1: \documentclass[aps,prl,showpacs,preprintnumbers,amsmath,assym,
2: twocolumn,superscriptaddress]{revtex4}
3: \usepackage{amsmath}
4:
5: \usepackage{graphicx}
6: \usepackage{epsfig}
7: \usepackage{bm}
8:
9: \begin{document}
10:
11: \begin{abstract}
12: We elucidate a long-standing puzzle about the non-equilibrium
13: universality classes describing self-organized criticality in sandpile
14: models. We show that depinning transitions of linear interfaces in
15: random media and absorbing phase transitions (with a conserved
16: non-diffusive field) are two equivalent languages to describe sandpile
17: criticality. This is so despite the fact that local roughening
18: properties can be radically different in the two pictures, as
19: explained here. Experimental implications of our work as well as
20: promising paths for future theoretical investigations are also
21: discussed.
22: \end{abstract}
23:
24: \title{
25: Absorbing states and elastic interfaces in random media: \\ two
26: equivalent descriptions of self-organized criticality}
27:
28: \author{Juan A. Bonachela}
29: \affiliation{
30: %Departamento de Electromagnetismo y F{\'\i}sica de la Materia and
31: Instituto de F{\'\i}sica Te{\'o}rica y Computacional Carlos I,
32: Facultad de Ciencias, Universidad de Granada, 18071 Granada, Spain}
33: \author{Hugues Chat\'e}
34: \affiliation{CEA -- Service de Physique de l'\'Etat Condens\'e,~CEN
35: ~Saclay,~91191~Gif-sur-Yvette,~France}
36: \author{Ivan Dornic}
37: \affiliation{
38: %Departamento de Electromagnetismo y F{\'\i}sica de la Materia and
39: Instituto de F{\'\i}sica Te{\'o}rica y Computacional Carlos I,
40: Facultad de Ciencias, Universidad de Granada, 18071 Granada, Spain}
41: \affiliation{CEA -- Service de Physique de l'\'Etat Condens\'e,~CEN
42: ~Saclay,~91191~Gif-sur-Yvette,~France}
43: \author{Miguel A. Mu\~noz}
44: \affiliation{
45: %Departamento de Electromagnetismo y F{\'\i}sica de la Materia and
46: Instituto de F{\'\i}sica Te{\'o}rica y Computacional Carlos I,
47: Facultad de Ciencias, Universidad de Granada, 18071 Granada, Spain}
48:
49: \date{\today}
50: \pacs{05.50.+q,02.50.-r,64.60.Ht,05.70.Ln}
51: \maketitle
52:
53: The concept of self-organized criticality (SOC) has been proposed to
54: account for the emergence of scale invariance in Nature
55: \cite{SOC}. Its main tenet is that in the presence of slow driving and
56: fast dissipation, acting at infinitely-separated time scales, many
57: systems self-organize (without any explicit tuning of parameters) to a
58: critical state \cite{GG}. Some archetypical examples of SOC are
59: provided by sandpile toy models, in which grains are slowly added,
60: locally redistributed on a fast timescale whenever an instability
61: threshold is overcome (generating avalanches of toppling events), and
62: finally dissipated at the open boundaries. Upon iteration, this
63: process leads to a critical steady state.
64:
65: SOC can be related to standard (non-equilibrium) critical phenomena by
66: defining the ``fixed energy ensemble'' \cite{fes-bak,FES} in which driving and
67: dissipation are switched off, so that the number of grains (or ``energy'') is
68: conserved. Using this quantity as a control parameter, a standard (i.e. non
69: self-organized) phase transition is observed: for large energy densities,
70: there is a finite density of active (toppling) sites, whereas the system ends
71: in a frozen (stable) state at low densities. It has been shown that the
72: critical point separating these two regimes occurs at the value of the energy
73: density at which the system self-organizes when subjected to slow-driving and
74: boundary dissipation \cite{FES,Oslo-AS}. Subsequent debate has attempted to
75: elucidate which non-equilibrium universality class stochastic sandpiles belong
76: to. As detailed below, two alternative solutions have been proposed.
77:
78: Sandpiles were first related to interfaces in random media
79: \cite{lim-map}. In this language, the interface height, $h(x,t)$ is
80: the number of times a given site $x$ has toppled up to time $t$, and
81: frozen states correspond to pinned interfaces. The resulting
82: pinning-depinning transition was argued to fall in the quenched
83: Edwards-Wilkinson or linear interface model (LIM) class, described by
84: \cite{LIM-renorm,Lesch}:
85: \begin{equation}
86: \partial_t h(x,t) = \nabla^{2}h(x,t) + F + \eta(x,h)
87: \label{LIM}
88: \end{equation}
89: where $F$ is a force and $\eta(x,h)$ is a quenched white noise. This
90: correspondence was recently proven exact between one particular sandpile model
91: \cite{Oslo} and one member of the LIM class \cite{Pruessner} but, in general,
92: it is only approximate, as some noise-correlations need to be
93: neglected to establish a full correspondence with Eq.(\ref{LIM}).
94:
95: Alternatively, sandpile models have been rationalized as systems exhibiting an
96: absorbing-state phase transition \cite{FES}. Indeed, in the fixed energy
97: ensemble, a stable configuration is one of the infinitely-many absorbing
98: states in which the system can be trapped forever, whereas activity never
99: ceases above the critical point. The corresponding universality class is
100: {\it not} the prominent directed percolation (DP) class,
101: %\cite{Reviews_AS},
102: but is characterized by the coupling of a DP-like activity field to a
103: conserved, non-diffusive, auxiliary field (the ``energy'')
104: \cite{FES,Romu}. Often called C-DP (or also Manna \cite{Manna}) class,
105: it is characterized by the following set of Langevin equations:
106: \begin{equation}
107: \label{CDP}
108: \begin{array}{rcl}
109: \partial_t \rho & = & a \rho - b \rho^2 + D_{\!\rho} \nabla^2 \rho
110: + \omega \rho\, \phi + \sigma \sqrt{\rho}\, \eta(x,t), \\
111: \partial_t \phi & = & D_{\!\phi} \nabla^2 \rho \mbox{ ,}
112: \end{array}
113: \end{equation}
114: \noindent
115: where $\rho$ is the activity field, $\phi$ the background energy field, and
116: $\eta(x,t)$ a Gaussian white noise \cite{FES,Romu}.
117:
118: The validity of both of these alternative pictures has been
119: (partially) backed by numerical measurements of critical exponents
120: but, in general, they have not been proven to be correct so far.
121: But, if both of the pictures are right, a remarkable consequence
122: follows: depinning transition of LIM-class interfaces in random media
123: and the C-DP class absorbing phase transition should be equivalent,
124: even if they look rather different (for instance, one involves
125: quenched disorder and the other does not). Here, we explore this issue and
126: the more general question of whether
127: any depinning interface universality class has an
128: equivalent absorbing phase transition class.
129:
130: Only a few works have approached the connections between these two
131: pictures. In \cite{DM}, interfaces were constructed from a DP class
132: model, using the cumulated local activity as the interface
133: height. Anomalously-rough interfaces, characterized by a positive
134: local-slope exponent $\kappa$ defined by $\langle (\nabla h)^2 \rangle
135: \sim t^{2 \kappa}$ \cite{kappa}, were found at criticality. These
136: anomalous interfaces are not related to any known interface class.
137: Focusing on SOC sandpiles, Alava and Mu\~noz \cite{MikkoMA} argued
138: heuristically that the LIM and C-DP classes could be identified with
139: each other (using also $h(x,t)=\int_0^{t}\rho(x,s)ds$ to relate
140: Eq.(\ref{LIM}) to Eq.(\ref{CDP})) although a one-to-one mapping could
141: not be rigorously established. However, this conclusion was later
142: challenged by Kockelkoren and Chat\'e (KC) \cite{KC} who found that
143: the $\kappa$ exponent takes completely different values for LIM
144: interfaces and for interfaces constructed from models in the C-DP
145: class. In particular, the constructed interfaces are anomalously-rough
146: below the upper critical dimension, i.e. for space dimensions $d<4$,
147: while LIM interfaces have $\kappa \leq 0$ (not anomalous) for
148: $d\ge2$. In $d=1$, both types of interfaces are anomalously rough, but
149: in a manifestly distinct manner, i.e. different values of $\kappa$
150: (Fig.~\ref{f1}). This led KC to conclude that LIM and C-DP classes
151: {\it cannot} be equivalent, even if all the other recorded
152: ``standard'' critical exponents (as $\theta$ and $z$, see definitions
153: below and table I, and others) take ``almost indistinguishable
154: values'' in these two problems, (which could, in principle, be
155: attributed to a numerical coincidence
156: \cite{KC}.)
157: \begin{table}
158: \caption{\label{t1} Some of the measured critical exponents
159: of the C-DP/LIM class vs space dimension $d$
160: %/(values compounded from)
161: \protect{\cite{FES,Lesch,KC,Dornic,Lubeck}}.
162: The A- and B-scaling values for $\kappa$ are from our own present
163: simulations.}
164: \begin{ruledtabular}
165: \begin{tabular}{clllll}
166: $d$ & $\theta$ & $z$ & $\kappa_A$ & $\kappa_B$ \\ \hline
167: 1 & 0.13(1) & 1.42(2) & 0.17(1) & 0.43(1) \\
168: 2 & 0.51(2) & 1.55(3) & $0^{-}$ & 0.25(2) \\
169: 3 & 0.77(3) & 1.78(5) & $<0$ & 0.12(3) \\
170: \end{tabular}
171: \end{ruledtabular}
172: \end{table}
173:
174: While the discrepancy in values of $\kappa$ is unquestionable,
175: simulating directly Eqs.(\ref{CDP}) (using the method in
176: \cite{Dornic}) we find all the other C-DP exponents to be
177: indistinguishable from their counterparts in the LIM class
178: (Table~\ref{t1}) as well as from the corresponding values in
179: stochastic sandpiles.
180:
181: In this Letter, we show that depinning transitions of LIM interfaces
182: and C-DP absorbing phase transitions are indeed two equivalent
183: descriptions of SOC sandpiles in spite of the discrepancies in
184: $\kappa$-values, which we explain. We show, using a combination of
185: numerical results and scaling arguments, that there is a unique
186: universality class and that differences in $\kappa$-values stem from
187: diverging {\it local} fluctuations, inherent to the absorbing state
188: picture, which do not affect other long-distance properties.
189:
190:
191: Let us start by clarifying the origin of the two possible values of
192: $\kappa$ and the scaling laws they obey by using simple scaling
193: arguments. First, since $h = \int {\rm d}t \rho(t)$ its scaling
194: dimension is $[h] \sim t^{1-\theta}$, where $\theta$ is the density
195: (or interface-velocity, recalling that $\partial_t h = \rho$) critical
196: time-decay exponent: $\langle \rho(t) \rangle
197: \sim t^{-\theta}$. This leads to $[\nabla h] \sim t^{-1/z +1 - \theta}$,
198: where $[\nabla^{-1}]\sim t^{1/z}$ defines the dynamical critical exponent $z$,
199: and therefore
200: \begin{equation}
201: \kappa_{A} = 1-\theta-\frac{1}{z} \;\;\;\;
202: {\rm(``A \;scaling ``\; from\; now\; on)}.
203: \label{A}
204: \end{equation}
205: For $d=1$, plugging the values $\theta \approx 0.13$ and $z \approx
206: 1.42$ of the LIM or C-DP class \cite{Lubeck,FES,Lesch,KC} into this
207: expression leads to $\kappa \approx 0.17$ which is indeed the value
208: measured for LIM class interfaces \cite{Lesch}. In higher dimensions,
209: this scaling law yields zero (with possible logarithmic corrections in
210: $d=2$) or negative $\kappa$ values, i.e. no anomalous scaling, as
211: indeed observed in simulations \cite{KC}.
212:
213: On the other hand, assuming that interface heights at adjacent sites are
214: asymptotically uncorrelated (this will be justified after) we have
215: $ \nabla h \sim \sqrt{h} \sim t^{(1-\theta)/2}$ and hence
216: \begin{equation} \kappa_{B} =\frac{1 - \theta}{2} \;\;\;\; {\rm(``B\;scaling'')}\;.
217: \label{B}
218: \end{equation}
219: KC observed that B-scaling is verified by many interfaces constructed
220: from microscopic models at absorbing phase transitions
221: \cite{KC}. Our own simulations (not shown) extend this result
222: to different sandpile models (simulated in the fixed energy ensemble):
223: the constructed interfaces of the Manna \cite{Manna}, Oslo
224: \cite{Oslo}, and Mohanty-Dhar \cite{MD} models show B-scaling at
225: criticality.
226: \begin{figure}
227: \includegraphics[width=70mm]{SNAPSHOTS_low.eps}
228: \caption{Typical one-dimensional interfaces for a system of $2^{15}$ sites
229: at $t=10^5$ from flat initial conditions. Top: Leschhorn automaton
230: (LIM class, $\kappa\approx0.17(1)$). Bottom: interface constructed
231: from the Manna sandpile \protect\cite{Manna} (C-DP class,
232: $\kappa\approx0.43(1)$).}
233: \label{f1}
234: \end{figure}
235:
236: To shed some light on the physical reason for the existence of two
237: different $\kappa$ values, let us consider the Leschhorn automaton, a
238: LIM-class model showing A-scaling. It is a discretization of
239: Eq.(\ref{LIM}): an integer-valued height advances at each site $x$
240: following:
241: \begin{equation}
242: h(x) \rightarrow h(x) +1 \;\;{\rm iff}\;\; \nabla^2 h +F+ \eta(x,h(x)) >0,
243: \label{lesch}
244: \end{equation}
245: where the (discretized) Laplacian is computed using the
246: nearest-neighbors of $x$, and $\eta=\pm 1$ with respective
247: probabilities $p$ and $1-p$. Consider now the Manna sandpile, a C-DP
248: class model whose local rule is: if two or more grains are present at
249: a given site, distribute two of them randomly to the nearest-neighbors
250: \cite{Manna,FES}. In this case, the interface $h(x)$ encoding the number
251: of times a site has toppled since $t=0$ shows B-scaling, as said
252: before. It can be can be expressed in terms of $z_{\rm in}(x)$ and
253: $z_{\rm out}(x)$, the cumulated number of particles respectively
254: received from and given to the nearest-neighbors. For example, in
255: $d=1$, $z_{\rm out}(x)= 2 h(x)$ while $z_{\rm in}$ can be expressed as
256: the sum of a mean flux $h(x+1) + h(x-1)$ plus a fluctuating part,
257: $\xi(x,h(x))$, indicating stochastic deviations from this mean
258: \cite{lim-map,MikkoMA,KC}. The toppling condition can be written as
259: $z_0(x)+z_{\rm in}(x) - z_{\rm out}(x) \geq 2$ (where $z_0(x)$ is the
260: initial number of grains) and thus ,expressed in terms of the
261: following advancement rule:
262: \begin{equation}
263: h(x) \rightarrow h(x)+1 \;\;{\rm iff}\;\; \nabla^2 h - 1 + z_0(x) +
264: \xi(x,h) > 0 \;.
265: \label{fictitious}
266: \end{equation}
267: This is very similar to Eq.(\ref{lesch}) but, as noticed in
268: \cite{KC}, there is a crucial difference: whereas in Eq.~(\ref{lesch})
269: $\eta(x,h(x))$ is a {\it bounded}, dichotomous, delta-correlated
270: noise, the noise term $\xi(x,h(x))$ in Eq.~(\ref{fictitious}) is a sum
271: of random variables (a unit is added or subtracted for each toppling)
272: whose amplitude, by virtue of the central limit theorem, behaves like
273: the square root of the average of $h(x+1)+h(x-1)$, and is therefore
274: {\it diverging} in time: $\langle \xi(h)^2 \rangle \sim
275: t^{1-\theta}$. In turn, this divergence has to be compensated by the
276: fluctuations of the Laplacian term in Eq.~(\ref{fictitious}) (since
277: the term $z_0(x)$ representing the initial condition should be
278: irrelevant in the long-time limit and is anyhow bounded). This is at
279: the origin of the strong fluctuations present in the constructed
280: interface (B scaling) but absent in the LIM class (see Fig.\ref{f1}).
281: \begin{figure}
282: \includegraphics[width=86mm]{KAPPA.eps}
283: \caption{Time series of interface squared-gradient and
284: activity-density (or velocity) at criticality.
285: (a) Eqs.(\ref{CDP}) using
286: $h$, (A-scaling, $\kappa \approx 0.17(1)$) or $\bar{h}$ (B-scaling,
287: $\kappa \approx 0.43(1)$ ). Parameters: $L=2^{15}$,
288: $b=w=\sigma^2/2=1$, $D=D_{E}=0.25$, time-mesh $0.1$; critical point
289: $a_{c}=0.86452(5)$. (b) Leschhorn automaton with $L=2^{15}$,
290: $F=0$, at the critical point $p=0.80078(5)$, and using $h$ as in
291: Eq.(\ref{LIM}) (A-scaling, $\kappa \approx 0.17(1)$) or $\tilde{h}$
292: as in Eq.(\ref{mod-LIM}) (B-scaling, $\kappa \approx 0.43(1)$) with
293: $p_c=0.76935(5)$.}
294: \label{f2}
295: \end{figure}
296: At this point, one clearly appreciates the qualitative difference between LIM
297: and C-DP-constructed interfaces, which occurs despite of the fact that both
298: classes share numerically-indistinguishable (standard) critical exponents, as
299: said before. We now show that this is not the end of the story, and
300: that we can construct A-scaling interfaces from C-DP class models, as well as
301: modify LIM-class models to obtain B-scaling.
302:
303:
304: Our first evidence showing that A-scaling and B-scaling can both be
305: compatible with a unique universality class was provided by numerical
306: integrations of Eqs.(\ref{CDP}). Constructing an interface, as before,
307: via $h(x,t) = \int{\rm d}s \rho(x,s)$ where $\rho$ is the continuous
308: activity field, we obtain clear A-scaling (Fig.~\ref{f2}), in contrast
309: with the B-scaling heretofore always observed with microscopic
310: models. Next, mimicking
311: microscopic models in which the interface advances by one {\it unit}
312: whenever a site is active, we constructed a different interface for
313: Eqs.(\ref{CDP}) through $\bar{h}(x,t) = \int_0^t {\rm
314: d}s\Theta(\rho(x,s))$ where $\Theta$ is the Heaviside step function;
315: this new interface advances by one unit whenever there is some
316: non-zero activity, regardless of its magnitude. Strikingly, B-scaling
317: is then observed (Fig.~\ref{f2}). Thus both A- and B-scaling
318: interfaces have been constructed at the {\it same} absorbing phase
319: transition point: they correspond to slightly different {\it
320: observables}. We have reached a similar conclusion for the Oslo
321: sandpile model by taking advantage of a recent result by Pruessner
322: \cite{Pruessner} who constructed an exact mapping between this particular
323: sandpile and the Leschhorn automaton. The local rule for $d=1$ is as
324: follows: distribute one grain to each nearest-neighbor whenever the
325: local height threshold is passed, this local threshold being randomly
326: reset to be $1$ or $2$ grains after each toppling. To achieve an exact
327: mapping, Pruessner showed that it is crucial to use the more
328: symmetrical $h^\dag(x)=h(x+1)+h(x-1)$, i.e. the accumulated number of
329: times a given site has been charged by its neighbors rather than the
330: accumulated activity at the site itself (the definition of
331: $h$). Indeed, using $h^\dag(x)$ for the Oslo model eliminates the
332: diverging noise in Eq.(\ref{fictitious}) and yields A-scaling in
333: numerical simulations, whereas using $h(x)$ we observe B-scaling
334: (results not shown). We have been able to extend easily this procedure
335: to other sandpiles as the Mohanty-Dhar one \cite{MD}. For other
336: sandpiles with less symmetric redistribution rules as, for instance,
337: the Manna one \cite{Manna} this can be much more complicated. In this
338: last, the $2$ toppling grains at any site can go to the same
339: neighbor. This introduces an extra noise that needs to be subtracted
340: by an appropriate (and intricate) definition of the height variable
341: (different from $h$ and $h^\dag$) to get rid of intrinsic local
342: fluctuations and disentangle the hidden A-scaling. These results show
343: that a unique universality class is compatible with different values
344: of $\kappa$, depending on microscopic details
345: %such as the local redistribution rule
346: and/or the definition of the height variable; {\it different $\kappa$
347: exponents correspond to different, though very similar,
348: observables}. Note also that the same definition of the interface can
349: lead to the two different types of scaling depending on the rules of
350: sandpile under study.
351:
352:
353: To close the loop, we now show that for standard interface
354: depinning transitions it is also possible to generate two different
355: $\kappa$ values without affecting other exponents. Let us define
356: %\begin{equation}
357: $\partial_{t}\tilde{h}(x,t)=\partial_{t}h(x,t) (1+\sigma(x,{h})),$
358: %\label{H_punto}
359: %\end{equation}
360: where the quenched noise $\sigma$ is $0$ or $1$ with probabilities $p$
361: and $1-p$, respectively, so every time the original interface
362: advances, a noise variable is added to $\tilde{h}$. Both interfaces
363: are related by $\tilde{h}(x,t)=h(x,t)+\tilde{\sigma}(x,\tilde{h})$
364: where $\tilde{\sigma}(x,\tilde{h})$ is an accumulated noise summing up
365: all values of $\sigma(x,h)$ at $x$ up to height $h$. In terms of
366: $\tilde{h}(x,t)$, Eq.(\ref{LIM}) becomes
367: \begin{equation}
368: %\begin{array}{ll}
369: \partial_{t}\tilde{h}=[\nabla^{2}(\tilde{h}(x,t)-
370: \tilde{\sigma}(x,\tilde{h}))+\eta(x,\tilde{h})]
371: \times [1+{\sigma}(x,\tilde{h})].
372: %\end{array}
373: \label{mod-LIM}
374: \end{equation}
375: Simulating Eq.(\ref{mod-LIM})
376: %({\it \`a la} Leschhorn
377: we observe B-scaling for the $\tilde{h}$-interface, while removing the
378: $\tilde{\sigma}$-noise we readily recover A-scaling for the
379: $h$-interface (Fig.~\ref{f2}), while all the other exponents coincide
380: for both interfaces. Na{\"i}ve power-counting for Eq.(\ref{mod-LIM})
381: shows that $\nabla^2 \tilde{\sigma}(x,\tilde{h})$ is an irrelevant
382: higher-order noise, and so is the term $\sigma(x,\tilde{h})$ added to
383: $1$ \cite{LIM-renorm}. Hence, upon coarse-graining, Eq.(\ref{mod-LIM})
384: flows towards the standard LIM renormalization group fixed point
385: \cite{LIM-renorm}, justifying that all {\it universal} critical
386: exponents should coincide in Eq.(\ref{LIM}) and Eq.(\ref{mod-LIM}) in
387: accordance with our numerical results. The inclusion of the extra
388: noise (a higher-order correction to scaling) is thus able to alter the
389: value of $\kappa$, intensifying anomalous behavior, but not standard
390: long-distance critical exponents, which are controlled by a unique
391: renormalization group fixed point. The two different values of
392: $\kappa$ correspond to two different height variables, $h$ and
393: $\tilde{h}$, differing by a diverging noise. A full understanding,
394: within the renormalization group perspective, of how local anomalous
395: roughening properties are affected by an otherwise irrelevant noise
396: remains a challenging task.
397:
398: The fact that a given universality class can be compatible with
399: different types of local roughening (amenable to experimental
400: analysis) is also of interest in general studies of fluctuating
401: interfaces \cite{kappa}. Indeed, one can show that $\chi$ and
402: $\chi_{loc}$ being respectively the global and local saturation
403: roughness exponents are related by $\chi = \chi_{loc} + z \kappa$
404: \cite{kappa}, which we have numerically verified for all models
405: studied here. Given that our results indicate that local roughening
406: properties, as encoded by $\kappa$ (or $\chi_{loc}$), can adopt
407: different values, depending on the presence or absence of local
408: fluctuations, while $\chi$, as all other non-local exponents, is
409: universal, it is intriguing that experiments on the propagation of
410: fracture cracks in wood \cite{fracture_exp} seem to lead to the
411: opposite conclusion. See also the nice recent experiments measuring
412: the exponents reported here for self-organized superconductors
413: \cite{Wijn}. More experimental
414: work along these lines would be most welcome, since experimental
415: realizations of absorbing phase transitions (even for the paradigmatic
416: DP class) are barely existing
417: \cite{Haye_DPexp}.
418:
419: On the theoretical side, since the C-DP class and the LIM class are
420: two faces of the same problem, both share the same upper critical
421: dimension $d_{\rm u}=4$. This result is in contradiction with the
422: perturbative approach for Eq.(\ref{CDP}) in \cite{FvW}. In fact,
423: field-theoretical treatments of the LIM class demand a functional
424: renormalization group calculation
425: %beyond the so-called Larkin length
426: \cite{LIM-renorm}, and one finds that the correlator of the quenched
427: noise develops a cusp in the $h$-variable. Given our results, it would
428: be very interesting to sort out the analogue of all this in the
429: absorbing phase transition picture, as well as attempting a
430: non-perturbative renormalization group approach for such a case.
431: %This
432: %correspondence could lead to a very interesting crossed-fertilization
433: %between these two research fields.
434:
435: In summary, depinning transitions of linear interfaces in random media
436: and absorbing phase transitions in the C-DP class are two equivalent
437: descriptions of sandpile self-organized critical points. This
438: clarifies the issue of universality in stochastic sandpiles and the
439: connection of SOC to standard non-equilibrium phase transitions and
440: opens the door to new and exciting research lines.
441:
442: %JAB and MAM acknowledge support from the MEC-FEDER project
443: %FIS2005-00791.
444: % and from Junta de Andaluc{\'\i}a as group FQM-165.
445: %ID warmly thanks the Instituto Carlos I
446: %for the kind hospitality and support.
447:
448:
449: \begin{thebibliography}{99}
450:
451: \bibitem{SOC} P. Bak, C. Tang and K. Wiesenfeld, Phys. Rev. Lett. {\bf 59},
452: 381 (1987). H. J. Jensen, {\it Self Organized Criticality}, Cambridge Univ.
453: Press, (1998). D. Dhar, Physica A {\bf 263}, 4 (1999).
454:
455: \bibitem{GG} G. Grinstein, in
456: %{\it Scale Invariance, Interfaces and Non-equilibrium Dynamics},
457: {\it NATO Advanced Study Institute, Series B: Physics}, vol. {\bf
458: 344}, A. McKane et al., Eds. (Plenum, New York, 1995).
459:
460: \bibitem{fes-bak} C. Tang and P. Bak, Phys. Rev. Lett. {\bf 60}, 2347 (1988).
461:
462: \bibitem{FES}
463: A. Vespignani {\it et al.}, %R. Dickman, M.A. Mu\~noz, S. Zapperi,
464: Phys. Rev. Lett. {\bf 81}, 5676 (1998); Phys. Rev. E {\bf 62}, 4564
465: (2000). R. Dickman {\it et al.}, % M. A. Mu{\~n}oz, A. Vespignani,
466: %and S. Zapperi,
467: % {\it Paths to Self-Organized-Criticality}
468: Braz. J. of Physics {\bf 30}, 27 (2000).
469: %M. A. Mu\~noz {\it et. al} %R. Dickman, R. Pastor-Satorras, A. Vespignani, and S. Zapperi,
470: % {\it Sandpiles and absorbing state
471: % phase transitions: recent results and open problems},
472: % in ''Modeling Complex Systems",
473: % Ed. J. and P. L. Garrido. AIP Conference Proceedings, vol. {\bf
474: % 574}, 102 (2001).
475: %Cond-mat/0011447.
476: %\bibitem{dvz} R. Dickman, {\it et. al.},
477: % A. Vespignani, and S. Zapperi,
478: % Phys. Rev. E {\bf 57}, 5095 (1998).
479:
480:
481: \bibitem{Oslo-AS}
482: K.~Christensen {\it et al.}, %Moloney NR, Peters O, and G. Pruessner,
483: %Avalanche behavior in an absorbing state Oslo model
484: Phys. Rev. E {\bf 70}, 067101 (2004).
485:
486:
487:
488: \bibitem{lim-map}
489: O. Narayan and A. A. Middleton, Phys. Rev. B {\bf 49}, 244
490: (1994). M. Paczuski and S. Boettcher, Phys. Rev. Lett. {\bf 77}, 111
491: (1996). M. J. Alava and K. B. Lauritsen, Europhys. Lett. {\bf 53},
492: 563 (2001).
493: % {\it ibid}, cond-mat/9903349.
494:
495: %\bibitem{lim-map-review}
496: %Self-organized criticality as a phase transition
497: %M. J. Alava, in {\it Advances in Condensed Matter and Statistical
498: %Physics}, Ed. E. Korutcheva and R. Cuerno, p. 69, Nova Science
499: %Publisher, 2004.
500:
501:
502:
503: \bibitem{LIM-renorm}
504: % RENORMALIZATION (1 loop):
505: T. Nattermann {\it et al.}, %S. Stepanow, L-H. Tang, and
506: H. Leschhorn, J. Phys. II {\bf 2}, 1483 (1992). H. Leschhorn {\it et
507: al.}, %T. Nattermann, S. Stepanow, and L-H. Tang, Ann. Phys. {\bf 7},
508: 1 (1997). O. Narayan and D. S. Fisher, Phys. Rev. B {\bf 48}, 7030
509: (1993).
510: %P.~Chauve, P.~Le~Doussal, and K.~.J.~Wiese, Phys. Rev. Lett. {\bf 86}, 1785 (2001).
511: %P.~Le~Doussal, K.~.J.~Wiese, and P.~Chauve, Phys. Rev. B. {\bf 66}, 174201 (2002).
512:
513:
514: \bibitem{Lesch} H. Leschhorn, Physica A {\bf 195}, 324 (1993).
515: % In higher dimensions:
516: %M. Jost and K.D. Usadel, Physica A {\bf 255}, 15 (1998)
517:
518: \bibitem{Oslo}
519: K. Christensen {\it et al.}, %A. Corral, V. Frette, J. Feder, and T. Jossang,
520: Phys. Rev. Lett. {\bf 77}, 107 (1996).
521: %V. Frette {\it et al.}, %K. Christensen, A. Malthe-Sorenssen, J. Feder, T. Jossang, and P. Meakin,
522: %Nature {\bf 379}, 49 (1996).
523:
524:
525: \bibitem{Pruessner} G. Pruessner, Phys. Rev. E {\bf 67}, 030301(R) (2003).
526:
527:
528: %\bibitem{Reviews_AS} H. Hinrichsen, Adv. Phys. {\bf 49}, 1 (2000).
529: %G. \'Odor, {\it Rev. Mod. Physics}, {\bf 76}, 663 (2004).
530:
531:
532:
533: \bibitem{Romu} M. Rossi, R. Pastor-Satorras, and A. Vespignani,
534: Phys. Rev. Lett. {\bf 85}, 1803 (2000).
535: %R. Pastor-Satorras and
536: % A. Vespignani, Phys. Rev. E {\bf 62}, R5875 (2000).
537:
538: \bibitem{Manna}
539: S. S. Manna, J. Phys. A {\bf 24}, L363 (1991).
540:
541:
542: \bibitem{DM}
543: R. Dickman and M. A. Mu\~noz,
544: % {\it Interface Scaling in the Contact Process},
545: Phys. Rev. E {\bf 62}, 7632 (2000).
546:
547: \bibitem{kappa}
548: %S. Das Sarma et al. Phys. Rev. E {\bf 49}, 122 (1994).
549: J. Krug, Phys. Rev. Lett. {\bf 72}, 2907 (1994)
550: %J. M. L\'opez, M. A. Rodriguez, and R. Cuerno, %''Superroughening versus intrinsic anomalous scaling of surfaces'',
551: %Phys. Rev. E {\bf 56}, 3993 (1997).
552: J.~M.~L\'opez, %''Scaling approach to calculate critical exponents in anomalous surface roughening''
553: Phys. Rev. Lett. {\bf 83}, 4594 (1999).
554: %J.~J.~Ramasco, J.~M.~L\'opez, and M.~A.~Rodriguez, %''Generic dynamic scaling in kinetic roughening'',
555: %Phys. Rev. Lett. {\bf 84}, 2199 (2000).
556:
557:
558: \bibitem{MikkoMA} M. Alava and M. A. Mu\~noz, Phys. Rev. E {\bf 65}, 026145 (2002).
559:
560: \bibitem{KC} J. Kockelkoren and H. Chat\'e, cond-mat/0306039, (2003).
561: %J. Kockelkoren, PhD Dissertation, Universit\'e Denis Diderot, Paris (2002).
562:
563: \bibitem{Dornic}
564: I. Dornic, H. Chat\'e, and M.~A. Mu\~noz,
565: Phys. Rev. Lett. {\bf 94}, 100601 (2005).
566:
567: \bibitem{Lubeck} S. L\"ubeck,
568: Int. J. of Mod. Phys. B{\bf 18}, 3977 (2004).
569: % S. L\"ubeck and A. Hutch, J. Phys. A. {\bf 35}, 4853 (2002).
570: %S. L\"ubeck and A. Hutch, Journal of Physics A {\bf 34}, L577 (2001).
571: %S. L\"ubeck, Phys. Rev. E {\bf 64}, 016123 (2001); Phys. Rev. E {\bf
572: %66}, 046114 (2002). S. L\"ubeck and P. C. Heger, Phys. Rev. Lett. {\bf
573: %90}, 230601 (2003); and Condmat/0309165.
574:
575:
576: \bibitem{MD}
577: P.K. Mohanty and D. Dhar, Phys. Rev. Lett. {\bf 89}, 104303
578: (2002). J. A. Bonachela {\it et al.},
579: % J.J. Ramasco, H. Chat\'e, I. Dornic, M. A. Mu\~noz,
580: Phys. Rev. E. {\bf 74}, 050102(R) (2006).
581:
582: \bibitem{fracture_exp}
583: S. Morel, {\it et al.}
584: % J. Schmittbuhl, J. M. L\'opez, and G. Valentin,
585: Phys. Rev. E {\bf 58}, 6999 (1998). J. M. L\'opez and J. Schmittbuhl,
586: Phys. Rev. E {\bf 57}, 6405 (1998).
587:
588:
589: \bibitem{Wijn} R. J. Wijngaarden, {\it et al.},
590: % M. S. Welling, C. M. Aegerter, and M. Menghini,
591: Eur. Phys. J. B {\bf 50}, 117 (2006).
592:
593: \bibitem{Haye_DPexp}
594: H.~Hinrichsen, %''On possible experimental realizations of directed
595: %percolation'',
596: Braz.~J.~Phys. {\bf 30}, 69 (2000).
597:
598: \bibitem{FvW}
599: F.~van~Wijland, %''Universality Class of Nonequilibrium Phase
600: %Transitions with Infinitely Many Absorbing States'',
601: Phys. Rev. Lett. {\bf 89}, 190602 (2002).
602:
603:
604: \end{thebibliography}
605:
606: \end{document}
607:
608:
609:
610: \bibitem{Lopezetal_PRL05}
611: J.~M.~L\'opez, M.~Castro, and R.~Gallego,
612: %''Scaling of local slopes, conservation laws, and anomalous roughening in surface growth'',
613: Phys. Rev. Lett. {\bf 94}, 166103 (2005).
614:
615:
616:
617: \bibitem{Lopezetal_PRE1997}
618: J. M. L\'opez, M. A. Rodriguez, and R. Cuerno, %''Superroughening versus intrinsic anomalous scaling of surfaces'',
619: Phys. Rev. E {\bf 56}, 3993 (1997).
620: J.~M.~L\'opez, %''Scaling approach to calculate critical exponents in anomalous surface roughening''
621: Phys. Rev. Lett. {\bf 83}, 4594 (1999).
622:
623: \bibitem{Lopez_PRL1999}
624: J.~M.~L\'opez, %''Scaling approach to calculate critical exponents in anomalous surface roughening''
625: Phys. Rev. Lett. {\bf 83}, 4594 (1999).
626:
627:
628: \bibitem{Lopezetal_PRL2000}
629: J.~J.~Ramasco, J.~M.~L\'opez, and M.~A.~Rodriguez, %''Generic dynamic scaling in kinetic roughening'',
630: Phys. Rev. Lett. {\bf 84}, 2199 (2000).
631:
632:
633: \bibitem{wood}
634:
635: The latter are related to the local
636: slope exponent because on general grounds $w(\ell,t) \sim
637: \sqrt{G(\ell,t)}$, where $G(\ell,t)= \langle [h(x+\ell,t)-h(x,t)]^2
638: \rangle$ is the height-difference correlation function between sites a
639: distance $\ell$ apart.% (a quantity also closely related to the
640: structure factor or so-called power spectrum of the interface).
641: