Research Article Dynamics of Numerics of Nonautonomous Equations with

advertisement
Hindawi Publishing Corporation
Journal of Applied Mathematics
Volume 2013, Article ID 645345, 11 pages
http://dx.doi.org/10.1155/2013/645345
Research Article
Dynamics of Numerics of Nonautonomous Equations with
Periodic Solutions: Introducing the Numerical Floquet Theory
Melusi Khumalo
Department of Mathematics, University of Johannesburg, P.O. Box 17011, Doornfontein 2028, South Africa
Correspondence should be addressed to Melusi Khumalo; mkhumalo@hotmail.com
Received 22 September 2012; Revised 2 March 2013; Accepted 26 March 2013
Academic Editor: Alberto Cabada
Copyright © 2013 Melusi Khumalo. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Nonautonomous systems with periodic solutions are encountered frequently in applications. In this paper, we will consider
simple systems whose solutions are periodic with a known period. Their transformation under linearized collocation methods
is investigated, using a technique called stroboscopic sampling, a discrete version of the well-known Poincaré map. It is shown that
there is an inextricable relationship between AN stability (or BN stability) of the numerical methods and the correct qualitative
behaviour of solutions.
1. Introduction
Let
π‘₯σΈ€  = 𝑓 (π‘₯, 𝑑) ,
π‘₯ (0) = π‘₯0 ,
(1)
where 𝑓 : 𝐼 ⊂ R × R → R, be a scalar ordinary differential
equation in which 𝑓(π‘₯, 𝑑) is a periodic function of 𝑑 with
prime period 𝑇.
The detailed dynamics of numerics for nonautonomous
ODEs has notably been lacking. Although any nonautonomous ODE can be transformed to an autonomous
one, thereby increasing the dimension by one, the familiar
dynamics of autonomous equations which is centered around
the notion of equilibrium points [1] is lost. In certain special
cases, this notion is replaced by that of periodicity. It is on
these special cases that we will focus our attention. Stuart
[2] proved using the bifurcation theory that for reactiondiffusion-convection equations, linearized instability implies
the existence of spurious periodic solutions. Our approach
differs from that of Stuart, who considered partial differential
equations. Nonautonomous ODEs where 𝑓 is periodic in 𝑑 are
very common in applications such as population dynamics
with seasonal parameters or periodically forced systems.
Under certain conditions on 𝑓, (1) has a unique 𝑇periodic solution [3]. We will assume that the solution is
approximated by a linearized one-point collocation method
as in Foster and Khumalo [4]. Our objective is to determine,
for each 𝑓 under consideration, whether the numerical
scheme has the same dynamical behaviour as the differential
equation. In particular, we will consider cases in which the
ODE has a unique, asymptotically stable periodic solution
and establish conditions under which the numerical methods
have the same dynamics. These special cases will take the
following form:
(i) 𝑓 linear: 𝑓(π‘₯, 𝑑) = π‘Ž(𝑑)π‘₯ + 𝑏(𝑑), where π‘Ž(𝑑) and 𝑏(𝑑)
are 𝐢1 𝑇-periodic functions of 𝑑,
(ii) 𝑓 nonlinear: 𝑓 = π‘Ž(𝑑)𝑔(π‘₯) + 𝑏(𝑑), where 𝑔(π‘₯) is a 𝐢2
nonlinear function of π‘₯.
The linearized one-point collocation methods for the
scalar nonautonomous equation (1) are given by
π‘₯𝑛+1 = (π‘₯𝑛 + β„Žπ‘“ (𝑑𝑛 , π‘₯𝑛 ) + 𝑐1 β„Ž2 (πœ•π‘“/πœ•π‘‘) (𝑑𝑛 , π‘₯𝑛 )
−𝑐1 β„Žπ‘₯𝑛 (πœ•π‘“/πœ•π‘₯) (𝑑𝑛 , π‘₯𝑛 ))
(2)
−1
× (1 − 𝑐1 β„Ž (πœ•π‘“/πœ•π‘₯) (𝑑𝑛 , π‘₯𝑛 )) ,
where 0 ≤ 𝑐1 ≤ 1.
For a discussion of collocation methods in general, see
Hairer et al. [5].
We begin with a description of the dynamical systems
theory approach, which will be used in determining the conditions under which the methods have the same dynamical
2
Journal of Applied Mathematics
behaviour as the differential equations. Upon establishing
these conditions, we compare them with those imposed by
nonautonomous stability theory.
1.1. Dynamical Systems Approach. In what follows, we will use
a technique known as stroboscopic sampling to reduce the
problem of determining existence and stability of periodic
solutions to existence and stability of fixed points.
Let π‘₯𝑛+1 = 𝑝(π‘₯𝑛 ; 𝑛; β„Ž) be the discrete system representing
the numerical method, applied with fixed stepsize, to the
nonautonomous differential equation.
Step 1. Using inductive arguments, write the method in the
form π‘₯𝑛+1 = πœ™(π‘₯0 ; 𝑛; β„Ž).
Step 2. Choose β„Ž such that the period 𝑇 = β„Žπ‘˜. Then, π‘₯π‘˜ =
πœ™(π‘₯0 ; π‘˜; β„Ž), and then we establish the new discrete system
𝑋𝑛+1 = πœ™(𝑋𝑛 ; π‘˜; β„Ž). This is known as stroboscopic sampling.
Step 3. The fixed points of the last system correspond to 𝑇periodic solutions of the method. These are determined with
their stability types.
The above procedure is analogous to the Poincaré map
of (1). Let Φ(𝑑, π‘₯0 ) be the 𝑇-periodic solution of (1), with the
starting value π‘₯(0) = π‘₯0 . Then, the Poincaré map of (1) is the
scalar mapping
Π : R 󳨀→ R;
π‘₯ 󳨃󳨀→ Φ (𝑇, π‘₯) .
β„Ž > 0 fixed) to a periodic solution for any starting value if and
only if
π‘˜−1
∑Μƒπ‘π‘Ÿ = 0,
where π‘π‘Ÿ = 𝑏(π‘Ÿβ„Ž) + β„Žπ‘1 𝑏󸀠 (π‘Ÿβ„Ž) for each π‘Ÿ.
𝑇
Proof. Assume that ]2 = ∫0 𝑏(𝑠)𝑑𝑠 = 0. Taking π‘Ž(𝑑) = 0 in
(5), we obtain
π‘₯𝑛+1 = π‘₯𝑛 + β„Ž {𝑏 (π‘›β„Ž) + β„Žπ‘1 𝑏󸀠 (π‘›β„Ž)} .
̃𝑏 = 𝑏 (π‘–β„Ž) + β„Žπ‘ 𝑏󸀠 (π‘–β„Ž)
𝑖
1
π‘₯ = π‘Ž (𝑑) π‘₯ + 𝑏 (𝑑) ,
for 𝑑 ≥ 0, π‘₯ (0) = π‘₯0 .
Π (π‘₯𝑛 ) = π‘₯𝑛 + β„Ž {𝑏 (π‘›β„Ž) + β„Žπ‘1 𝑏󸀠 (π‘›β„Ž)} ,
If ]1 = ∫0 π‘Ž(𝑑)𝑑𝑑 < 0, then (4) has a unique 𝑇-periodic
asymptotically stable solution. If π‘Ž(𝑑) = 0, then (4) has a
unique 𝑇-periodic solution that is asymptotically stable if
𝑇
]2 = ∫0 𝑏(𝑠)𝑑𝑠 = 0 (Hale and KocΜ§ak, [3]).
Now, the linearized one-point collocation methods are
given by
π‘₯𝑛+1 = π‘₯𝑛 +
2.1. Linear Case with π‘Ž(𝑑) = 0
Theorem 1. Suppose that a linearized one-point collocation
method is used to solve the linear nonautonomous differential
equation (4) with π‘Ž(𝑑) = 0. The method tends (as 𝑛 → ∞,
(9)
𝑛
π‘₯𝑛+1 = π‘₯0 + β„Ž∑Μƒπ‘π‘Ÿ .
(10)
π‘Ÿ=0
Fix π‘˜ ∈ N such that 𝑇 = β„Žπ‘˜. Consider the π‘˜th iterate of π‘₯0
under Π:
π‘₯π‘˜ = Ππ‘˜ (π‘₯0 ) = π‘₯0 +
𝑇 π‘˜−1Μƒ
∑𝑏
π‘˜ π‘Ÿ=0 π‘Ÿ
(11)
and the related iteration, which corresponds to stroboscopic
sampling
𝑋𝑛+1 = 𝑋𝑛 +
𝑇 π‘˜−1Μƒ
∑𝑏 ,
π‘˜ π‘Ÿ=0 π‘Ÿ
(12)
where 𝑋0 = π‘₯0 . If the summation on the right-hand side
of (12) is zero, then the discrete system is fixed at 𝑋0 for
all 𝑛, which corresponds to a periodic solution. If it is
nonzero, then the stroboscopic iteration has no fixed point
and diverges.
Remark 2. The second term on the right-hand side of (12) can
be viewed as an application of the left rectangular quadrature
rule to the integral
β„Ž
{π‘Ž (π‘›β„Ž) π‘₯𝑛 + 𝑏 (π‘›β„Ž)
1 − β„Žπ‘1 π‘Ž (π‘›β„Ž)
+β„Žπ‘1 [π‘ŽσΈ€  (π‘›β„Ž) π‘₯𝑛 + 𝑏󸀠 (π‘›β„Ž)]} .
(5)
(8)
so that for a given value of π‘₯0 ,
(4)
𝑇
for 𝑖 = 0, 1, 2, . . .
and denote
2. Linear Case
σΈ€ 
(7)
We define
(3)
Suppose that 𝑓 = π‘Ž(𝑑)π‘₯ + 𝑏(𝑑), where π‘Ž and 𝑏 are 𝑇-periodic
functions of 𝑑. Then, the linear nonautonomous differential
equation (1) becomes
(6)
π‘Ÿ=0
𝑇
𝑇
∫ (𝑏 (𝑠) + 𝑐1 𝑏󸀠 (𝑠)) 𝑑𝑠,
π‘˜
0
(13)
and (12) can be written as the simple map
𝑋𝑛+1 = 𝑋𝑛 + 𝑐,
(14)
󡄨
󡄨
𝑇 σ΅„¨σ΅„¨σ΅„¨σ΅„¨π‘˜−1Μƒ 󡄨󡄨󡄨󡄨
|𝑐| = 󡄨󡄨 ∑ π‘π‘Ÿ 󡄨󡄨 .
π‘˜ σ΅„¨σ΅„¨σ΅„¨π‘Ÿ=0 󡄨󡄨󡄨
(15)
where
Journal of Applied Mathematics
3
2
or,
1
π‘₯𝑛+1 = π‘₯𝑛 (
0
−1
𝑇𝑐
𝑛𝑇
𝑛𝑇
𝑇
[𝑏 ( ) + 1 𝑏󸀠 ( )] .
+
π‘˜ + 𝑇𝑐1
π‘˜
π‘˜
π‘˜
π‘₯π‘–π‘˜
−2
−3
(18)
Proceeding in a manner analogous to the above, we can show
by induction that
−4
−5
𝑛+1
−6
π‘₯𝑛+1 = π‘₯0 (
−7
−8
π‘˜ + 𝑇 (𝑐1 − 1)
)
π‘˜ + 𝑇𝑐1
0
20
40
60
80
100
120
π‘Ÿ
𝑛
π‘˜ + 𝑇 (𝑐1 − 1) Μƒ
𝑇
+
) 𝑏𝑛−π‘Ÿ ,
∑(
π‘˜ + 𝑇𝑐1 π‘Ÿ=0
π‘˜ + 𝑇𝑐1
140
𝑑
Figure 1: Numerical results for (12) with 𝑐1 = 1/2. π‘˜ = 4 (× ×), and
π‘˜ = 5 (—), π‘˜ = 10 (– – –).
The above theorem can then be restated as follows.
Suppose that a linearized one-point collocation method is
used to solve the linear nonautonomous differential equation
(4) with π‘Ž(𝑑) = 0. The method tends (as 𝑛 → ∞, β„Ž > 0 fixed)
to a periodic solution for any starting value if and only if the
rectangular quadrature rule, used to approximate the integral
in (13), gives a zero.
π‘˜ + 𝑇 (𝑐1 − 1)
)
π‘˜ + 𝑇𝑐1
(19)
where each ̃𝑏𝑖 is defined by (8).
Denoting the right-hand side of (19) by Π(π‘₯𝑛 ), we can
perform stroboscopic sampling and consider
π‘˜
π‘˜ + 𝑇 (𝑐1 − 1)
)
π‘₯π‘˜ = Π (π‘₯0 ) = π‘₯0 (
π‘˜ + 𝑇𝑐1
π‘˜
π‘Ÿ
𝑇 π‘˜−1 π‘˜ + 𝑇 (𝑐1 − 1) Μƒ
+
) π‘π‘˜−π‘Ÿ−1
∑(
π‘˜ + 𝑇𝑐1 π‘Ÿ=0
π‘˜ + 𝑇𝑐1
(20)
and the associated map
π‘˜
𝑋𝑛+1 = 𝑋𝑛 (
Illustration. We examine the stroboscopic sampling of the
numerical solution of the differential equation π‘₯σΈ€  = cos 𝑑𝑒sin 𝑑 ,
π‘₯(0) = 1. Figure 1 shows the numerical results for π‘˜ = 4, 5,
and 10 with 𝑐1 = 0.5. For π‘˜ = 4, the method diverges quicker
from the periodic solution than for π‘˜ = 5. For π‘˜ = 10, the
divergence is negligible.
π‘˜ + 𝑇 (𝑐1 − 1)
)
π‘˜ + 𝑇𝑐1
π‘Ÿ
𝑇 π‘˜−1 π‘˜ + 𝑇 (𝑐1 − 1) Μƒ
+
) π‘π‘˜−π‘Ÿ−1 .
∑(
π‘˜ + 𝑇𝑐1 π‘Ÿ=0
π‘˜ + 𝑇𝑐1
(21)
Equation (21) is just the linear map
2.2. Linear Case with π‘Ž(𝑑) = −1. If π‘Ž(𝑑) is a negative constant,
(4) could be scaled in such a way that π‘Ž(𝑑) = −1. Then, clearly,
𝑇
]1 = ∫0 π‘Ž(𝑠)𝑑𝑠 < 0, and the differential equation has a unique,
asymptotically stable periodic solution.
Theorem 3. Suppose that a linearized one-point collocation
method is used to solve the linear nonautonomous differential
equation (4) with π‘Ž(𝑑) = −1. Then, for fixed 𝑐1 and π‘˜, the
method admits a unique periodic solution that is asymptotically
stable, provided
𝑋𝑛+1 = 𝑐𝑋𝑛 + 𝑑
(22)
with
π‘˜
𝑐=(
π‘˜ + 𝑇 (𝑐1 − 1)
) ,
π‘˜ + 𝑇𝑐1
π‘Ÿ
𝑇 π‘˜−1 π‘˜ + 𝑇 (𝑐1 − 1) Μƒ
) π‘π‘˜−π‘Ÿ−1 .
𝑑=
∑(
π‘˜ + 𝑇𝑐1 π‘Ÿ=0
π‘˜ + 𝑇𝑐1
(23)
The map has a single fixed point,
π‘˜>
𝑇
(1 − 2𝑐1 ) .
2
(16)
Proof. If π‘Ž(𝑑) = −1, (5) simplifies to
π‘₯𝑛+1
𝑇𝑐
𝑇
𝑛𝑇
𝑛𝑇
= π‘₯𝑛 +
[−π‘₯𝑛 + 𝑏 ( ) + 1 𝑏󸀠 ( )] , (17)
π‘˜ + 𝑇𝑐1
π‘˜
π‘˜
π‘˜
𝑋∗ =
𝑑
,
1−𝑐
(24)
which is asymptotically stable if and only if |𝑐| < 1; that is,
π‘˜>
𝑇 (1 − 2𝑐1 )
,
2
and the result is established.
(25)
4
Journal of Applied Mathematics
The following results are simple consequences of the
above theorem.
Corollary 4. The linearized one-point collocation method
with any 𝑐1 ∈ [1/2, 1] applied to the linear nonautonomous
differential equation with π‘Ž(𝑑) = −1 admits a unique periodic
solution that is asymptotically stable for all π‘˜ > 0.
Corollary 5. The explicit Euler method for the linear nonautonomous differential equation with π‘Ž(𝑑) = −1 admits a
unique, asymptotically stable periodic solution if and only if
π‘˜ > 𝑇/2.
We now attempt to bound |𝑋∗ |. The following lemma
gives a bound on the solution Φ(𝑑) of (4).
Lemma 6. Let 𝑏(𝑑) be a 𝐢1 𝑇-periodic function of 𝑑, and
assume that (4) with π‘Ž(𝑑) = −1 has a unique 𝑇-periodic
solution. There exist a number 𝑀 > 0 such that |𝑏(𝑑)|, |𝑏󸀠 (𝑑)| ≤
𝑀 and the 𝑇-periodic solution, Φ(𝑑), satisfies
|Φ (𝑑)| ≤ 𝑀
Observe that
𝑇
󡄨󡄨̃ 󡄨󡄨
σ΅„¨σ΅„¨π‘π‘Ÿ 󡄨󡄨 ≤ 𝑀1 = 𝑀 (1 + 𝑐1 ) ,
󡄨 󡄨
π‘˜
π‘Ÿ
|𝑑| ≤
=
π‘˜−1
π‘˜ + 𝑇 (𝑐1 − 1)
𝑇
⋅ 𝑀1 ⋅ ∑ (
)
π‘˜ + 𝑇𝑐1
π‘˜ + 𝑇𝑐1
π‘Ÿ=0
𝑇
⋅ 𝑀1
π‘˜ + 𝑇𝑐1
π‘˜−1
⋅ {1 +
π‘˜ + 𝑇 (𝑐1 − 1)
π‘˜ + 𝑇 (𝑐1 − 1)
)
[1 − (
𝑇
π‘˜ + 𝑇𝑐1
]}
π‘˜
= 𝑀1 ⋅ [1 − (
π‘˜ + 𝑇 (𝑐1 − 1)
) ]
π‘˜ + 𝑇𝑐1
= 𝑀1 (1 − 𝑐) .
(30)
Therefore,
(26)
for 𝑑 → ∞.
Proof. The boundedness of 𝑏(𝑑) and 𝑏󸀠 (𝑑) follows from the
periodicity and continuity of both functions.
Let πœ™(𝑑) be a periodic solution of (4) with π‘Ž(𝑑) = −1. Then,
πœ™(𝑑) satisfies the inequality
𝑇𝑐
󡄨󡄨 ∗ 󡄨󡄨 𝑀1 (1 − 𝑐)
= 𝑀1 = 𝑀 (1 + 1 ) .
󡄨󡄨𝑋 󡄨󡄨 ≤
1−𝑐
π‘˜
(31)
Hence, 𝑋∗ has essentially the same bound as the periodic
solution.
Example 8. Consider the linear nonautonomous equation
σΈ€ 
−πœ™ − 𝑀 ≤ πœ™ ≤ −πœ™ + 𝑀 ∀ 𝑑.
(27)
This implies that
𝑒−𝑑 (π‘₯0 + 𝑀) − 𝑀 ≤ πœ™ ≤ 𝑒−𝑑 (π‘₯0 − 𝑀) + 𝑀;
(28)
see [3]. Thus, πœ™(𝑑) is bounded for 𝑑 ≥ 0; therefore, it
approaches a 𝑇-periodic solution Φ(𝑑). Taking 𝑑 → ∞ in
(28) establishes the lemma.
Μ‚ ≤ 𝑀. Then, we have
From the above lemma, we have |π‘₯|
the following theorem.
Theorem 7. Let 𝑋∗ , given by (24), denote the fixed point
generated by the stroboscopic sampling of the numerical solution of the linear nonautonomous differential equation with
π‘Ž(𝑑) = −1 by a linearized one-point collocation method. Then,
the inequality
π‘₯σΈ€  = −π‘₯ + sin 𝑑,
π‘₯ (0) = 0,
(32)
which has solution π‘₯(𝑑) = (1/2)[sin 𝑑 − cos 𝑑 + 𝑒−𝑑 ]. Figure 2
shows the numerical results (stroboscopic sampling) of the
linearized implicit midpoint method (𝑐1 = 1/2) with π‘˜ = 3
and π‘˜ = 10. Figure 3 shows the results of the explicit Euler
method (𝑐1 = 0) with π‘˜ = 3 and π‘˜ = 10. In these experiments,
𝑇 = 2πœ‹; hence, the convergence to a unique periodic solution
is expected for π‘˜ > πœ‹(1 − 2𝑐1 ).
2.3. Linear Case with π‘Ž(𝑑) = −1 + πœ–πœŒ(𝑑). Let
π‘Ž (𝑑) = −1 + πœ–πœŒ (𝑑) ,
(33)
where πœ– ≥ 0 is a constant and 𝜌(𝑑) is a 𝑇-periodic function
𝑇
of 𝑑. We assume that −𝑇 + πœ– ∫0 𝜌(𝑠)𝑑𝑠 < 0 so that (4) has a
unique asymptotically stable periodic solution.
holds.
2.3.1. Linearized One-Point Collocation Methods. We establish conditions under which a one-point collocation method
with fixed 𝑐1 and step-size β„Ž > 0 exhibits the same dynamical
behaviour as the nonautonomous linear differential equation
with π‘Ž(𝑑) = −1 + πœ–πœŒ(𝑑).
Proof. Since 𝑏(𝑑) is continuous, there is a number 𝑀 such that
|𝑏(𝑑)|, |𝑏󸀠 (𝑑)| ≤ 𝑀 for all 𝑑 ∈ R. Then, for large 𝑑, the solution
Μ‚ < 𝑀.
|Φ(𝑑)| < 𝑀, and hence |π‘₯|
Notation. In what follows, we will denote 𝑏𝑛 = 𝑏(π‘›β„Ž), 𝑏𝑛󸀠 =
𝑏󸀠 (π‘›β„Ž), πœŒπ‘› = 𝜌(π‘›β„Ž), and πœŒπ‘›σΈ€  = 𝜌(π‘›β„Ž). Here, as before, 𝑇 = β„Žπ‘˜
(π‘˜ ∈ N).
𝑇𝑐
󡄨󡄨 ∗ 󡄨󡄨
󡄨󡄨𝑋 󡄨󡄨 ≤ 𝑀 (1 + 1 )
π‘˜
(29)
Journal of Applied Mathematics
5
0
We can write the above as
π‘₯𝑛+1 = 𝑇𝑛 π‘₯𝑛 + 𝐻𝑛̃𝑏𝑛 := Π (π‘₯𝑛 ) ,
(36)
where
−0.5
β„Ž (−1 + πœ–πœŒπ‘› ) + 𝑐1 β„Ž2 πœ–πœŒπ‘›σΈ€ 
,
1 − β„Žπ‘1 (−1 + πœ–πœŒπ‘› )
π‘₯π‘–π‘˜
𝑇𝑛 = 1 +
𝐻𝑛 =
−1
β„Ž
,
1 − β„Žπ‘1 (−1 + πœ–πœŒπ‘› )
(37)
̃𝑏 = 𝑏 + 𝑐 β„Žπ‘σΈ€  .
𝑛
𝑛
1 𝑛
−1.5
0
20
40
60
80
100
120
140
Proceeding by induction, we establish that
𝑑
𝑛
𝑛
𝑛
𝑖=0
𝑖=0
𝑗=𝑖+1
π‘₯𝑛+1 = π‘₯0 ∏𝑇𝑖 + ∑𝐻𝑖̃𝑏𝑖 ∏ 𝑇𝑗 ,
Figure 2: Numerical results of (32) using linearized implicit
midpoint method: π‘˜ = 3 (—) and π‘˜ = 10 (– – –).
(38)
from which we deduce that
400
π‘˜−1
π‘˜−1
π‘˜−1
𝑖=0
𝑖=0
𝑗=𝑖+1
π‘₯π‘˜ = π‘₯0 ∏𝑇𝑖 + ∑ 𝐻𝑖̃𝑏𝑖 ∏ 𝑇𝑗 := Ππ‘˜ (π‘₯0 ) .
300
The discrete system that corresponds to stroboscopic
sampling is the linear system
200
100
π‘₯π‘–π‘˜
𝑋𝑛+1 = 𝑐 (πœ–) 𝑋𝑛 + 𝑑 (πœ–) ,
0
(40)
where
−100
π‘˜−1
𝑐 (πœ–) = ∏𝑇𝑖 ,
−200
−300
(39)
𝑖=0
0
20
40
60
80
100
120
140
𝑑
π‘˜−1
π‘˜−1
𝑖=0
𝑗=𝑖+1
(41)
𝑑 (πœ–) = ∑ 𝐻𝑖̃𝑏𝑖 ∏ 𝑇𝑗 .
Figure 3: Numerical results of (32) using the explicit Euler method:
π‘˜ = 3 (—) and π‘˜ = 10 (– – –).
This system has a unique fixed point, 𝑋∗ , given by (24). It
is asymptotically stable if and only if |𝑐(πœ–)| < 1; that is,
|∏π‘˜−1
𝑖=0 𝑇𝑖 | < 1, or
Theorem 9. Suppose that a linearized one-point collocation
method is used to solve a linear nonautonomous differential
equation with π‘Ž(𝑑) = −1 + πœ–πœŒ(𝑑). Then, for fixed 𝑐1 and
π‘˜, the method will admit a unique periodic solution that is
asymptotically stable, provided
󡄨󡄨 π‘˜−1
2 σΈ€  󡄨󡄨
󡄨󡄨
󡄨
󡄨󡄨∏ [1 + β„Ž (−1 + πœ–πœŒπ‘– ) + 𝑐1 β„Ž πœ–πœŒπ‘– ]󡄨󡄨󡄨 < 1.
󡄨󡄨
󡄨󡄨
1
−
β„Žπ‘
(−1
+
πœ–πœŒ
)
󡄨󡄨 𝑖=0
󡄨󡄨
1
𝑖
󡄨󡄨 π‘˜−1
2 σΈ€  󡄨󡄨
󡄨
󡄨󡄨
󡄨󡄨∏ [1 + π‘˜π‘‡ (−1 + πœ–πœŒπ‘– ) + 𝑐1 𝑇 πœ–πœŒπ‘– ]󡄨󡄨󡄨 < 1.
󡄨󡄨
󡄨󡄨
2 − π‘˜π‘‡π‘ (−1 + πœ–πœŒ )
π‘˜
󡄨󡄨
󡄨󡄨 𝑖=0
1
𝑖
(34)
Proof. From (5), we deduce that the linearized collocation
methods, applied to the nonautonomous linear ODE with
π‘Ž(𝑑) given by (33), are
π‘₯𝑛+1 = [1 +
β„Ž (−1 + πœ–πœŒπ‘› ) + 𝑐1 β„Ž2 πœ–πœŒπ‘›σΈ€ 
] π‘₯𝑛
1 − β„Žπ‘1 (−1 + πœ–πœŒπ‘› )
β„Ž
+
⋅ [𝑏𝑛 + 𝑐1 β„Žπ‘π‘›σΈ€  ] .
1 − β„Žπ‘1 (−1 + πœ–πœŒπ‘› )
(35)
(42)
Substituting β„Ž = 𝑇/π‘˜ gives the result.
2.3.2. Examples. For each of the three special values of 𝑐1 ,
𝜌(𝑑) = sin 𝑑, and the increasing values of πœ–, we determined,
using (42), the minimum value of π‘˜ such that each method
has dynamical behaviour that is the same as that of the
differential equation. The results are illustrated in Figure 4.
For πœ– < 2, the explicit Euler method is the most restrictive
of the three (i.e., comparatively larger minimum values of π‘˜
must be used to obtain dynamical behaviour that is the same
as that of the differential equation). However, as πœ– is increased,
the Explicit Euler method outperforms the linearized implicit
Euler method by becoming less restrictive than that method
for πœ– ≥ 3. For πœ– ≥ 3, the explicit Euler and linearized midpoint
6
Journal of Applied Mathematics
where β„Ž = 𝑇/π‘˜. If
60
V (πœ–) =
50
Min(π‘˜)
(46)
then
40
30
20
10
0
1 + β„Žπ‘1 − β„Žπ‘1 πœ–π‘€2 + β„Žπœ–π‘€2 + β„Ž2 𝑐1 πœ–π‘€2 − β„Ž
,
1 + β„Žπ‘1 − β„Žπ‘1 πœ–π‘€2
0
2
4
6
8
10
πœ€
Figure 4: Least π‘˜ for unique periodic solution. The explicit Euler
(– – –), linearized implicit midpt (—), and the linearized implicit
Euler (× ×).
methods give comparable results, and for those values of πœ–,
the linearized implicit Euler method becomes more and more
restrictive in comparison to the other two.
Finally, we develop a bound on |𝑋∗ |. For the comparison
purposes, we present the following lemma, which can be
proved in a manner analogous to Lemma 6.
Lemma 10. Let 𝑏(𝑑), 𝜌(𝑑) be 𝐢1 𝑇-periodic functions of 𝑑. There
exist numbers 𝑀, 𝑀2 > 0 such that |𝑏(𝑑)|, |𝑏󸀠 (𝑑)| ≤ 𝑀, |𝜌(𝑑)|,
|πœŒσΈ€  (𝑑)| ≤ 𝑀2 , and the solution Φ(𝑑) satisfies
𝑀
|Φ (𝑑)| ≤ 󡄨󡄨
󡄨
󡄨󡄨1 − πœ–π‘€2 󡄨󡄨󡄨
(43)
as 𝑑 → ∞.
Μ‚ ≤ 𝑀/|1 − πœ–π‘€2 |.
From the lemma, we have |π‘₯|
Theorem 11. Let 𝑋∗ , given by (24), be the fixed point generated
by the stroboscopic sampling of the numerical solution of the
linear nonautonomous differential equation with π‘Ž(𝑑) = −1 +
πœ–πœŒ(𝑑) by a linearized collocation method. Then, the following
inequality holds:
𝑀 (1 + 𝑇𝑐1 /π‘˜)
󡄨󡄨 ∗ 󡄨󡄨
󡄨󡄨𝑋 󡄨󡄨 ≤ 󡄨󡄨
󡄨.
󡄨󡄨1 − β„Žπ‘1 πœ–π‘€2 − πœ–π‘€2 󡄨󡄨󡄨
(44)
1
Proof. Since 𝑏(𝑑) and 𝜌(𝑑) are 𝐢 and periodic, there are
numbers 𝑀, 𝑀2 such that |𝑏(𝑑)|, |𝑏󸀠 (𝑑)| ≤ 𝑀 and |𝜌(𝑑)|,
|πœŒσΈ€  (𝑑)| ≤ 𝑀2 for all 𝑑 ∈ R.
For each 𝑖,
󡄨
󡄨
󡄨󡄨 󡄨󡄨 󡄨󡄨󡄨 1 + β„Žπ‘1 + β„Žπœ–π‘€2 − β„Ž 󡄨󡄨󡄨
󡄨,
󡄨󡄨𝑇𝑖 󡄨󡄨 ≤ 󡄨󡄨
󡄨󡄨 1 + β„Žπ‘1 − β„Žπ‘1 πœ–π‘€2 󡄨󡄨󡄨
β„Ž
󡄨󡄨 󡄨󡄨
󡄨󡄨𝐻𝑖 󡄨󡄨 ≤ 󡄨󡄨
󡄨,
󡄨󡄨1 + β„Žπ‘1 − β„Žπ‘1 πœ–π‘€2 󡄨󡄨󡄨
󡄨󡄨̃ 󡄨󡄨
󡄨󡄨𝑏𝑖 󡄨󡄨 ≤ 𝑀1 = 𝑀 (1 + 𝑐1 β„Ž) ,
󡄨 󡄨
(45)
󡄨󡄨
σ΅„¨σ΅„¨π‘˜−1
󡄨󡄨
β„Ž
π‘˜−𝑖−1 󡄨󡄨󡄨
󡄨
⋅
𝑀
⋅
V
∑
|𝑑 (πœ–)| ≤ 󡄨󡄨
󡄨󡄨
󡄨
󡄨
1 󡄨
󡄨󡄨
󡄨󡄨 𝑖=0
󡄨󡄨1 + β„Žπ‘1 − β„Žπ‘1 πœ–π‘€2 󡄨󡄨󡄨
󡄨
󡄨
󡄨󡄨
󡄨󡄨
β„Žπ‘€1
1 − Vπ‘˜
󡄨󡄨
󡄨󡄨
󡄨
󡄨󡄨
= 󡄨󡄨
⋅
󡄨 󡄨
󡄨󡄨1 + β„Žπ‘1 − β„Žπ‘1 πœ–π‘€2 󡄨󡄨󡄨 󡄨󡄨󡄨 β„Ž (1 − 𝑐1 β„Žπœ–π‘€2 − πœ–π‘€2 ) 󡄨󡄨󡄨 (47)
󡄨
󡄨
⋅ 󡄨󡄨󡄨1 + β„Žπ‘1 − β„Žπ‘1 πœ–π‘€2 󡄨󡄨󡄨
󡄨󡄨
󡄨󡄨
1 − Vπ‘˜
󡄨
󡄨󡄨
󡄨󡄨 .
= 𝑀1 ⋅ 󡄨󡄨󡄨󡄨
󡄨󡄨 (1 − 𝑐1 β„Žπœ–π‘€2 − πœ–π‘€2 ) 󡄨󡄨󡄨
On the other hand, we deduce from the definition of 𝑐(πœ–)
that
|1 − 𝑐 (πœ–)|
󡄨󡄨
π‘˜ 󡄨󡄨
󡄨󡄨
1 + β„Žπ‘1 − β„Ž + β„Žπœ–π‘€2 − β„Žπ‘1 πœ–π‘€2 + β„Ž2 𝑐1 πœ–π‘€2 󡄨󡄨󡄨
󡄨
≥ 󡄨󡄨󡄨1 − (
) 󡄨󡄨󡄨 .
1 + β„Žπ‘1 − β„Žπ‘1 πœ–π‘€2
󡄨󡄨
󡄨󡄨
󡄨
󡄨
(48)
Hence,
𝑀 (1 + 𝑐1 β„Ž)
𝑀1
󡄨󡄨 ∗ 󡄨󡄨
󡄨󡄨𝑋 (πœ–)󡄨󡄨 ≤ 󡄨󡄨
󡄨󡄨 = 󡄨󡄨
󡄨,
1
−
𝑐
1
−
𝑐1 β„Žπœ–π‘€2 − πœ–π‘€2 󡄨󡄨󡄨
β„Žπœ–π‘€
−
πœ–π‘€
󡄨󡄨
󡄨󡄨
1
2
2 󡄨󡄨
(49)
which is identical to the inequality (44).
If we substitute πœ– = 0 in (49), we obtain (31) as expected.
Here, as well, the bound for 𝑋∗ (πœ–) is the same as that of the
periodic solution as β„Ž → 0.
We would like to obtain a relationship between the
dynamical approach study and stability analysis. We introduce a natural stability criterion for the differential equation
as well as any numerical method used to discretize it.
2.4. Conditional AN Stability and AN Stability. We consider
the problem of determining a criterion for some sort of
“controlled behaviour” of the solutions of the methods. We
adopt a linear stability criterion that is based on the scalar
test equation
π‘₯σΈ€  = π‘Ž (𝑑) π‘₯,
(50)
where π‘Ž(𝑑) ∈ C. If Re(π‘Ž(𝑑)) < 0 for all 𝑑 ∈ [𝛽1 , 𝛽2 ], then
π‘₯ (𝑑 + β„Ž) = 𝐾π‘₯ (𝑑) ,
|𝐾| ≤ 1
(51)
for all π‘₯ ∈ [𝛽1 , 𝛽2 ] and β„Ž > 0.
Definition 12. A numerical method is said to be conditionally
AN stable for some β„Ž > 0 if, when applied to the test equation
(50) with Re(π‘Ž(𝑑)) < 0, for all 𝑑,
Μƒ 󡄨󡄨󡄨
Μƒ (β„Ž) π‘₯𝑛 , 󡄨󡄨󡄨󡄨𝐾
(52)
π‘₯𝑛+1 = 𝐾
󡄨 (β„Ž)󡄨󡄨 ≤ 1
holds for all 𝑛 ∈ N.
Journal of Applied Mathematics
7
If this condition is satisfied for all β„Ž > 0, then the method
is AN stable.
For a discussion of AN stability, see Lambert [6] and
Stuart and Humphries [7]
The following simple result gives a condition under which
the linearized one-point collocation methods are conditionally AN stable.
Theorem 13. Assuming real π‘Ž(𝑑), the linearized one-point
collocation methods are conditionally AN stable if and only if
(1 − 2𝑐1 ) π‘Ž (𝑑) + 𝑐1 β„Žπ‘ŽσΈ€  (𝑑) ≥ −
2
β„Ž
(53)
and π‘Ž(𝑑) + 𝑐1 β„Žπ‘ŽσΈ€  (𝑑) ≤ 0.
Proof. Use the above test equation in (2). On the other hand,
if (2) is not satisfied, the method fails the stability criterion
and is not AN stable.
Examples:
(1) if π‘Ž(𝑑) = −1, then the linearized one-point collocation
methods are conditionally AN stable if and only if
β„Ž
(1 − 2𝑐1 ) ≤ 1.
2
(54)
It is easy to observe that the methods are AN stable if
𝑐1 ≥ 1/2,
(2) if π‘Ž(𝑑) = −1 + πœ–πœŒ(𝑑), where πœ–, 𝜌(𝑑) ∈ C, and
πœ–(𝜌(𝑑) + 𝑐1 β„ŽπœŒσΈ€  (𝑑)) ≤ 1, then the linearized one-point
collocation methods are conditionally AN stable if
and only if
2
−1 + πœ–πœŒ (𝑑) + 𝑐1 (β„Žπœ–πœŒσΈ€  (𝑑) + 2 − 2πœ–πœŒ (𝑑)) ≥ − .
β„Ž
(55)
We have proved the existence of a relationship between
the linear stability theory of the collocation methods and
the existence and asymptotic stability of periodic solutions,
identified via stroboscopic sampling. This relationship is
stated in the following theorem.
Theorem 14. Suppose that a linearized one-point collocation
method is used to solve a linear nonautonomous equation of the
form discussed in Sections 2.2 or 2.3 which has periodic coefficients and possesses a unique, asymptotically stable periodic
solution. Then, the following conditions are equivalent:
(a) the method is AN stable,
(b) the method yields the same dynamics as the differential
equation.
The last theorem is very significant, since it gives us a
bridge connecting standard stability theory with dynamical
systems. Naturally, we would like to find out if there is a
corresponding result for the nonlinear case, which we now
consider.
3. Nonlinear Case
We consider the nonlinear equation,
π‘₯σΈ€  = π‘Ž (𝑑) 𝑔 (π‘₯) + 𝑏 (𝑑) ,
for 𝑑 ≥ 0, π‘₯ (0) = π‘₯0 ,
(56)
where 𝑏(𝑑) is a 𝑇-periodic function of 𝑑 and 𝑔(π‘₯) is a 𝐢2
nonlinear function of π‘₯.
Massera [8] proved that if a nonlinear equation of the
form (56) has the uniqueness property with respect to the
initial conditions, the existence of a bounded solution implies
the existence of a 𝑇-periodic solution.
3.1. Linearized One-Point Collocation Methods. The linearized collocation methods, applied to (56), are given by
π‘₯𝑛+1 = π‘₯𝑛 +
β„ŽΜƒ
π‘Žπ‘› 𝑔 (π‘₯𝑛 )
β„ŽΜƒπ‘π‘›
β„Ž+
,
σΈ€ 
1 − β„Žπ‘Žπ‘› 𝑐1 𝑔 (π‘₯𝑛 )
1 − β„Žπ‘Žπ‘› 𝑐1 𝑔󸀠 (π‘₯𝑛 )
(57)
where ̃𝑏𝑛 = 𝑏𝑛 + 𝑐1 β„Žπ‘π‘›σΈ€  and π‘ŽΜƒπ‘› = π‘Žπ‘› + 𝑐1 β„Žπ‘Žπ‘›σΈ€  .
We perform a simplification on the third term of (57) that
takes the form of evaluating the derivative of 𝑔 at the starting
value, instead, at each step. The resulting method, that will
be referred to as a simplified linearized one-point collocation
method, is
π‘Žπ‘› 𝐺 (π‘₯𝑛 ; 𝑛) + 𝐻𝑛̃𝑏𝑛 ,
π‘₯𝑛+1 = π‘₯𝑛 + β„ŽΜƒ
(58)
where
𝐻𝑛 =
β„Ž
,
1 − β„Žπ‘Žπ‘› 𝑐1 𝑔󸀠 (π‘₯0 )
𝑔 (π‘₯)
.
𝐺 (π‘₯; 𝑛) =
1 − β„Žπ‘Žπ‘› 𝑐1 𝑔󸀠 (π‘₯)
(59)
3.2. The Dynamical Systems Approach. We would like to take
the dynamical systems approach and determine the conditions under which (58), applied to the differential equation
(56), yields the same dynamics as the continuous system.
We rewrite (58) as
β„ŽΜƒ
π‘Žπ‘› 𝐺 (π‘₯𝑛 ; 𝑛) = π‘₯𝑛+1 − π‘₯𝑛 − 𝐻𝑛̃𝑏𝑛 .
(60)
Inductively, we can show that
𝑛
𝑛
π‘Ÿ=0
π‘Ÿ=0
π‘₯𝑛+1 = π‘₯0 + ∑π»π‘ŸΜƒπ‘π‘Ÿ + β„Ž∑π‘ŽΜƒπ‘Ÿ 𝐺 (π‘₯π‘Ÿ ; π‘Ÿ) .
(61)
We choose an integer π‘˜ such that 𝑇 = β„Žπ‘˜. Sampling
stroboscopically in the iteration above, we get
π‘˜−1
π‘˜−1
π‘Ÿ=0
π‘Ÿ=0
π‘₯π‘˜ = π‘₯0 + ∑ π»π‘ŸΜƒπ‘π‘Ÿ + β„Ž ∑ π‘ŽΜƒπ‘Ÿ 𝐺 (π‘₯π‘Ÿ ; π‘Ÿ)
(62)
and associate this with the discrete system
π‘˜−1
π‘˜−1
π‘Ÿ=0
π‘Ÿ=0
Μƒπ‘Ÿ ; π‘Ÿ) ,
𝑋𝑛+1 = 𝑋𝑛 + ∑ π»π‘ŸΜƒπ‘π‘Ÿ + β„Ž ∑ π‘ŽΜƒπ‘Ÿ 𝐺 (𝑋
(63)
8
Journal of Applied Mathematics
where
Proof. Finding possible fixed points of (63), hence periodic
solutions of the methods, is the same as finding zeros of (67).
This is equivalent to solving the nonlinear system:
Μƒ0 = 𝑋𝑛 ,
𝑋
π‘Ÿ−1
π‘Ÿ−1
𝑖=0
𝑖=0
Μƒπ‘Ÿ = 𝑋𝑛 + ∑π»π‘ŸΜƒπ‘π‘Ÿ + β„Ž∑π‘ŽΜƒπ‘– 𝐺 (𝑋
̃𝑖 ; 𝑖)
𝑋
for π‘Ÿ = 1, 2, . . . , π‘˜ − 1.
(64)
The fixed points of (63) correspond to periodic solutions
of (56). Fixed points are points, 𝑋∗ , such that
π‘˜−1
π‘˜−1
π‘Ÿ=0
π‘Ÿ=0
Μƒπ‘Ÿ ; π‘Ÿ) = 0,
∑ π»π‘ŸΜƒπ‘π‘Ÿ + β„Ž ∑ π‘ŽΜƒπ‘Ÿ 𝐺 (𝑋
π‘Ž1 𝐺 (π‘₯Μƒ1 ; 1) − β„ŽΜƒ
π‘Ž2 𝐺 (π‘₯Μƒ2 ; 2) − ⋅ ⋅ ⋅
− β„ŽΜƒ
π‘Ž0 𝐺 (π‘₯∗ ; 0) − β„ŽΜƒ
− β„ŽΜƒ
π‘Žπ‘˜−1 𝐺 (π‘₯Μƒπ‘˜−1 ; π‘˜ − 1) − 𝐻0̃𝑏0 − 𝐻1̃𝑏1 − ⋅ ⋅ ⋅
− π»π‘˜−1Μƒπ‘π‘˜−1 = 0,
π‘Ž0 𝐺 (π‘₯∗ ; 0) − 𝐻0̃𝑏0 = 0,
π‘₯Μƒ1 − π‘₯∗ − β„ŽΜƒ
π‘Ž0 𝐺 (π‘₯∗ ; 0) − β„ŽΜƒ
π‘Ž1 𝐺 (π‘₯Μƒ1 ; 1) − 𝐻0̃𝑏0 − 𝐻1̃𝑏1 = 0,
π‘₯Μƒ2 − π‘₯∗ − β„ŽΜƒ
(65)
..
.
π‘Ž0 𝐺 (π‘₯∗ ; 0) − β„ŽΜƒ
π‘Ž1 𝐺 (π‘₯Μƒ1 ; 1) − β„ŽΜƒ
π‘Ž2 𝐺 (π‘₯Μƒ2 ; 2) − ⋅ ⋅ ⋅
π‘₯Μƒπ‘˜−1 − π‘₯∗ − β„ŽΜƒ
where
π‘Ÿ−1
π‘Ÿ−1
𝑖=0
𝑖=1
Μƒπ‘Ÿ = 𝑋∗ + ∑𝐻𝑖̃𝑏𝑖 + β„Ž∑π‘ŽΜƒπ‘– 𝐺 (𝑋
̃𝑖 ; 𝑖)
𝑋
− β„ŽΜƒ
π‘Žπ‘˜−2 𝐺 (π‘₯Μƒπ‘˜−2 ; π‘˜ − 2) − 𝐻0̃𝑏0 − 𝐻1̃𝑏1 − ⋅ ⋅ ⋅
for π‘Ÿ = 1, 2, . . . , π‘˜ − 1,
− π»π‘˜−2Μƒπ‘π‘˜−2 = 0.
(69)
(66)
and 𝑋0 = 𝑋∗ . Define the sequence of functions 𝐹1 (π‘₯),
𝐹2 (π‘₯), . . . by
The above system is of the form F(x) = 0, where F is a
nonlinear function of
π‘₯∗ = π‘₯Μƒ0
π‘₯Μƒ1
x = ( π‘₯Μƒ2 ) .
..
.
π‘˜−1
πΉπ‘˜ (π‘₯) = ∑ [Μƒ
π‘Žπ‘Ÿ β„ŽπΊ (π‘₯Μƒπ‘Ÿ ; π‘Ÿ) + π»π‘ŸΜƒπ‘π‘Ÿ ] ,
(67)
π‘Ÿ=0
(70)
π‘₯Μƒπ‘˜−1
where π‘₯Μƒ0 = π‘₯ and
π‘Ÿ−1
π‘Žπ‘– β„ŽπΊ (π‘₯̃𝑖 ; 𝑖) + 𝐻𝑖̃𝑏𝑖 ] ,
π‘₯Μƒπ‘Ÿ = π‘₯ + ∑ [Μƒ
(68)
To prove existence, it is sufficient to show that the
Jacobian matrix, J(F), of F is nonsingular. Now,
𝑖=0
for π‘Ÿ = 1, 2, . . . , π‘˜ − 1.
The theorem below gives conditions under which the
simplified linearized one-point collocation methods, applied
to (56), exhibit dynamical behaviour that is the same as the
differential equation.
Theorem 15. Assume that the differential equation (56) has a
unique solution. If
(i) 𝑔󸀠󸀠 (π‘₯)𝑔(π‘₯) ≤ 0,
(ii) 𝑐1 π‘ŽσΈ€  (𝑑) < −π‘Ž(𝑑)/β„Ž for all 𝑑,
π‘Ž(𝑑)|,
(iii) Max 𝐺󸀠 (π‘₯; 𝑑) ≤ 1/β„Ž|Μƒ
(iv) |πΉπ‘˜σΈ€  (π‘₯)| < 1 for all π‘˜ ∈ N, where πΉπ‘˜ (π‘₯) is given by (67),
then a simplified one-point collocation method has a periodic
solution for any π‘˜ = 𝑇/β„Ž; this solution is unique and asymptotically stable.
𝑑0
−1 + 𝑑0
(−1 + 𝑑0
J (F) = (−1 + 𝑑0
𝑑1
1
𝑑1
𝑑1
𝑑2
0
1
𝑑2
𝑑3
0
0
1
..
.
⋅ ⋅ ⋅ π‘‘π‘˜−2 π‘‘π‘˜−1
⋅⋅⋅
0
0
0
⋅⋅⋅
0 )
0
⋅⋅⋅
0 ) , (71)
(−1 + 𝑑0 𝑑1 𝑑2 ⋅ ⋅ ⋅ π‘‘π‘˜−2 π‘‘π‘˜−1
1 )
π‘Žπ‘– 𝐺󸀠 (π‘₯̃𝑖 ; 𝑖) for each 𝑖. We perform one elewhere 𝑑𝑖 = −β„ŽΜƒ
mentary row operation: row 1 → row 1–row π‘˜. The matrix
becomes
1
−1 + 𝑑0
(−1 + 𝑑0
(−1 + 𝑑0
0
1
𝑑1
𝑑1
0
0
1
𝑑2
0
0
0
1
..
.
⋅⋅⋅
⋅⋅⋅
0
0
0 −1 + π‘‘π‘˜−1
0
0
⋅⋅⋅
0
)
).
⋅⋅⋅
0
(−1 + 𝑑0 𝑑1 𝑑2 ⋅ ⋅ ⋅ π‘‘π‘˜−2 π‘‘π‘˜−1
1
)
It is easy to see that the above matrix is nonsingular.
(72)
Journal of Applied Mathematics
9
To prove uniqueness, let π‘₯1 and π‘₯2 be two fixed points of
(63). Then, from (67),
+ π»π‘˜−1Μƒπ‘π‘˜−1 = 0,
π‘Žπ‘˜−1 𝐺 (π‘₯2 + πΉπ‘˜−1 (π‘₯2 ) ; π‘˜ − 1)
πΉπ‘˜ (π‘₯2 ) = πΉπ‘˜−1 (π‘₯2 ) + β„ŽΜƒ
+ π»π‘˜−1Μƒπ‘π‘˜−1 = 0.
0.8
0.6
(73)
(74)
0.4
π‘₯π‘–π‘˜
π‘Žπ‘˜−1 𝐺 (π‘₯1 + πΉπ‘˜−1 (π‘₯1 ) ; π‘˜ − 1)
πΉπ‘˜ (π‘₯1 ) = πΉπ‘˜−1 (π‘₯1 ) + β„ŽΜƒ
1
−0.4
−0.6
σΈ€ 
σΈ€ 
(𝛼1 ) + β„ŽΜƒ
π‘Žπ‘˜−1 𝐺󸀠 (𝛼2 ; π‘˜ − 1) (1 + πΉπ‘˜−1
(𝛼1 ))}
(π‘₯1 − π‘₯2 ) {πΉπ‘˜−1
−0.8
= 0,
(76)
80
100
120
140
0.6
0.4
0.2
0
−0.2
−0.4
3.3. Numerical Experiments. Consider the nonlinear nonautonomous ODE
π‘₯ (0) = 1.
60
0.8
In the above equations, 𝛼1 is between π‘₯1 and π‘₯2 , and 𝛼2 is
between π‘₯1 + πΉπ‘˜−1 (π‘₯1 ) and π‘₯2 + πΉπ‘˜−1 (π‘₯2 ).
From the hypotheses of the theorem, we have that 𝐹󸀠 < 0,
σΈ€ 
π‘Žπ‘˜−1 𝐺󸀠 (𝛼2 ; π‘˜ − 1) ≥ 0. Thus,
𝐺 > 0, π‘ŽΜƒπ‘˜−1 < 0, and 1 ≥ 1 + β„ŽΜƒ
π‘₯1 = π‘₯2 .
The periodic solution is asymptotically stable since |1 +
𝐹󸀠 | < 1. This completes the proof.
π‘₯ (𝑑) = −π‘₯ + sin 𝑑,
40
1
π‘₯π‘–π‘˜
+ β„ŽΜƒ
π‘Žπ‘˜−1 𝐺󸀠 (𝛼2 ; π‘˜ − 1) } = 0.
20
Figure 5: Numerical results for (77) with 𝑐1 = 0. π‘˜ = 20.
which may alternatively be written as
σΈ€ 
(π‘₯1 − π‘₯2 ) {πΉπ‘˜−1
(𝛼1 ) (1 + β„ŽΜƒ
π‘Žπ‘˜−1 𝐺󸀠 (𝛼2 ; π‘˜ − 1))
0
𝑑
(75)
3
0
−0.2
Subtracting (74) from (73) and using the mean value theorem
gives
σΈ€ 
0.2
−0.6
(77)
−0.8
0
20
40
60
80
100
120
140
𝑑
The numerical results (stroboscopic sampling) of each of
the three cases are given in Figures 5, 6, and 7.
The methods give comparable results, but the implicit
midpoint converges faster to the periodic solutions than the
other two.
Figure 6: Numerical results for (77) with 𝑐1 = 1/2. π‘˜ = 20.
1
Definition 16. Let π‘₯(𝑑) and π‘₯(𝑑) be any two solutions of the
differential equation π‘₯σΈ€  = 𝑓(π‘₯, 𝑑), satisfying initial conditions
π‘₯(0) = πœ‚, π‘₯(0) = πœ‚, πœ‚ =ΜΈ πœ‚. Then, if
σ΅„©σ΅„©
σ΅„© σ΅„©
σ΅„©
(78)
σ΅„©σ΅„©π‘₯ (𝑑2 ) − π‘₯ (𝑑2 )σ΅„©σ΅„©σ΅„© ≤ σ΅„©σ΅„©σ΅„©π‘₯ (𝑑1 ) − π‘₯ (𝑑1 )σ΅„©σ΅„©σ΅„©
0.8
0.6
0.4
0.2
π‘₯π‘–π‘˜
3.4. Nonlinear Stability Theory. We wish to establish conditions under which numerical methods for the solution of (56)
behave in a “controlled” manner. We will view such controlled
behaviour in the system meaning that neighbouring solution
curves get closer and closer together as 𝑑 increases. This
concept, called contractivity, is discussed, for instance, in
Lambert [6]. As in the linear case, we will contrast the
conditions for stability with the existence and uniqueness of
a periodic solution in the numerical methods.
We briefly state the concepts of contractivity and conditional BN stability.
0
−0.2
−0.4
−0.6
−0.8
0
20
40
60
80
100
120
𝑑
Figure 7: Numerical results for (77) with 𝑐1 = 1. π‘˜ = 20.
140
10
Journal of Applied Mathematics
holds under the Rπ‘š norm β€– ⋅ β€– for all 𝑑1 , 𝑑2 such that 𝛽1 ≤ 𝑑1 ≤
𝑑2 ≤ 𝛽2 , the solutions of the system are said to be contractive
in [𝛽1 , 𝛽2 ].
The discrete analog of the above definition is given below.
Definition 17. Let {π‘₯𝑛 } and {π‘₯𝑛 } be two numerical solutions
generated by a numerical method with different starting
values. Then, if
σ΅„© σ΅„©
σ΅„©
σ΅„©σ΅„©
σ΅„©σ΅„©π‘₯𝑛+1 − π‘₯𝑛+1 σ΅„©σ΅„©σ΅„© ≤ σ΅„©σ΅„©σ΅„©π‘₯𝑛 − π‘₯𝑛 σ΅„©σ΅„©σ΅„© ,
0 ≤ 𝑛 ≤ 𝑁,
(79)
Subtracting (84) from (83) and using the mean value
theorem gives
󡄨
󡄨 󡄨
󡄨 󡄨
󡄨󡄨
π‘Žπ‘› 𝐺󸀠 (πœ‰π‘› ; 𝑛)󡄨󡄨󡄨󡄨 ,
󡄨󡄨π‘₯𝑛+1 − π‘₯𝑛+1 󡄨󡄨󡄨 = 󡄨󡄨󡄨(π‘₯𝑛 − π‘₯𝑛 )󡄨󡄨󡄨 ⋅ 󡄨󡄨󡄨󡄨1 + β„ŽΜƒ
where πœ‰π‘› lies between π‘₯𝑛 and π‘₯𝑛 .
The solutions generated by the methods are contractive if
and only if |1 + β„ŽΜƒ
π‘Žπ‘› 𝐺󸀠 (πœ‰π‘› ; 𝑛)| ≤ 1. Assuming π‘ŽΜƒ(𝑑) < 0 for all 𝑑
σΈ€ 
and 𝐺 (π‘₯; 𝑑) ≥ 0 for all π‘₯ and 𝑑, the methods are conditionally
BN stable if and only if
𝐺󸀠 (π‘₯; 𝑑) ≤
the numerical solutions are said to be contractive for 𝑛 ∈
[0, 𝑁].
Definition 18. The system π‘₯σΈ€  = 𝑓(π‘₯, 𝑑) is dissipative in [𝛽1 , 𝛽2 ]
if
βŸ¨π‘“ (π‘₯, 𝑑) − 𝑓 (π‘₯, 𝑑) , π‘₯ − π‘₯⟩ ≤ 0
2
.
β„ŽΜƒ
π‘Ž (𝑑)
(86)
Remark 20. If we let 𝑔(π‘₯) = π‘₯ (the linear case), then
condition (86) collapses to (53), which is the condition for
conditional AN stability.
(80)
holds for all π‘₯, π‘₯ ∈ R and for all 𝑑 ∈ [𝛽1 , 𝛽2 ].
It is easy to show that the solutions of a dissipative
system are contractive under the norm induced by the inner
product in (80). It is desirable that a numerical method,
used with fixed stepsize β„Ž > 0 to solve a dissipative system,
gives contractive solutions. This brings us to the concepts of
conditional BN stability, BN stability and nonlinear nonautonomous stability criteria.
Definition 19. If a numerical method, applied with fixed
steplength β„Ž > 0 to (56) satisfying (80), generates contractive
solutions, the method is said to be conditionally BN stable.
If the method generates contractive solutions when applied
with any β„Ž > 0, then it it is BN stable (Lambert [6]).
The concepts of AN and BN stability are equivalent for
the nonconfluent Runge-Kutta methods.
To determine the conditional BN stability of the methods,
we use the scalar test system
π‘₯σΈ€  = π‘Ž (𝑑) 𝑔 (π‘₯) ,
(85)
(81)
where, as before, π‘Ž(𝑑) ∈ C and 𝑔(π‘₯) ∈ 𝐢1 . This system is
dissipative if
π‘Ž (𝑑) βŸ¨π‘” (π‘₯) − 𝑔 (π‘₯) , π‘₯ − π‘₯⟩ = π‘Ž (𝑑) 𝑔󸀠 (πœ‰) |π‘₯ − π‘₯|2 ≤ 0, (82)
where πœ‰ lies between π‘₯ and π‘₯ and ⟨⋅, ⋅⟩ is an inner product in
R. The condition is satisfied if π‘Ž(𝑑) ≤ 0 for all 𝑑 and 𝑔󸀠 (π‘₯) ≥ 0
for all π‘₯. Therefore, the existence and uniqueness of a periodic
solution in (56) is a sufficient condition for the dissipativity of
the system.
Now, we determine the conditions under which the
linearized one-point collocation methods are conditionally
BN stable. Applying the methods to the test system (81) for
two different initial conditions gives
π‘Žπ‘› 𝐺 (π‘₯𝑛 ; 𝑛) ,
π‘₯𝑛+1 = π‘₯𝑛 + β„ŽΜƒ
(83)
π‘Žπ‘› 𝐺 (π‘₯𝑛 ; 𝑛) .
π‘₯𝑛+1 = π‘₯𝑛 + β„ŽΜƒ
(84)
3.5. Discussion. We have, in Theorem 15, established sufficient conditions for the simplified linearized one-point collocation methods to exhibit the same dynamics as the nonlinear
nonautonomous ODE. Conditions (i) and (ii) are required
for conditional BN stability as well, but condition (iii) is not
necessary for the existence of a unique, asymptotically stable
solution. In fact, we can come to the same conclusion as in
Theorem 15 if, in (75),
π‘Ž (𝑑) 𝐺󸀠 (π‘₯; 𝑑)] + β„ŽΜƒ
π‘Ž (𝑑) 𝐺󸀠 (π‘₯; 𝑑) =ΜΈ 0
𝐹󸀠 (π‘₯) [1 + β„ŽΜƒ
(87)
for all π‘₯ and 𝑑. The following theorem shows that condition
(87) is satisfied if the method is conditionally BN stable.
Theorem 21. Suppose that a simplified linearized one-point
collocation method is used to solve a nonlinear nonautonomous
equation of the form (56) with periodic coefficients which has
a unique, asymptotically stable periodic solution. Then, the
method yields a unique periodic solution that is asymptotically
stable if it is conditionally BN stable.
Proof. Assume the conditional BN stability. To prove the
theorem, it is sufficient to establish condition (87). From
conditional BN stability, we have −1 ≤ 1 + β„ŽΜƒ
π‘ŽπΊσΈ€  ≤ 1. Recall
σΈ€ 
that −1 ≤ 𝐹 < 0 from condition (iv). Therefore,
π‘ŽπΊσΈ€  ≤ 𝐹󸀠 (1 + β„ŽΜƒ
π‘ŽπΊσΈ€  ) + β„ŽΜƒ
π‘ŽπΊσΈ€  ≤ −𝐹󸀠 + β„ŽΜƒ
π‘ŽπΊσΈ€  .
𝐹󸀠 + β„ŽΜƒ
(88)
Since 0 < −𝐹󸀠 ≤ 1, if −1 ≤ 1 + β„ŽΜƒ
π‘ŽπΊσΈ€  < 0,
π‘ŽπΊσΈ€  ) + β„ŽΜƒ
π‘ŽπΊσΈ€  ≤ 1 + β„ŽΜƒ
π‘ŽπΊσΈ€  < 0.
𝐹󸀠 (1 + β„ŽΜƒ
(89)
If 0 ≤ 1 + β„ŽΜƒ
π‘ŽπΊσΈ€  ≤ 1,
π‘ŽπΊσΈ€  ≤ 𝐹󸀠 (1 + β„ŽΜƒ
π‘ŽπΊσΈ€  ) + β„ŽΜƒ
π‘ŽπΊσΈ€  ≤ β„ŽΜƒ
π‘ŽπΊσΈ€  < 0.
𝐹󸀠 + β„ŽΜƒ
(90)
This proves the theorem.
Remark 22. The above theorem is somewhat similar to
Theorem 14 (in the linear case) if we replace the concept of
conditional BN stability by conditional AN stability.
Journal of Applied Mathematics
4. Conclusion
We already knew that numerical methods can introduce spurious behaviours into the solution for autonomous equations.
By concentrating on linear and nonlinear nonautonomous
equations with unique periodic solutions and discretizing
them using one-point collocation methods, we were interested in the existence of periodic solutions in the numerical
methods.
We found that the results obtained from the dynamical
systems approach are closely linked to those that are imposed
by standard stability analysis. It has been shown that for
linear and nonlinear nonautonomous differential equations
of the form considered in this chapter, there is a relationship
between conditional AN or BN stability of a one-point collocation method and the method yielding the same dynamical
behaviour as the differential equation under consideration.
References
[1] A. Iserles, “Stability and dynamics of numerical methods for
nonlinear ordinary differential equations,” IMA Journal of
Numerical Analysis, vol. 10, no. 1, pp. 1–30, 1990.
[2] A. Stuart, “Linear instability implies spurious periodic solutions,” IMA Journal of Numerical Analysis, vol. 9, no. 4, pp. 465–
486, 1989.
[3] J. K. Hale and H. Koçak, Dynamics and Bifurcations, vol. 3,
Springer, New York, NY, USA, 1991.
[4] A. Foster and M. Khumalo, “Transformation of local bifurcations under collocation methods,” Journal of the Korean
Mathematical Society, vol. 48, no. 6, pp. 1101–1123, 2011.
[5] E. Hairer, S. P. Nørsett, and G. Wanner, Solving Ordinary
Differential Equations. I, vol. 8, Springer, Berlin, Germany, 2nd
edition, 1993.
[6] J. D. Lambert, Numerical Methods for Ordinary Differential
Systems, John Wiley & Sons, New York, NY, USA, 1991.
[7] A. M. Stuart and A. R. Humphries, Dynamical Systems and
Numerical Analysis, vol. 2, Cambridge University Press, New
York, NY, USA, 1996.
[8] J. L. Massera, “The existence of periodic solutions of systems of
differential equations,” Duke Mathematical Journal, vol. 17, pp.
457–475, 1950.
11
Advances in
Operations Research
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Advances in
Decision Sciences
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Mathematical Problems
in Engineering
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Journal of
Algebra
Hindawi Publishing Corporation
http://www.hindawi.com
Probability and Statistics
Volume 2014
The Scientific
World Journal
Hindawi Publishing Corporation
http://www.hindawi.com
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
International Journal of
Differential Equations
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Volume 2014
Submit your manuscripts at
http://www.hindawi.com
International Journal of
Advances in
Combinatorics
Hindawi Publishing Corporation
http://www.hindawi.com
Mathematical Physics
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Journal of
Complex Analysis
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
International
Journal of
Mathematics and
Mathematical
Sciences
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com
Stochastic Analysis
Abstract and
Applied Analysis
Hindawi Publishing Corporation
http://www.hindawi.com
Hindawi Publishing Corporation
http://www.hindawi.com
International Journal of
Mathematics
Volume 2014
Volume 2014
Discrete Dynamics in
Nature and Society
Volume 2014
Volume 2014
Journal of
Journal of
Discrete Mathematics
Journal of
Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com
Applied Mathematics
Journal of
Function Spaces
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Optimization
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
Download