1: \documentclass[twocolumn,superscriptaddress,showpacs,preprintnumbers,amsmath,amssymb,aps,pre]{revtex4}
2: %\documentclass[galley,superscriptaddress,showpacs,preprintnumbers,amsmath,amssymb,aps,pre]{revtex4}
3: %\documentclass[preprint,superscriptaddress,showpacs,preprintnumbers,amsmath,amssymb,aps,pre]{revtex4}
4: \usepackage{graphicx}% Include figure files
5: \usepackage{dcolumn}% Align table columns on decimal point
6: \usepackage{bm}% bold math
7: \begin{document}
8: \title{Similarity of fluctuations in correlated systems: The case of seismicity\footnote{Published in Physical Review E {\bf72}, 041103 (2005).}}
9: \author{P. A. Varotsos}
10: \email{pvaro@otenet.gr}
11: \affiliation{Solid State Section, Physics Department, University of Athens, Panepistimiopolis, Zografos 157 84, Athens, Greece}
12: \affiliation{Solid Earth Physics Institute, Physics Department, University of Athens, Panepistimiopolis, Zografos 157 84, Athens, Greece}
13: \author{N. V. Sarlis}
14: \affiliation{Solid State Section, Physics Department, University of Athens, Panepistimiopolis, Zografos 157 84, Athens, Greece}
15: \author{H. K. Tanaka}
16: \affiliation{Earthquake Prediction Research Center, Tokai University 3-20-1, Shimizu-Orido, Shizuoka 424-8610, Japan}
17: \author{E. S. Skordas}
18: \affiliation{Solid Earth Physics Institute, Physics Department, University of Athens, Panepistimiopolis, Zografos 157 84, Athens, Greece}
19: \begin{abstract}
20: We report a similarity of fluctuations in equilibrium critical phenomena and non-equilibrium
21: systems, which is based on the concept of natural time.
22: The world-wide seismicity as well as that of San Andreas fault
23: system and Japan are analyzed. An order parameter is chosen and its fluctuations relative to the standard
24: deviation of the distribution are studied. We find that the scaled distributions fall on the same curve,
25: which interestingly exhibits, over four orders of magnitude, features similar to those in several
26: equilibrium critical phenomena ( e.g., 2D Ising model) as well as in non-equilibrium systems
27: (e.g., 3D turbulent flow).
28: \end{abstract}
29: \pacs{05.40.-a, 91.30.Dk, 89.75.Da, 89.75.-k}
30: \maketitle
31:
32: \section{introduction}
33: Recently, a great interest has been focused on the fluctuations of correlated systems in
34: general and of critical systems in particular\cite{BRA98,BRA00,BRA01,ZHE01,BRA00R1,BRA00R2,BRA00R3,ZHE03,CLU04}.
35: Bramwell, Holdsworth and Pinton (BHP)\cite{BRA98}, in an experiment of
36: a closed turbulent flow, found that the ({\em normalized})
37: probability distribution function (PDF) of
38: the power fluctuations has the same functional form as that of the magnetization ($M$) of the
39: finite-size 2D (two-dimensional) XY equilibrium model in the critical region below the
40: Kosterlitz-Thouless transition temperature (Magnetic ordering is then described by the {\em order parameter} $M$).
41: The normalized PDF, denoted by $P(m)$, is defined by introducing the
42: reduced magnetization\cite{BRA98} $m=(M-\langle M\rangle )/\sigma$, where $\langle M\rangle$
43: denotes the mean and $\sigma$ the standard
44: deviation. For both systems, BHP found that while the high end ($m>0$) of the distribution
45: has\cite{BRA98} a Gaussian shape the asymptote of which was later clarified\cite{BRA01} to have
46: a double exponential
47: form, a distinctive exponential tail appears towards the low end ($m<0$) of
48: the distribution. The latter tail, which will be hereafter simply called, for the sake of convenience,
49: ``exponential tail'', provides the main
50: region of interest\cite{BRA98}, since it shows that the probability for a rare fluctuation, e.g., of
51: greater than six standard deviations from the mean, is almost five orders of magnitude higher than in
52: the Gaussian case. Subsequent independent simulations
53: \cite{BRA00,BRA01,ZHE01,ZHE03,CLU04} showed that a variety of highly
54: correlated (non equilibrium as well as equilibrium)
55: systems, under certain conditions, exhibit approximately the ``exponential tail''.
56:
57: Earthquakes do exhibit complex correlations in space, time and magnitude, e.g. \cite{BAK02,COR04,ABE04}.
58: It has been
59: repeatedly proposed (see Ref.\cite{SOR04} and references therein) that the occurrence of earthquakes (cf. mainshocks) can be considered as
60: a critical point (second-order phase change), but alternative models based on first-order phase transitions have
61: been also forwarded which are probably more applicable, see Ref.\cite{RUN03} and references therein.
62: (Such a diversity also exists for the brittle rupture which is a phenomenon closely
63: related to earthquakes. Buchel and Sethna\cite{BUC97} have associated brittle rupture with
64: a first-order transition and a similar view has been also expressed in Refs.\cite{ZAP97,KUN99}. On the
65: other hand, Gluzman and Sornette\cite{GLU01} later suggested that it is
66: analogous to a critical point phenomenon.) Both approaches lead to scaling laws or
67: power-law distributions for the dynamical variables (second-order transition demonstrate
68: scaling near a critical point, whereas first-order transitions demonstrate scaling when
69: the range of interactions is large (mean-field condition), as is the case with elastic
70: interactions\cite{RUN03}). However, the question on whether
71: earthquakes exhibit an ``exponential tail'', has not yet been clarified. This might be due to the major difficulty
72: of choosing an order parameter in the case of earthquakes(EQs). Following the wording of Ref.\cite{SET92}, we note that
73: in general such a choice is an art, since usually it's a new phase which we do not understand yet, and guessing the
74: order parameter is a piece of figuring out what's going on. The scope of the present paper is twofold: to propose
75: an order parameter for the case of EQs and then examine whether an ``exponential tail'' appears.
76: We find that our scope is achieved only {\em if} we analyze the series of earthquakes in the
77: natural time-domain\cite{VAR01,VAR02,VAR02A,VAR03A,VAR03B,VAR04,VAR05,VAR05B}.
78:
79:
80: In order to serve the aforementioned scope, the present paper is organized as follows:
81: In Section II, we explain how the power spectrum of the seismicity in natural time can be obtained.
82: An order parameter for EQs is proposed in Section III. In the light of this proposal,
83: and without using any adjustable parameter, we show in Section IV that
84: the normalized distribution of the long term seismicity for different seismic areas fall on a universal curve. It consists of two segments the one of which exhibits the ``exponential tail''. Interestingly, a further investigation of the latter segment in Section V, reveals that it is similar to that observed in several equilibrium critical phenomena (e.g., 2D Ising, 3D Ising) and in non-equilibrium systems (e.g., 3D turbulent flow). A brief discussion follows in Section VI, while Section VII summarizes our main conclusions. Two Appendices provide clarifications
85: on some points discussed in the main text.
86:
87: \section{the seismicity in natural time}
88:
89: In a time series consisting of $N$ events, the {\em natural time} $\chi_k = k/N$ serves as an index\cite{VAR01,VAR02}
90: for the occurrence of the $k$-th event. It is, therefore, smaller than, or equal to, unity.
91: For the analysis of seismicity, the evolution of the pair ($\chi_k, E_k$)
92: is considered\cite{VAR01,VAR02A,VARBOOK,TAN04}, where $E_k$ denotes the seismic energy released during the
93: $k$-th event see Fig.\ref{fg1} (cf. This energy -which is itself proportional to the seismic moment $M_0$
94: and hence we can use in the vertical axis of Fig.\ref{fg1}(b) either $E_k$ or $(M_0)_k$ is
95: related\cite{EPAPS} to the magnitude M through $E\propto 10^{c {\rm M}}$, where $c$ is a constant around 1.5).
96: The following continuous function $F(\omega )$ was introduced\cite{VAR01,VAR02,VAR02A}:
97: %\begin{equation}
98: $F(\omega)=\sum_{k=1}^{N} E_{k}
99: \exp \left( i \omega \frac{k}{N} \right)$
100: %\end{equation}
101: where $\omega =2 \pi \phi$, and $\phi$ stands for the
102: {\it natural frequency}. We normalize $F( \omega )$
103: by dividing it by $F(0)$,
104: \begin{equation}
105: \Phi(\omega)=\frac{\sum_{k=1}^{N} E_{k}
106: \exp \left( i \omega \frac{k}{N} \right)}{\sum_{n=1}^{N} E_{n}}=
107: \sum_{k=1}^{N} p_k \exp \left( i \omega \frac{k}{N} \right)
108: \label{equ1}
109: \end{equation}
110: where $p_k=E_{k}/\sum_{n=1}^{N}E_{n}$. A kind of normalized
111: power spectrum $\Pi(\omega)$ can now be defined:
112: $\Pi(\omega)=\left| \Phi(\omega) \right|^2$.
113:
114: For a Seismic Electric Signals (SES) activity, which is a sequence of low frequency
115: ($\leq 1$ Hz) electric pulses emitted when the stress in the
116: focal area approaches\cite{VAR86,VAR03C} a {\em critical } value,
117: we have shown (for details see Ref.\cite{VAR01}, see also \cite{VAR02}) that the following relation holds\cite{VAR01,VAR02,VAR02A}
118: \begin{equation}
119: \Pi ( \omega ) = \frac{18}{5 \omega^2}
120: -\frac{6 \cos \omega}{5 \omega^2}
121: -\frac{12 \sin \omega}{5 \omega^3}.
122: \label{fasma}
123: \end{equation}
124:
125: We focus on the properties of $\Pi(\omega)$ or $\Pi(\phi)$ for natural frequencies $\phi$ less than 0.5, since in
126: this range of $\phi$, $\Pi(\omega)$ or $\Pi(\phi)$ reduces\cite{VAR01,VAR02,VAR02A,VARBOOK}
127: to a {\em characteristic function} for the
128: probability distribution $p_k$ in the context of probability theory. According to the probability
129: theory, the moments of a distribution and hence the distribution itself can be approximately determined once
130: the behavior of the characteristic function of the distribution is known around zero. For
131: $\omega \rightarrow 0$, Eq.(\ref{fasma}) leads to\cite{VAR01,VAR02,VARBOOK}:
132: \begin{equation}
133: \label{eq4}
134: \Pi (\omega )\approx 1-0.07 \omega^2
135: \end{equation}
136: which reflects (see Appendix \ref{apb}) that the variance of $\chi$ is given by:
137: \begin{equation}
138: \label{eq5}
139: \kappa_1=\langle \chi^2 \rangle -\langle \chi \rangle ^2=0.07.
140: \end{equation}
141: In Section IV, we will investigate whether Eq.(\ref{fasma}) holds for EQs.
142:
143:
144:
145: \section{the order parameter proposed}
146: We now proceed to choose the order parameter, assuming that a mainshock may be
147: considered as the new phase.
148: We take advantage of the experimental fact\cite{VAR03C} that several hours to a few
149: months before a mainshock an SES activity is recorded,
150: and focus our attention on the evolution of the seismicity (in the candidate area) during
151: the period from the SES detection until the mainshock.
152: If we set the natural time for the seismicity zero at the initiation
153: of the concerned SES activity, we form time series of seismic events in natural time (see Fig.\ref{fg1}) for
154: various time windows as the number $N$ of consecutive (small) EQs increases. When computing $\Pi (\phi )$
155: (as well as $\kappa_1$, see below) for each of the time windows, we find that,
156: in the range $0<\phi \leq 0.5$,
157: it approaches, as $N$ increases from 6 to some value less than (or equal to) 40, to that given by Eq.(\ref{fasma})
158: (or the $\kappa_1$-value becomes equal to 0.07, see Eq.(\ref{eq5})). The coincidence occurs {\em only} a few
159: hours to a few days before the mainshock. (In simple words, before
160: a mainshock a sequence of earthquakes occurs, which obeys Eq.(\ref{fasma}) and this process will be called
161: {\em single correlated process}.) When the mainshock occurs (the new phase), $\Pi (\phi )$
162: abruptly increases to approximately unity (for details see Ref.\cite{EPAPS}) and $\kappa_1$ becomes almost zero.
163: This can be visualized in the
164: example depicted in Fig.\ref{fg3}, where we plot the $\kappa_1$-value versus the number of EQs after the SES
165: detection on April 18, 1995 (see Refs.\cite{VAR02,VAR03C}) until the occurrence of the M=6.6
166: mainshock on May 13, 1995 at $40.2^o N, 21.7^oE$. This figure shows that the $\kappa_1$-value becomes $\kappa_1\approx 0.07$ after the 11th EQ (see also Ref.\cite{EPAPS}), while upon the mainshock the $\kappa_1$-value
167: {\em abruptly} decreases to $\kappa_1 \approx 9 \times 10^{-5}$.
168: Such a behavior has been verified\cite{VAR01,VAR02A} for
169: several major EQs and points to the conclusion that $\Pi (\phi )$ for small $\phi$, or $\kappa_1$, could be considered
170: as an order parameter.
171:
172: \section{universal curve for seismicity}
173:
174: The properties of the power spectrum for the long term seismicities in natural time can be
175: studied by means of the following
176: procedure: First, calculation of $\Pi(\phi)$ was made for an event taking time
177: windows from 6 to 40 consecutive events
178: (for the reasons explained in Section III; the choice of the precise value of the upper limit, up to 100 or so, is not found decisive\cite{VAR01,VAR02A,TAN04}). And second,
179: this process was performed for all the events by scanning the whole catalogue. The following
180: data from two different areas, i.e., San Andreas fault system and Japan, have been analyzed:
181: First, the EQs that occurred during the period 1973-2003 within the area $N_{32}^{37} W_{114}^{122}$
182: using the Southern California Earthquake catalogue (hereafter called SCEC). Second, the
183: EQs within $N_{25}^{46} E_{125}^{146}$ for the period 1967-2003 using the Japan Meteorological
184: Agency catalogue (hereafter simply called ``Japan''). The thresholds M$\geq$2.0 and M$\geq$3.5 have
185: been considered for SCEC and Japan, respectively, for the sake of data completeness\cite{EPAPS}.
