1}_{i})\right]\;,\nonumber \\
156: \omega^{(4)}_i&=& \left\{\textstyle\frac{1}{2}\left[1-\Theta(H^{i-1}_{i})\right]+
157: \Theta(H^{i-1}_{i})\right\}\left[1-\Theta(H^{i+1}_{i})\right]\;.
158: \nonumber
159: \end{eqnarray}
160: where $H^{i\pm s}_{i\pm k}=(h_{i\pm s}-h_{i\pm k})/a_{\perp}$,
161: and $\Theta(z)$ is the unit step function defined as
162: $\Theta(z)=1$ for $z\ge 0 $ and $\Theta(z)= 0$ for $z<0$.
163: The representation of the step function can be expanded as
164: $\Theta(z)=\sum_{k=0}^{\infty} c_k z^k$ providing that $z$ is
165: smooth. In any discrete model there is in principle an infinite
166: number of nonlinearities, but at long wavelengths the higher order
167: derivatives can be neglected using scaling arguments, since one
168: expect affine interfaces over a long range of scales, and then one
169: is usually concerned with the form of the relevant terms. Thus,
170: keeping the expansion of the step function to first order in his
171: argument and replacing the expansion Eq.~(\ref{Eq.rules1}), Eq.~(\ref{Eq.3}) can be
172: written as
173: \begin{equation}\label{Eq.dh-dt}
174: \frac{dh_i}{dt}= \frac{a_{\perp}(1- p)}{\tau_B} + \frac{a_{\perp}
175: p}{\tau_A} \left( 1 + c_1\,\frac{\Delta^2 h_i}{a_\perp}\right)+\eta_i(t)\;,
176: \end{equation}
177: where $\Delta^2 h_i= h_{i+1}- 2 h_i +h_{i+1}\simeq a_\parallel^2\;\partial^2 h/\partial x^2\rfloor_{h_i}$, and $a_{\parallel}$ is the horizontal lattice spacing. Replacing the rates
178: given by Eq.~(\ref{Eq.tau}) in Eq.~(\ref{Eq.dh-dt}) and
179: using a standard coarse-grain approach \cite{Vvedensky,lidia} the continuous
180: equation for this CGM is
181: \begin{equation}\label{Eq.c1}
182: \frac{dh}{dt}= F(p) + \nu(p)\,\frac{\partial^2 h}{\partial\,x^2}+
183: \eta(x,t)\;,
184: \end{equation}
185: where $h=h(x,t)$ and
186: \begin{eqnarray}\label{Eq.coef1}
187: F(p) &=& \frac{a_{\perp}}{\tau_0} \left[(1-p)^2 + p^2\right]\;,\\
188: \nu(p) &=& 2\;c_1\frac{a_\parallel^2}{\tau_0}\, p^2\;. \nonumber
189: \end{eqnarray}
190: The noise covariance is given by
191: \begin{equation}\label{Eq.cov}
192: \left\{\eta(x,t) \eta(x',t') \right\} = D(p)\;\delta(x -
193: x') \delta (t - t')\;,
194: \end{equation}
195: where
196: \begin{equation}\label{Eq.dif}
197: D(p)=a_\parallel\,a_\perp\, F(p)\;.
198: \end{equation}
199: Equations (\ref{Eq.coef1}) and Eq.~(\ref{Eq.dif}) shows that the quadratic dependance on
200: the coefficients of the continuous equation, arises naturally
201: as a feature of the CGM and is due to the existence
202: of different average time rates.
203:
204: The second model is a mixture of RDSR with probability $1-p$ ($B$
205: process) and ballistic deposition (BD) with probability $p$ ($A$
206: process) \cite{reis}. The evolution rules for RDSR are $\omega_i^{j}$, with $j=2,3,4$
207: [see Eq.~(\ref{Eq.rules1})]. In the BD model, the incident
208: particle follows a straight trajectory and sticks to the surface at the column $i$.
209: The height in the column $i$ is increased in
210: $\max[h_i+1,h_{i+1},h_{i-1}]$. If this process is done with
211: probability $p$ (A process), the rules can be summarized as:
212: \begin{eqnarray}\label{Eq.k2}
213: \omega^{(5)}_i&=&\Theta(H^i_{i+1})\; \Theta(H^i_{i-1})\;,\\
214: \omega^{(6)}_i&=&H^{i+1}_i\left[1-\Theta(H^i_{i+1})\right]\left[1-\Theta(H^{i-1}_{i+1})\right]\;,\nonumber\\
215: \omega^{(7)}_i&=& H^{i-1}_i\left[1-\Theta(H^i_{i-1})\right]\left[1-\Theta(H^{i+1}_{i-1})\right]\;,\nonumber\\
216: \omega^{(8)}_i&=&\textstyle\frac{1}{2}\; \delta(H^{i+1}_{i-1},0)\;
217: \left\{H^{i+1}_i\left[1-\Theta(H^i_{i+1})\right]\right.\nonumber\\
218: &&\hspace{14ex}+\left.H^{i-1}_i\left[1-\Theta(H^i_{i-1})\right]\right\}\;,\nonumber
219: \end{eqnarray}
220: where $\delta(z,0)=\Theta(z)+\Theta(-z)-1$ is the Kronecker delta.
221: Following the steps leading to Eq.~(\ref{Eq.c1}) the evolution
222: equation for this process can be written as:
223:
224: \begin{equation}\label{Eq.c2}
225: \frac{dh}{dt}= F(p) + \nu(p) \;\frac{\partial^2 h}{\partial x^2}+
226: \lambda(p) \left(\frac{\partial h}{\partial x}\right)^2 +
227: \eta(x,t)
228: \end{equation}
229: where
230: \begin{eqnarray}\label{Eq.coef2}
231: F(p) &=& \frac{a_{\perp}}{\tau_0} \left[(1-p)^2 + c_0^2 \; p^2\right]\;,\nonumber\\
232: \nu(p) &=& \frac{a_\parallel^2}{\tau_0} \left[\textstyle\frac{1}{2}\,p^2 (1-c_0-2 c_0 c_1)+2 c_1 (1-p)^2\right],\\
233: \lambda(p)&=&\frac{a_\parallel^2}{\tau_0\,a_\perp}p^2\,c_1\,\left(5-4
234: c_0 -c_1\right)\;.\nonumber
235: \end{eqnarray}
236: The covariance of noise and $D(p)$ is given by Eq.~(\ref{Eq.cov}) and Eq.~(\ref{Eq.dif}), respectively.
237: Notice that we have to change $p$ by $1-p$ in
238: all the above equations for RDSR, because in the first model RDSR
239: is a kind $A$ process and now is a kind B process.
240: Equation~(\ref{Eq.coef2}) shows again that quadratic dependance on
241: the coefficients of the continuous equation. The quadratic dependance
242: of $\lambda$ on $p$, found by Chame and Aar$\tilde{a}$o
243: Reis \cite{reis}, is a general feature of the CGM.
244:
245: As both models have an EW behavior, it is expected that in that regime
246: the following generalized scaling ansatz \cite{natterman,albano},
247: \begin{equation}\label{Eq.scal}
248: W^2(p,L,t) \sim L^{2 \alpha}\;[D(p)/\nu(p)]\;
249: f\left(\nu(p)\,t/L^z\right)\;,
250: \end{equation}
251: where $f(u)\sim u^{2\beta }$ for $u\ll 1$ and $f(u)\sim
252: \mbox{const}$ for $u \gg 1$. Moreover, the second model is
253: represented by a mixture of EW and KPZ universality classes. In
254: the early time regime $W(t)\sim t^{\beta_{RDSR}}$, while a
255: crossover to a KPZ, with $\beta_{KPZ}= 1/3$ and
256: $\alpha\equiv\alpha_{KPZ}=1/2$, is expected in the intermediate
257: regime before the saturation. Thus, for the KPZ regime we propose
258: the following generalization \cite{amar} of the scaling behavior
259: of the width
260: \begin{equation}\label{Eq.scal-kpz}
261: W^2(p,L,t)\!\sim L^{2\alpha} [D(p)/\nu(p)]\,f\!\left(\lambda(p)
262: \sqrt{D(p)/\nu(p)}\,t/L^z\right)\,,
263: \end{equation}
264: where $z=3/2$, and $f(u) \sim u^{2
265: \beta_{KPZ} }$ for $u \ll 1$ and $f(u) \sim\mbox{const}$ for $u
266: \gg 1$.
267:
268: In order to test our analytical result and the proposed scalings, we
269: perform a numerical integration of Eq.~(\ref{Eq.c1}) and
270: Eq.~(\ref{Eq.c2}), and compute $W^2$ for both models.
271: \begin{figure}[h]
272: \includegraphics[width=5.5cm,height=5cm]{fig1a.eps}
273: \includegraphics[width=5.5cm,height=5cm]{fig1b.eps}
274: \caption{
275: (a) Log-log plot of $W^2\nu(p)/D(p)$ for the RDSR/RD model as function of
276: $\nu(p)\,t$ for $L=128$. The different symbols represent different values of $p$,
277: $p=0.04$ ($\circ$), $p=0.08$ ($\Box$), $p=0.016$ ($\bigtriangleup$), $p=0.32$ ($+$), and
278: $p=0.64$ ($*$). Here we used $C=2.58$ and $b=0.2$ as parameters of the
279: $\Theta$-function representation. The dashed lines are used as guides to show the RD regime with $2 \beta = 1$ and
280: the EW regime with $2 \beta = 0.5$.
281: (b) Log-log plot of $W^2 \nu(p) /D(p)$ for the BD/RDSR model as function of
282: $\nu(p)\,t$ for $L = 1024$. The symbols represent the same as in Fig.(1a).
283: Here we used $C=0.18$ and $b=0.5$. The collapse of the curves at the earlier stage clearly shows the EW
284: behavior ($2 \beta = 0.5$). After this stage the curves split and undergoes a slow
285: crossover to the KPZ behavior ($2 \beta = 0.66$). The dashed lines are used as guides to show the EW regime with $2 \beta = 1/2$ and the KPZ regime. The slope showed here is $2\beta=0.61$.\label{f1.1}}
286: \end{figure}
287: Notice that in order to numerically integrate the continuous equation, we
288: do need a continuum representation of the $\Theta$-function to
289: numerically compute the coefficients $c_0$ and $c_1$ related to
290: the ones of the continuous equations.
291: To perform the numerical integration, we chose the shifted
292: hyperbolic tangent \cite{Predota} as the continuous representation
293: of $\Theta$-function defined as $\Theta(x)=\{1+ \tanh [C (x
294: +b)]\}/2$, where $b$ is the shift and $C$ is a parameter that
295: allows to recover the $\Theta$ in the limit $C \to \infty$. The numerical integration
296: was made in short lattices using a discretized version of the
297: continuous equations Eq.(\ref{Eq.c1}) and Eq.~(\ref{Eq.c2}). The
298: results in large systems and the details of the integration are
299: beyond the scope of this letter and will be published elsewhere.
300:
301: \begin{figure}
302: \includegraphics[width=6cm,height=5cm]{fig2.eps}
303: \caption{Log-log plot of $\tilde{W}^2$ as function of $\tilde t$ as defined in the text, for $p=0.16$ ($\circ$), $p=0.32$ ($\Box$) and $p=0.64$ ($\bigtriangleup$). The empty symbols correspond to $L=512$ and the
304: filled ones to $L=1024$ . The collapse of the curves on the saturation regime using
305: $z=3/2$ shows that the curves saturates with a KPZ behavior as expected.\label{f1.2}}
306: \end{figure}
307:
308: For the first model, Horowitz et al. \cite{albano} presented their
309: data from simulations plotting the scaling relation $W/L^\alpha
310: p^{-\delta}$ vs $t/L^z p^{-y}$. Clearly, their $\delta=1$ and
311: $y=2$ is related to our $\nu(p)$ and $D(p)$ [see
312: Eq.~(\ref{Eq.coef1}) and Eq.~(\ref{Eq.dif})]. In
313: Fig.~{\ref{f1.1}(a) we plot $W^2\,\nu(p)\,/D(p)$ as function of
314: $\nu(p)\,t$ for the that model for different values of $p$ and
315: $L=128$. This figure represent the same as in \cite{albano}
316: after coarse-graining. The agreement with the results of our
317: numerical integration, the numerical simulation \cite{albano} and
318: the scaling presented in Eq.~(\ref{Eq.scal}) is excellent. On the
319: other hand, for the second model, Chame and Aar$\tilde{a}$o Reis
320: \cite{reis} did not present the result for $W$. They studied the
321: crossover from EW to KPZ using an indirect method because of the
322: slow convergence of the discrete model to KPZ behavior. The crossover is
323: well represented in our Fig.~{\ref{f1.1}(b), where we plot the
324: same as in Fig.~{\ref{f1.1}(a) but for the second model. It is
325: clear the collapse of the curves in the EW regime. In the
326: intermediate regime the KPZ behavior appears thus, it is expected
327: that Eq.~(\ref{Eq.scal-kpz}) holds in that regime. In
328: Fig.~{\ref{f1.2} we plot
329: $\tilde{W}^2=W^2\,\nu(p)/[L^{2\alpha}D(p)]$ as function of
330: $\tilde{t}=\lambda(p)\sqrt{D(p)/\nu(p)}\;t/L^z$ for three
331: different values of $p$ using $z=3/2$. As $p$ increases, the KPZ
332: behavior appears earlier, but independent of $p$ all the curves
333: saturate as a KPZ. The agreement with Eq.~(\ref{Eq.scal-kpz}) is
334: excellent in the saturation regime. The departure in the intermediate regime is due
335: to a slow crossover to the KPZ and to finite size effects.
336:
337:
338: Finally, notice that the quadratic dependence of the coefficients
339: of the continuous equation on $p$ is independent of the CGM
340: considered, because it is due to two different rates of deposition
341: given by Eq.~(\ref{Eq.tau}). This dependence is totally generally,
342: as shown from Eq.~(\ref{Eq.tau}) to Eq.~(\ref{Eq.k1}).
343:
344: In summary, we demonstrate that the coefficient of the continuous equation have
345: quadratic dependance on $p$. This feature is universal for all the CGM model and
346: is due to the competition between different average time rate. We propose
347: generalized scaling for the model that reproduce the scaling behavior in each
348: regime. The numerical integration of the continuous equation are in excellent
349: agreement with the propose scalings and the numerical simulation of the models.
350:
351: Acknowledgements: We thanks ANPCyT and UNMdP (PICT 2000/1-03-08974)
352: for the financial support.
353:
354: \begin{thebibliography}{99}
355:
356: \bibitem{family}F. Family, J. Phys. A: Math. Gen. {\bf 19}, L441 (1986).
357: \bibitem{barabasi}A.-L. Barab\'{a}si and H. E. Stanley, {\it Fractal Concepts in
358: Surface Growth}, Cambridge Univ. Press, New York (1995); P.Meakin, {\it Fractals, scaling and growth far from equilibrium}, Cambridge Univ. Press, Cambridge (1998).
359: \bibitem{albanobact} E. V. Albano, R. C. Salvarezza, L. V\'azquez and A. J. Arvia, Phys. Rev. B {\bf 59}, 7354 (1999).
360: \bibitem{clar} S. Clar, B. Drossel, and F. Schwabl, J. Phys.: Condens. Matter {\bf 8}, 6803 (1996).
361: \bibitem{shapir} Y. Shapir, S. Raychaudhuri, D. G. Foster, and J. Jorne, Phys. Rev. Lett. {\bf 84}, 3029 (2000).
362: \bibitem{albano} C. M. Horowitz, R. A. Monetti and E. V. Albano, Phys. Rev. E {\bf 63}, 066132 (2001).
363: \bibitem{pellegrini1} Y.P. Pellegrini and R. Jullien, Phys. Rev. Lett. {\bf 64}, 1745 (1990); {\it ibid.}
364: Phys. Rev. A {\bf 43}, 920 (1991).
365: \bibitem{reis} A. Chame and F. D. A. Aar$\tilde{a}$o Reis , Phys. Rev. E {\bf 66}, 051104 (2002).
366: \bibitem{VK} N. G. Van Kampen, {\it Stochastic Processes in Physics and
367: Chemistry}, North-Holland, Amsterdam (1981).
368: \bibitem{Vvedensky} D. D. Vvedensky, Phys. Rev. E {\bf 67}, 025102(R) (2003).
369: \bibitem{lidia} L. A. Braunstein, R. C. Buceta, C. D. Archubi and G. Costanza, Phys. Rev. E {\bf 62}, 3920 (2000).
370: \bibitem {foot}Notice that if the model involves only the nearest
371: neighbors the evolution equation contains only the first moment \protect\cite{Vvedensky2}.
372: \bibitem{Vvedensky2}D. D. Vvedensky, A. Zangwill, C. N. Luse, and M. R Wilby, Phys Rev E {\bf 48}, 852 (1993); G. Costanza, Phys. Rev. E {\bf 55}, 6501 (1997).
373: \bibitem{natterman}T. Nattermann and Lei-Han Tang, Phys. Rev. A {\bf 45}, 7156 (1992).
374: \bibitem{amar} J. G. Amar and F. Family, Phys. Rev. A {\bf 45}, R3373 (1992).
375: \bibitem{Predota} M. P\v{r}edota and M. Kotrla, Phys. Rev. E {\bf 54}, 3933 (1996).
376: \end{thebibliography}
377: \end{document}
378:
379: