1: %\documentclass[aps,prl,preprint,groupedaddress,graphics]{revtex4}
2: %\usepackage{latexsym}
3:
4: \documentclass[a4paper,onecolumn,showpacs]{revtex4}
5: %\documentclass[aps,twocolumn,groupedaddress]{revtex4}
6: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7:
8: \usepackage{amsmath}
9: \usepackage{graphicx}
10: %TCIDATA{OutputFilter=LATEX.DLL}
11: %TCIDATA{LastRevised=Mon Mar 04 15:19:28 2002}
12: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
13: %TCIDATA{Language=American English}
14: %TCIDATA{CSTFile=revtex.cst}
15: \usepackage[english]{babel}
16: \usepackage[latin1]{inputenc}
17: \usepackage[T1]{fontenc}
18:
19:
20: \begin{document}
21: \title{Study of universality crossover in the contact process}
22: \author{Wellington G. \surname{Dantas}}
23: \email{wgd@if.uff.br}
24: \author{J\"urgen F. \surname{Stilck}}
25: \email{jstilck@if.uff.br}
26: \affiliation{Instituto de F\'{\i}sica, Universidade Federal
27: Fluminense, \\
28: Campus da Praia Vermelha, \\
29: Niter\'oi, RJ, 24.210-340, Brazil.}
30:
31: \date{\today}
32:
33: \pacs{05.70.Ln, 02.50.Ga, 64.60.Cn,64.60.Kw}
34:
35: \begin{abstract}
36: We consider a generalization of the contact process stochastic model,
37: including an additional autocatalitic process. The phase diagram of this model
38: in the proper two-parameter space displays a line of transitions between an
39: active and an absorbing phase which starts at the critical point of the
40: contact process and ends at the transition point of the voter model. Thus, a
41: crossover between the directed percolation and the compact percolation
42: universality classes is observed at this latter point. We study this crossover
43: by a variety of techniques. Using supercritical series expansions analyzed
44: with partial differential approximants, we obtain precise estimates of the
45: crossover behavior of the model. In particular, we find an estimate for the
46: crossover exponent $\phi=2.00 \pm 0.02$. We also show arguments that support
47: the conjecture $\phi=2$.
48: \end{abstract}
49:
50: \maketitle
51:
52: \section{Introduction}
53: \label{intro}
54: The phase transitions exhibited by stochastic models with absorbing states
55: have attracted much attention in recent years, particularly in order to
56: identify and understand the aspects which determine the universality classes
57: in those models. Most of these models have not been solved exactly, but a
58: variety of approximations allow quite conclusive results regarding their
59: critical properties. Stochastic models are, of course, well fitted for
60: simulations, but closed form approximations and other analytical approaches
61: have also been useful in investigating their behavior \cite{md99}.
62:
63: One of the simplest and most studied model of this type is the contact process
64: (CP), which was conceived as a simple model for the spreading of an
65: epidemic and proven to display a continuous transition between the absorbing
66: and an active state, even in one dimension \cite{h74}. Actually, it was found
67: that the CP is equivalent to other models such as Schl\"ogl's lattice model
68: for autocatalytic chemical reactions \cite{s72} and Reggeon Field Theory (RFT)
69: \cite{gt}. The CP belongs to the direct percolation (DP) universality class,
70: together with others models such as the Ziff-Gulari-Barshad model of catalysis
71: \cite{zgb86} and branching and annihilating walks with an odd offspring
72: \cite{tty92}. The {\em DP conjecture} states that all phase transitions between
73: an active and an absorbing state in models with a scalar order parameter,
74: short range interactions and no conservation laws belong to this class
75: \cite{j81}. This conjecture was verified in all cases studied so far
76: \cite{h00}.
77:
78: Here we study a generalization of the CP, with an additional parameter, so
79: that the CP transition point becomes a critical line. Since the symmetry
80: properties of this generalized model are the same of the CP, it is expected
81: that this critical line should belong to the DP universality class. However,
82: at one point of this line the model is equivalent to the zero temperature
83: Glauber model \cite{g63}, also called the voter model \cite{l85}, which
84: displays a spin
85: inversion (or particle-hole) symmetry and therefore belongs to another
86: universality class. Thus the critical line in the phase diagram of the
87: generalized model starts at the CP model and ends at the voter model, a
88: crossover between the two universality classes being observed. The voter model
89: belongs to the compact percolation universality class, and also corresponds to
90: a limiting point in the phase diagram of the Domany-Kinzel cellular automaton,
91: where an exact solution is possible \cite{dk84}. Thus, the exact critical
92: exponents are known for this model. In a study of models with several
93: absorbing states \cite{h97} simulational results are shown for the model we
94: are considering here and a study of the crossover between direct percolation
95: and compact direct percolation may be found in \cite{mdh96}, motivated by
96: the possibility to explain the non-universality in models with several
97: absorbing states as a surface effect. Also, the shape of the critical line
98: close to the CDP endpoint in the Domany-Kinzel automaton was studied in detail
99: \cite{kt95,tikt95,ikb01}, and these results are compared to our findings in
100: the conclusion. Some physical motivation for the model
101: we are studying here
102: might be given. In the contact process, the additional term might be
103: understood as an enhancement of the possibility of a sick individual to
104: recover proportional to the number of its first neighbor which are
105: healthy. However, our motivation to study the model is centered on its
106: simplicity and the universality class crossover present in its phase diagram.
107:
108: In section \ref{model} we define the model and explain how supercritical
109: series expansions may be obtained for it. The coefficients of the two-variable
110: series for the survival probability up to order 25 are given. Section \ref{res}
111: contains the description and the results of the Pad\'e and PDA estimates for
112: the model, with emphasis on the multicritical behavior in the voter model
113: limit. Final discussions and the conclusion may be found in section
114: \ref{conclu}.
115:
116: \section{Definition of the model and calculation of the coefficients of the
117: supercritical series for the survival probability}
118: \label{model}
119: The model is defined on a one-dimensional lattice with $N$ sites and periodic
120: boundary conditions. Each site is occupied either by a particle A or a
121: particle B, no holes are allowed. The microscopic state of the model may thus
122: be described by the set of binary variables
123: $\eta=(\eta_1,\eta_2,\ldots,\eta_N)$, where $\eta_i=0$ or $1$ if site $i$ is
124: occupied by particles B or A, respectively.
125:
126: The model evolves in time according to the following Markovian rules:
127: \begin{enumerate}
128: \item A site $i$ of the lattice is chosen at random.
129: \item If the site is occupied by a particle B, it becomes occupied by a
130: particle A with a transition rate equal to $p_a n_A/2$, where $n_A$ is the
131: number of A particles in the sites which are first neighbors to site $i$.
132: \item If site $i$ is occupied by a particle A, it may become occupied by a
133: particle B through two processes:
134: \begin{itemize}
135: \item Spontaneously, with a transition rate $p_c$.
136: \item Through an autocatalytic reaction, with a rate $p_bn_B/2$, where $n_B$
137: is the number of B particles in the sites which are first neighbors to site
138: $i$.
139: \end{itemize}
140: \end{enumerate}
141:
142: We define the time in such a way the the non-negative parameters $p_a$, $p_b$,
143: and $p_c$ obey the normalization $p_a+p_b+p_c=1$. We may then discuss the
144: behavior of the model in the $(p_a,p_c)$ plane without loss of generality.
145:
146: The probability $P(\eta,t)$ to find the system in state $\eta$ at
147: time $t$ obeys the master equation
148: \begin{eqnarray}
149: \label{eq1}
150: \frac{\partial P(\eta,t)}{\partial t} =
151: \sum_{i=1}^N [w_i(\eta^i)P(\eta^i,t)
152: -w_i(\eta)P(\eta,t)]
153: \label{me}
154: \end{eqnarray}
155: where $\eta^i$ corresponds to the following configuration
156: \begin{eqnarray}
157: \label{eq2}
158: \eta^i\equiv(\eta_1,...,1-\eta_i,...,\eta_N)
159: \end{eqnarray}
160: and $w_i(\eta)$ is the transition rate of the model, given by
161: \begin{eqnarray}
162: \label{eq3}
163: w_{i}(\eta) =
164: \frac{\mu}{2}(1-\gamma\eta_i)\sum_{\delta}
165: \eta_{i+\delta}+\eta_i,
166: \end{eqnarray}
167: where $\mu = p_a/(1-p_a)$, $\gamma = (1-p_c)/p_a$, and the sum is
168: over first neighbors of site $i$.
169:
170: It may be useful to remark that this model may be mapped to a spin
171: system if we describe sites occupied by A and B particles by an Ising
172: spin variables $\sigma_i=1$ and $\sigma_i=-1$, respectively.