186: By plotting for a given value of $\phi$ the observed probability
187: $P [\Pi(\phi)]$ versus $\Pi(\phi )$ (two such examples are given in Figs.\ref{fg2}(a) and \ref{fg2}(b) for
188: $\phi=0.05$ and $\phi=0.4$, respectively), we find that a local
189: maximum occurs at a value of $\Pi(\phi)$ hereafter called $\Pi_p(\phi)$ (see also Appendix \ref{apc}).
190: This lies very close (see Fig.\ref{fg2}(c))
191: to the value $\Pi_{th}(\phi )$ obtained theoretically, i.e., estimated from Eq.(\ref{fasma}).
192: The validity of Eq.(\ref{fasma}) for various $\phi$-values, in the range $0<\phi \leq 0.5$,
193: can be now visualized in Fig.\ref{fg2}(d), where we see that $\Pi_p(\phi)$-values versus
194: $\phi$ for both SCEC and Japan do not differ
195: by more than 1\% from the $\Pi_{th}(\phi )$-values (cf. this difference is more or less comparable to the estimation error of $\Pi_p(\phi)$, for details see Appendix \ref{apc}).
196:
197:
198:
199:
200: We now plot, in Fig.\ref{fg4}, the quantity $\sigma P(X)$ versus
201: $(X-\langle X\rangle)/\sigma$ where $X$ stands for
202: $\Pi(\phi)$ and $\langle \Pi(\phi) \rangle$ and $\sigma$ refer to the mean value and the standard deviation of
203: $\Pi(\phi)$ (recall that the
204: calculations should be done for small $\phi$-values,
205: e.g., $\phi$=0.05, since we assume here $\phi \rightarrow 0$, for the
206: reasons explained in Section II). One could alternatively plot $\sigma_{\kappa_1}P(\kappa_1)$ versus
207: $(\langle \kappa_1 \rangle -\kappa_1)/\sigma_{\kappa_1}$,
208: where $\langle \kappa_1 \rangle$ and $\sigma_{\kappa_1}$ now refer to the mean value and
209: the standard deviation of $\kappa_1$. The results
210: in Fig. \ref{fg4}, for both areas, fall on the {\em same} curve. This log-linear plot clearly consists of two
211: segments: The segment to the left shows a decrease of $P(X)$ almost by five orders of
212: magnitude, while the upper right segment has an almost constant $P(X)$ (Obviously, the latter segment
213: deviates from the general behavior of the BHP distribution -as it was summarized in Section I-
214: but from thereon we put emphasis on the left segment since our main interest here concerns the ``exponential tail'').
215: The feature of this
216: plot is strikingly reminiscent of the one obtained by Bak et al.\cite{BAK02} (see their Fig. 4) on
217: different grounds, using EQs in California only. More precisely, they measured $P_{S,l}(T)$,
218: the distribution of waiting times $T$, between EQs occurring within range $l$ whose
219: magnitudes are greater than $M\equiv \log S$. They then plotted $T^\alpha P_{S,l}(T)$ versus
220: $TS^{-b}l^d$ and found
221: that, for a {\em suitable choice} of the exponent $\alpha$ (i.e., $\alpha=1$),
222: the Gutenberg-Richter law exponent $b$
223: (i.e., $b$=1) and the spatial dimension $d$ (i.e., fractal dimension $d=1.2$)
224: all the data collapse onto
225: a {\em single} curve which is similar to that of Fig.\ref{fg4}. Recall, however,
226: that Fig.\ref{fg4}
227: was obtained here
228: without considering at all the waiting time distribution and without the suitable choice of any
229: parameter. After a further inspection of Fig.\ref{fg4}, the following points have been clarified:
230:
231: First, the rapidly decaying part (i.e., the left segment), which is consistent with
232: an almost exponential decaying function over almost four orders of magnitude, remains
233: practically unchanged, upon {\em randomizing} the data (``shuffling''\cite{VAR04}). (Some changes
234: do occur in the right part, associated with aftershocks, see also below.) This can be seen in the
235: inset of Fig. \ref{fg4}, where for the sake of clarity only the results from the data of Japan (the
236: original as well as the ``shuffled'' ones) are depicted.
237:
238: %\begin{widetext}
239: Second, the feature of the plot of Fig. \ref{fg4} is not altered upon changing either the seismic
240: region or the time-period (provided that the latter does not include aftershocks {\em only}, see
241: below). As an example, Fig.\ref{fg5}(b) shows that three different regions A, B, C in Japan (depicted in Fig.\ref{fg5}(a)),
242: as well as the whole Japan, result in almost identical plots.
243:
244: Third, the ``upturn branch'' in the upper right part of Fig. \ref{fg4} arises from the presence of
245: aftershocks. It disappears (see the
246: crosses in Fig. \ref{fg6}) when, in Japan, for example, we delete the EQs with M$\leq$5.7 (and
247: hence drastically reduce the number of aftershocks), but it does not, when deleting
248: EQs with smaller threshold, i.e., M$<$4.0; the latter can be also visualized in the SCEC
249: example of Fig. \ref{fg6}, where we give the results for M$\geq$4.0 (cf. this threshold still allows the
250: presence of a reasonable number of aftershocks).
251:
252:
253: Fourth, if we consider the relevant results for the worldwide seismicity (WWS) by
254: taking a large magnitude threshold, i.e., M$>$5.7 (so that for the data to be complete\cite{EPAPS}),
255: we find (see Fig.\ref{fg6} that will be further discussed below) that they fall
256: onto the {\em same} curve with the results of
257: both Japan and SCEC.
258:
259: \section{does a universal behaviour exist for diverse systems?}
260: We now compare in Fig.\ref{fg6} the aforementioned results of seismicity with those obtained in some equilibrium critical systems (e.g., see Ref.\cite{ZHE03}). We first recall that the PDF in the {\em critical} regime depends on $K = 1/T$ and the length $L$
261: through a scaling variable $s\equiv L^{1/\nu} (K-K_c)/K_c$, where $K_c = 1/T_c$ and $T_c$ denotes the critical
262: temperature (the quantity $s^\nu$ provides the ratio of the lattice size and the {\em correlation length} at
263: $K$). In Fig. \ref{fg6}, we include numerical results of the 2D Ising model for $s=8.72 (L=128,
264: K=0.4707)$ and $s=17.44 (L=256, K=0.4707)$. Here, $X$ stands for $M$. These $s$-values were
265: intentionally selected, because\cite{ZHE03} for $s\geq 8.72$ for the 2D Ising model, the $P(m, s)$'s of a
266: number of critical models (i.e., 2D XY, 2D Ising, 3D Ising, 2D three-state Potts) share the
267: {\em same} form (up to a constant factor of $s$), which interestingly exhibits an exponential-like left-tail ($m<0$).
