1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: \documentstyle[aps]{revtex}
3: %----------------------------------------------------------------
4: \begin{document}
5: \draft
6: \title{
7: Multifractal Behavior of the Two-Dimensional Ising Model
8: at small spatio-temporal scales
9: \cite{N}}
10: \author{
11: W. Sakikawa and O. Narikiyo}
12: \address{
13: Department of Physics,
14: Kyushu University,
15: Fukuoka 810-8560,
16: Japan}
17: \date{
18: December, 2001}
19: \maketitle
20: %----------------------------------------------------------------
21: \begin{abstract}
22: The distribution of the fractal dimension
23: of the two-dimensional Ising model
24: at the critical temperature
25: measured by the Monte-Carlo simulation is discussed.
26: At small spatio-temporal scales
27: it exhibits a multifractal behavior
28: and is well fitted by a non-Gaussian distribution
29: derived from the R\'enyi or Tsallis entropy.
30: \vskip 8pt
31: \noindent
32: {\it Keywords:} multifractal, non-equilibrium, large deviation,
33: R\'enyi entropy, Tsallis entropy
34: \end{abstract}
35: %----------------------------------------------------------------
36: \vskip 18pt
37: Recently a large-deviation analysis\cite{FT}
38: for the Monte-Carlo data
39: of the two-dimensional Ising model at the critical temperature
40: has been attempted.
41: Inspired by this attempt we perform a multifractal analysis
42: for the same model in this Short Note.
43: Here we do not distinguish the large-deviation and the multifractal
44: schemes.\cite{EM}
45:
46: Multifractal behaviors of the Ising model are expected
47: for random or chaotic lattices,\cite{OY}
48: while the model on a regular lattice is understood
49: in terms of the mono-fractal analysis.\cite{IS}
50: Even for the model on a regular lattice,
51: we can expect multifractal behaviors when we observe it
52: at small spatio-temporal scales.
53: This is a view point of the large-deviation scheme.\cite{FI}
54:
55: In the following we measure the distribution
56: of the fractal dimension of the model on the square lattice
57: at the critical temperature and show that the data is fitted
58: by a non-Gaussian distribution
59: derived from the R\'enyi or Tsallis entropy.
60: Such a non-Gaussian behavior originates from
61: the multifractality of the spin states
62: and the $q$-index appearing in the distribution
63: serves as a measure for the multifractality.
64: On the other hand
65: the non-Gaussian behavior is observed
66: at small spatio-temporal scales
67: as expected from the large-deviation scheme
68: so that the $q$-index also serves as a measure
69: for the degree of non-equilibrium.
70:
71: We measure the fractal dimension of the Ising model
72: on the square lattice,
73: %----------------------------------------------------------------
74: \begin{equation}
75: H = -J \sum_{<ij>}^{L \times L} \sigma_i \sigma_j,
76: \end{equation}
77: %----------------------------------------------------------------
78: where we adopt the ordinary notation,
79: $<ij>$ means the nearest neighbor bond
80: and $L$ is the number of the spins in $x$- or $y$-direction.
81: The fractal dimension,\cite{IS}
82: %----------------------------------------------------------------
83: \begin{equation}
84: D = \ln[M(1)/M(b)]/\ln(b),
85: \end{equation}
86: %----------------------------------------------------------------
87: is determined by the Monte-Carlo data at the critical temperatute,
88: $T_{\rm c} = 2J / \ln(1+\sqrt{2}) k_{\rm B}$.
89: Here $b$ is the scaling parameter for the coarse-graining
90: and $L \times L$ spins are transformed into $(L/b)\times(L/b)$ block spins.
91: $M(b)$ is the total spin at the level of $b$ of the coarse-graining.
92: The Monte-Carlo up-dating is done with the Metropolis algorithm.
93: The simulation time, $t$, is represented
94: by the unit of the usual Monte-Carlo step (MCS).
95: In the following
96: the measurement is done during $ \tau < t < \tau + T$.
97: The simulation is performed on our personal computer
98: with Pentium-4 CPU.
99:
100: Some typical data for the unnormalized distribution
101: of the fractal dimension are shown in Fig.\ 1
102: where $D$ is measured at each Monte-Carlo steps
103: and the total number of the measurement is $T$.
104: The simulation data are well fitted by the distribution function,
105: %----------------------------------------------------------------
106: \begin{equation}
107: P(D) = P_0 \cdot [1 + (q-1)(D - D_0)^2/2\sigma^2]^{1/(1-q)},
108: \end{equation}
109: %----------------------------------------------------------------
110: which is derived by the maximum-entropy principle
111: using the R\'enyi or Tsallis entropy\cite{Tsallis,AA,SB}
112: so that the distribution exhibits a multifractality of the system.
113: The Tsallis entropy is non-extensive,
114: while the R\'enyi entropy is extensive.
115: The functional form of the distribution is independent of
116: the choice of the entropy, the R\'enyi or Tsallis,\cite{AA}
117: so that our result is independent of the extensivity of the entropy.
118: Here $q$ represents the multifractality of the system investigated
119: and the mean value, $D_0 \sim 1.875$, is the one
120: expected from the mono-fractal analysis.\cite{IS}
121: The limiting case with $q \rightarrow 1$
122: corresponds to the Gaussian distribution
123: which is parabolic in the $D$-$\ln P$ plane.
124: The $q$-index for the solid curve is determined
125: by the least-square fitting.
126: In our scheme
127: $\sigma$ is also determined by the data fitting.
128: The peak value of the unnormalized distribution is $P_0$.
129:
130: In Fig.\ 2
131: the system-size dependence of the $q$-index is shown.
132:
133: In Fig.\ 3
134: the simulation-time dependence of the $q$-index is shown.
135:
136: From the data in Figs.\ 2 and 3
137: we can read a tendency that $q \rightarrow 1$
138: at large spatio-temporal scales.
139: The value, $q=1$, is expected
140: for the equilibrium state in the thermodynamic limit.
141: Our data at small spatio-temporal scales
142: reflect a deviation from this limit
143: and exhibit non-equilibrium nature.
144: A similar analysis has already been done
145: for the experimantal data for turbulene.\cite{Tsallis}
146:
147: In conclusion
148: we numerically observed a multifractal behavior
149: of the two-dimensional Ising model
150: at small spatio-temporal scales.
151: The $q$-index serves as a measure not only for multifractality
152: but also non-equilibrium in this case.
153: %----------------------------------------------------------------
154: \begin{references}
155:
156: \bibitem{N}
157: This paper has been submitted to the Short Note section
158: of J. Phys. Soc. Jpn.
159:
160: \bibitem{FT}
161: H.\ Fujisaka and H.\ Tutu: Unpublished work
162: where the coarse-grained order-parameter measured
163: by the Monte-Carlo simulation is analyzed
164: in terms of the entropy function of the large-deviation theory.
165:
166: \bibitem{EM}
167: C.\ J.\ G.\ Evertsz and B.\ B.\ Mandelbrot:
168: {\it Chaos and Fractals} H-O.\ Peitgen, H.\ J\"urgens and D.\ Saupe
169: (Springer-Verlag, New York, 1992) p.921.
170:
171: \bibitem{OY}
172: T.\ Olson and A.\ P.\ Young:
173: Phys. Rev. B {\bf 60} (1999) 3428 and references therein.
174:
175: \bibitem{IS}
176: N.\ Ito and M.\ Suzuki:
177: Prog. Theor. Phys. {\bf 77} (1987) 1391.
178:
179: \bibitem{FI}
180: H.\ Fujisaka and M.\ Inoue:
181: Phys. Rev. A {\bf 39} (1989) 1376.
182:
183: \bibitem{Tsallis}
184: C.\ Tsallis:
185: {\it Nonextensive Statistical Mechanics and Its Applications}
186: S.\ Abe and Y.\ Okamoto (Springer-Verlag, Berlin, 2001) p.3.
187:
188: \bibitem{AA}
189: T.\ Arimitsu and N.\ Arimitsu:
190: Physica A {\bf 295} (2001) 177.
191:
192: \bibitem{SB}
193: B.\ K.\ Shivamoggi and C.\ Beck:
194: J. Phys. A: Math. Gen. {\bf 34} (2001) 4003.
195:
196: \end{references}
197: \vskip 100pt
198: %----------------------------------------------------------------
199:
200: \begin{figure}
201: \caption{
202: The distripution of the fractal dimension
203: for a randomly-produced initial state
204: with $L=125$, $b=5$ and $\tau = 10^5$. }
205: \label{fig:1}
206: \end{figure}
207:
208: \begin{figure}
209: \caption{
210: The system-size dependence of the $q$-index
211: averaged over 10 randomly-produced initial states
212: with $b=5$, $\tau = 10^5$ and $T = 10^4$. }
213: \label{fig:2}
214: \end{figure}
215:
216: \begin{figure}
217: \caption{
218: The simulation-time dependence of the $q$-index
219: averaged over 10 randomly-produced initial states
220: with $L=125$, $b=5$ and $T = 10^5$. }
221: \label{fig:3}
222: \end{figure}
223:
224: %----------------------------------------------------------------
225: \end{document}
226: %----------------------------------------------------------------
227: \bye
228: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
229: