• Keine Ergebnisse gefunden

Autonomous Dynamical Systems, Explicitly Time Dependent First Integrals, and Gradient Systems

N/A
N/A
Protected

Academic year: 2022

Aktie "Autonomous Dynamical Systems, Explicitly Time Dependent First Integrals, and Gradient Systems"

Copied!
2
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

696 Note

Autonomous Dynamical Systems, Explicitly Time Dependent First Integrals, and Gradient Systems

Willi-Hans Steeb

International School for Scientific Computing, University of Johannesburg, Auckland Park 2006, South Africa

Reprint requests to W.-H. S.;

E-mail:steebwilli@gmail.com Z. Naturforsch.66a,696 – 697 (2011) DOI: 10.5560/ZNA.2011-0030

Received June 6, 2011 / revised July 1, 2011

We study first integrals and first-order autonomous sys- tems of differential equations. Some of the concepts for ex- plicitly time-independent first integrals are extended to ex- plicitly time-dependent first integrals. Several applications are given.

Key words:Dynamical System; Time Dependent First Integrals; Gradient Systems.

We investigate first-order autonomous systems of differential equations

duj

dt =fj(u), j=1,2, . . . ,n, (1) and first integrals in particular explicitly time-depend- ent first integrals [1]. It is assumed that the functions fj:Rn→Rare analytic. If the autonomous system (1) admits a first integralI, it can be written as [2]

du

dt =S(u)∇I(u), (2)

where ∇I(u) := (∂I/∂u1, . . . ,∂I/∂un)T, ST(u) =

−S(u)is a skew-symmetricn×nmatrix andTdenotes transpose. This representation is important when we discretize a dynamical system (1) and want to preserve the first integral. As an example consider

du1

dt =cu1+c23u2u3, (3a) du2

dt =cu2+c31u3u1, (3b) du3

dt =cu3+c12u1u2, (3c)

wherec23,c31,c126=0, andcdescribes the damping. If c=0, then the system admits the first integral I1= (c31u21c23u22)/2 and the dynamical system can be written as

 du1/dt du2/dt du3/dt

=

0 −u3 0

u3 0 c12u1/c23 0 −c12u1/c23 0

·

I1/∂u1

I1/∂u2

I1/∂u3

.

(4)

Note that the system also admits the independent first integral(c12u21c23u23)/2, and the dynamical system can be reconstructed using Nambu mechanics. Ifc6=0, thenI1,I2are not first integrals anymore.

Now many dissipative dynamical systems such as the Lorenz model and the Rikitake-two disc dynamo admit, depending on the bifurcation parameters, ex- plicitly time-dependent first integrals. These type of first integrals are of the form [3–9]

f(u(t))eλt. (5)

The dynamical system given above withc6=0 admits first integrals of this form. Thus the concepts given above to cast the dynamical system (1) into the form (2) has to be extended. We show with an example that the concept to write system (1) if a first integral exists as a gradient system (2) can be extended to explicitly time-dependent first integrals of the type given by (5).

To study such cases we extend the autonomous system (1) to

du

dτ =f(u), dun+1

dτ =1. (6)

Thus we settun+1andun(τ=0) =0. The first in- tegrals (5) then take the formf(u1, . . . ,un)eλun+1.

As an example consider the autonomous system du1

dt =u1(1+au2+bu3), (7a) du2

dt =u2(1−au1+cu3), (7b) du3

dt =u3(1−bu1−cu2), (7c)

c

2011 Verlag der Zeitschrift f¨ur Naturforschung, T¨ubingen·http://znaturforsch.com

(2)

W.-H. Steeb·Dynamical Systems, Time-Dependent First Integrals, Gradient Systems 697 where a,b,c∈R with the first integral I(t,u(t)) =

(u1+u2+u3)e−t. Thus we consider the autonomous system

du1

dτ =u1(1+au2+bu3), (8a) du2

dτ =u2(1−au1+cu3), (8b) du3

dτ =u3(1−bu1cu2), (8c) du4

dτ =1. (8d)

This system can be written in the form

 du1/dτ du2/dτ du3/dτ du4/dτ

=

0 s12eu4 s13eu4 s14eu4

−s12eu4 0 s23eu4 s24eu4

−s13eu4 −s23eu4 0 s34eu4

−s14eu4 −s24eu4 −s34eu4 0

·

I/∂u1

I/∂u2

I/∂u3

I/∂u4

(9) withI(u) = (u1+u2+u3)e−u4,∂I/∂u1=∂I/∂u2=

I/∂u3=e−u4,∂I/∂u4=−(u1+u2+u3)e−u4, and s12=1

3u1−1

3u2+au1u2, (10a) s13=1

3u1−1

3u3+bu1u3, (10b) s23=1

3u2−1

3u3+cu2u3, (10c)

wheres14=s24=s34=−1/3. Thus from the example it is obvious how to extend the concept to explicitly time-dependent first integrals.

The approach can be extended if there are more than one explicitly first integral. Consider the dynamical system

du1

dt =cu1+c234u2u3u4, (11a) du2

dt =cu2+c134u1u3u4, (11b) du3

dt =cu3+c124u1u2u4, (11c) du4

dt =cu4+c123u1u2u3. (11d) This is an extension of system (3) to higher dimen- sions. This system has been studied for c=0 from a Lie algebraic and integrability point of view by Steeb [10]. Such a system forc=0 appears from the self-dual Yang–Mills equation by exact reduction [11].

Forc=0, we have the three first integrals c134u21−c234u22, c124u22c134u23,

c123u23−c124u24.

(12)

Including dampingc6=0 provides the explicitly time- dependent first integrals

e−ct(c134u21c234u22), e−ct(c124u22−c134u23), e−ct(c123u23c124u24).

(13)

[1] W.-H. Steeb, Nonlinear Workbook, 5th edition. World Scientific, Singapore 2011.

[2] G. R. W. Quispel and G. S. Turner, J. Phys. A: Math.

Gen.29, L341 (1996).

[3] W.-H. Steeb, J. Phys. A: Math. Gen.15, L380 (1982).

[4] M. Kus, J. Phys. A: Math. Gen.16, L689 (1983).

[5] F. Schwarz and W.-H. Steeb, J. Phys. A: Math. Gen.17, L819 (1984).

[6] K. Kowalski and W.-H. Steeb, Nonlinear Dynamical Systems and Carleman Linearization, World Scientific, Singapore 1991.

[7] M. A. Almeida and I. C. Moreira, J. Phys. A: Math.

Gen.25, L669 (1992).

[8] M. A. F. Sanju´an, J. L. Valero, and M. G. Velarde, Il Nuovo Cim.13D, 913 (1991).

[9] W.-H. Steeb, Z. Naturforsch.49a, 751 (1994).

[10] W.-H. Steeb, Int. J. Geom. Meth. Mod. Phys.7, 405 (2010).

[11] W.-H. Steeb and N. Euler, Il Nuovo Cim.106B, 1059 (1991).

Referenzen

ÄHNLICHE DOKUMENTE

Example 3.3.1. Choose the Euclidean– norm. In the following we prove that there exist exponential rates such that the given system is M–hyperbolic with this disturbed family

In Section 6, we present the Casati-Prosen map on the torus within a large class of classically reversible maps and formulate conjectures about the nature and distribution of

We get an interdisciplinary framework to understand self-organizing complex systems and to ask for applications in organic computing.. 1 From Linear to

In some branches of the control theory, the problem of constructing a common Lyapunov function for a family of dynamical systems turns up. 00-01-00641 and Grant of Scientific

Economic model predictive control without terminal constraints for optimal periodic behavior. On the role of dissipativity in economic model

We employ a general phase condition and truncate the boundary-value problem to a finite interval by using on both ends the technique of asymptotic boundary conditions; see, for

We investigate macroscopic dynamical quantum tunneling (MDQT) in the driven Duffing oscillator, characteristic for Josephson junction physics and nanomechanics.. Under

Reducing the underlying complex network to a model system as simple as possible and combining nonlinear and stochastic effects, we are able to describe the complex dynamical