UNCLASSIFIED
Defense Technical Information Center
Compilation Part Notice
ADPO10915
TITLE: Entropy Dynamics Associated with
Self-Organization
DISTRIBUTION: Approved for public release, distribution unlimited
This paper is part of the following report:
ITLE: Paradigms of Complexity. Fractals and
Structures in the Sciences
To order the complete compilation report, use: ADA392358
The component part is provided here to allow users access to individually authored sections
f proceedings, annals, symposia, ect. However, the component should be considered within
he context of the overall compilation report and not as a stand-alone technical report.
The following component part numbers comprise the compilation report:
ADP010895 thru ADP010929
UNCLASSIFIED
229
ENTROPY DYNAMICS ASSOCIATED WITH SELF-ORGANIZATION
R. I. ZAINETDINOV
Moscow State University of Railway Communication,15 Obraztsov Street,Moscow, 101475, Russia
E-mail: zri@hotmail,com
A general model linking dynamics of informational entropy and the self-organization process in an
open, steady-state, nonequilibrium system is proposed. Formulas for dynamics of the informational
entropy flow and its rate are developed with respect to random process of influences exerted upon a
system. It was revealed that the open system responds to a strong change of conditions by steep
growth of the informational entropy flow up to a maximum value at the critical point associated with
the self-organization process. An example of self-organization during elasto-plastic deformation of
metal is considered.
1
Introduction
In this paper, we consider an open, steady-state, nonequilibrium, active system to be a
dissipative system. The spectrum of such systems is quite broad. If an exchange of energy
and matter with environment is structured, we can designate it as a transfer of information
(entropy). In this broad sense, information exchange is not necessarily limited to "intelligent" systems [1]. Presence of an energy flux from an external source to a system and
the dissipation of energy on external environment are the preconditions of activity in any
system. Because of this, the evolution of open active systems does not necessarily lead
towards the equilibrium state. On the contrary, open systems may be involved in processes of self-organization, which result in more complicated and more advanced structures.
As is known [8,9] for open systems the variation of the entropy dS for an interval of
time dt can be decomposed into a sum of two components deS and diS, with quite
different physical meanings. deS is the entropy flow, which depends on the processes of
matter and energy exchange between system and environment. diS is the entropy
production, caused by irreversible processes inside the system. If conditions deS<O and
IdeSl>d4S are observed, the certain stages of temporal evolution in the open system can
occur at a general downturn of entropy dS<O.
It is necessary to note that such a situation is possible only far from equilibrium, as in
an equilibrium state the member diS always prevails. It means that a system is so far from
its equilibrium that the linear laws no longer apply; nonlinear terms become important.
Self-organization is the "supercritical" phenomenon. Nevertheless far from equilibrium,
the system may still evolve to some steady state. In far from equilibrium conditions,
various types of self-organization processes may occur.
According to the traditional interpretation of entropy, as a measure of disordering
(uncertainty) of a system, it should be considered that, if the disorder decreases at the
expense of entropy return in the course of evolution, the system evolves into more
complicated and more advanced structures [8].
We present a general model linking dynamics of informational entropy and the selforganization process in an open system. We analyze the local zone's behavior in different
"modes of being" [3]. Formulas for dynamics of the entropy flow and its rate are obtained
with respect to process of influences exerted upon a system. Finally we describe a few
examples in order to elucidate the received results.
230
2
Definitions and Mathematical Formulation
Let an open system be in a steady (stable), nonequilibrium state. Further, let us assume
that in this state the entropy production is compensated for by a negative entropy flow
diS = -dS,
(1)
i.e., the system gives back so much entropy as is produced inside the system. Eq. (1)
establishes a condition of current balance between the entropy flow through the system
and its production inside the system. Current balance is understood as a stationary (not
time-dependent) nonequilibrium state of an open system, stable in relation to small
deviations. The system in this state is actually in dynamic balance with its environment.
The entropy production inside the system in a certain sense characterizes exhaustion of
the system's lifespan. Formulas, describing dynamics of the entropy flow and its rate, will
be deduced below. According to Eq. (1), these terms can be extended to the internal
entropy production inside the system in a steady (stable) nonequilibrium state.
All real processes are irreversible and unbalanced in some degree. Local gradients of
temperature, chemical potential, pressure can exist only in a nonequilibrium system. Let
X be an important local parameter, determining longevity, lifetime, load-carrying capacity
or any other property of vital significance for an open system in a local zone (hereafter we
call X the determiningparameter).Further, let us label a zone of an increased gradient of
the determining parameter as the local dissipative zone (LDZ). Such a local zone limits
the lifetime of a system. As a rule, these zones correspond to places of the most probable
failures during its lifetime. In the course of intensive operating, the number of LDZ can
grow. The growth dynamics of the LDZ number is an important property of the behavior
of dissipative system; however, in this paper this question is not considered. Here we
investigate the evolution of a system based on processes, occurring only in one LDZ.
There are at least two auto-regulating mechanisms of energy dissipation, which act
inside a system. We consider the behavior of a LDZ by using the model of a bistable
element. A bistable element (Fig. 1) has two stable states (down and up), in each of which
it can exist for a rather long time. Let us denote the mean value of the determining
parameter of a system in the down state as X0 and in up state as X 1. External influences or
internal changes in a system can result in transition of the bistable element from one state
to another. It was assumed that the transition from down state to up state is caused by
such an external influence exerted upon a system, at which determining parameter X (the
average over a volume of LDZ) exceeds mean value of the threshold level Xth.
In each particular case, the physical content of determining parameter, its threshold
level and criterion of transition from one state to another are defined by the type and
natures of the system examined and depend on the statement of a problem. In Sec. 5 we
shall consider interpretation of this concept for elasto-plastic deformation of metal in the
local zone of a load-carrying structure.
For studying the behavior of the LDZ of an open system, the mathematical apparatus
of Markovian stochastic processes was used. Solution of the differential Kolmogorov
equations for the probabilities Po(t) and P1 (t) of respectively, the down and up states of the
bistable element was obtained [11]. The case of a homogeneous Markovian process were
considered, that is the transition rates 4(t) and v(t) respectively, from the up state to the
down state and vice versa do not depend on time: 4(t)=pL; v(t)=v. Hereafter we call
L= v/R the regime parameter. Depending on the regime parameter, a., three typical
"modes of being" [3] the LDZ are considered: light at a<l, v<4, Po>Pj, i.e., the LDZ is in
231
up
I
Up
v(t)
I
)
DownJ
Figmre 1. Graph of transition in the bistable element modeling the LDZ behavior.
the down state for a long period of time; symmetric at oL=1, v=g, Po= P1 = 0.5; heavy at
cL>1, v>ýt, PI>Po, i.e., the LDZ is in the up state for a long period of time.
3
General Results
Let us consider the behavior of the statistical properties of the determining parameter X
for the examined LDZ in the course of evolution. Time dependence of mean value X(t) of
the determining parameter was determined under the obtained formula
-X(t)= [x0 +of,- a(Xj- Xo) e -16']/(I+ a),
(2)
where 1 - v + gt. At transient stage, the mean value has the tendency to grow or decrease
depending on the initial conditions, as shown in Figure 2 by solid line. Curve 1 occurs
when the initial state is the down one and curve 2 - when the initial state is the up. The
dashed line corresponds to the steady level of the mean value of determining parameter.
One can observe that both lines 1 and 2 approached the steady level at stationary stage.
The second of the principal statistical properties we studied is variance of the determining parameter X. The time course of a variance Dx(t) was described by term
DXQ)
=
(x,- X0 )2 [po (t)- P2(t)].
(3)
The graph of time courses of the variance Dx(t) for heavy and light modes of the system
being is shown in Figure 3. Since in the initial moment of time Po(to)=1, we have
Dx(to)=0. At the stationary stage (at t--->o) we have
DsT
=
(X1 - XX) (_
+
a) 2 .
(4)
For heavy mode of being at a> 1 in a critical point, corresponding to a moment
tb
IllIa-1)
(5)
the function (3) has a maximum. The light mode of the LDZ being when a<l (See line 2
in Figure 3) is characterized by absence of the critical point and by stabilization of the
variance on the level DST in the course of evolution towards a stationary stable state.
232
x,!
2
L
F-
ST
X0
Time t
0
Figure 2. The time courses of the mean value of determining parameter (solid lines) from down (1) and up (2)
initial states. Steady level is denoted by a dashed line.
DmaxDx(t) .::I .
DST
...
.
'.
..
0t
tb
Time t
Figure 3. Thetime courses of the variance Dx(t) for heavy (1) and light (2) modes of being (solid lines). Steady
level Dsr is denoted by a dashed line. The coordinates of the critical point are labeled as Dm_ and th.
Let us assume that the random process Y(t) of dynamic influences with probability
distribution function (PDF) F(Y), is exerted upon a system. This process produces a
similar random process X(t) of changing the determining parameter X in the LDZ.
Further, let us assume the PDF for instantaneous values of the random process is F(X). A
dependence of regime parameter ox on the PDF of this process is obtained in the form
233
F(Xth) = Po = 1/(l+o.), 1 - F(Xh) = Pl = oW(l+o),
(6)
where F(Xth) is the value of PDF F(X) at the threshold level of Xth, Po and P1 are final
probabilities of the down and up states corresponding to the stationary stage. In Sec. 5 we
shall obtain functions connecting regime parameter oc and statistical properties (mean
value and variance) of normal (Gaussian) random process exerted upon a system.
The main point of interest of the foregoing analysis is the possibility of computing
and forecasting the time course of entropy. Note that various interpretations of entropy
are internally linked quite closely [7,8]. The physical entropy of a system coincides with
the thermodynamic entropy S. The informational entropy H is connected to them by a
ratio [7]:
S =kH In 2,
(7)
where k is the Bolzmann constant. We consider the informational entropy, being the
measure of uncertainty, and equal to the amount of information (according to Shannon)
required for removing this uncertainty. For the bistable element modelling the LDZ, the
informational entropy H is determined from the equation [7]:
1
H =-EPj(t)log 2 PQ(t).
(8)
j=0
The minimum value of entropy H = 0 corresponds to the degeneration of a stochastic
system into a rigid determinate system. For open self-organizing systems the maximum
Hmx = 1 corresponds to a moment of bifurcation, when there is the destruction of pattern
(microstructure) exhausting its dissipative abilities, and resulting in an emergence of new
pattern at other hierarchical levels. Taking into account decomposition of entropy into a
sum of two components and Eq. (1) for current balance, we have considered dynamics of
the informational entropy flow in the LDZ of a system and obtained the analytical
dependence on of the informational entropy flow on time in the following form:
H(t) - - (1l+oa)In
)a
2 {1 +aea
'
ln(I1+ae(. l+a ). +0 - e-t)ln[a l} +
(9)
The time courses of the entropy flow H(t) is shown in Figure 4. Analysis of Eq. (9) shows
that in heavy mode of the LDZ being (at a>1) at the time tb, determining by Eq. (5), the
maximum entropy flow is reached, H(tb)= 1. It can be shown that this moment corresponds
to the condition of equal probabilities (maximum uncertainty) of keeping the LDZ in the
down and up states Po(tb) = PI(tb) = 0.5. Note that this moment tb corresponds to the critical point of maximum variation Dxmax of parameter X. That means, some fluctuations get
amplified up to a macroscopic scale. In this way, fluctuations - environmental perturbations or eigenfluctuations - may drive the system into a completely new state and thus
become the driving force of system development ("order through fluctuation" [9]).
Instabilities at the same time can break symmetry, that is, bifurcations occur: the system
may choose among two states, though determined by causality. The selection of the future
path of development is unpredictable now [5]. However, at least we can predict the
moment of maximum uncertainty and endeavor to take precautions against unfavorable
paths of development.
After passing the system through the critical point, which is a stochastic analogue of
the bifurcation point [8,12], the entropy flow decreases and, leaving the transient stage,
stabilizes on the steady level Hs:
234
H(t)
H st
---
---
------
0.5
Time t
tb
Figure 4. The time courses of the informational entropy flow H(t) for heavy (1) and light (2) modes of being
(solid lines). Steady level Ht is denoted by the dashed line. The coordinates of the critical point are labeled as
H. and bt.
dH
dt
0
_.H
tb
Time t
Figure 5. The time courses of the informational entropy flow rate dH/dt for heavy (1) and light (2) modes of
being.
235
H,
+a)/a
I
=n[(
I /ln2.
(10)
According to the traditional interpretation of entropy, it means that because of entropy
outflow in the course of evolution, the disorder decreases. A system is structured in
response to the heavy being mode by self-organization of more complicated and more
advanced patterns. This is an attribute of the system's adaptation to the random process
characterized by the regime parameter cX>1.
The light being mode of the LDZ when oL<1 (See line 2 in Figure 4) is characterized
by absence of the stochastic analog of the bifurcation point and by stabilization of the
entropy flow on the level Ht during the period of exit from the transient to stationary
stage.
Special interest is attracted by dynamics of the entropy flow rate in the course of the
system's evolution. The rate is defined as the derivation of the entropy flow function as:
dH/ _ e-1I
I2
dt
I
L
.
(--"1)
+ae-
Plot of the entropy flow rate with time is shown in Figure 5. Analyzing the function (11)
allows us to conclude that the response of a system to the heavy being mode (Oc>1) by a
rapid increase of the entropy flow takes place simultaneously with the reduction of the
entropy flow rate to zero at the moment of time tb. Hereafter the rate of entropy flow
becomes negative, passes through a minimum, and aspires to zero, when the transient
process approaches the steady (stable) stage.
4
Influence of the Variation of Conditions
Consider now the behavior of the LDZ with the variation of external and/or internal
conditions of a system. The main point of interest of the foregoing analysis is the
possibility of computing and forecasting a system response (namely the time course of the
informational entropy) to a change of the being conditions. Denote values of the
quantities, corresponding to a new set of conditions letters with an asterisk The change of
conditions may lead to both the modification of the regime parameter a* and the
probabilities Po* and P1 *. Timing of time t* started again at the moment of variation of
the conditions. Solution of the differential Kolmogorov equations for the probabilities
Po*(t*) and P1*(t*) was obtained [11].
An important role is played by analysis of the functions of entropy flow H*(t*) and
its rate dH*/dt* as a response of an open system to a sudden change of the external and/or
internal conditions. Using Eq. (8), a mathematical expression for the time course of the
informational entropy flow H*(t*) under new conditions was obtained in following form:
H *(t*)=
rr
1
.j.1[(1+a
1n2(1 + a)(1 + ct) L
a
_•eav
+ a) + (a*
l1
1(+
a)+ (a*-otee-'n* ++
(1+ aX)+ a*)
+ (a-a *kPý
+Ija*(1 +a)+(a~a*)&,j1n..a*(I +(Ia)
+aX1-+-a*)
(12)
(1
236
A function of the informational entropy flow rate after a variation of conditions of the
system existence can be expressed as:
*
dH*
dt*
*t*
V*#+(V
*
vpt-v*
"P e-''t*n v*fI+(V *-v* P)e- 0**
-
L[*f+(v*iU-vPl*)e-"
61n2
(13)
j
The initial rate of the entropy flow is equal to
dH *(t*)
dt *
v
,Io
/I*-V*
PVna.
(14)
fi ln2
It is evident that the obtained Eq. (12) to Eq. (14) depend on the combination of the
previous and new being conditions.
Figure 6 shows the graphs of the informational entropy flow H* and its rate dH*/dt*
for the case when being conditions become heavier (cc<l, 0*>1). On the first time
interval (O<t<1.4 sec) when (x<l, the LDZ exists in an easy being mode, so the entropy
flow is stabilized at a level Ht, corresponding to this mode. At the time t = 1.4 sec, being
conditions become heavier (ax*>l). An open system responds to a strong change of conditions by a rapid increase of the entropy flow from the stationary level ,Lt, obtained
under previous being conditions, to the maximum value H*(tb*) = 1. At the time tb*, the
rate of entropy flow sharply falls up to zero. The mathematical expression for the moment
of time tb* was obtained in the following form:
t* = -- 1 In (a*-1Xca + 1)
(15)
One can note that it depends on a combination of the previous and new being conditions.
In a time tb* after transition in a heavier being mode (cc*>1), the LDZ passes across a
bifurcation point. This point is associated with destruction of the pattern of the first
hierarchical level, exhausting its dissipative possibilities, and emergence of a new appropriate pattern, corresponding to the changed being conditions. Leaving on the second
level of hierarchy after the jump of development, the LDZ enters an evolutionary stage of
development. There is the rather slow stabilization of the entropy flow during this stage at
the expense of saturation by the information up to a level I = 1- H,,*, which corresponds
to a new mode of system existence. In other words, the LDZ adapts to new being conditions by perfection of structure. In this case, the entropy flow rate (See Fig. 6) gets a
negative value, passes through a minimum and, remaining negative, aspires to zero, when
the transient process reaches the stationary stage with a new steady level of the entropy
flow I-It*:
H St
a*)(a *)ITa* /In2,
Inn(1+
(16)
adequate to new conditions.
In passage through the critical point at time tb* the variance D*x of the systems determining parameter X is maximum. The system is characterized at this stage by the highest
degree of disordering, with the random fluctuations manifested on the macroscopic level.
After passage through the critical point, the variance D x is stabilized on a new stationary
level corresponding to the changed being conditions.
237
H(t),
0.8
_
bit
0.6
_
0.4
0.2
0
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
22
2.4
2.6
1.6
1.8
2
2.2
2.4
2.6
Time t, sec
b) 7
dH,6
bit/see4
t1
-
3
_
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Time t, sec
Figure 6. Response of a system to a change of the being conditions. The time courses of the informational
entropy flow H(t) (a) and its rate dH/dt (b) for the case when mode of being becomes heavier (ax< 1, a*>I). The
coordinates ofthe critical point are labeled as H*m.x and t%,.
There is no place for more detailed examination of behavior in the course of time of the
statistic properties of the systems determining parameter X after changing the being
conditions. Let us note only, that there are possible both the cases of monotonous increase
or decrease of the variance D*x and the cases of emerging the characteristic peak on a
curve of variance D*x. The characteristic peak is similar to that shown in Figure 3 and
corresponds to a critical point. Passing through such a typical peak and stabilizing the
variance D*x on a new steady level D*ST confirm completion of the period of running in
("bum-in") of a system and its adaptation to new changed conditions. When testing the
system, it is possible to measure and record the time history of the mean value and
238
variance of determining parameter in the vicinity of some LDZs. We believe that on a
basis of analysis of processes, occurring in the LDZ, it should be possible to speed up the
reception of data concerning the termination of the transition to new state as well as
adaptation of system to changed conditions. In Sec. 6 we shall consider the plots, representing the time evolution of the mean value and the variance for process of nucleation
in model system involving multiple steady states [8,12].
An open system of any nature comes back in a steady stable state due to the inflow of
information from the outside or/and redistribution of the informational entropy between
hierarchical levels of the system [2].
After passage through the bifurcation point, the entropy flow decreases in accordance
with an information accumulation; that means appropriate increase of an organization
level during a system development. At each hierarchical level of a system's evolution at
the end of self-organization process, when the "architecture" of the system was basically
defined and becomes saturated by the information, the entropy curve is gradually straightened, displaying the transition of a system to the evolutionary stage of development. A
growth in the degree of organization for any system has a limit, area of saturation, determined by the limited opportunities of an information accumulation in the given structure
at a given hierarchical level.
Such a picture of temporal evolution of an open system with respect to variation of
the being conditions agrees with the synergetic approach to processes of self-organization
in nonequilibrium dissipative systems [1,4,8,91. Obtained analytical dependencies reveal a
quantitative ratio, reflecting the dynamics of the entropy flow and its rate, which shows
the evolution of open systems in the course of its life cycle under variation of being
conditions. The derived laws, in our opinion, are applicable for open nonequilibrium
systems of various natures: technical, economic, biological, ecological, social, etc. The
discussed model and obtained dependencies make it possible on a unified basis to
describe the whole life cycle of a system, including passage across a sequence of
bifurcation points ("jumps" of development) and evolutionary stages of development at
each hierarchical level. Transition to a new level of development goes from disorder to
order, through the phenomenon of instability in the bifurcation points, where a system has
a number of options to diverge in several directions.
In the following, we shall give a few examples in order to elucidate the received
results.
5
Example 1. Self-Organization during Elasto-Plastic Deformation of Metal
Here we consider an example of self-organization process during elasto-plastic deformation of metal in local zones of load-carrying structures. From the thermodynamic and
synergetic points of view, a material undergoing plastic deformation is an open system
brought far from equilibrium conditions [4,6,10]. In this case, the LDZ is represented by
the zone of stress (strain) concentration. Let us accept as determining parameter X,
limiting the lifetime of the load-carrying structure, stress intensity ai, determined from
the distortion energy (Huber-Von Mises-Hencky) theory. It was assumed that transition
from down state, corresponding to the elastic behavior of metal in the LDZ, in up (the
plastic yielding state) is caused by such an external influence, exerted upon the system,
under which the determining parameter exceeds a threshold level equal to the mean value
of the yield strength a5 , that is: cai > cs.
The main point of this example is to show how the random loading process exerted
upon the load-carrying structure is linked with the structural response (namely the time
239
course of entropy) to a change of the loading conditions. Let us accept the hypothesis that
instantaneous values of the loading process have a normal distribution with the following
statistical properties: mean value (m and variance D, of the stresses. Using Eq. (6), the
relations that link the regime parameter (x with these statistical properties and also with
the threshold level (mean value of yield stress as) were obtained in the following form:
a=[Ias--.
a=[1I-_,(a
Ji-rl I
aj)-
1
at acm<as,
(17)
at a'm>S,
(18)
where
(
I I'
1(Z)=
72ý;r:
2-+ dxZ=-
Dý
(19)
The dependence of cc on a. and D, for low-carbon steel was computed from Eq. (17),
(18) and plotted in Figure 7. We used this approach for a load-carrying welded structure,
namely the pivoting section of a gondola car body. Examination of the time history of
stress in elements of the gondola car, including directly the welded joints, carried out
using the automated system for amplitude-spectral analysis, shows that the PDF of the
instantaneous values of the random process is normal.
In the course of elasto-plastic deformation in metals, a number of dislocation patterns
(microstructures) are formed. One after the other ball, cellular, persistent slip band, quasiamorphous microstructures arise and are destroyed, consistently replacing each other on a
background of existing grains' boundaries. Further increase of load results in formation of
the crack origins in a quasi-amorphous zone and growth of their density. The spontaneous
emergence of a quasi-amorphous microstructure corresponds to achievement of a
maximum disorder in this local zone, at which point the thermodynamic entropy is
maximum and equal to the enthalpy of melting. All these transitions are supervised by
achievement of a maximum level of the entropy flow [4]. Under action of an energy flux,
pumped up by the stochastic loading process in the LDZ, the deformation ability of metal
at the lowest hierarchical structural level is exhausted. Getting through a critical point of
bifurcation, metal passes to a higher level of the pattern hierarchy to microstructure,
having the higher dissipative characteristic [4]. This transition between patterns on the
microscopic level is reflected by the curves of the entropy flow and its rate, obtained
before. Returning to Figure 6, note that at the first time stage (0<t<l.4 sec) the regime
parameter (x was determined by using Eq. (17) to be equal to 0.25. It corresponds to the
vertical (gross) dynamic load of the gondola car. The value 0*=5 corresponds to the total
effect of the vertical and longitudinal inertial (in collision of cars) loads. That mode of the
LDZ being matches the transition of metal in plastic state. In Figure 6, one can see that a
load-carrying structure responds to the heavy conditions by a rapid increase of the entropy
flow from the stationary level It, obtained under vertical load, to the maximum H*,x at
the time tb*. This critical point is associated with the destruction of the pattern exhausting
its dissipative abilities, and the emergence of new microstructure at another hierarchical
level corresponding to plastic state of the LDZ. The local zone becomes structured,
responding to the heavy being mode by self-organization of more advanced patterns.
240
a)
4~
6
0
50
100
150
200
1/
2
250
am, MPa
300
350
400
450
500
450
500
b)
cc 10
6
121
10\
7
9
_
2
0,
0
50
100
150
200
250
Da, 'MPa
300
350
400
2
Figure 7. Change of the regime parameter o, as a fimction of the statistical properties (mean value am, (a) and
variance D. (b)) of Gaussian random process of dynamic stresses: 1 - 5) D. is equal to 50, 100, 150, 200, 300
MPa 2 respectively, 6 - 12) ac, is equal to 100, 200, 300, 350, 400, 450, 500 MPa respectively. Line ac = 1,
correspondingto threshold level a. , as= 310 MPa, is denoted by a dashed line.
6
Example 2. Nicolis and Prigogine's Model
A process of nucleation caused by formation of germs in model system involving multiple
steady states is considered by Nicolis and Prigogine [8,12]. The plots, representing the
time evolution of the average value and the variance for this numerical model, are
indicated in Figure 8, which was adopted from [12].
241
a) X
al
2"b)
Qx
2.//.
3
L_
_ _
/(
_.
-
1-1
0
200
100
Time t
300
0
10)
Trret
230
30D
Figure 8. Evolution of the mean value X (a) and variance Dx (b) of fluctuations for the Nicolis-Prigogine's
model. Adopted from 112].
We can see the process of the model system transition from initial state to another stable
state corresponding to a higher level of mean value. During the process the mean value of
determining parameter monotonously increases (See Fig. 8a). The variance begins to
increase sharply as soon as sufficiently large new domain appears in the system through
fluctuations. Appearance of the typical peak on the curve Dx(t) and subsequent stabilization of the variance on a new steady level (See Fig. 8b) lead to a conclusion that the
transition period is finished
The plots of Figure 8 correspond to analytical dependencies, Eqs. (2), (3) of mean
value X(t) and variance Dx(t) of determining parameter X with time (See Figures 2 and
3). We believe that on a basis of the analysis of processes, occurring in the LDZ, it should
be possible to speed up the reception of data concerning the termination of the transition
as well as adaptation of system to new conditions.
7
Conclusions
The various kinds of open systems (technical, ecological, biological, socio-economic,
etc.) respond to a strong change of being conditions in the same way: by steep growth of
the entropy flow up to a maximum value at the critical point. According to the traditional
interpretation of entropy, it means that the disordering and chaos in the system increase at
this stage of its evolution. The critical point (being the stochastic analogue of a bifurcation point) associates with the self-organization process, that is the destruction of
pattern of the previous hierarchical level, exhausting its possibilities, and resulting in the
emergence of new more complicated and more advanced patterns, corresponding to the
changed being conditions. An open system of any nature comes back to a steady stable
state due to inflow of the information from the outside or/and redistribution of the
informational entropy between the hierarchical levels of system.
242
Obtained mathematical expressions for the time course of the informational entropy
flow make it possible to predict the moment that the critical point will be reached. That is
the moment of maximum uncertainty, instability, and chaos in the system when small
fluctuations become amplified up to a macroscopic scale. It may drive the open system
into a completely new state and thus become the driving force of the system's
development.
Acknowledgements
The author is grateful to Prof. I. V. Gadolina, Prof. W. I. Griffith, and Prof. R. St.John for
reading the manuscript and giving valuable comments.
REFERENCES
1. Alonso M. (ed.), Organization and Change in Complex Systems (Paragon House,
New York, 1990).
2. Auger P., Dynamics and Thermodynamics in Hierarchically Organized Systems:
Applications in Physics,Biology and Economics (Pergamon, Oxford, 1989).
3. Ingarden R. S., Kossakowsky A. and Ohya M., Information Dynamics and Open
Systems: Classical and Quantum Approach (Kluwer Academic Publishers,
Dordrecht, 1997).
4. Ivanova V. S., Synergetics: Strength and Fractureof Metallic Materials (Cambridge
University Press, 1998).
5. Krohn W., Kuppers G. and Nowotny H. (eds.), Selforganization: Portraitof a
Scientific Revolution (Kluwer Academic Publishers, Dordrecht, 1990).
6. Kubin L. P. and Lepinoux J., The Dynamic Organization of Dislocation Structures,
In Proc. of VIII Intern. Conf of Strength Metals and Alloys (Tampere, 1988) 1 pp.
35-59,
7. Nicolis J. S., Dynamics of Hierarchical Systems: An Evolutionary Approach
(Springer, Berlin, 1986).
8. Nicolis G. and Prigogine I., Self-Organization in Nonequilibrium Systems: From
Dissipative Structures to Order through Fluctuations(Wiley, New York, 1977).
9. Prigogine I., From Being to Becoming: Time and Complexity in the PhysicalSciences
(W.H.Freeman and Company, New York, 1980).
10. Seeger A., Thermodynamics of Open Systems, Self-organization, and Crystal
Plasticity, In Proc. of VIII Intern. Conf of Strength Metals and Alloys (Tampere,
1988) 1 pp. 463-468.
11. Zainetdinov R. I., Dynamics of Informational Entropy Associated with SelfOrganisation Process in Open System, Chaos, Solitons & Fractals 9 (1999) pp.
1425-1435.
12. I-rnomic F., [Ipnroami 14. Ho3uatnue cno:lcnoeo: Beetenue. flep. c airiI. (Mi4p,
Mocxma, 1990).