173: In these variables,
174: the transition rate will be given by
175: \begin{eqnarray}
176: \label{eq6}
177: w_{i}(\sigma) =
178: \frac{\alpha}{2}\left[1+\beta \sigma_i-\frac{1}{2}(\epsilon
179: \sigma_i+\xi)
180: \sum_{\delta}\sigma_{i+\delta}\right],
181: \end{eqnarray}
182: where $\alpha =(p_a+p_b+2p_c)/2,\beta=(p_a-p_b-2p_c)/(pa+p_b+2p_c),
183: \epsilon=(p_a+p_b)/(p_a+p_b+2p_c)$, and
184: $\xi=(p_a-p_b)/(p_a+p_b+2p_c)$.
185:
186: In two particular cases, this model corresponds to well known models. If we
187: make $p_b=0$ or $\gamma=1$ contact
188: process is recovered \cite{h74}:
189: \begin{eqnarray}
190: \label{eq7}
191: w_{i}^{(CP)}(\eta)=\frac{\mu}{2}(1-\eta_i)\sum_{\delta}
192: \eta_{i+\delta}+\eta_i.
193: \end{eqnarray}
194: If now we take $p_a=p_b$ and $p_c=0$ in the spin formulation of the
195: model, the zero temperature linear Glauber model \cite{o03}, also
196: known as the voter model, is recovered \cite{l85}:
197: \begin{eqnarray}
198: w_{i}^{(LGM)}(\sigma) = \frac{\alpha}{2}\left[1-\frac{1}{2}\sigma_i
199: \sum_{\delta}\sigma_{i+\delta}\right].
200: \end{eqnarray}
201:
202: This model has been studied using mean-field approximations \cite{dts03}, as
203: well as simulations \cite{h97,dts03}. For $p_c>0$, the stationary state at low
204: values of $p_A$
205: corresponds to the absorbing state, where the density of A particles
206: $\rho_A=<N_A>/N$ vanishes. As $p_a$ is increased, a continuous phase
207: transition occurs and an active stationary state ($\rho_A>0$) is stable at
208: high values of $p_a$. Thus a critical line is present in the phase diagram
209: starting at $(p_a=0.767325(6),p_c=1-p_a=0.232674(4))$ \cite{dj91} (contact
210: process), and ending at $(p_a=1/2,p_c=0)$ (linear Glauber model), where the
211: transition is discontinuous and between two absorbing states ($\rho_A=1$ and
212: $\rho_B=1$). In figure \ref{f1} results from mean-field calculations and
213: simulations for the phase diagram are shown \cite{dts03}. While it is expected
214: that the critical exponents at the whole critical line are the ones of the
215: directed percolation (DP) universality class, a crossover to the compact
216: directed percolation (CDP) universality class exponents happens as $p_c$
217: vanishes \cite{h00}. Thus, we may recognize the point $(p_a=1/2,p_c=0)$ of the
218: phase diagram as a multicritical point. Using supercritical series expansions,
219: for the survival probability, we will study the multicritical singularity in
220: the neighborhood of this point.
221:
222: Now let us develop a two-variable supercritical series expansion for the
223: model. We follow closely the operator formalism presented in the paper by
224: Jensen and Dickman on series for the CP process and related models
225: \cite{dj91}. We may then represent the microscopic configurations of the
226: lattice by the direct product of kets
227: \begin{equation}
228: | \eta \rangle= \bigotimes_i | \eta_i \rangle,
229: \end{equation}
230: which are defined to be orthonormal
231: \begin{equation}
232: \langle \eta | \eta^\prime \rangle = \prod_i \delta_{\eta_i,\eta_i\prime}.
233: \end{equation}
234: Now we may define A particles creation and annihilation operators for the site
235: $i$:
236: \begin{eqnarray}
237: A_i^\dagger |\eta_i \rangle &=&(1-\eta_i)|\eta_i+1\rangle,\nonumber \\
238: A_i |\eta_i\rangle &=& \eta_i|\eta_i-1\rangle.
239: \end{eqnarray}
240: In this formalism, the state of the system at time $t$ is
241: \begin{equation}
242: |\psi(t) \rangle = \sum_{\{\eta\}} p(\eta,t) |\eta \rangle.
243: \end{equation}
244: If we define the projection onto all possible states as
245: \begin{equation}
246: \langle \; | \equiv \sum_{\{\eta\}} \langle\eta|,
247: \end{equation}
248: the normalization of the state of the system may be expressed as $\langle \;
249: |\psi \rangle =1$. In this notation, the master equation for the evolution of
250: the state of the system (Eq. \ref{me}) is
251: \begin{equation}
252: \frac{d |\psi(t)\rangle}{d t}=S|\psi(t)\rangle.
253: \label{me1}
254: \end{equation}
255: The evolution operator $S$ may be expressed in terms of the creation and
256: annihilation operators as $S=\lambda S_0 + V$ where
257: \begin{eqnarray}
258: S_0 &=& \sum_i [\alpha(2-A_{i-1}^\dagger A_{i-1} +
259: A_{i+1}^\dagger A_{i+1})+1](A_i-A_i^\dagger A_i), \\
260: V &=& \sum_i (A_i^\dagger + A_i^\dagger A_i -1)(A_{i-1}^\dagger A_{i-1} +
261: A_{i+1}^\dagger A_{i+1}),
262: \end{eqnarray}
263: and the new parameters
264: \[
265: \lambda \equiv \frac{2p_c}{p_a}
266: \]
267: and
268: \[
269: \alpha \equiv \frac{p_b}{2p_c}
270: \]
271: were introduced.
272:
273: We notice that the operator $S_0$ only annihilates A particles (transitions $A
274: \to B$), while the operator $V$ acts in the opposite way, generating
275: transitions $B \to A$. Thus, for small values of the parameter $\lambda$ the
276: creation of A particles is favored, and the decomposition above is convenient
277: for a supercritical perturbation expansion. Let us show explicitly the effect
278: of each operator on a configuration $(\mathcal{C})$.
279: \begin{equation}
280: S_0({\mathcal{C}})=\alpha \sum_i
281: ({\mathcal{C}}^\prime_i)+2\alpha \sum_j
282: ({\mathcal{C}}^\prime_j) + \sum_k
283: ({\mathcal{C}}^\prime_k)- [\alpha(r_1+2r_2)+r] ({\mathcal{C}}),
284: \label{as0}
285: \end{equation}
286: where the first sum is over the $r_1$ sites with A particles and one B
287: neighbor,
288: the second sum is over the $r_2$ sites with A particles and two B neighbors
289: and the third
290: sum is over all $r$ sites with A particles of the configuration
291: $(\mathcal{C})$. Configuration $({\mathcal{C}}_i^{\prime})$ is obtained
292: replacing
293: the A particle at site $i$ by a B particle. The action of operator $V$ is
294: \begin{equation}
295: V({\mathcal{C}})=\sum_i ({\mathcal{C}}^{\prime \prime}_i)+2 \sum_j
296: ({\mathcal{C}}^{\prime \prime}_j) -(q_1+2q_2)({\mathcal{C}}),
297: \label{av}
298: \end{equation}
299: where the first sum is over the $q_1$ sites with B particles and one A
300: neighbor,
301: the second sum is over the $q_2$ sites with B particles and two A
302: neighbors. Configuration $({\mathcal{C}}_i^{\prime \prime})$ is obtained
303: replacing the B particle at site $i$ in configuration $(\mathcal{C})$ by a A
304: particle.
305:
306: To obtain a supercritical expansion for the ultimate survival probability of A
307: particles, we start by remembering that in order to access the long-time
308: behavior of a quantity, it is useful to consider its Laplace transform. For
309: instance, the Laplace transform of the state of the system is
310: \begin{equation}
311: |\tilde{\psi}(s) \rangle = \int_0^\infty e^{-st} |\psi(t)\rangle,
312: \label{lt}
313: \end{equation}
314: and inserting the formal solution $|\psi(t)\rangle =e^{St} |\psi(0)\rangle$ of
315: the master equation \ref{me1}, we find
316: \begin{equation}
317: |\tilde{\psi}(s) \rangle = (s-S)^{-1} |\psi(0)\rangle.
318: \label{tpsi}
319: \end{equation}
320: The stationary state $|\psi(\infty) \rangle \equiv \lim_{t \to \infty} |\psi(t)
321: \rangle$ may then be found noticing that
322: \begin{equation}
323: |\psi(\infty) \rangle = \lim_{s \to 0} s |\tilde{\psi}(s) \rangle,
324: \end{equation}
325: which may be obtained integrating \ref{lt} by parts. A perturbative expansion
326: may be obtained assuming that $|\tilde{\psi}(s) \rangle$ may be expanded in
327: powers of $\lambda$ and using \ref{tpsi},
328: \begin{equation}
329: |\tilde{\psi}(s) \rangle = |\tilde{\psi}_0 \rangle+\lambda |\tilde{\psi}_1
330: \rangle +\lambda^2 |\tilde{\psi}_2 \rangle + \cdots = \frac{1}{s- V -\lambda
331: S_0 } |\psi(0) \rangle.
332: \end{equation}
333: Since
334: \begin{equation}
335: \frac{1}{s- V -\lambda S_0 }= \frac{1}{s-V} \left[ 1 + \lambda \frac{1}{s-V}
336: S_0
337: + \lambda^2 \frac{1}{(s-V)^2} S_0^2 + \cdots \right],
338: \end{equation}
339: we arrive at
340: \begin{eqnarray}
341: |\tilde{\psi}_0 \rangle &=& \frac{1}{s-V} |\psi(0)\rangle \nonumber \\
342: |\tilde{\psi}_1 \rangle &=& \frac{1}{s-V} S_0 |\tilde{\psi}_0 \rangle
343: \label{rr} \\
344: |\tilde{\psi}_2 \rangle &=& \frac{1}{s-V} S_0 |\tilde{\psi}_1 \rangle
345: \nonumber \\
346: &\vdots&
347: \end{eqnarray}
348: The action of the operator $(s-V)^{-1}$ on an arbitrary configuration
349: $({\mathcal{C}})$ may be found noting that
350: \begin{equation}
351: (s-V)^{-1} ({\mathcal{C}})=s^{-1} ({\mathcal{C}})+ \frac{V}{s(s-V)}
352: ({\mathcal{C}}),
353: \end{equation}
354: and using the expression \ref{av} for the action of the operator $V$, we get
355: \begin{equation}
356: (s-V)^{-1} ({\mathcal{C}})= s_q \left\{ ({\mathcal{C}}) + (s-V)^{-1} \left[
357: \sum_i
358: ({\mathcal{C}}^{\prime \prime}_i)+2 \sum_j ({\mathcal{C}}^{\prime \prime}_j)
359: \right] \right\},
360: \label{sv}
361: \end{equation}
362: where the first sum is over the $q_1$ sites with B particles and one A
363: neighbor, the second sum is over the $q_2$ sites with B particles and two A
364: neighbors, and we define $s_q \equiv 1/(s+q)$, where $q=q_1+2 q_2$.
365:
366: It is convenient to adopt as the initial configuration a translational
367: invariant one with a single A particle (periodic boundary conditions are
368: adopted). Now we may notice in the recursive expression \ref{sv} that the
369: operator $(s-V)^{-1}$ acting on any configuration generates an infinite set of
370: configurations, and thus we are unable to calculate $| \tilde{\psi} \rangle$ in
371: a closed form. We may, however, calculate the extinction probability
372: $\tilde{p}(s)$, which corresponds to the coefficient of the vacuum state
373: $|0\rangle$. As happens also for the models related to the CP studied in
374: \cite{dj91} configurations with more than $j$ particles only contribute at
375: orders higher than $j$, and since we are interested in the ultimate survival
376: probability for A particles $P_\infty=1-\lim_{s \to 0} s \tilde{p}(s)$, $s_q$
377: may be replaced by $1/q$ in Eq. \ref{sv}. To illustrate the procedure, we will
378: perform the explicit calculation of the series for $\lim_{s \to
379: 0}s\tilde{p}(s)$ up to third
380: order in $\lambda$. We furthermore represent a configuration denoting by
381: $\bullet$ a site occupied by an A particle and by $\circ$ a site occupied by
382: a B particle. Thus $(\bullet \circ \bullet)$ denotes the translationally
383: invariant configuration $\sum_i A_i^\dagger A_{i+2}^\dagger |0 \rangle$. The B
384: particles situated to the left of the leftmore A particle in a configuration
385: and the ones situated to the right
386: of the rightmost A particle are omitted in the representation of this
387: configuration. The vacuum
388: state will be
389: represented by $(0)$. As stated above, at $t=0$ the
390: system will be supposed to be in the configuration $|\psi(0) \rangle = \sum_i
391: A_i^\dagger |0 \rangle = (\bullet)$. Keeping configurations with up to three
392: particles,
393: and omitting the global factor $1/s$, the first of the recursion relations
394: Eqs. \ref{rr} leads to
395: \begin{equation}
396: |\tilde{\psi}_0 \rangle=\frac{1}{2}[(\bullet)+(\bullet
397: \bullet)+(\bullet\bullet\bullet)+ \cdots].
398: \end{equation}
399: The next step is the calculation of
400: $|\tilde{\phi}_0\rangle=S_0|\tilde{\psi}_0\rangle$. Now we need to keep only
401: configurations with up to two particle. The result is
402: \begin{equation}
403: |\tilde{\phi}_0\rangle=\frac{1}{2}[(2\alpha+1)(0)+(\bullet)+
404: \bullet\circ\bullet)+\cdots].
405: \end{equation}
406: Now we obtain $|\tilde{\psi}_1\rangle=(s-V)^{-1} |\tilde{\phi}_0\rangle$ for
407: $s=0$, resulting in
408: \begin{equation}
409: |\tilde{\psi}_1\rangle=\frac{1}{2}(2\alpha+1)(0)+\frac{1}{4}(\bullet)+
410: \frac{1}{4}(\bullet\bullet)+\frac{1}{8}(\bullet\circ\bullet)+\cdots.
411: \end{equation}
412: From this point on we will only show the results of each step, up to the
413: relevant numbers of A particles:
414: \begin{eqnarray}
415: |\tilde{\phi}_1\rangle &=& \frac{1}{4}(2\alpha+1)(0)+\frac{1}{4}
416: (2\alpha+2)(\bullet)+\cdots, \nonumber \\
417: |\tilde{\psi}_2\rangle &=& \frac{1}{4}(2\alpha+1)(0)+\frac{1}{8}
418: (2\alpha+2)(\bullet)+\cdots, \nonumber \\
419: |\tilde{\phi}_2\rangle &=& \frac{1}{8}(2\alpha+2)(2\alpha+1)(0)+\cdots,
420: \nonumber \\
421: |\tilde{\psi}_3\rangle &=&
422: \frac{1}{8}(2\alpha+2)(2\alpha+1)(0)+\cdots. \nonumber
423: \end{eqnarray}
424: The first coefficients of the ultimate survival probability will then be given
425: by
426: \begin{equation}
427: P_\infty=1-\frac{1}{2}(2\alpha+1)\lambda+-\frac{1}{4}(2\alpha+1)\lambda^2 +
428: \frac{1}{8}(2\alpha+2)(2\alpha+1) \lambda^3 + \cdots
429: \end{equation}
430:
431: The algebraic operations above may be easily performed in a computer using a
432: proper algorithm. The
433: configurations are expressed as binary numbers and the coefficients as double
434: precision variables. Although we have tried to do the calculation representing
435: the coefficients as rational numbers, thus avoiding any roundoff errors since
436: all calculations were done with integers, we found that the denominators
437: increase very rapidly with the order of the calculations, and thus we were
438: unable to perform the calculations this way up to a reasonable order. With
439: rather modest computational resources (Athlon MP2200, double processor, 1Gb
440: memory) it is not difficult to calculate the coefficients up to order 25. The
441: required processing time amounts to about 6 hours, the limiting factor is
442: actually the memory required for the calculation. The maximum number of terms
443: (polynomials in $\alpha$) amounts to more than $44 \times 10^6$. We define the
444: coefficients $b_{i,j}$ as:
445: \begin{equation}
446: P_\infty=1-\frac{1}{2}(2\alpha+1) \lambda - \sum_{i=2}^{25} \sum_{j=0}^{i-1}
447: b_{i,j} \lambda^i \alpha^j,
448: \end{equation}
449: and they are listed in Table \ref{coefs}. Up to order 24, our results are
450: numerically coincident with the supercritical series expansion for the
451: ultimate survival probability of the contact process \cite{dj91}, in the
452: particular case $\alpha=0$ (we remark that the variable $\lambda$ in the
453: supercritical expansion for the ultimate survival probability in reference
454: \cite{dj91} is half the variable $\lambda$ we use here).
455:
456: \section{Analysis of the series}
457: \label{res}
458: Let us consider initially the one variable series for fixed values of
459: $\alpha$
460: \begin{equation}
461: P_\infty=\sum_{i=0}^{25} a_i(\alpha) \lambda^i.
462: \end{equation}
463: As a preliminary test, we may apply the ratio method \cite{g89} to these
464: series. The
465: results for $r_i=a_i/a_{i-1}$ as functions of $1/i$ are depicted in Figure
466: \ref{rm}. In the case $\alpha=0$ (circles), the asymptotic linear behavior
467: $r_i \approx 1,6489(1-0.7231/i)$,
468: obtained from precise estimates for $\lambda_c$ and the exponent $\beta$ for
469: the contact process \cite{dj91}. In the figure it is apparent that the ratios
470: approach the asymptotic limit as $i$ is increased, as a matter of fact this
471: approach is close to linear in $1/i$. Thus, we may infer that for this model
472: the singularity which is closer to the origin is actually the physical
473: singularity. As the value of $\alpha$ is increased, one may notice that the
474: asymptotic linear behavior in $1/i$ for the ratios occurs only for higher
475: values of $i$, and thus it will be increasingly difficult to obtain precise
476: estimates for higher values of $\alpha$.
477:
478: Even in cases where the singularity of physical interest is the one closest to
479: the origin, the d-log Pad\'e approximants usually lead to better estimates
480: than the ratio method \cite{g89}. The approximants are defined as ratios of
481: two polynomials $P_L(\lambda)$ and $Q_M(\lambda)$:
482: \begin{equation}
483: F_{LM}(\lambda)=\frac{P_L(\lambda)}{Q_M(\lambda)}=\frac{\sum_{i=0}^Lp_i
484: \lambda^i}{1+\sum_{j=1}^M q_j \lambda^j}.
485: \label{pa}
486: \end{equation}
487: The series for $\frac{d}{d \lambda} \ln P_\infty (\lambda)$ for fixed values of
488: $\alpha$ are substituted in the defining equation \ref{pa} and the
489: coefficients of the polynomials are chosen such that the identity is true up
490: to the order of the available series expansion. Thus approximants with $L+M
491: \leq 24$ may be built with the available series. Usually diagonal ($L=M$) or
492: close to diagonal approximants furnish better results, so we restricted our
493: calculations to these cases. The estimate for the critical value of $\lambda$
494: is found among the poles of the approximant, the estimate for the critical
495: exponent $\beta$ will be the residue at this pole.
496:
497: We thus built approximants with $\alpha$ ranging between 0 and 40, estimating
498: the critical value of the parameter $\lambda$ as well as the critical exponent
499: $\beta$. Although for small values of $\alpha$ the results are very good, with
500: estimates of $\beta$ comparable to the best ones in the literature for the
501: contact process, as the value of $p_c$ is decreased we notice a growing
502: dispersion of the estimates for the exponent and for $p_c<0.01$ most of the
503: approximants lead even to ill conditioned systems of linear equations for the
504: coefficients, and therefore we were not able to obtain estimates in this
505: region. We made additional one-variable investigations, such as obtaining
506: approximants for $(\lambda_c-\lambda)\frac{d}{d\lambda} P_s(\lambda)=\beta$
507: for several values of $\lambda_c$ and searching for the intercept of the curves
508: $\beta(\lambda_c)$ \cite{dj91} and non-homogeneous Pad\'e approximants
509: \cite{g89,nha}, and although some improvements of the estimates may be
510: obtained in
511: certain cases, the situation does not change qualitatively. In figure
512: \ref{cline} the Pad\'e estimates for the critical line are displayed.
513:
514: Actually, the increasing dispersion of the estimates as the parameter
515: $p_c$ approaches zero is not a surprise, since as was mentioned above in this
516: limit the model corresponds to the voter model which is in the compact
517: directed percolation (CDP) universality class, whose exponents are different
518: from the ones of the contact process, which belongs to the directed
519: percolation (DP) universality class. For the voter model, the exponent $\beta$
520: of the order parameter is equal to zero (a discontinuity in the order parameter
521: occurs at the transition) \cite{h00}, but the exponent for the survival
522: probability $\beta^\prime$ is distinct from $\beta$ and equal to 1
523: \cite{dt95}. Therefore a crossover to from the DP to the CDP universality class
524: occurs as $p_c \to 0$, and it is known that in such situations the reduction
525: of two-variable series to one variable leads to very poor estimates
526: \cite{fk77}. So we analyzed the series without reducing the problem to one
527: variable, and to our knowledge the best results for two-variable series
528: applied to multicritical phenomena in the literature were obtained using the
529: partial differential approximants (PDA's) \cite{fk77}. They may be regarded as
530: a
531: generalization to two variables of the d-log Pad\'e approximants. The defining
532: equation of the approximants is
533: \begin{equation}
534: P_{\mathbf L}(x,y)F(x,y)=Q_{\mathbf M}(x,y)\frac{\partial F(x,y)}{\partial x} +
535: R_{\mathbf N}(x,y) \frac{\partial F(x,y)}{\partial y},
536: \label{pda}
537: \end{equation}
538: where $P$, $Q$, and $R$ are polynomials in the variables $x$ and $y$ with the
539: set of nonzero coefficients ${\mathbf L}$, ${\mathbf M}$, and ${\mathbf N}$,
540: respectively. The coefficients
541: of the polynomials are obtained through substitution of the series expansion
542: for the quantity which is going to be analyzed
543: \begin{equation}
544: f(x,y)=\sum_{k,k^\prime=0} f(k,k^\prime)x^k y^{k^\prime}
545: \end{equation}
546: into the defining equation \ref{pda} and requiring the equality to hold for a
547: set of indexes defined as ${\mathbf K}$. Again this procedure leads to a system
548: of linear
549: equations for the coefficients of the polynomials, and since the coefficients
550: $f_{k,k^\prime}$ of the series are known for a finite set of indexes this sets
551: an upper limit to the number of coefficients in the polynomials. Since the
552: number of equations has to match the number of unknown coefficients, we must
553: have that the numbers of elements in each set satisfy $K=L+M+N-1$ (one
554: coefficient is fixed arbitrarily). An additional issue,
555: which is not present in the one-variable case, is the symmetry of the
556: polynomials. Two frequently used options are the triangular and the
557: rectangular arrays of coefficients. The choice of these symmetries is related
558: to the symmetry of the series itself \cite{s90}. Below we discuss the solution
559: we adopted in the present case for this point.
560:
561: Let us suppose that quantity represented by the series is expected to have a
562: multicritical behavior at a point $(x_c,y_c)$, described by
563: \begin{equation}
564: f(x,y) \approx |\Delta \widetilde{x}|^{-\gamma}Z\left( \frac{|\Delta
565: \widetilde{y}|}{|\Delta \widetilde{x}|^\phi} \right),
566: \label{mcs}
567: \end{equation}
568: where
569: \begin{equation}
570: \Delta \widetilde{x}=(x-x_c )-(y-y_c)/e_2,
571: \end{equation}
572: and
573: \begin{equation}
574: \Delta \widetilde{y}=(y-y_c)-e_1(x-x_c).
575: \end{equation}
576: Here $\gamma$ is the critical exponent of the quantity described by $f$ when
577: $\Delta \widetilde{y}=0$, $e_1$ and $e_2$ are the scaling slopes \cite{fk77}
578: and
579: $\phi$ is the crossover exponent. The function $Z(z)$ is singular for one or
580: more values of its argument, corresponding to the critical line(s) incident on
581: the multicritical point. Once the coefficients of the defining polynomials are
582: obtained, the estimated location of the multicritical point corresponds to the
583: common zero of the polynomials $Q_M$ and $R_N$. This may be seen substituting
584: the scaling form \ref{mcs} in the defining equation \ref{pda} of the
585: approximant. The exponents and scaling
586: slopes may also be obtained directly from the polynomials, without
587: integrating the partial differential equation. A detailed discussion of the
588: algorithm, as well as computer codes, may be found in \cite{s90}.
589:
590: Before proceeding with the analysis of the series, it is convenient to perform
591: a change of variables, since the multicritical point in the original variables
592: is located at $\alpha \to \infty$. We thus express the series in the variables
593: \begin{eqnarray*}
594: x = \lambda = \frac{2p_c}{p_a}\\
595: y = \alpha\lambda = \frac{1-p_a-p_c}{p_a}.
596: \end{eqnarray*}
597: In these new variables the multicritical point is located at $x=0$, $y=1$, and
598: the survival probability may be written as
599: \begin{equation}
600: P_{\infty}=1-\frac{1}{2}x-y-xF(x,y),
601: \label{ns}
602: \end{equation}
603: where
604: \begin{equation}
605: F(x,y)=\sum_{i=1}^{24}\sum_{j=0}^i b_{i+1,j}x^{i-j}y^j
606: \end{equation}
607: is represented by a series with triangular symmetry, which is very convenient
608: to be analyzed using PDA's. The number of approximants which may be obtained
609: from the series is very big, so we restricted ourselves to approximants with
610: the number of elements in $\mathbf{M}$ close to the number of elements in
611: $\mathbf{N}$. The
612: polynomials had the same triangular symmetry as the series, but in most cases
613: some higher order elements of the polynomials were set equal to zero in order
614: to
615: match the number of unknowns to the number of equations in the set of linear
616: equations for the coefficients. This is a rather standard procedure and is
617: discussed in detail by Styer \cite{s90}. Even at
618: rather low orders, we found a reasonable agreement between most of the
619: estimates from different approximants. Finally,
620: we considered a set of 42 approximants which use the elements of the series
621: for $F(x,y)$ with highest orders $i$ between 15 and 24. Discarding some
622: approximants which generated estimates which were rather far away of the
623: general trend, finally we used a set of 36 approximants to obtain the
624: estimates.
625:
626: In figure \ref{mcp} the estimates for the location of the multicritical point
627: are displayed. We notice that the estimates are very close to the exactly known
628: values $x=0$ and $y=1$. The estimated values for $x_c$ and $y_c$ are $(0.4 \pm
629: 1.8) \times 10^{-6}$ and $y=1.0000 \pm 0.0003$. The exponent $-\gamma$ in the
630: multicritical scaling form \ref{mcs} corresponds to the exponent
631: $\beta^\prime$ of the CDP universality class. The estimated value is equal to
632: $1.00 \pm 0.01$, which agrees very well with the exact value $\beta^\prime=1$
633: \cite{h00}. Finally, the crossover exponent was estimated as $\phi=2.01 \pm
634: 0.03$, and thus the mean field value for this exponent ($\phi=2$
635: \cite{dts03})
636: is within the confidence interval of our estimate. The estimates for
637: $\beta^\prime$ and $\phi$ are shown in figure \ref{betaphi}. We also obtained
638: biased PDA's, fixing the other parameters at their known values and calculating
639: improved estimates for $\phi$. This procedure resulted in the estimate
640: $\phi=2.00 \pm 0.02$. We thus conclude that our estimates for $\phi$ are very
641: close to the classical value of the crossover exponent. The estimates for the
642: slopes of the scaling axes show a rather broad distribution, a significant
643: majority
644: of the approximants provide quite large values for $e_1$, while $e_2$ is
645: typically much smaller. This suggests that $\Delta \widetilde{y}=x$ and it is
646: reasonable to choose $\Delta \widetilde{x}=1-y$, since it corresponds to the
647: weak direction, parallel to the critical line at the multicritical point. In
648: the limit of the voter model ($x=0$) we notice that the series \ref{ns}
649: reduces to the exact result $P_\infty=1-y$.
650:
651: One way to actually estimate values of the quantity which is described by a
652: PDA is to integrate the equation using the method of characteristics. A
653: timelike variable $\tau$ is defined and a family of curves $(x(\tau),y(\tau))$
654: (the characteristics) is considered. These curves are defined by the
655: equations
656: \begin{eqnarray}
657: \frac{dx}{d \tau} &=& Q_M(x(\tau),y(\tau)),\nonumber\\
658: \frac{dy}{d\tau} &=& R_N(x(\tau),y(\tau)).
659: \label{char}
660: \end{eqnarray}
661: Along such a curve, the defining equation of the PDA \ref{pda} leads to an
662: ordinary differential equation for $F$.
663: \begin{equation}
664: \frac{dF}{d\tau}=P_L(x(\tau),y(\tau))F,
665: \end{equation}
666: which may readily be integrated, providing the value of $F$ at the points of
667: the characteristics, once we know this value at a initial point. Our efforts
668: to obtain estimates for the survival probability, particularly close to the
669: multicritical point, using this procedure were not very
670: successful. Nevertheless, it is worth to mention that an estimate for the
671: critical line may be obtained this way. The critical line, which connects the
672: point which corresponds to the CP to the multicritical point of the voter
673: model transition is a line of singularities, and it is not difficult to show
674: that such a line is one of the characteristics defined by equations \ref{char}.
675: Therefore, the characteristic whose initial point is located on the transition
676: point of the CP, which is known with great accuracy, is an estimate for the
677: critical line. We therefore chose a particular approximant with estimates
678: close to the mean values. The number of elements in the sets for this
679: approximant are $K=300$, $L=171$, and $M=N=65$. In figure \ref{cline} this
680: characteristic curve is shown, and a
681: nice agreement with other estimates of the critical line may be observed.
682:
683: \section{Conclusion and discussion}
684: \label{conclu}
685: We may notice that
686: the coefficients $b(i,i-1)$, for $i=1,2,\ldots$ are equal to $1/2$, a
687: result
688: which is indicated by the coefficients below but may be shown to be true in
689: general using the operator formalism above. Therefore, we may sum these set
690: of terms in the series, obtaining
691: \begin{equation}
692: P_\infty=(1-y) \left[1-\frac{1}{2}\frac{x}{(1-y)^2}- \ldots \right].
693: \end{equation}
694: Now if we compare this expression with the multicritical scaling form
695: \ref{mcs}, we may recognize between braces the two first terms of a
696: Taylor expansion of the multicritical scaling function $Z(z)$, where
697: the variable
698: $z$ is identified as $\frac{x}{(1-y)^2}$. This agrees with the estimates
699: obtained
700: from the PDA's. Moreover, remembering that for $y=0$ the series
701: represents the CP,
702: the function $Z(z)$ may be recognized as the survival probability of the CP,
703: which was studied in great detail \cite{dj91} and found to have a singularity
704: at $z_0=0.6064$ with the exponent $\beta^\prime=\beta=0.276486$. As expected,
705: the
706: critical line is characterized by the same exponent of the CP, that is, is
707: in the DP universality class. The critical line corresponds to $z=z_0$, that is
708: \begin{equation}
709: x=z_0(1-y)^2,
710: \end{equation}
711: and this curve is shown in figure \ref{cline}. It is interesting to notice
712: that
713: the agreement of this estimate of the critical line with the other two,
714: obtained from
715: Pad\'e and partial differential approximants, is quite good even far away from
716: the multicritical point. The multicritical scaling form with the
717: identification of the scaling variable $z$ above is exact in the limit $p_c=0$
718: (biased voter model) and reproduces the supercritical series expansion for the
719: CP, when $y=0$. It does, however, not reproduce the two-variable supercritical
720: series expansion for the full model we considered here away from these
721: limiting cases.
722:
723: The Domany Kinzel probabilistic cellular automaton
724: \cite{dk84} in part of its two-dimensional phase diagram corresponds to a
725: synchronous update version of the contact process, and as stated above has the
726: synchronous update voter model as a endpoint of the critical line. It is
727: believed, although to our knowledge not proven, that if in a particular model
728: the update procedure is changed from synchronous to asynchronous, the
729: critical exponents do not change.Some bounds for the critical line are
730: presented in \cite{kt95}. Although an upper bound for this line due to Liggett
731: \cite{l94} is quadratic close to the CDP endpoint, the lower bounds are
732: linear, and thus the asymptotic behavior of the critical line is not fixed by
733: those bounds. The critical line is studied in more detail by simulations and
734: series expansions in \cite{tikt95}, and based on this results the authors
735: conjectured a quadratic asymptotic behavior of the critical line, consistent
736: with $\phi=2$.Finally, a more detailed series analysis is done in
737: \cite{ikb01}, but since one-variable Pad\'e approximants were used, the
738: results are not precise in the region close to the CDP point. Thus, there are
739: indications that the crossover exponent has the same value in both models,
740: and if these indications are correct, the invariance of this multicritical
741: exponent with respect to the update procedure is verified in this particular
742: case.
743:
744: In conclusion, the analysis of the series for the ultimate survival
745: probability using PDA's lead to quite precise
746: estimates for the multicritical exponents, and these estimates, together with
747: the possibility to sum the terms of the two-variable series which are linear
748: in $x$ allowed us to conjecture the exact form of the multicritical scaling
749: expression. The multicritical scaling function $Z(z)$ is known as a series
750: expansion up to order 25.
751:
752: \begin{acknowledgments}
753: We thank Prof. Ronald Dickman for many helpful discussions, and one of the
754: referees for calling out attention to references \cite{kt95}
755: -\cite{ikb01}. We
756: also thank Professor M\'ario J. de Oliveira for a critical reading of the
757: manuscript. This
758: research was partially supported by the Brazilian agencies CAPES, FAPERJ
759: and CNPq, particularly through the project PRONEX-CNPq-FAPERJ/171.168-2003.
760: \end{acknowledgments}
761:
762:
763:
764:
765: \begin{thebibliography}{99}
766: \bibitem{md99}J. Marro and R. Dickman, {\it Nonequilibrium Phase Transitions in
767: Lattice Models} (Cambridge University Press, Cambridge, 1999)
768: \bibitem{h74} T.E. Harris, \emph{Ann. Probab.} \textbf{2}, 969 (1974).
769: \bibitem{s72} F. Schl\"ogl, \emph{Z. Phys.} \textbf{253}, 147 (1972).
770: \bibitem{gt} P. Grassberger and A. De La Torre, \emph{Ann.Phys.}
771: \textbf{122}, 373 (1979).
772: \bibitem{zgb86} R. M. Ziff, E. Gulari, and Y. Barshad, \emph{Phys. Rev. Lett.}
773: \textbf{56}, 2553 (1986).
774: \bibitem{tty92} H. Takayasu, A. Tretyakov, and A. Yu, \emph{Phys. Rev. Lett.}
775: \textbf{68}, 3060 (1992).
776: \bibitem{j81} H. K. Janssen, \emph{Z. Phys. B}, \textbf{42}, 151 (1981);
777: P. Grassberger, \emph{Z. Phys. B}, \textbf{47}, 365 (1982).
778: \bibitem{h00} H. Hinrichsen, \emph{Adv. Phys.}\textbf{49}, 815 (2000).
779: \bibitem{g63} R. J. Glauber, \emph{J. Math. Phys.} \textbf{4}, 294 (1963).
780: \bibitem{o03} M. J. de Oliveira, \emph{Phys. Rev. E} \textbf{67} 066101
781: (2003).
782: \bibitem{l85} T. M. Ligget, \emph{Interacting particle systems}, Springer,
783: Berlin (1985).
784: \bibitem{dk84}E. Domany and W. Kinzel, \emph{Phys. Rev. Lett.} \textbf{53},
785: 311 (1984).
786: \bibitem{h97}H. Hinrichsen, \emph{Phys. Rev. E} \textbf{55}, 219 (1997).
787: \bibitem{mdh96}J. F. F. Mendes, R. Dickman, and H. Herrmann,
788: \emph{Phys. Rev. E} \textbf{54}, R3071 (1996).
789: \bibitem{kt95}M. Katori and H. Tsukahara, \emph{J. Phys. A: Math. Gen.}
790: \textbf{28}, 3935 (1995).
791: \bibitem{tikt95}A. Yu. Tretyakov, N. Inui, M. Katori, and H. Tsukahara,
792: \emph{cond-mat/9509061} (1995).
793: \bibitem{ikb01}N. Inui, M. Katori and F. M. Bhatti, \emph{J. Phys. Soc.
794: Jpn.}
795: \textbf{70}, 359 (2001).
796: \bibitem{dts03} W. G. Dantas, A. Ticona, and J. F. Stilck,
797: \emph{Braz. J. Phys.} \textbf{35},536 (2005).
798: \bibitem{dj91} R. Dickman and I. Jensen, \emph{Phys. Rev. Lett.} \textbf{67},
799: 2391 (1991); I. Jensen and R. Dickman, \emph{J. Stat. Phys.} \textbf{71}, 89
800: (1993).
801: \bibitem{g89}A. J. Guttmann in \emph{Phase Transitions and Critical
802: Phenomena}, vol. 13, ed. by C. Domb and J. L. Lebowitz (Academic Press,
803: London, 1989).
804: \bibitem{nha} F. D. A. Aar\~ao Reis and R. Rieira,\emph{Phys. Rev. E}
805: \textbf{49}, 2579 (1994).
806: \bibitem{dt95}R. Dickman and A. Y. Tretyakov, \emph{Phys. Rev. E} \textbf{52},
807: 3218 (1995).
808: \bibitem{fk77}M. E. Fisher and R. M. Kerr, \emph{Phys. Rev. Lett.}
809: \textbf{32}, 667 (1977).
810: \bibitem{s90}D. F. Styer, \emph{Comp. Phys. Comm.} \textbf{61}, 374 (1990).
811: \bibitem{l94}T. M. Liggett, preprint (1994).
812: \end{thebibliography}
813:
814: \begin{center}
815: \begin{table}
816: \begin{tabular}{|c|c|c|}
817: \hline
818: $i$&$j$&$b_{i,j}$\\
819: \hline
820: 1&1& 0.10000000000000000000D+01\\
821: 1&0& 0.50000000000000000000D+00\\
822: \hline
823: 2&1& 0.50000000000000000000D+00\\
824: 2&0& 0.25000000000000000000D+00\\
825: \hline
826: 3&2& 0.50000000000000000000D+00\\
827: 3&1& 0.75000000000000000000D+00\\
828: 3&0& 0.25000000000000000000D+00\\
829: \hline
830: 4&3& 0.50000000000000000000D+00\\
831: 4&2& 0.15000000000000000000D+01\\
832: 4&1& 0.11875000000000000000D+01\\
833: 4&0& 0.28125000000000000000D+00\\
834: \hline
835: 5&4& 0.50000000000000000000D+00\\
836: 5&3& 0.25000000000000000000D+01\\
837: 5&2& 0.35234375000000000000D+01\\
838: 5&1& 0.18867187500000000000D+01\\
839: 5&0& 0.34375000000000000000D+00\\
840: \hline
841: 6&5& 0.50000000000000000000D+00\\
842: 6&4& 0.37500000000000000000D+01\\
843: 6&3& 0.81665039062499928945D+01\\
844: 6&2& 0.73991699218749946709D+01\\
845: 6&1& 0.29899902343749985789D+01\\
846: 6&0& 0.44726562499999982236D+00\\
847: \hline
848: 7&6& 0.50000000000000000000D+00\\
849: 7&5& 0.52499999999999982236D+01\\
850: 7&4& 0.16280639648437500000D+02\\
851: 7&3& 0.21964294433593751776D+02\\
852: 7&2& 0.14622192382812501776D+02\\
853: 7&1& 0.47470703125000008881D+01\\
854: 7&0& 0.60223388671874964472D+00\\
855: \hline
856: 8&7& 0.50000000000000000000D+00\\
857: 8&6& 0.69999999999999982236D+01\\
858: 8&5& 0.29241512298583977269D+02\\
859: 8&4& 0.54546823501586914062D+02\\
860: 8&3& 0.52912919998168961299D+02\\
861: 8&2& 0.27873636245727544391D+02\\
862: 8&1& 0.75799846649169939638D+01\\
863: 8&0& 0.83485031127929723027D+00\\
864: \hline
865: 9&8& 0.50000000000000000000D+00\\
866: 9&7& 0.90000000000000017763D+01\\
867: 9&6& 0.48670531988143883594D+02\\
868: 9&5& 0.11948751342296599631D+03\\
869: 9&4& 0.15729669463192959000D+03\\
870: 9&3& 0.11890031438403647712D+03\\
871: 9&2& 0.51820946525644346891D+02\\
872: 9&1& 0.12126680864228140954D+02\\
873: 9&0& 0.11814667913648822050D+01\\
874: \hline
875: 10&9& 0.50000000000000000000D+00\\
876: 10&8& 0.11249999999999982236D+02\\
877: 10&7& 0.76407402418553784784D+02\\
878: 10&6& 0.23825064151361545761D+03\\
879: 10&5& 0.40643453331768384373D+03\\
880: 10&4& 0.41173626079900831342D+03\\
881: 10&3& 0.25488795003760951196D+03\\
882: 10&2& 0.94770005067848632762D+02\\
883: 10&1& 0.19463028091937289332D+02\\
884: 10&0& 0.16988672076919923981D+01\\
885: \hline
886: \end{tabular}
887: \end{table}
888: \end{center}
889: \begin{center}
890: \begin{table}
891: \begin{tabular}{|c|c|c|}
892: \hline
893: 11&10& 0.50000000000000000000D+00\\
894: 11&9& 0.13750000000000004440D+02\\
895: 11&8& 0.11453289534710346941D+03\\
896: 11&7& 0.44150459107663433400D+03\\
897: 11&6& 0.94395713065082347270D+03\\
898: 11&5& 0.12251638531684152511D+04\\
899: 11&4& 0.10067004627495166335D+04\\
900: 11&3& 0.52744546391776081506D+03\\
901: 11&2& 0.17102490738153186100D+03\\
902: 11&1& 0.31313301122887615690D+02\\
903: 11&0& 0.24775957118665874467D+01\\
904: \hline
905: 12&11& 0.50000000000000000000D+00\\
906: 12&10& 0.16500000000000039079D+02\\
907: 12&9& 0.16534952380310286024D+03\\
908: 12&8& 0.77136775289458459070D+03\\
909: 12&7& 0.20158736418235778664D+04\\
910: 12&6& 0.32476491678046719435D+04\\
911: 12&5& 0.33907974883298468427D+04\\
912: 12&4& 0.23413097838386356386D+04\\
913: 12&3& 0.10630662385637921207D+04\\
914: 12&2& 0.30555868520944353683D+03\\
915: 12&1& 0.50452588505226367843D+02\\
916: 12&0& 0.36488812342264482779D+01\\
917: \hline
918: 13&12& 0.50000000000000000000D+00\\
919: 13&11& 0.19499999999999982236D+02\\
920: 13&10& 0.23139824792517162954D+03\\
921: 13&9& 0.12840079302994218402D+04\\
922: 13&8& 0.40209683861219263079D+04\\
923: 13&7& 0.78496665288886928735D+04\\
924: 13&6& 0.10096110998534653102D+05\\
925: 13&5& 0.88003157466952828258D+04\\
926: 13&4& 0.52368347467192908339D+04\\
927: 13&3& 0.20980007339714172864D+04\\
928: 13&2& 0.54172526170510240106D+03\\
929: 13&1& 0.81489527352144666139D+02\\
930: 13&0& 0.54293656084851633636D+01\\
931: \hline
932: 14&13& 0.49999986588954925537D+00\\
933: 14&12& 0.22749997384846243342D+00\\
934: 14&11& 0.31544379761375296311D+03\\
935: 14&10& 0.20523899291719041038D+04\\
936: 14&9& 0.75794568674053097723D+04\\
937: 14&8& 0.17597122905598574504D+05\\
938: 14&7& 0.27244155488205645809D+05\\
939: 14&6& 0.29082737294330742727D+05\\
940: 14&5& 0.21732772717178829857D+05\\
941: 14&4& 0.11357034171626549934D+05\\
942: 14&3& 0.40697753516544512564D+04\\
943: 14&2& 0.95363664836240946698D+03\\
944: 14&1& 0.13169793741584487900D+03\\
945: 14&0& 0.81325422193071297272D+01\\
946: \hline
947: \end{tabular}
948: \end{table}
949: \end{center}
950: \begin{center}
951: \begin{table}
952: \begin{tabular}{|c|c|c|}
953: \hline
954: 15&14& 0.50000000000000000000D+00\\
955: 15&13& 0.26249998256564137655D+02\\
956: 15&12& 0.42048775894871077696D+03\\
957: 15&11& 0.31693914938186833474D+04\\
958: 15&10& 0.13620079986016961903D+05\\
959: 15&9& 0.37037632746719575393D+05\\
960: 15&8& 0.67783286393305548500D+05\\
961: 15&7& 0.86632679492839894663D+05\\
962: 15&6& 0.78910368357158162666D+05\\
963: 15&5& 0.51571902934753062197D+05\\
964: 15&4& 0.24020987390678723016D+05\\
965: 15&3& 0.77874939092949517771D+04\\
966: 15&2& 0.16707029021016392533D+04\\
967: 15&1& 0.21330513440327356633D+03\\
968: 15&0& 0.12275012836144905126D+02\\
969: \hline
970: 16&15& 0.50000000000000000000D+00\\
971: 16&14& 0.29999999329447643248D+02\\
972: 16&13& 0.54975682655068318638D+03\\
973: 16&12& 0.47510546519798584341D+04\\
974: 16&11& 0.23493882045888385690D+05\\
975: 16&10& 0.73897182789227748855D+05\\
976: 16&9& 0.15755031904987298219D+06\\
977: 16&8& 0.23687090513090347521D+06\\
978: 16&7& 0.25718644037330271601D+06\\
979: 16&6& 0.20406058122209760341D+06\\
980: 16&5& 0.11843376024372214061D+06\\
981: 16&4& 0.49739262723985895320D+05\\
982: 16&3& 0.14718788444797093362D+05\\
983: 16&2& 0.29113789219292591781D+04\\
984: 16&1& 0.34558659823607458250D+03\\
985: 16&0& 0.18620961415130533822D+02\\
986: \hline
987: 17&16& 0.50000000000000000000D+00\\
988: 17&15& 0.33999999329447789797D+02\\
989: 17&14& 0.70671335354464535072D+03\\
990: 17&13& 0.69402158602317101099D+04\\
991: 17&12& 0.39112396145099372901D+05\\
992: 17&11& 0.14079076808597497105D+06\\
993: 17&10& 0.34548310262719810204D+06\\
994: 17&9& 0.60241714803092252239D+06\\
995: 17&8& 0.76636402935447742734D+06\\
996: 17&7& 0.72224826262886052674D+06\\
997: 17&6& 0.50729839736110484693D+06\\
998: 17&5& 0.26473279696452087783D+06\\
999: 17&4& 0.10122975226631107936D+06\\
1000: 17&3& 0.27555943803143052583D+05\\
1001: 17&2& 0.50567133817482794455D+04\\
1002: 17&1& 0.56084260997641584012D+03\\
1003: 17&0& 0.28405733950687368505D+02\\
1004: \hline
1005: \end{tabular}
1006: \end{table}
1007: \end{center}
1008: \begin{center}
1009: \begin{table}
1010: \begin{tabular}{|c|c|c|}
1011: \hline
1012: 18&17& 0.50000000000000000000D+00\\
1013: 18&16& 0.38249999329447760487D+02\\
1014: 18&15& 0.89504578674400576687D+03\\
1015: 18&14& 0.99102626626136949283D+04\\
1016: 18&13& 0.63122994123535232091D+05\\
1017: 18&12& 0.25768737165396164989D+06\\
1018: 18&11& 0.72043302015833452500D+06\\
1019: 18&10& 0.14399790608479312581D+07\\
1020: 18&9& 0.21166673049189457245D+07\\
1021: 18&8& 0.23292641415436174945D+07\\
1022: 18&7& 0.19371723268429082764D+07\\
1023: 18&6& 0.12198708033512473125D+07\\
1024: 18&5& 0.57806709777733367161D+06\\
1025: 18&4& 0.20284284513882400169D+06\\
1026: 18&3& 0.51115204972041548003D+05\\
1027: 18&2& 0.87477997069442139377D+04\\
1028: 18&1& 0.91057218719788650673D+03\\
1029: 18&0& 0.43520782874229171355D+02\\
1030: \hline
1031: 19&18& 0.49999999999999911182D+00\\
1032: 19&17& 0.42749999329447687657D+02\\
1033: 19&16& 0.11186774875974003773D+04\\
1034: 19&15& 0.13869278218084735243D+05\\
1035: 19&14& 0.99114912109387987015D+05\\
1036: 19&13& 0.45525829782828974856D+06\\
1037: 19&12& 0.14375483791065075678D+07\\
1038: 19&11& 0.32614349098596422393D+07\\
1039: 19&10& 0.54767322733604206774D+07\\
1040: 19&9& 0.69423928844342714938D+07\\
1041: 19&8& 0.67230154917519895363D+07\\
1042: 19&7& 0.50002124204980198385D+07\\
1043: 19&6& 0.28527370161475179344D+07\\
1044: 19&5& 0.12379379646484691690D+07\\
1045: 19&4& 0.40134571279240418562D+06\\
1046: 19&3& 0.94145168822797273833D+05\\
1047: 19&2& 0.15093606456186623887D+05\\
1048: 19&1& 0.14798263077956907984D+04\\
1049: 19&0& 0.66930218067927942371D+02\\
1050: \hline
1051: 20&19& 0.50000000000000000000D+00\\
1052: 20&18& 0.47499999329447701867D+02\\
1053: 20&17& 0.13817585309510260671D+04\\
1054: 20&16& 0.19064302620655780629D+05\\
1055: 20&15& 0.15187519641466302289D+06\\
1056: 20&14& 0.77950921221482092349D+06\\
1057: 20&13& 0.27591465361185987248D+07\\
1058: 20&12& 0.70457626117426288558D+07\\
1059: 20&11& 0.13386942665323722234D+08\\
1060: 20&10& 0.19328829379718762027D+08\\
1061: 20&9& 0.21502873943167273296D+08\\
1062: 20&8& 0.18574978781954090578D+08\\
1063: 20&7& 0.12487211289613961984D+08\\
1064: 20&6& 0.65099543472194438820D+07\\
1065: 20&5& 0.26049731385271019945D+07\\
1066: 20&4& 0.78478644916613982118D+06\\
1067: 20&3& 0.17219353528307646428D+06\\
1068: 20&2& 0.25969733685492251140D+05\\
1069: 20&1& 0.24071073326750083154D+04\\
1070: 20&0& 0.10337399908888862398D+03\\
1071: \hline
1072: \end{tabular}
1073: \end{table}
1074: \end{center}
1075: \begin{center}
1076: \begin{table}
1077: \begin{tabular}{|c|c|c|}
1078: \hline
1079: 21&20& 0.49999999999999955591D+00\\
1080: 21&19& 0.52499999329447701867D+02\\
1081: 21&18& 0.16886733349218010502D+04\\
1082: 21&17& 0.25785991340403939808D+05 \\
1083: 21&16& 0.22768166104993685650D+06 \\
1084: 21&15& 0.12978288769859755902D+07 \\
1085: 21&14& 0.51154797613905129693D+07 \\
1086: 21&13& 0.14596468511940512868D+08 \\
1087: 21&12& 0.31124330624461475913D+08 \\
1088: 21&11& 0.50710951504426553526D+08 \\
1089: 21&10& 0.64100799162967723177D+08 \\
1090: 21&9& 0.63467922470529245515D+08 \\
1091: 21&8& 0.49453964458225394551D+08 \\
1092: 21&7& 0.30321310892616786247D+08 \\
1093: 21&6& 0.14550387399195248150D+08 \\
1094: 21&5& 0.54012506504483805969D+07 \\
1095: 21&4& 0.15195548211633345125D+07 \\
1096: 21&3& 0.31313135707735888502D+06 \\
1097: 21&2& 0.44571407092920996007D+05 \\
1098: 21&1& 0.39157750249797601327D+04 \\
1099: 21&0& 0.15998830271630990473D+03 \\
1100: \hline
1101: 22&21& 0.50000000000000088817D+00 \\
1102: 22&20& 0.57749999329447438967D+02 \\
1103: 22&19& 0.20440335494663091076D+04 \\
1104: 22&18& 0.34373382551268485407D+05 \\
1105: 22&17& 0.33466104195644446051D+06 \\
1106: 22&16& 0.21070967788898307126D+07 \\
1107: 22&15& 0.91945254038400960894D+07 \\
1108: 22&14& 0.29128975460417066756D+08 \\
1109: 22&13& 0.69213935104954478205D+08 \\
1110: 22&12& 0.12623609387336516274D+09 \\
1111: 22&11& 0.17963375603295976823D+09 \\
1112: 22&10& 0.20164422581600076611D+09 \\
1113: 22&9& 0.17971976143434213568D+09 \\
1114: 22&8& 0.12747436997852454876D+09 \\
1115: 22&7& 0.71815965145384188517D+08 \\
1116: 22&6& 0.31918963164481914951D+08 \\
1117: 22&5& 0.11049035068083481458D+08 \\
1118: 22&4& 0.29157961094487756525D+07 \\
1119: 22&3& 0.56650954312689849601D+06 \\
1120: 22&2& 0.76372602458216514165D+05 \\
1121: 22&1& 0.63802269924577696968D+04 \\
1122: 22&0& 0.24876519640924477094D+03 \\
1123: \hline
1124: \end{tabular}
1125: \end{table}
1126: \end{center}
1127: \begin{center}
1128: \begin{table}
1129: \begin{tabular}{|c|c|c|}
1130: \hline
1131: 23&22& 0.50000015522128578027D+00 \\
1132: 23&21& 0.63250003986086227314D+02 \\
1133: 23&20& 0.24526847667386335594D+04 \\
1134: 23&19& 0.45219068649736344767D+05 \\
1135: 23&18& 0.48319032233047343183D+06 \\
1136: 23&17& 0.33439431430492136954D+07 \\
1137: 23&16& 0.16069927289546637183D+08 \\
1138: 23&15& 0.56208076057486060506D+08 \\
1139: 23&14& 0.14791153728125157051D+09 \\
1140: 23&13& 0.29990967758279261090D+09 \\
1141: 23&12& 0.47669571268868358160D+09 \\
1142: 23&11& 0.60118669013109657939D+09 \\
1143: 23&10& 0.60634053785517059154D+09 \\
1144: 23&9& 0.49107989528314268667D+09 \\
1145: 23&8& 0.31953776898326688993D+09 \\
1146: 23&7& 0.16647521279074794620D+09 \\
1147: 23&6& 0.68894669993720274447D+08 \\
1148: 23&5& 0.22337519908203442575D+08 \\
1149: 23&4& 0.55496488699702037905D+07 \\
1150: 23&3& 0.10196528668928661609D+07 \\
1151: 23&2& 0.13050250676997474652D+06 \\
1152: 23&1& 0.10386065610378922841D+05 \\
1153: 23&0& 0.38696067609363534955D+03 \\
1154: \hline
1155: 24&23& 0.49999999999999706901D+00 \\
1156: 24&22& 0.69000000457884933524D+02 \\
1157: 24&21& 0.29197000139992891121D+04 \\
1158: 24&20& 0.58774468615505739421D+05 \\
1159: 24&19& 0.68638127471201864082D+06 \\
1160: 24&18& 0.51981592602471167197D+07 \\
1161: 24&17& 0.27383018956536124832D+08 \\
1162: 24&16& 0.10521292804690998146D+09 \\
1163: 24&15& 0.30494730611982263646D+09 \\
1164: 24&14& 0.68324606257180731105D+09 \\
1165: 24&13& 0.12048047382312678799D+10 \\
1166: 24&12& 0.16938688975594596186D+10 \\
1167: 24&11& 0.19158522496513690214D+10 \\
1168: 24&10& 0.17528882189096098187D+10 \\
1169: 24&9& 0.13003099701647686803D+10 \\
1170: 24&8& 0.78129739240902704722D+09 \\
1171: 24&7& 0.37851359502279247060D+09 \\
1172: 24&6& 0.14654979817717892487D+09 \\
1173: 24&5& 0.44688725216130675832D+08 \\
1174: 24&4& 0.10490868838890698988D+08 \\
1175: 24&3& 0.18288174674782467832D+07 \\
1176: 24&2& 0.22289668577994086184D+06 \\
1177: 24&1& 0.16948463518651144532D+05 \\
1178: 24&0& 0.60509199550607846163D+03 \\
1179: \hline
1180: \end{tabular}
1181: \end{table}
1182: \end{center}
1183: \begin{center}
1184: \begin{table}
1185: \begin{tabular}{|c|c|c|}
1186: \hline
1187: 25&24& 0.50000051317104317050D+00 \\
1188: 25&23& 0.75000019616897146690D+02 \\
1189: 25&22& 0.34503863852011384949D+04 \\
1190: 25&21& 0.75555516468201169288D+05 \\
1191: 25&20& 0.96061396122189321999D+06 \\
1192: 25&19& 0.79292390146647733217D+07 \\
1193: 25&18& 0.45593449242937929000D+08 \\
1194: 25&17& 0.19157385641492643557D+09 \\
1195: 25&16& 0.60860240979789494986D+09 \\
1196: 25&15& 0.14987944522389162749D+10 \\
1197: 25&14& 0.29148047590757961700D+10 \\
1198: 25&13& 0.45381694251898121450D+10 \\
1199: 25&12& 0.57125179306761664221D+10 \\
1200: 25&11& 0.58520654724976619576D+10 \\
1201: 25&10& 0.48967910880832077324D+10 \\
1202: 25&9& 0.33495648941633358042D+10 \\
1203: 25&8& 0.18689764057712466183D+10 \\
1204: 25&7& 0.84597759965914463009D+09 \\
1205: 25&6& 0.30764609873623349756D+09 \\
1206: 25&5& 0.88524517163033653588D+08 \\
1207: 25&4& 0.19689926825686265843D+08 \\
1208: 25&3& 0.32638605561858318182D+07 \\
1209: 25&2& 0.37947216762104512000D+06 \\
1210: 25&1& 0.27603268426234146559D+05 \\
1211: 25&0& 0.94519638900755200694D+03 \\
1212: \hline
1213: \end{tabular}
1214: \caption{Coefficients $b_{i,j}$ of the supercritical series expansion for the
1215: ultimate survival probability of A particles.}
1216: \label{coefs}
1217: \end{table}
1218: \end{center}
1219:
1220:
1221: \begin{center}
1222: \begin{figure}[h!]
1223: \includegraphics[scale=0.7]{diagramafases.eps}
1224: \vspace{0.4cm}
1225: \caption{Phase diagram of the model obtained from mean-field approximations
1226: and simulations. The physical region is $p_a+p_c \leq 1$ and the line
1227: $p_a+p_c=1$ corresponds to the contact process. In the region labeled RI the
1228: absorbing phase (all particles of type B) is stable, while in the region RII
1229: an active phase is stable, with a nonzero density of particles A.}
1230: \label{f1}
1231: \end{figure}
1232: \end{center}
1233:
1234: \begin{center}
1235: \begin{figure}
1236: \includegraphics[scale=0.7]{rm.eps}
1237: \caption{Ratio of successive coefficients of the series expansion in $\lambda$
1238: for fixed values of $\alpha$. The results shown are for $\alpha=0$ (circles),
1239: $\alpha=1$, $\alpha=2$, $\alpha=3$ $\alpha=5$, and $\alpha=10$ (triangles
1240: pointing down). The solid line shows the expected asymptotic behavior for
1241: $\alpha=0$ (contact process).}
1242: \label{rm}
1243: \end{figure}
1244: \end{center}
1245:
1246:
1247: \begin{center}
1248: \begin{figure}
1249: \includegraphics[scale=0.5]{compsim.eps}
1250:
1251: \vspace{1cm}
1252:
1253: \includegraphics[scale=0.5]{compcarac.eps}
1254: \caption{Estimates for the critical line. Curve (a) shows results from
1255: simulations (triangles) and Pad\'e approximants (circles). Curve (b) displays
1256: the characteristic which starts at the CP (full line) and simulation results
1257: (triangles). In curve (c) the full line is again the characteristic and the
1258: values which follow from the multicritical scaling form are represented by
1259: crosses.}
1260: \label{cline}
1261: \end{figure}
1262: \end{center}
1263:
1264: \begin{center}
1265: \begin{figure}
1266: \includegraphics[scale=0.7]{aglomeradopc.eps}
1267: \caption{Estimates for the location of the multicritical point provided by the
1268: set of PDA's.}
1269: \label{mcp}
1270: \end{figure}
1271: \end{center}
1272:
1273: \begin{center}
1274: \begin{figure}
1275: \includegraphics[scale=0.7]{espalhaexp.eps}
1276: \caption{Estimates for the values of the exponents $\beta^\prime$ and $\phi$.}
1277: \label{betaphi}
1278: \end{figure}
1279: \end{center}
1280: \end{document}
1281: