Professional Documents
Culture Documents
LETTERS
VOLUME 58 23 M A R C H 1987 NUMBER 12
Phase Organization
Chao Tang, Kurt Wiesenfeld, and Per Bak
Physics Department, Brookhaven National Laboratory, Upton, New York 11973
Susan Coppersmith
Physics Department, Princeton University, Princeton, New Jersey 08544
and
Peter Littlewood
A T& T Bell Laboratories, Murray Hill, New Jersey 07974
(Received 24 December 1986)
We describe the selection mechanism for novel pattern formation in a driven chain of nonlinear oscil-
lators. Phase organization arises as a consequence of dynamical selection of minimally stable states.
Understanding the behavior of complex (many- behavior. Systems where few-EDF models appear inade-
degrees-of-freedom) systems continues to present a great quate include spin-glasses, arrays of coupled maps, 2
challenge to physicists. The standard statistical mechan- spatio-temporal chemical oscillators, 3 neural networks, 4
ics is a powerful theory for equilibrium and uniform sys- large-aspect-ratio fluid convention, 5 and a video feed-
tems; however, for nonequilibrium systems possessing back system. 6 In this Letter, we study a high-EDF sys-
spatio-temporal structure there is no general theory tem displaying complex dynamics and interesting organ-
indeed, it is not even clear what quantities might izing behavior. In contrast with these other complex sys-
characterize the physics of such systems. Recently a tems, ours is simple enough to completely characterize
large step has been taken towards the understanding of theoretically. Our hope is that this will stimulate pro-
these complex systems, in large part because of the study gress toward a general characterization of high-EDF be-
of simple low-dimensionality discrete dynamical equa- havior.
tions. It is now realized that (1) simple nonlinear maps The model consists of an array of N balls of mass m,
can have very rich time series and very complex struc- each connected to its neighbors by springs with force
tures can be produced from very simple algorithms; (2) constant K. The array is subject to a sinusoidal potential
there exist universality classes, such that very different and driven by a time-periodic square-wave force E(t).
systems possess identical scaling properties. The hope is The equation of motion is
that the study of these simple deterministic models can
help us pose and answer more general questions in com- m'y = yyj + Kiyj + i lyj+yj-O Asin(27ryj)+E,
plex systems; this approach is in some sense the obverse
; = 1,2,...,7V, (1)
of the statistical point of view.
To date, significant progress has been made in under- where yj is the position of the yth ball, y is the damping
standing complex systems whose behavior can be cap- constant, and A the amplitude of the potential. In the
tured by few "effective degrees of freedom" ( E D F ) 1 ; in static case (m=y=E = 0 ) , Eq. (1) is the Frenkel-
contrast, very little is known in general about high-EDF Kontorova model which has been studied extensively. 7
Of particular interest for our purpose is that the says that when the pulse is off each ball falls directly into
Frenkel-Kontorova model has a very large number its nearest potential minimum, while Eq. (2a) says that
(~~eN) of stable configurationsin our problem there is during the pulse the balls are depinned from their mini-
a similarly large number of stable time-periodic station- ma and their motion is determined by the pulse force F
ary states and a hierarchy of metastable states. and the nearest-neighbor elastic interaction leading in
What is a phase-organized state? It is a time-periodic general to a net drift current. That Eq. (2) captures the
state with the property shown in Fig. 1: A snapshot at essential physics of Eq. (1) is supported by extensive nu-
an instant just before the pulse is turned off shows the merical simulations of both the iterative map and the or-
majority of balls sitting at maxima of the periodic poten- dinary differential equations; a derivation of a closely re-
tial. The phase of 0 of a ball is its distance from the lated map for m = 0 is possible within the context of per-
nearest-potential minimum; a configuration is phase or- turbation theory. 1 0 The chief effect of the approxima-
ganized when 0 = y for most of the balls at the falling tion leading to Eq. (2) is that simulations of the mapping
edge of a pulse. Remarkably, in the appropriate param- typically lead to more perfect phase organization than do
eter regime (weak springs and large damping), simula- the corresponding ordinary differential equations. In
tions reveal that the phase-organized states are always what follows, we focus on the iterative map since its
reached for sufficiently random initial conditions even dynamical properties can be completely characterized by
though these states are measure zero (as TV goes to analytic means.
infinity) with respect to the total number of stable states. Without loss of generality, we take F to be an integer
This phenomenon was first discovered in the context of and rewrite Eq. (2) as
the pulse-duration memory effect in charge-density
waves, 8 where experiments show a macroscopic manifes- yj(n + \)=yj(n)+F + mn\[k&yj(n)]. (3)
tation of this microscopic organization. We will return Applying the operator <2 to both sides and denoting the
to this point briefly later on. "curvature" Dyj(n) as Cjin) yields
Our goal is to understand the selection mechanism
leading to phase organization. We accomplish this by C / ( / i + l ) - C / ( / i ) + 2)nint[fcC / (/i)]. (4)
writing down a "Poincare m a p " for Eq. ( l ) . By studying
the set of fixed points we recognize a "topological" selec- Equation (4) is a nonlinear diffusion equation on the lat-
tion for marginally stable states and identify the subset tice phase space ZN. In contrast with linear diffusion,
of minimally stable states as the phase-organized states. the competition between diffusion (pulse on) and local
The dynamics then selects for these latter states, and re- relaxation (pulse off) allows the chain to stably maintain
veals the existence of a hierarchy of metastable, "al- nonzero curvature configurations. The fixed-point condi-
most" phase-organized configurations. tion Cj (n + 1) = Cj (n) implies
In the limit of strong damping, weak coupling, and
nint[kCj(n)]=m+lj\ j = 1,2, . . . ,7V, (5)
high field (y^>m, E > A^>K), we approximate Eq. (1)
by the mapping 9 where m and / are integers. For periodic or free bound-
Zj in) =yj in) + klyj + \ (n ) 2yj {n)+yj-\{n)]+F ary conditions, only m = = / = 0 is allowed, so that the set
of fixed-point solutions is given by
=yj(n) + kyj(n)+Fy (2a)
-\/2k<Cj<\/2k, 7 = 1 , 2 , . . . ,7V. (6)
yj(n + \) ^ n i n t b y O O ] , (2b)
Since k is small, this defines a large (compact) hyper-
where yj(n) and Zjin) are the positions of the yth ball cube in {Cj} space. Any point in or on this cube is an al-
just before the /2th pulse is turned on and off, respective- lowed solution 11 and represents a stable time-periodic
ly, and nintGc) = i n t ( x + 0.5) is the nearest integer of x. solution of Eq. (1); it can be proven that there are no at-
The parameters k and F are the effective nearest- tractors outside this cube. The origin is located at the
neighbor coupling and external force, respectively, and center of the cube and is the most stable configuration, in
are related to the parameters in Eq. ( l ) . 9 Equation (2b) that it requires the largest (finite) perturbation to violate
FIG. 1. Snapshot of a phase-organized solution of Eq. (1) just before the field is switched off, showing the positions of the balls
relative to the sinusoidal potential; TV = 55, ra = l, 7 = 100, K = 1, A = 100, " = 305, r o n = 5 , Toff = 2, free boundary conditions. Only
an eight-ball portion of the chain is shown.
1162
V O L U M E 58, N U M B E R 12 PHYSICAL REVIEW LETTERS 23 M A R C H 1987
1163
V O L U M E 58, N U M B E R 12 PHYSICAL REVIEW LETTERS 23 M A R C H 1987
corner states, since only the edges of a cluster can have since the training pulses may bring the C D W directly to
significant change under Eq. (4). This slow "leaking" one of the stable corner states. Then, decreasing k (i.e.,
process towards a stable state takes a time OiN/k). shortening the pulse) simply expands the stable hyper-
We now discuss some ramifications of the picture cube, so that the system is already in a stable config-
developed above. uration and no relearning will occur, though increasing k
First, the fact that the dynamics selects configurations instead will give the familiar effect.
of minimal stability has significant implications for the In conclusion, we have studied a periodically pulsed,
statistical mechanics of this system. The usual approach weakly coupled dissipative many-body system. This
would be to seek some global free energy, e.g., the total high-EDF system displays spontaneous self-organization
elastic energy of the chain, and assume a probability with a selection mechanism which can be fully charac-
density that peaks at the energy minimum. In fact, pre- terized. This collective behavior is not due to a long-
cisely the opposite situation applies here: The most range correlation (like that in second-order phase transi-
stable (minimum-energy) configuration is the state least tions), but rather due to a very local dynamical rule; we
likely to be observed; the minimally stable (maximum- expect this rule to play a role in a wide class of complex
energy) configurations should be most heavily weighted. dynamical systems.
Next, we can extend our analysis to include the effect We thank Vic Emery and Laszio Mihaly for useful
of quenched randomness modeled by the addition of ran- discussions. This work was supported by the Division of
dom constants / to Eq. (2a). In fact, it was simulations Materials Sciences, U.S. Department of Energy, under
on this random system that first detected the phenom- Contract No. DE-AC02-76CH00016.
enon of phase organization. 9 This complication is
nonessential to the problem since the rigid translation
yj-*yj-Pj> zj^zj-Pj,
!See, e.g., Chaos, edited by Hao Bai-Lin (World Scientific,
where Singapore, 1984).
Proceedings of the Conference on Spatio-Temporal Coher-
kpj=aj ence, Los Alamos, New Mexico, 1986, Physica (Amsterdam)
eliminates the a,. (This is analogous to the nonessential D (to be published).
3
nature of randomness in the ID spin-glass problem. 1 2 ) See, e.g., Y. Oono and M. Kohomoto, Phys. Rev. Lett. 55,
2927 (1985).
The corner states still are selected, and are phase organ- 4
J. J. Hopfield, Proc. Nat. Acad. Sci. U.S.A. 79, 2554
ized. In effect, the a / s change the potential from (1985).
sinusoidal to irregular, but the selected states are still 5
M. S. Heutmaker and J. P. Gollub, Phys. Rev. A 35, 242
those where the balls sit at potential maxima at the in- (1987).
stant the pulse is turned off. 6
J. P. Crutchfield, Physica 10D, (Amsterdam) 229 (1984).
7
Finally, we revisit the original realization of this mod- P. Bak, Rep. Prog. Phys. 45, 587 (1982).
8
el, namely the pulse-duration memory effect in charge- R. M. Fleming and L. F. Schneemeyer, Phys. Rev. B 33,
density waves (CDW's). A discussion of the relevance of 2930 (1986); S. E. Brown, G. Griiner, and L. Mihaly, Solid
Eqs. (1) and (2) to CDW's appears in Ref. 9. Recall State Commun. 57, 165 (1986).
9
that periodic voltage pulses cause the CDW to "learn" S. N. Coppersmith and P. B. Littlewood, Phys. Rev. B (to
the length of the pulse. For a "dirty" CDW, the pulses be published).
10
bring the system to some metastable state, so that chang- S. N. Coppersmith, to be published.
11
For periodic boundary conditions on yj, there is an extra
ing the pulse duration (i.e., changing k in our model)
constraint on Eq. (6), i.e., X / Q = 0 . Thus, the allowed solu-
causes the C D W to "relearn" since the structure of the tions are the intersection of the hypercube with this hyper-
metastable state space has changed. This relearning pro- plane; however, this constraint does not change the general pic-
cess has been observed experimentally. However, for a ture for large TV.
clean enough C D W we can expect a different behavior, l2
D. Mattis, Phys. Lett. 56A, 421 (1976).
1164