Abstract
Consider a conservative Hamiltonian system. One may specify the state of such a system by giving all the position and momentum coordinates (q,p) of a point in the system phase space, and the time evolution of the system is described by a trajectory lying on a surface described by the conservation of energy in the phase space. A dynamical system is said to be ergodic, if left to itself for long enough, it will pass in an erratic manner close to nearly all the dynamical states compatible with conservation of energy.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
This picture appears to be supported by the spectacular success of the predictions of statistical mechanics which is predicated on the basic supposition that the Hamiltonian system with a large number of degrees of freedom in question is equally likely to be found, at each instant, at any point on the energy surface in the phase space.
- 2.
The F i ’s may then be viewed as new momenta generated by some canonical transformation (the constancy of the new momenta immediately leads to the integration of Hamilton’s equations for the system).
- 3.
An n-dimensional torus is defined as
$$ T^n=S^1\times S^1\times\cdots \times S^1 \quad(n\ \textit{products}) $$where,
$$ S^1= \bigl\{ (x_1, x_2) | x^2_1+x^2_2=1 \bigr\} . $$ - 4.
It may be noted that the time average often depends on initial conditions, so it may vary from trajectory to trajectory. Nonetheless, there may be cases, like when the dynamical system has an attractor, with families of trajectories possessing the same time averages.
- 5.
For a regular motion, this result would be sensitive to the choice of initial conditions because, if one chooses the ensemble of initial values such that they are bunched together, then due to the regularity of the motion, they will remain bunched together. As a result, the ensemble samples and the averages computed by them will not be statistically independent, and the result (4.14) will not hold. For a mixing motion (see Sect. 5.9), however, the result (4.14) would be insensitive to the choice of initial conditions because, the mixing motion spreads the samples of the ensemble uniformly over the energy surface even if they are very close to each other initially. As a result, these samples and the averages computed by them will be statistically independent, and the result (4.14) holds.
- 6.
For a one-dimensional map
$$ x_{n+1}= M(x_n) $$with an attractor A and a basin of attraction B (see Chap. 6) and a natural measure μ, (4.14) takes the form
$$\lim_{T\rightarrow\infty}\frac{1}{T}\int^T_{n=0}f \bigl(M^{[n]}(x_0) \bigr) = \int f(x)\,d \mu(x), $$which states that the time average of f(x) over an orbit emanating from an initial condition x 0 on B is equal to the natural-measure weighted average of f(x) over A.
- 7.
This follows from the fact that, for an n degree-of-freedom system, the n-dimensional tori foliate the (2n−1)-dimensional energy surface if they are capable of being the boundaries of the latter region, i.e., are of dimension at least (2n−2), so
$$n\geq (2n-2 ) $$or
$$n\leq2. $$ - 8.
A periodic motion can ensue in a system of coupled nonlinear oscillators as a consequence of the frequency locking effect (originally discovered by Christiaan Huygens (1673)) This refers to a self-synchronization of two coupled nonlinear oscillators so as to make their basic frequencies commensurate and to keep the system locked in this state over a range of parameters. One example of this effect is the locking of the moon’s rotation rate to its orbital period caused by the dissipative tidal forces, which leads to the same side of the moon always facing the earth!
- 9.
A compact set is of Lebesgue measure zero if, for any ε>0, it can be covered with a finite number of intervals of total length less than ε.
- 10.
The existence of a non-integrable Hamiltonian H in the neighborhood of every integrable Hamiltonian H 0 implies that non-integrable Hamiltonians are dense in the set of analytic Hamiltonians.
- 11.
However, this is, generally, not the case; most multi-dimensional nonlinear systems are not integrable. In such systems, chaotic trajectories are densely distributed among the regular trajectories and have finite measure in the phase space. Perturbation theory cannot describe the complexity of this chaotic motion; formally the series diverge (see below). Nonetheless, one may still use the perturbation theory to obtain solutions that in some sense “approximate” the actual motion in nonlinear systems. If the actual trajectory is chaotic or involves significant change in topology, then the perturbation solution may still approximate the motion in some coarse-grained sense, as when the chaotic motion is confined to a thin separatrix layer bounded by regular motion (Lichtenberg and Lieberman 1992).
- 12.
A subset A of a set S is said to be dense in S, if and only if the closure of A is S, the closure of a set A being the union of the set A with all its limit points. A point P is a limit point of a set A, if there exists an infinite sequence of distinct points x 1,x 2,… in A such that lim n→∞ x n =P. A set A is said to be closed if it contains all of its limit points.
- 13.
That is, given an irrational number k, in any neighborhood of k, namely, [k−ε,k+ε], one can always find a rational number no matter how small ε is. Indeed, every irrational number can be thought of as the limit of a sequence of rational numbers.
- 14.
A subset A of a set S is said to be nowhere dense in S if the complement of the closure of A is dense there.
- 15.
These numbers can be expressed as solutions of quadratic equations with integer coefficients.
- 16.
These continued-fraction representations can be generated as follows:
-
(a)
\(x=\sqrt{3}\) can be written as
$$\begin{aligned} x&=1+\frac{2}{1+x} \\ & =1+\cfrac{2}{1+1+\cfrac{2}{1+1+\cfrac{2}{1+}}} \end{aligned}$$or
$$\sqrt{3}=1+\cfrac{1}{1+\cfrac{1}{2+\cfrac{1}{1+\cdots}}} $$ -
(b)
\(x=-1+\sqrt{2}\) can be written as
$$\begin{aligned} x&=\frac{1}{2+x} \\ &=\cfrac{1}{2+\cfrac{1}{2+\cfrac{1}{2+\cdots}}} \end{aligned}$$Therefore,
$$\sqrt{2}=1+\cfrac{1}{2+\cfrac{1}{2+\cfrac{1}{2+\cdots}}} $$ -
(c)
\(x=\frac{1+\sqrt{5}}{2}\) can be written as
$$\begin{aligned} x&=1+\frac{x}{1+x}=1+\cfrac{1}{1+\cfrac{1}{x}} \\ &=1+\cfrac{1}{1+\cfrac{1}{1+\cfrac{1}{1+\cdots}}} \end{aligned}$$Therefore, \(\frac{\sqrt{5}-1}{2}\), called the golden mean number, is given by
$$\frac{\sqrt{5}-1}{2}=\frac{1}{1} +\cfrac{1}{1+\cfrac{1}{1+\cdots}} $$
Thus, the golden mean number has the simplest possible continued fraction expansion.
It is of interest to note that the continued fraction for \(\frac{\sqrt{5}-1}{2}\) is also generated by the Fibonacci numbers F n , which are defined by
$$F_{n+1}=F_n+F_{n-1};\qquad F_0=0,\qquad F_1=1;\quad n=0,1,2,\ldots $$via a sequence of rationals W n ,
$$W_n\equiv\frac{F_n}{F_{n+1}}=\frac{F_n}{F_n + F_{n-1}}= \frac{1}{1+W_{n-1}}=\cfrac{1}{\underbrace{1+\cfrac{1}{1+\cdots}}_{n\ \mathrm{times}}} $$which converge towards
$$W^\ast = \lim_{n\to\infty} W_n= \frac{1}{1+W^\ast} $$from which
$$W^\ast_{1,2}=\frac{\sqrt{5}\pm 1}{2}. $$Since 1>W ∗>1/2, \(W^{\ast}_{1}\) is a spurious root and has to be discarded. The above derivation also shows that irrational numbers may be represented by an elementary limit process resulting from a fixed-point equation.
-
(a)
- 17.
Note that this is a crude overestimate of the “measure” of destroyed tori because we have included separately rationals r/s whose deleted neighborhoods overlap.
- 18.
The Toda potential
$$ V(r)=\frac{a}{b}e^{-br}+ a r $$in the limit a⇒∞, b⇒0, with ab finite, reduces to the linear spring potential
$$ V(r)\approx \frac{1}{2} ab r^2 $$while in the other limit a⇒0, b⇒∞, with ab finite, describes the interaction among hard spheres.
- 19.
Otherwise, the overall problem may significantly modify the width of each resonance region so that the resonance regions do not actually overlap despite the prediction of the single-resonance calculation to the contrary.
- 20.
However, efforts to produce a magnetically confined plasma have been considerably hampered by dynamic plasma instabilities which degrade the plasma confinement.
References
V.I. Arnol’d, Mathematical Methods of Classical Mechanics (Springer, Berlin, 1978)
R. Balescu, M. Vlad, F. Spineau, Tokamak: a Hamiltonian twist map for magnetic field lines in a toroidal geometry. Phys. Rev. E 58, 951 (1998)
J.R. Cary, Lie transforms and their use in Hamiltonian perturbation theory. Phys. Rep. 79, 129 (1981)
G. Casati, J. Ford, Stochastic transition in the unequal-mass Toda lattice. Phys. Rev. A 12, 1702 (1975)
C. Cerjan, W.P. Reinhardt, Critical point analysis of instabilities in Hamiltonian systems. J. Chem. Phys. 71, 1819 (1979)
B. Chirikov, A universal instability of many-dimensional oscillator systems. Phys. Rep. 52, 263 (1979)
E.C. daSilva, I.L. Caldas, R.L. Viana, M.A.F. Sanjuan, Escape patterns, magnetic footprints and homoclinic tangles due to ergodic magnetic limiters. Phys. Plasmas 9, 4917 (2002)
D.F. Escande, Stochasticity in classical Hamiltonian systems: universal aspects. Phys. Rep. 121, 165 (1985)
E. Fermi, J. Pasta, S. Ulam, Studies of nonlinear problems. Los Alamos Sci. Lab. Report LA-1940 (1955)
J. Ford, J. Waters, Computer studies of energy sharing and ergodicity for nonlinear oscillator systems. J. Math. Phys. 4, 1293 (1963)
H. Goldstein, C. Poole, J. Safko, Classical Mechanics, 3rd edn. (Pearson, Upper Saddle River, 2002), Chap. 10
J.M. Greene, A method of determining a stochastic transition. J. Math. Phys. 20, 1183 (1979)
M. Hénon, C. Heiles, The applicability of the third integral of motion: some numerical experiments. Astron. J. 69, 73 (1964)
E.A. Jackson, Nonlinearly-coupled oscillators: I. Perturbation theory, ergodic problem. J. Math. Phys. 4, 551 (1963a)
E.A. Jackson, Nonlinearly-coupled oscillators: II. Comparison of theory with computer solutions. J. Math. Phys. 4, 686 (1963b)
H. Kabakow, A perturbation procedure for nonlinear oscillations. Ph.D. thesis, California Institute of Technology (1968)
J. Kevorkian, Resonance in a weakly-nonlinear system with slowly-varying parameters. Stud. Appl. Math. 62, 23 (1980)
A.J. Lichtenberg, M.A. Lieberman, Regular and Chaotic Dynamics (Springer, Berlin, 1992)
P. Lochak, Hamiltonian perturbation theory: periodic orbits, resonances and intermittency. Nonlinearity 6, 885 (1993)
B.K. Shivamoggi, R.K. Varma, Internal resonances in nonlinearly-coupled oscillators. Acta Mech. 72, 111 (1988)
M. Tabor, The onset of chaos in dynamical systems. Adv. Chem. Phys. 46, 73 (1981)
M. Toda, Instabilities of trajectories of the lattice with cubic nonlinearity. Phys. Lett. A 48, 335 (1974)
A.H.P. van der Burgh, Studies in the asymptotic theory of nonlinear resonances. J. Sound Vib. 49, 93 (1976)
F. Verhulst, Discrete symmetric dynamical systems at the main resonances. Philos. Trans. R. Soc. A 290, 435 (1979)
G.H. Walker, J. Ford, Amplitude instability and ergodic behavior for conservative nonlinear oscillator systems. Phys. Rev. 188, 416 (1969)
R.B. White, Resistive instabilities and field line reconnection, in Handbook of Plasma Physics I, ed. by M.N. Rosenbluth, R.Z. Sagdeev (North-Holland, Amsterdam, 1983)
E.T. Whittaker, Analytical Dynamics (Cambridge University Press, Cambridge, 1944)
Author information
Authors and Affiliations
Appendix: The Problem of Internal Resonances in Nonlinearly-Coupled Systems
Appendix: The Problem of Internal Resonances in Nonlinearly-Coupled Systems
The problem of internal resonances in nonlinearly-coupled oscillator systems is of much interest in connection with redistribution of energy among the various natural modes of such a system. This energy sharing is usually brought about by resonant interactions among the natural modes of the system. The nature of the couplings among the latter plays a crucial role in such interactions.
Systems such as the one considered by Fermi et al. (1955) (see Chap. 7), namely, a system of one-dimensional weakly-nonlinearly-coupled chain of oscillators with the Hamiltonian of the form
ε being chosen sufficiently small (so that the nonlinear terms can be treated as a small perturbation on a primarily linear problem), are non-ergodic for small values of ε and are amenable to analysis using a perturbation theory. One property of such systems is that there is, in general, no effective energy sharing without an internal resonance. One would therefore desire to analyze a nonlinear system of equations possessing internal resonances and show that the latter lead to an effective exchange of energy.
Jackson (1963a,b) and Ford and Waters (1963) made such attempts for a problem of two coupled oscillators. In a straightforward perturbation theory, as we saw in Sect. 4.5, the internal resonances lead to the problem of small divisors, which are associated with terms having \((\sum^{N}_{k=1}n_{k}\omega_{k} )^{-1}\) as factors. Jackson (1963a,b) tried to remove the small divisors in the solution by introducing shifts in the frequencies ω k so that the corrected frequencies were no longer commensurable. However, the frequency shifts arise whether or not small divisors in the solution exist so that this approach cannot describe adequately what happens during an internal resonance—such as the manner in which energy is exchanged among the oscillators.
In order to describe the internal resonances adequately, it is obvious that such a perturbation theory should include modulations of the amplitudes of the oscillators along with any corrections to the frequencies ω k —which were most clearly seen in the numerical calculations of Ford and Waters (1963). Some perturbation theories for treating the problem of internal resonances in a system of two nonlinearly-coupled oscillators adequately have been given by Kabakow (1968), van der Burgh (1976), Verhulst (1979), and Kevorkian (1980). These studies revealed that whereas the actions of the individual oscillators are nearly constant when the system does not show an internal resonance, the total action of the uncoupled system is found to be nearly constant when the system undergoes an internal resonance and allows for a significant redistribution of action between two oscillators. The problem of higher-order internal resonances was considered by Shivamoggi and Varma (1988). The approximate invariant mentioned above to exist for the lowest-order internal resonance was found to continue to hold for the higher-order internal resonances.
Let us consider a system of two nonlinearly-coupled oscillators given by the generalized Hénon-Heiles Hamiltonian,
where ε is a small parameter characterizing the weakness of the couplings between the two oscillators.
Hamilton’s equations for the two oscillators then follow from (4.166),
Let us look for solutions of the form,
where t n =ε n t are the slow-time scales that characterize the slow variations introduced by the weak couplings among the oscillators, and
Note that (4.169) expresses the fact that the solutions of equations (4.167) and (4.168), for ε≪1, are very nearly equal to the set of harmonics represented by the first terms in the expansions of (4.169), which they would be, if ε=0. The perturbations induced by the terms of O(ε) in equations (4.167) and (4.168) may then be expected to show up as slow variations in the A’s and θ’s and as higher harmonics in the solution through the u k ’s and ν k ’s.
In order to fully specify the solution, one needs to specify some initial conditions. However, we will not do this keeping in mind that the periodic solutions derived in the following correspond to very special initial conditions.
Using (4.169), equations (4.167) and (4.168) give
By equating the coefficients of sinϕ 1, cosϕ 1, sinϕ 2, cosϕ 2 and the rest to zero separately, one obtains from equations (4.170) and (4.171) to O(ε):
On solving equation (4.172)–(4.177), one obtains
where the A’s and θ’s are constants to O(ε). Thus, in general, the oscillators move as if they were uncoupled effectively, and there is no appreciable energy sharing among them. Note, however, that this solution breaks down when ω 2=2ω 1, which leads to small divisors in (4.178). This condition corresponds to a first-order internal resonance in the system which, as we will see below, leads to considerable energy sharing in the system.
Let us now try to determine whether there are any higher-order resonances in this system. Towards this end, let us first assume that the lowest-order resonance is inoperative, i.e., ω 2≠2ω 1. Then, to O(ε 2), equations (4.170) and (4.171) give
Using (4.172), (4.173), (4.175), (4.176) and (4.178), equations (4.179) and (4.180) become
from which, one obtains
Inspection of (4.185) and (4.188) shows that in the O(ε)2 problem, one has a second-order internal resonance at ω 2=ω 1. This corresponds to the case considered by Hénon and Heiles (1964).
Next, let us determine the internal resonances of the O(ε)3 problem. First, let us assume that there are no internal resonances in the O(ε) and O(ε)2 problems, i.e., ω 2≠2ω 1 and ω 2≠ω 1. Then, to O(ε 3), equations (4.170) and (4.171) give
Using (4.172), (4.173), (4.175), (4.176), (4.178), (4.183)–(4.188), and proceeding as before, one obtains from equations (4.189) and (4.190),
Inspection of (4.193) and (4.196) shows that in the O(ε 3) problem, one has third-order internal resonances at \(\omega_{2}=\frac{2}{3}\omega_{1}\) and ω 2=4ω 1.
These higher-order resonances, as is confirmed in the following, become weaker successively, so they do not lead to as much exchange of energy among the oscillators as the first-order resonance does. Let us first consider the latter case.
(i) First-Order Internal Resonance (ω 2=2ω 1)
Let us write the O(ε) terms on the right in equations (4.170) and (4.171) as follows:
Using these expressions, equations (4.170) and (4.171) give
Solving equations (4.199) and (4.202), one obtains
Observe that (4.203) no longer exhibits any small divisors.
Also, one obtains from equations (4.198) and (4.201),
or
which dictates the manner in which action is exchanged among the oscillators under an internal resonance. Note that (4.205) is only an approximate constant of motion, because it has been derived under conditions of weak nonlinearities in the coupled system, and this constant of motion will cease to exist when the nonlinearities become strong. If one introduces the action J k by
then, (4.205) becomes
This result was first given by Kabakow (1968) and van der Burgh (1976), and is similar to the “adelphic” integral discussed by Whittaker (1944). That the action exchange at the first-order resonance (ω 2=2ω 1) follows according to E 1+E 2=const=E, where E k ≡J k ω k , was demonstrated by the computer calculations of Ford and Waters (1963), as shown in Fig. 4.6.
(ii) Second-Order Internal Resonance (ω 2=ω 1)
Let us write
on the right hand sides in equations (4.181) and (4.182); one then obtains,
Solving equations (4.210) and (4.211), one obtains
Next, one obtains from equations (4.209) and (4.212),
or
which is the same as the one, namely (4.207), deduced for the first-order resonance. This integral was also found by Verhulst (1979) and Kevorkian (1980) and was confirmed by the numerical calculations of Ford and Waters (1963)—see Fig. 4.7.
Comparison of (4.209) and (4.212) with (4.198) and (4.201) shows that the action exchange at the first-order resonance is much greater than that at the second-order resonance. The numerical calculations of Ford and Waters (1963), also showed that (see Fig. 4.7) the action exchange at the second-order resonance was weaker than that at the first-order resonance.
(iii) Third-Order Resonance (\(\omega_{2}=\frac{2}{3}\omega_{1}\) and ω 2=4ω 1)
For the resonance \(w_{2} = \frac{2}{3} \omega_{1}\), let us write
on the right hand sides of equations (4.189) and (4.190); one obtains therefrom,
One obtains from equations (4.220) and (4.222),
or
or
as before, in (4.207) and (4.218).
Next, for the resonance ω 2=4ω 1, let us write
on the right hand sides of equations (4.189) and (4.190); one obtains therefrom,
One obtains from equations (4.227) and (4.229),
or
or
as before, in (4.207), (4.218) and (4.225). It thus appears that this result remains valid for resonances of all order in the system (4.166).
Rights and permissions
Copyright information
© 2014 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Shivamoggi, B.K. (2014). Integrable Systems. In: Nonlinear Dynamics and Chaotic Phenomena: An Introduction. Fluid Mechanics and Its Applications, vol 103. Springer, Dordrecht. https://doi.org/10.1007/978-94-007-7094-2_4
Download citation
DOI: https://doi.org/10.1007/978-94-007-7094-2_4
Publisher Name: Springer, Dordrecht
Print ISBN: 978-94-007-7093-5
Online ISBN: 978-94-007-7094-2
eBook Packages: Physics and AstronomyPhysics and Astronomy (R0)