268: An inspection of Fig.\ref{fg6} shows that our 2D Ising results almost coincide (cf. this can be safely checked only for the left segment, i.e., $m<0$)
269: with those of seismicity, i.e., Japan, SCEC and WWS (cf. Some disparity which appears in the upper right
270: part of SCEC only, might be attributed to the selection of the magnitude threshold for seismicity, recall the third
271: point mentioned in Section IV). This coincidence (which seems to be
272: better for $s=17.44$) reveals that the seismicity, irrespective of the seismic area we consider,
273: exhibits -over four orders of magnitude- fluctuations of the order parameter similar to those in
274: several critical systems as well as in 3D turbulent flow.
275:
276: \section{Discussion}
277:
278: It is of interest to see how the scaled distributions look like in
279: the frame of the present analysis, if one generates surrogate data
280: either by means of a simple Poisson model or by the
281: Gutenberg-Richter law and compare the results to those deduced
282: from actual seismicity data.
283:
284: In Fig.\ref{fgPoi}, we present the linear-linear plot (Fig. \ref{fgPoi}(a)) as well as
285: the log-linear plot (Fig. \ref{fgPoi}(b)) of $\sigma P(X)$ versus $(X- \langle X
286: \rangle)/ \sigma$ where $X$ stands for $\Pi(\phi)$ for $\phi
287: \approx 0$, for surrogate data of EQs for which their $(M_o)_{k}$
288: obey a simple Poisson rule for
289: various mean values $\lambda$ lying between 5 and 200.
290: In the same figure we insert the results for Japan (M $\geq 3.5$)
291: and SCEC (M $\geq 2.0$) already discussed in Fig.\ref{fg4}. Although we
292: find that upon decreasing $\lambda$ the surrogate data move closer
293: to the real data, however a satisfactory agreement between them
294: cannot be supported.
295:
296: In Fig. \ref{fgGuRi} we repeat the procedure followed in Fig. \ref{fgPoi}, but now the
297: surrogate data are produced on the basis of the Gutenberg-Richter
298: law, i.e., that the (cumulative) number of EQs with magnitude
299: greater than M (occurring in a specified region and time) is
300: given by
301: \begin{equation}
302: \label{aguri}
303: N(> {\rm M}) \approx 10^{-b{\rm M}}.
304: \end{equation}
305: It is currently
306: considered\cite{RUN03} that $b$ is generally a constant varying
307: only slightly from region to region being approximately in the
308: range $0.8 \leq b \leq 1.2$. For Japan and SCEC we find on the basis of
309: Eq.(\ref{aguri}) $b\approx 1.05$. Note that in Fig.\ref{fgGuRi}, surrogate data
310: are intentionally produced for a variety of $b$ values in the
311: range $b=0.5$ to 2.0. An inspection of this figure leads to the
312: following conclusions: First, the curves of the surrogate data
313: marked with $b=0.5$ to $b=0.9$ significantly differ from that of
314: the real data. Second, for $b$-values larger than 1 and smaller than 1.4,
315: the curves of the surrogate data have a general
316: feature more or less similar to the curve of the real data.
317: However, none of these $b$-values in the surrogate data can lead to a curve
318: coinciding to the one obtained from the real data.
319:
320: In other words,
321: the scaled distribution, deduced within the frame of the present
322: analysis, reveals for the real data an extra complexity when compared to
323: the surrogate data even if the latter are produced with
324: $b$-values comparable to the experimental ones.
325:
326:
327:
328:
329: \section{conclusions}
330: The main conclusions could be summarized as follows:
331:
332: (1)The analysis of the seismicity in the natural time-domain reveals that
333: $\Pi(\phi)$ (for small $\phi$) or $\kappa_1$, may be considered as an order parameter.
334:
335: (2)If we study the order parameter fluctuations relative to the standard deviation of its distribution,
336: the following two facts emerge (without making use of {\em any} adjustable
337: parameter):
338:
339: First, the scaled distributions of different seismic areas (as well as that of the world wide seismicity)
340: fall on the {\em same} curve ({\em universal}).
341:
342: Second, this curve exhibits an ``exponential tail'' form similar to that
343: observed in certain non-equilibrium
344: systems (e.g., 3D turbulent flow) as well as in several(e.g., 2D Ising, 3D Ising, 2D XY)
345: equilibrium critical phenomena.
346:
347:
348: \appendix
349: \section{derivation of equation (\ref{eq5})}
350: \label{apb}
351: The Taylor expansion, around $\omega$=0, of the relation $\Pi(\omega)=\left| \Phi(\omega) \right|^2$ using
352: Eq(1) reveals that \cite{VAR01}
353: \begin{equation}
354: \Pi(\omega)=1-\kappa_1\omega^2+\kappa_2\omega^4+\kappa_3\omega^6+\kappa_4\omega^8+ \ldots
355: \end{equation}
356: where
357: \begin{equation}
358: \kappa_1=- \left. \frac{1}{2}\frac{d^2 \Pi(\omega)}{d\omega^2}\right|_{\omega=0}.
359: \end{equation}
360: We now consider
361: \begin{equation}
362: \frac{d^2\Pi(\omega)}{d\omega^2}={ \Phi}^{*}(\omega) \frac{d^2\Phi(\omega )}{d\omega^2}+
363: \Phi(\omega)\frac{d^2\Phi ^*(\omega)}{d\omega^2}+2 \frac{d\Phi(\omega)}{d\omega}
364: \frac{d\Phi ^*(\omega)}{d\omega}
365: \end{equation}
366: and taking into account that $\Phi(\omega)\equiv \sum_{k} p_k \exp ( i\omega \chi_k )$, with $\Phi(0)=1$, we find:
367: \begin{eqnarray}
368: &\kappa_1&=-\frac{1}{2} \left[ -\sum_{k}p_k \chi_k^2-\sum_{k}p_k \chi_k^2+2 \left( \sum_{k}p_k \chi_k \right)^2 \right] \nonumber \\
369: &=&\langle \chi^2 \rangle - \langle \chi \rangle^2,
370: \end{eqnarray}
371: where $\langle \chi^n \rangle = \sum_{k}p_k \chi_k^n$.
372:
373: Expanding Eq.(\ref{fasma}), around $\omega=0$, we have:
374: \begin{equation}
375: \Pi(\omega)=1-0.07 \omega^2+\ldots
376: \end{equation}
377: and hence:
378: \begin{equation}
379: \kappa_1=\langle \chi^2\rangle - \langle \chi\rangle^2=0.07
380: \end{equation}
381: which is just Eq.(\ref{eq5}).
382:
383:
384: \section{the procedure to determine the maximum in $P[\Pi(\phi)]$ versus $\Pi(\phi)$}
385: \label{apc}
386: The calculation of $\Pi(\phi)$ was made, as mentioned in Section IV, for an event taking time windows from 6 to 40
387: consecutive events and this process was performed for all
388: the events by scanning the whole catalogue. This
389: procedure resulted in the calculation, for each $\phi$-value
390: and each catalogue, of more than $10^6$ $\Pi(\phi)$-values, whose
391: probability density function (PDF) was determined by
392: using the computer code {\tt histogram} of Ref.\cite{heg99} with a
393: number of bins proportional to $N^{1/3}$, where $N$ is the
394: number of $\Pi(\phi)$-values (cf. this point, i.e., that the
395: number of bins should be proportional to $N^{1/3}$, is
396: discussed in Ref. \cite{mer99}). This method resulted in the PDFs
397: shown in Fig.\ref{fg7} as well as
398: in those depicted in Figs.\ref{fg2}(a),\ref{fg2}(b), \ref{fg4}, \ref{fg5}(b), and \ref{fg6}.
399: Due to the
400: intrinsic fluctuations of the values of the PDF (because
401: $N$ is still finite), a direct determination of the value
402: $\Pi_p(\phi)$ where the PDF maximizes, just by simply taking the
403: maximum value of the calculated PDF, may lead to
404: erroneous values of $\Pi_p(\phi)$. One should consider the general
405: trend of the PDF as a whole, which can definitely show a
406: more accurate and stable value of $\Pi_p(\phi)$. Thus, the
407: procedure we applied for the determination of $\Pi_p(\phi)$ was as
408: follows: For each $\phi$-value, a region $[a,b]$ around the
409: maximum was selected (examples are shown in Figs.\ref{fg7}(a) and
410: (b)) and then a cubic polynomial, $p(x)=a+bx+cx^2+dx^3$, was
411: used to fit the PDF values in this region. (Close enough
412: to the maximum, a parabolic fit could be also good since
413: $f(x)=f_{max}-|f^{''}_{max} | (x-x_{max})^2/2$, but in view of the PDF
414: asymmetry the cubic polynomial used, provides a better
415: approximation in the whole region $[a,b]$.) The value of
416: $\Pi_p(\phi)$ was determined through the direct maximization of this
417: cubic polynomial, i.e, $\Pi_p(\phi)=\frac{-2c-\sqrt{4c^2-12bd}}{6d}$.
418: The values of $\Pi_p(\phi)$ shown
419: in Fig.\ref{fg2}(c) have been obtained by means
420: of such a procedure. Finally, we note that, due to the
421: fitting procedure involved and the relative arbitrariness
422: in the definition of [a,b], the estimation error of $\Pi_p(\phi)$
423: is more or less comparable to its percentage deviation
424: from $\Pi_{th}(\phi)$, depicted in Fig.\ref{fg2}(d). Thus,
425: we can state that $\Pi_p(\phi)$ and $\Pi_{th}(\phi)$ are experimentally
426: indistinguishable, which strengthens the statement that
427: Eq.(2) -which has been used for
428: estimating $\Pi_{th}(\phi)$- holds for EQs.
429: %\bibliography{prlop2}
430: %\bibliographystyle{apsrev}
431: \begin{thebibliography}{35}
432: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
433: \expandafter\ifx\csname bibnamefont\endcsname\relax
434: \def\bibnamefont#1{#1}\fi
435: \expandafter\ifx\csname bibfnamefont\endcsname\relax
436: \def\bibfnamefont#1{#1}\fi
437: \expandafter\ifx\csname citenamefont\endcsname\relax
438: \def\citenamefont#1{#1}\fi
439: \expandafter\ifx\csname url\endcsname\relax
440: \def\url#1{\texttt{#1}}\fi
441: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
442: \providecommand{\bibinfo}[2]{#2}
443: \providecommand{\eprint}[2][]{\url{#2}}
444:
445: \bibitem[{\citenamefont{Bramwell et~al.}(1998)\citenamefont{Bramwell,
446: Holdsworth, and Pinton}}]{BRA98}
447: \bibinfo{author}{\bibfnamefont{S.~T.} \bibnamefont{Bramwell}},
448: \bibinfo{author}{\bibfnamefont{P.~C.~W.} \bibnamefont{Holdsworth}},
449: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~F.}
450: \bibnamefont{Pinton}}, \bibinfo{journal}{Nature (London)}
451: \textbf{\bibinfo{volume}{396}}, \bibinfo{pages}{552} (\bibinfo{year}{1998}).
452:
453: \bibitem[{\citenamefont{Bramwell1 et~al.}(2000)\citenamefont{Bramwell1,
454: Christensen, Fortin, Holdsworth, Jensen, Lise, L\'{o}pez, Nicodemi, Pinton,
455: and Sellitto}}]{BRA00}
456: \bibinfo{author}{\bibfnamefont{S.~T.} \bibnamefont{Bramwell1}},
457: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Christensen}},
458: \bibinfo{author}{\bibfnamefont{J.~Y.} \bibnamefont{Fortin}},
459: \bibinfo{author}{\bibfnamefont{P.~C.~W.} \bibnamefont{Holdsworth}},
460: \bibinfo{author}{\bibfnamefont{H.~J.} \bibnamefont{Jensen}},
461: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Lise}},
462: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{L\'{o}pez}},
463: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Nicodemi}},
464: \bibinfo{author}{\bibfnamefont{J.-F.} \bibnamefont{Pinton}},
465: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Sellitto}},
466: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{84}},
467: \bibinfo{pages}{3744} (\bibinfo{year}{2000}).
468:
469: \bibitem[{\citenamefont{Bramwell1
470: et~al.}(2001{\natexlab{a}})\citenamefont{Bramwell1, Holdsworth, Peysson,
471: Portelli, and Sellitto}}]{BRA01}
472: \bibinfo{author}{\bibfnamefont{J.~Y.} \bibnamefont{Bramwell1},
473: \bibfnamefont{S.~T. nad~Fortin}}, \bibinfo{author}{\bibfnamefont{P.~C.~W.}
474: \bibnamefont{Holdsworth}}, \bibinfo{author}{\bibfnamefont{J.~F.}
475: \bibnamefont{Peysson}, \bibfnamefont{S.~Pinton}},
476: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Portelli}}, \bibnamefont{and}
477: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Sellitto}},
478: \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{63}},
479: \bibinfo{pages}{041106} (\bibinfo{year}{2001}{\natexlab{a}}).
480:
481: \bibitem[{\citenamefont{Zheng and Trimper}(2001)}]{ZHE01}
482: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Zheng}} \bibnamefont{and}
483: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Trimper}},
484: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{87}},
485: \bibinfo{pages}{188901} (\bibinfo{year}{2001}).
486:
487: \bibitem[{\citenamefont{Bramwell1
488: et~al.}(2001{\natexlab{b}})\citenamefont{Bramwell1, Christensen, Fortin,
489: Holdsworth, Jensen, Lise, L\'{o}pez, Nicodemi, Pinton, and
490: Sellitto}}]{BRA00R1}
491: \bibinfo{author}{\bibfnamefont{S.~T.} \bibnamefont{Bramwell1}},
492: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Christensen}},
493: \bibinfo{author}{\bibfnamefont{J.~Y.} \bibnamefont{Fortin}},
494: \bibinfo{author}{\bibfnamefont{P.~C.~W.} \bibnamefont{Holdsworth}},
495: \bibinfo{author}{\bibfnamefont{H.~J.} \bibnamefont{Jensen}},
496: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Lise}},
497: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{L\'{o}pez}},
498: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Nicodemi}},
499: \bibinfo{author}{\bibfnamefont{J.-F.} \bibnamefont{Pinton}},
500: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Sellitto}},
501: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{87}},
502: \bibinfo{pages}{188902} (\bibinfo{year}{2001}{\natexlab{b}}).
503:
504: \bibitem[{\citenamefont{Watkins et~al.}(2002)\citenamefont{Watkins, Chapman,
505: and Rowlands}}]{BRA00R2}
506: \bibinfo{author}{\bibfnamefont{N.~W.} \bibnamefont{Watkins}},
507: \bibinfo{author}{\bibfnamefont{S.~C.} \bibnamefont{Chapman}},
508: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Rowlands}},
509: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{89}},
510: \bibinfo{pages}{208901} (\bibinfo{year}{2002}).
511:
512: \bibitem[{\citenamefont{Bramwell1 et~al.}(2002)\citenamefont{Bramwell1,
513: Christensen, Fortin, Holdsworth, Jensen, Lise, L\'{o}pez, Nicodemi, Pinton,
514: and Sellitto}}]{BRA00R3}
515: \bibinfo{author}{\bibfnamefont{S.~T.} \bibnamefont{Bramwell1}},
516: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Christensen}},
517: \bibinfo{author}{\bibfnamefont{J.~Y.} \bibnamefont{Fortin}},
518: \bibinfo{author}{\bibfnamefont{P.~C.~W.} \bibnamefont{Holdsworth}},
519: \bibinfo{author}{\bibfnamefont{H.~J.} \bibnamefont{Jensen}},
520: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Lise}},
521: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{L\'{o}pez}},
522: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Nicodemi}},
523: \bibinfo{author}{\bibfnamefont{J.-F.} \bibnamefont{Pinton}},
524: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Sellitto}},
525: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{89}},
526: \bibinfo{pages}{208902} (\bibinfo{year}{2002}).
527:
528: \bibitem[{\citenamefont{Zheng}(2003)}]{ZHE03}
529: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Zheng}},
530: \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{67}},
531: \bibinfo{pages}{026114} (\bibinfo{year}{2003}).
532:
533: \bibitem[{\citenamefont{Clusel et~al.}(2004)\citenamefont{Clusel, Fortin, and
534: Holdsworth}}]{CLU04}
535: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Clusel}},
536: \bibinfo{author}{\bibfnamefont{J.~Y.} \bibnamefont{Fortin}},
537: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{P.~C.~W.}
538: \bibnamefont{Holdsworth}}, \bibinfo{journal}{Phys. Rev. E}
539: \textbf{\bibinfo{volume}{70}}, \bibinfo{pages}{046112}
540: (\bibinfo{year}{2004}).
541:
542: \bibitem[{\citenamefont{Bak et~al.}(2002)\citenamefont{Bak, Christensen, Danon,
543: and Scanlon}}]{BAK02}
544: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Bak}},
545: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Christensen}},
546: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Danon}}, \bibnamefont{and}
547: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Scanlon}},
548: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{88}},
549: \bibinfo{pages}{178501} (\bibinfo{year}{2002}).
550:
551: \bibitem[{\citenamefont{Corral}(2004)}]{COR04}
552: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Corral}},
553: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{92}},
554: \bibinfo{pages}{108501} (\bibinfo{year}{2004}).
555:
556: \bibitem[{\citenamefont{Abe and Suzuki}(2004)}]{ABE04}
557: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Abe}} \bibnamefont{and}
558: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Suzuki}},
559: \bibinfo{journal}{Europhys. Lett.} \textbf{\bibinfo{volume}{65}},
560: \bibinfo{pages}{581} (\bibinfo{year}{2004}).
561:
562: \bibitem[{\citenamefont{Sornette}(2004)}]{SOR04}
563: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Sornette}},
564: \emph{\bibinfo{title}{Critical Phenomena in Natural Science}}
565: (\bibinfo{publisher}{Springer}, \bibinfo{address}{Berlin},
566: \bibinfo{year}{2004}), \bibinfo{edition}{2nd} ed.
567:
568: \bibitem[{\citenamefont{Rundle et~al.}(2003)\citenamefont{Rundle, Turcotte,
569: Shcherbarkov, Klein, and Sammis}}]{RUN03}
570: \bibinfo{author}{\bibfnamefont{J.~B.} \bibnamefont{Rundle}},
571: \bibinfo{author}{\bibfnamefont{D.~L.} \bibnamefont{Turcotte}},
572: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Shcherbarkov}},
573: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Klein}}, \bibnamefont{and}
574: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Sammis}},
575: \bibinfo{journal}{Rev. Geophys.} \textbf{\bibinfo{volume}{41}},
576: \bibinfo{pages}{1019} (\bibinfo{year}{2003}).
577:
578: \bibitem[{\citenamefont{Buchel and Sethna}(1997)}]{BUC97}
579: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Buchel}} \bibnamefont{and}
580: \bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Sethna}},
581: \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{55}},
582: \bibinfo{pages}{7669} (\bibinfo{year}{1997}).
583:
584: \bibitem[{\citenamefont{Zapperi et~al.}(1997)\citenamefont{Zapperi, Ray,
585: Stanley, and Vespignani}}]{ZAP97}
586: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Zapperi}},
587: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Ray}},
588: \bibinfo{author}{\bibfnamefont{H.~E.} \bibnamefont{Stanley}},
589: \bibnamefont{and}
590: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Vespignani}},
591: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{78}},
592: \bibinfo{pages}{1408} (\bibinfo{year}{1997}).
593:
594: \bibitem[{\citenamefont{Kun and Herrmann}(1999)}]{KUN99}
595: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Kun}} \bibnamefont{and}
596: \bibinfo{author}{\bibfnamefont{H.~J.} \bibnamefont{Herrmann}},
597: \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{59}},
598: \bibinfo{pages}{2623} (\bibinfo{year}{1999}).
599:
600: \bibitem[{\citenamefont{Gluzman and Sornette}(2001)}]{GLU01}
601: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Gluzman}} \bibnamefont{and}
602: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Sornette}},
603: \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{63}},
604: \bibinfo{pages}{066129} (\bibinfo{year}{2001}).
605:
606: \bibitem[{\citenamefont{Sethna}(1992)}]{SET92}
607: \bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Sethna}}, in
608: \emph{\bibinfo{booktitle}{1991 Lectures in Complex Systems, Santa Fe
609: Institute Studies in the Sciences of Complexity, Proc. Vol. XV}}, edited by
610: \bibinfo{editor}{\bibfnamefont{L.}~\bibnamefont{Nagel}} \bibnamefont{and}
611: \bibinfo{editor}{\bibfnamefont{D.}~\bibnamefont{Stein}}
612: (\bibinfo{publisher}{Addison-Wesley}, \bibinfo{address}{New York},
613: \bibinfo{year}{1992}).
614:
615: \bibitem[{\citenamefont{Varotsos et~al.}(2001)\citenamefont{Varotsos, Sarlis,
616: and Skordas}}]{VAR01}
617: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
618: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
619: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~S.}
620: \bibnamefont{Skordas}}, \bibinfo{journal}{Practica of Athens Academy}
621: \textbf{\bibinfo{volume}{76}}, \bibinfo{pages}{294} (\bibinfo{year}{2001}).
622:
623: \bibitem[{\citenamefont{Varotsos
624: et~al.}(2002{\natexlab{a}})\citenamefont{Varotsos, Sarlis, and
625: Skordas}}]{VAR02}
626: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
627: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
628: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~S.}
629: \bibnamefont{Skordas}}, \bibinfo{journal}{Phys. Rev. E}
630: \textbf{\bibinfo{volume}{66}}, \bibinfo{pages}{011902}
631: (\bibinfo{year}{2002}{\natexlab{a}}).
632:
633: \bibitem[{\citenamefont{Varotsos
634: et~al.}(2002{\natexlab{b}})\citenamefont{Varotsos, Sarlis, and
635: Skordas}}]{VAR02A}
636: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Varotsos}},
637: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Sarlis}}, \bibnamefont{and}
638: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Skordas}},
639: \bibinfo{journal}{Acta Geophys. Pol.} \textbf{\bibinfo{volume}{50}},
640: \bibinfo{pages}{337} (\bibinfo{year}{2002}{\natexlab{b}}).
641:
642: \bibitem[{\citenamefont{Varotsos
643: et~al.}(2003{\natexlab{a}})\citenamefont{Varotsos, Sarlis, and
644: Skordas}}]{VAR03A}
645: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
646: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
647: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~S.}
648: \bibnamefont{Skordas}}, \bibinfo{journal}{Phys. Rev. E}
649: \textbf{\bibinfo{volume}{67}}, \bibinfo{pages}{021109}
650: (\bibinfo{year}{2003}{\natexlab{a}}).
651:
652: \bibitem[{\citenamefont{Varotsos
653: et~al.}(2003{\natexlab{b}})\citenamefont{Varotsos, Sarlis, and
654: Skordas}}]{VAR03B}
655: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
656: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
657: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~S.}
658: \bibnamefont{Skordas}}, \bibinfo{journal}{Phys. Rev. E}
659: \textbf{\bibinfo{volume}{68}}, \bibinfo{pages}{031106}
660: (\bibinfo{year}{2003}{\natexlab{b}}).
661:
662: \bibitem[{\citenamefont{Varotsos et~al.}(2004)\citenamefont{Varotsos, Sarlis,
663: Skordas, and Lazaridou}}]{VAR04}
664: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
665: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
666: \bibinfo{author}{\bibfnamefont{E.~S.} \bibnamefont{Skordas}},
667: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.~S.}
668: \bibnamefont{Lazaridou}}, \bibinfo{journal}{Phys. Rev. E}
669: \textbf{\bibinfo{volume}{70}}, \bibinfo{pages}{011106}
670: (\bibinfo{year}{2004}).
671:
672: \bibitem[{\citenamefont{Varotsos
673: et~al.}(2005{\natexlab{a}})\citenamefont{Varotsos, Sarlis, Skordas, and
674: Lazaridou}}]{VAR05}
675: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
676: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
677: \bibinfo{author}{\bibfnamefont{E.~S.} \bibnamefont{Skordas}},
678: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{M.~S.}
679: \bibnamefont{Lazaridou}}, \bibinfo{journal}{Phys. Rev. E}
680: \textbf{\bibinfo{volume}{71}}, \bibinfo{pages}{011110}
681: (\bibinfo{year}{2005}{\natexlab{a}}).
682:
683: \bibitem[{\citenamefont{Varotsos
684: et~al.}(2005{\natexlab{b}})\citenamefont{Varotsos, Sarlis, Tanaka, and
685: Skordas}}]{VAR05B}
686: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
687: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
688: \bibinfo{author}{\bibfnamefont{H.~K.} \bibnamefont{Tanaka}},
689: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~S.}
690: \bibnamefont{Skordas}}, \bibinfo{journal}{Phys. Rev. E}
691: \textbf{\bibinfo{volume}{71}}, \bibinfo{pages}{032102}
692: (\bibinfo{year}{2005}{\natexlab{b}}).
693:
694: \bibitem[{\citenamefont{Varotsos
695: et~al.}(2003{\natexlab{c}})\citenamefont{Varotsos, Sarlis, and
696: Skordas}}]{VAR03C}
697: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
698: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
699: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~S.}
700: \bibnamefont{Skordas}}, \bibinfo{journal}{Phys. Rev. Lett.}
701: \textbf{\bibinfo{volume}{91}}, \bibinfo{pages}{148501}
702: (\bibinfo{year}{2003}{\natexlab{c}}).
703:
704: \bibitem[{EPA()}]{EPAPS}
705: \eprint{See EPAPS Document No. [E-PLEEE8-72-058510] for additional information. This
706: document may be retrieved via the EPAPS homepage
707: (\url{http://www.aip.org/pubservs/epaps.html}) or from \url{ftp.aip.org} in
708: the directory /epaps/. See the EPAPS homepage for more information.}
709:
710: \bibitem[{\citenamefont{Varotsos}(2005)}]{VARBOOK}
711: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Varotsos}},
712: \emph{\bibinfo{title}{The Physics of Seismic Electric Signals}}
713: (\bibinfo{publisher}{TerraPub}, \bibinfo{address}{Tokyo},
714: \bibinfo{year}{2005}).
715:
716: \bibitem[{\citenamefont{Tanaka et~al.}(2004)\citenamefont{Tanaka, Varotsos,
717: Sarlis, and Skordas}}]{TAN04}
718: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Tanaka}},
719: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Varotsos}},
720: \bibinfo{author}{\bibfnamefont{N.~V.} \bibnamefont{Sarlis}},
721: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~S.}
722: \bibnamefont{Skordas}}, \bibinfo{journal}{Proc. Japan Acad. Ser. B}
723: \textbf{\bibinfo{volume}{80}}, \bibinfo{pages}{283} (\bibinfo{year}{2004}).
724:
725: \bibitem[{\citenamefont{Varotsos and Alexopoulos}(1986)}]{VAR86}
726: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Varotsos}} \bibnamefont{and}
727: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Alexopoulos}},
728: \emph{\bibinfo{title}{Thermodynamics of Point Defects and their Relation with
729: Bulk Properties}} (\bibinfo{publisher}{North Holland},
730: \bibinfo{address}{Amsterdam}, \bibinfo{year}{1986}).
731:
732: \bibitem[{\citenamefont{Schultka and Manousakis}(1994)}]{STR94}
733: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Schultka}} \bibnamefont{and}
734: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Manousakis}},
735: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{49}},
736: \bibinfo{pages}{12071} (\bibinfo{year}{1994}).
737:
738: \bibitem[{\citenamefont{Hegger et~al.}(1999)\citenamefont{Hegger, Kantz, and
739: Schreiber}}]{heg99}
740: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Hegger}},
741: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Kantz}}, \bibnamefont{and}
742: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Schreiber}},
743: \bibinfo{journal}{CHAOS} \textbf{\bibinfo{volume}{9}}, \bibinfo{pages}{413}
744: (\bibinfo{year}{1999}).
745:
746: \bibitem[{\citenamefont{Mercik and Weron}(1999)}]{mer99}
747: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Mercik}} \bibnamefont{and}
748: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Weron}},
749: \bibinfo{journal}{Phys. Rev. E} \textbf{\bibinfo{volume}{60}},
750: \bibinfo{pages}{7343} (\bibinfo{year}{1999}).
751:
752: \end{thebibliography}
753:
754: \begin{figure}
755: \includegraphics{Figure1}% Here is how to import EPS art
756: \caption{\label{fg1} How a series of seismic events in conventional time (a) is read in the natural time (b).
757: This example refers to the first 11 small earthquakes (cf. the month/date is marked on the horizontal axis in (a)) that occurred after the SES activity recorded on April 18,1995 and preceded the mainshock (M=6.6) of May 13,1995.}
758: \end{figure}
759:
760: \begin{figure}
761: \includegraphics{Fig2}% Here is how to import EPS art
762: \caption{\label{fg3} How the variance $\kappa_1$ evolves event by event during the following
763: period: from the detection\cite{VAR03C} of the SES activity on April 18, 1995 until
764: the occurrence of the M=6.6 mainshock(labelled 18) on May 13, 1995.
765: All the EQs used in the calculation are tabulated in
766: Ref.\cite{EPAPS}. (cf. the first 11 EQs -out of 18- are those depicted in Fig.\ref{fg1}(a)). }
767: \end{figure}
768:
769:
770: \begin{figure}
771: \includegraphics{Fig3}% Here is how to import EPS art
772: \caption{\label{fg2} (Color online) Validity of Eq.(\ref{fasma}) for SCEC and Japan.
773: We first determine (see also Appendix B) for SCEC(circles) and Japan(crosses), for each $\phi$-value, the value $\Pi_p(\phi)$
774: at which $P[\Pi(\phi)]$ maximizes. Two such examples are shown in (a) and (b) for
775: $\phi=0.05$ and $\phi=0.4$, respectively.
776: In (c), we plot the
777: resulting $\Pi_p(\phi)$ values versus $\phi$ for SCEC(circles) and Japan(crosses); the solid line corresponds to Eq.(\ref{fasma}). The percentage difference
778: between $\Pi_p(\phi)$ and $\Pi_{th} (\phi )$ (obtained from Eq.(\ref{fasma})) is shown in (d).}
779: \end{figure}
780:
781:
782:
783: \begin{figure}
784: \includegraphics{Figure4}% Here is how to import EPS art
785: \caption{\label{fg4} (Color online) Universality of the probability density function of
786: $\Pi(\phi )$ for EQs in the natural
787: time-domain. The log-linear plot of $\sigma P(X)$ versus $(X-\langle X \rangle )/\sigma$,
788: where $X$ stands for $\Pi(\phi )$ for $\phi \approx 0$.
789: Crosses and circles correspond to Japan (M$\geq$3.5) and SCEC
790: (M$\geq$2.0), respectively. The inset depicts the corresponding results for the
791: ``shuffled'' data
792: (black curve) and the original data (red crosses) in Japan. The same graph is obtained for
793: three different regions in Japan (see Fig.\ref{fg5}).}
794: \end{figure}
795:
796: %\begin{widetext}
797: \begin{figure}
798: \includegraphics{Figure5}% Here is how to import EPS art
799: \caption{\label{fg5} (Color online) The same as Fig. \ref{fg4}, but
800: for the regions A (red), B (green) and C (blue) in Japan (b). A
801: map of these regions is shown in (a).}
802: \end{figure}
803: %\end{widetext}
804:
805: \begin{figure}
806: \includegraphics{Figure6}% Here is how to import EPS art
807: \caption{\label{fg6}(Color online) The common feature of fluctuations in different
808: correlated systems. The log-linear plot
809: of $\sigma P(X)$ versus $(X-\langle X\rangle)/\sigma$ for
810: WWS(triangles), Japan(crosses) and SCEC(circles). The magnitude
811: threshold M$>$5.7 for WWS and Japan (while M$\geq$4.0 for SCEC)
812: was used see the text. Furthermore, the dotted curve shows the
813: results obtained for the 2D XY model (with\cite{STR94} inverse
814: Kosterlitz-Thouless transition temperature $K_{KT}\approx1.2$) ($X=\sqrt{M_x^2+M_y^2}$), $K=2.0$ for $L=10
815: (N=100)$ which has been shown\cite{BRA98} to describe the
816: experimental results for 3D turbulent flow. The results of the 2D Ising
817: model $K=0.4707$ (while $K_c\approx0.4407$), either for
818: $L=128$(dashed) or $L=256$(solid line), are also plotted.}
819: \end{figure}
820:
821: \begin{figure}
822: \includegraphics{Figure7}% Here is how to import EPS art
823: \caption{\label{fgPoi} (Color online) The linear-linear (a) and log-linear (b) plots
824: of $\sigma P(X)$ versus $(X-\langle X \rangle )/\sigma$, where $X$ stands for $\Pi(\phi)$ for $\phi \approx 0$. The results for Japan (M$\geq 3.5$) and SCEC (M$\geq 2.0$) are plotted along
825: with those deduced from a series of independent and identicaly distributed $(M_0)_k$ sampled
826: from a Poisson distribution with mean value 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100 and 200:
827: from the bottom to the top in the maxima appearing in (a), respectively, and
828: from the upper to the lower left branch in (b), respectively.}
829: \end{figure}
830:
831:
832: \begin{figure}
833: \includegraphics{Figure8}% Here is how to import EPS art
834: \caption{\label{fgGuRi} (Color online) The linear-linear (a) and log-linear (b) plots
835: of $\sigma P(X)$ versus $(X-\langle X \rangle )/\sigma$, where $X$ stands for $\Pi(\phi)$ for $\phi \approx 0$. The results for Japan ($M\geq 3.5$) and SCEC ($M\geq 2.0$) are plotted along
836: with those deduced from shuffled artificially generated EQ data obeying the Gutenberg-Richter law for
837: various values of the exponent $b=$0.5, 0.7, 0.9, 1.0, 1.1, 1.2, 1.3, 1.5 and 2.0 from the lower to the
838: upper curve at the value $(X-\langle X \rangle )/\sigma \approx -0.5$ in (a), and from the upper to the lower curve at the value $(X-\langle X \rangle )/\sigma \approx -4$ in (b).}
839: \end{figure}
840:
841: \begin{figure}
842: \includegraphics{Figure9}% Here is how to import EPS art
843: \caption{\label{fg7} (Color online) Two examples of the procedure used in the determination of $\Pi_p(\phi)$.
844: (a) For SCEC at $\phi=0.03$, (b) For Japan at $\phi=0.4$. The (red) curves in each case, show the cubic polynomial fit which was used in the range $[a,b]$ around the maximum. The (red) arrow indicates the position of $\Pi_p(\phi)$.}
845: \end{figure}
846:
847:
848:
849: \end{document}
850: