Volume 2009, Article ID 923809,29pages doi:10.1155/2009/923809
Research Article
Simple-Zero and Double-Zero
Singularities of a Kaldor-Kalecki Model of Business Cycles with Delay
Xiaoqin P. Wu
Department of Mathematics, Computer & Information Sciences, Mississippi Valley State University, Itta Bena, MS 38941, USA
Correspondence should be addressed to Xiaoqin P. Wu,xpaul wu@yahoo.com Received 12 August 2009; Accepted 2 November 2009
Recommended by Xue-Zhong He
We study the Kaldor-Kalecki model of business cycles with delay in both the gross product and the capital stock. Simple-zero and double-zero singularities are investigated when bifurcation parameters change near certain critical values. By performing center manifold reduction, the normal forms on the center manifold are derived to obtain the bifurcation diagrams of the model such as Hopf, homoclinic and double limit cycle bifurcations. Some examples are given to confirm the theoretical results.
Copyrightq2009 Xiaoqin P. Wu. 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.
1. Introduction
In the last decade, the study of delayed differential equations that arose in business cycles has received much attention. The first model of business cycles can be traced back to Kaldor 1 who used a system of ordinary differential equations to study business cycles in 1940 by proposing nonlinear investment and saving functions so that the system may have cyclic behaviors or limit cycles, which are important from the point of view of economics. Kalecki 2introduced the idea that there is a time delay for investment before a business decision.
Krawiec and Szydłowski3–5incorporated the idea of Kalecki into the model of Kaldor by proposing the following Kaldor-Kalecki model of business cycles:
dYt
dt αIYt, Kt−SYt, Kt, dKt
dt IYt−τ, Kt−qKt,
1.1
whereY is the gross product, K is the capital stock,α > 0 is the adjustment coefficient in the goods market, q ∈ 0,1 is the depreciation rate of capital stock, IY, Kand SY, K
are investment and saving functions, and τ ≥ 0 is a time lag representing delay for the investment due to the past investment decision. This model has been studied extensively by many authors; see 6–11. Several authors also discussed similar models 12–14 and established the existence of limit cycles.
Considering that past investment decisions6also influence the change in the capital stock, Kaddar and Talibi Alaoui15extended the model1.1by imposing delays in both the gross product and capital stock. Thus adding the same delay to the capital stock K in the investment function IY, K of the second equation of Sys.1.1leads to the following Kaldor-Kalecki model of business cycles:
dYt
dt αIYt, Kt−SYt, Kt, dKt
dt IYt−τ, Kt−τ−qKt.
1.2
As in3; also see10,16,17, using the following saving and investment functionsS andI, respectively,
SY, K γY, IY, K IY−βK, 1.3
whereβ >0 andγ∈0,1are constants, we obtain the following system:
dYt dt α
IYt−βKt−γYt , dKt
dt IYt−τ−βKt−τ−qKt.
1.4
Kaddar and Talibi Alaoui15studied the characteristic equation of the linear part of Sys.
1.4at an equilibrium point and used the delayτas a bifurcation parameter to show that the Hopf bifurcation may occur under some conditions asτpasses some critical values. However, they did not obtain the stability of the bifurcating limit cycles and the direction of the Hopf bifurcation. Wang and Wu18further studied Sys.1.4and gave a more detailed discussion of the distribution of the eigenvalues of the characteristic equation which has a pair of purely imaginary roots. They derived the normal forms on the center manifold for sys.1.4to give the direction of the Hopf bifurcation and the stability of the bifurcating limit cycles for some critical values ofτ.
However, under certain conditions, the characteristic equation of the linear part of Sys.1.4may have a simple-zero root, a double-zero root, or a simple zero root and a pair of purely imaginary roots. In this paper, simple-zerofoldand double-zeroBogdanov-Takens singularities for Sys.1.4and their corresponding dynamical behaviors are investigated by usingkandτ as bifurcation parameterswherekis defined inSection 2. The discussion of zero-Hopf singularity will be addressed in a coming paper.
The rest of this manuscript is organized as follows. InSection 2, a detailed presentation is given for the distribution of eigenvalues of the linear part of Sys.1.4at an equilibrium point in thek, τ-parameter space. InSection 3, the theory of center manifold reduction for general delayed differential equationsDDEsis briefly introduced. In Sections4and5, center
manifold reduction is performed for Sys.1.4; and hence, the normal forms for simple-zero and double-zero singularities are obtained on the center manifold, respectively. InSection 6, the normal forms for the double-zero singularity are used to predict the bifurcation diagrams such as Hopf, homoclinic, and double limit cycle bifurcations for the original Sys. of1.4.
Finally in Section 7, some numerical simulations are presented to confirm the theoretical results.
2. Distribution of Eigenvalues
Throughout the rest of this paper, we assume that
α, β >0, q, γ∈0,1, andIsis a nonlinearC4 function, 2.1 and thatY∗, K∗is an equilibrium point of Sys.1.4. LetI∗IY∗,u1Y−Y∗,u2K−K∗, andis Is Y∗−I∗. Then Sys.1.4can be transformed as
du1t dt α
iu1t−βu2t−γu1t , du2t
dt iu1t−τ−βu2t−τ−qu2t.
2.2
Let the Taylor expansion ofiat 0 be
iu ku i2u2 i3u3 O
|u|4
, 2.3
where
ki0 IY∗, i2 1
2i0 1
2IY∗, i3 1
3!i0 1
3!IY∗. 2.4 The linear part of Sys.2.2at0,0is
du1t
dt α
k−γ
u1t−βu2t , du2t
dt ku1t−τ−βu2t−τ−qu2t,
2.5
and the corresponding characteristic equation is Δλ≡λ2 Aλ B
βλ C
e−λτ0, 2.6
where
Aq−α k−γ
, B−αq
k−γ
, Cαβγ. 2.7
Forτ 0,2.6becomes
λ2 A β
λ B C0. 2.8
Define
k∗ βγ
q γ, k∗∗ q β
α γ. 2.9
Theorem 2.1. Letτ 0. Ifk <min{k∗, k∗∗}, then all roots of 2.8have negative real parts, and hence Y∗, K∗ is asymptotically stable. Ifk > min{k∗, k∗∗}, then2.8has a positive root and a negative root, and hence,Y∗, K∗is unstable.
Now assumeτ >0. ClearlyΔ0 0 if and only ifkk∗. Next we always assume that kk∗. It is easy to attain
Δλ 2λ q−αβγ
q βe−λτ−
βλ C τe−λτ, Δλ 2−2βτe−λτ βτ2λe−λτ Cτ2e−λτ.
2.10
Defineτ∗ q2 qβ−αβγ/αβγq. Then we have that,
Δ0 αβγ
q τ∗−τ, Δ0
ττ∗ q4−β2q2 α2β2γ2
αβγq2 . 2.11
Define
fx x2 βx−αβγ, gx x2−β2x α2β2γ2. 2.12
Hence iffq ≤0,τ∗ ≤0, and henceΔ0< 0, and iffq> 0,τ∗ > 0, and henceΔ0 0 if and only ifτ τ∗. AlsoΔ0|ττ∗/0 if and only ifgq2/0. Thus we obtain the following result.
Lemma 2.2. Suppose thatkk∗. Then the following are considered.
iIfτ∗≤0, then2.6has a simple root 0 for allτ >0.
iiLetτ∗>0. Then the following are given.
aEquation2.6has a simple root 0 if and only ifτ /τ∗,
bEquation2.6has a double root 0 if and only ifττ∗andgq2/0.
Letωi ω > 0be a purely imaginary root of2.6. After plugging it into 2.6and separating the real and imaginary parts, we have that
ω2 αβγ αβγcosωτ βωsinωτ, q2−αβγ
q ωαβγsinωτ−βωcosωτ. 2.13
Adding squares of two equations yields
ω2 g q2
q2 0. 2.14
Then2.14 has a nonzero solution if and only if gq2 < 0 and does not have a nonzero solution if and only ifgq2≥0. Ifgq2<0, from2.14, we solveωas follows:
ωω0 ≡ 1 q −g
q2
, 2.15
and from2.13, we solve cosω0τ, sinω0τas:
cosω0τ −q2ω20 αβγω20 qαγ
αβγ ω20 qβ
α2γ2 ω20 ≡a, sinω0τ q2αγω0−α2βγ2ω0 qαβγω0 qω30
qβ
α2γ2 ω20 ≡b.
2.16
Define
δ
⎧⎨
⎩
arccosa, if b≥0,
2π−arccosa, if b <0. 2.17
From2.16, we obtain
τ τj≡ 1 ω0
δ 2jπ
, j 0,1,2, . . . . 2.18
Clearly ifβ >2αγ, thengx 0 has two positive roots, and ifβ≤2αγ, thengx≥0.
Now, underkk∗, we impose the following conditions:
H1β≤2αγ,τ∗≤0, H2β≤2αγ,τ∗>0,τ /τ∗, H3β≤2αγ,τ∗>0,ττ∗,
H4β >2αγ,τ∗>0,τ /τ∗, gq2≥0,
H5β >2αγ,τ∗>0,τ /τ∗, gq2<0, H6β >2αγ,τ∗>0,ττ∗, gq2≥0, H7β >2αγ,τ∗>0,ττ∗, gq2<0.
Based onLemma 2.2, we have the following result.
Lemma 2.3. Suppose thatkk∗and 0< q <1. Then the following are obtained.
iUnder one of the conditions (H1), (H2), and (H4),2.6has a simple zero root and does not have other roots in the imaginary axis.
iiUnder the condition (H5),2.6has a simple zero root and a pair of purely imaginary roots
±ω0iin the imaginary axis ifτ τj,j 0,1,2, . . . .
iiiUnder one of the conditions (H3) and (H6), then2.6has a double root 0 and does not have other roots in the imaginary axis.
ivUnder the condition (H7),2.6has a double zero root and a pair of purely imaginary roots
±ω0iin the imaginary axis ifτ∗τjfor somej.
Now we use the roots offx 0,gx 0 to give a more detailed discussion for the roots of2.6. Define
q0 1 2
−β β2 4αβγ
, q1 1
2
β2− β4−4α2β2γ2
, q2 1
2
β2 β4−4α2β2γ2
.
2.19
Clearly q0 is the positive root offx 0 and q1,q2 are two positive roots of gx 0 if β > 2αγ. Note thatfx ≤0 if 0 < x ≤ q0,andfx > 0 ifx > q0,gx≥ 0 if 0 < x ≤ q1, or x≥ q2,thengx< 0 ifq1 < x < q2. Also note that as well as ifβ >2αγ,q02 < q1. In fact it is based on the following calculation:
q1−q02 1 2
β2− β4−4α2β2γ2
−1 4
−β β2 4αβγ 2
β
2 β2 4αβγ − β2−4α2γ2−2αγ
2αβγ
β− β2−4α2γ2 β2 4αβγ β2−4α2γ2 2αγ
>0.
2.20
Thus for β > 2αγ, we always have q0 < √q1 < √q2. Noting that q ∈ 0,1, we have the following result.
Lemma 2.4. Letβ >2αγ. Then the following are given.
iSuppose thatq0 ≥1. Then for 0< q <1, then2.6has a simple zero root and does not have roots in the imaginary axis.
iiSuppose thatq0<1≤ √q1<√q2. If 0< q≤q0, then2.6has a simple zero root and does not have roots in the imaginary axis. And ifq0 < q <1,2.6has a double zero root and does not have roots in the imaginary axis.
iiiSuppose thatq0<√q1<1<√q2. If 0< q≤q0, then2.6has a simple zero root and does not have roots in the imaginary axis. Ifq0< q≤ √q1, then2.6has a double zero root and does not have roots in the imaginary axis. And if√q1< q <1, then2.6has a double zero root and has a pair of purely imaginary roots.
ivSuppose that√q2 ≥1. Then if 0< q ≤q0, then2.6has a simple zero root and does not have roots in the imaginary axis. Ifq0 < q ≤ √q1, then2.6has a double zero root and does not have roots in the imaginary axis. If√q1 < q <√q2, then2.6has a double zero root and has a pair of purely imaginary roots whenτ∗τjfor somej. And if√q2≤q <1, 2.6has a double zero root and does not have a pair of purely imaginary roots.
Defineλτ στ iωτto be the root of2.6such thatστj 0 andωτj ω0. Then we have the following result.
Lemma 2.5. Suppose thatkk∗andgq2<0. Thenστj>0.
Proof. Differentiating2.6with respect toτ yields dλ
dτ −1
2λ q−α k−γ
eλτ β λβ
λ αγ −τ
λ, 2.21
and a simple calculation gives
Re dλ
dτ −1
ττj
α2β2γ2 q2
−β2 q2 2ω20 β2q2
α2γ2 ω20 q2β2−α2β2γ2−q4 β2q2
α2γ2 ω20 , 2.22
which gives
Sign Re dλ
dτ −1
ττj
Sign
−g q2
1, 2.23
thus completing the proof.
Next we discuss the distribution of other roots of2.6. We need the following lemma due to Ruan and Wei19.
Lemma 2.6. Consider the exponential polynomial P
λ, e−λτ
pλ qλe−λτ, 2.24
wherep,qare real polynomials such that degq < degpandτ ≥ 0. As τ varies, the sum of the order of zeros ofPλ, e−λτon the open right half-plane can change only if a zero appears on or crosses the imaginary axis.
Lemma 2.7. Letkk∗andτ >0. Then, the following are obtained.
iIfq > q0, then all roots of 2.6except 0 and purely imaginary roots have negative real parts,
iiIf 0< q≤q0, then2.6has at least one positive root.
Proof. Note that, forτ 0, ifq > q0orq2 qβ > αβγ,Δλ 0 has a zero root and a negative root. Using Lemmas2.2and2.6, we obtain claimi. Forτ0,Δλ 0 has a zero root and a positive root if 0< q≤q0orq2 qβ≤αβγ. Forτ >0, let
fλ Δλ
λ λ A βe−λτ B Ce−λτ
λ . 2.25
Also noting thatB C0 whenkk∗, we have that
λ→lim0fλ A β−Cτ 1 q
q2 qβ−αβγ
−αβγτ
<0, 2.26
and limλ→ ∞fλ ∞. This proves the second part of the lemma and completes the proof of the lemma.
3. Center Manifold Reduction
In this section, we briefly summarize the theory of center manifold reduction for general DDEs. The material is mainly taken from20,21. Consider the following DDE:
dx dt L
μ xt G
xt, μ
, 3.1
wherex∈C−τ,0,Rn,μ∈Rp. This equation is equivalent to dx
dt L μ
xt G xt, μ
, dμ
dt 0, 3.2
which can be written as
dX
dt LXt FXt, 3.3
whereX x, μT,FXt Gxt,0T, andLdiagL,0. DefineX ∈C:C−τ,0,Rn p with supreme norm andXt∈Cis defined byXtθ Xt θ,−τ≤θ≤0;L:C → LRn pis
a bounded linear operator; andF:C → Cis aCk k≥2function withF0 0,DF0 0.
Consider the following linear system:
Xt ˙ LXt. 3.4
Since L is a bounded linear operator, then L can be represented by a Riemann-Stieltjes integral
Lϕ 0
−τdηθϕθ, ∀ϕ∈C, 3.5
by the Riesz representation theorem, where ηθ θ ∈ −τ,0 is an n p × n p matrix function of bounded variation. LetA0 be the infinitesimal generator for the solution semigroup defined by Sys.3.4such that
A0ϕϕ,˙ DA0
ϕ∈C1−τ,0,Rn p: ˙ϕ0 0
−τdηθϕθ
. 3.6
Define the bilinear form betweenCandC∗ C0, τ,Rn p∗ whereRn p∗ is the space of all rown p-vectorsby
ψ, ϕ
ψ0ϕ0− 0
−τ
θ
0
ψξ−θdηθϕξdξ, ∀ψ∈C∗, ∀ϕ∈C. 3.7
The adjoint ofA0is defined byA∗0as
A0∗ψ −ψ,˙ D A0∗
ϕ∈C1
0, τ,Rn p∗
: ˙ψ0 − 0
−τψ−θdηθ
. 3.8
In our setting,3.3hasptrivial components. Assume that the characteristic equation of3.3 has eigenvalue zero with multiplicity 2pand all other eigenvalues have negative real parts.
ThenLhas a generalized eigenspaceP which is invariant under the flow3.4. LetP∗be the space adjoint withP inC∗. ThenCcan be decomposed asC P ⊕QwhereQ {ϕ ∈ C : ψ, ϕ0,∀ψ∈P∗}. Choose the basesΦandΨforP andP∗, respectively, such that
Ψ,ΦI, Φ ΦJ,˙ Ψ ˙ −JΨ, 3.9
whereJis Jordan matrix associated with the eigenvalue 0.
To consider Sys.3.3, we need to enlarge the spaceCto the followingBC:
BC
ϕ:−τ,0 → Rn p:ϕis continuous on −τ,0, ∃lim
θ→0−ϕθ∈Rn p
. 3.10
The elements ofBCcan be expressed asψϕ X0αwithϕ∈C, α∈Rn p,and
X0θ
⎧⎨
⎩
0, −τ ≤θ <0,
I, θ0, 3.11
whereIis then×nidentity matrix. Define the projectionπ:BC → Pby π
ϕ X0α Φ
Ψ, ϕ
Ψ0α
. 3.12
Then the enlarged phase spaceBCcan be decomposed asBC P ⊕kerπ.LetX Φx y withx∈R2pandy∈Q1{ϕ∈Q: ˙ϕ∈C}. Then3.3can be decomposed as
˙
xJx Ψ0F
Φx y ,
˙
yAQ1y I−πX0F
Ψx y
, 3.13
whereAis an extension of the infinitesimal generatorA0fromC1toBC, defined by
A0ϕϕ˙ X0
Lϕ−ϕ0˙
⎧⎪
⎪⎨
⎪⎪
⎩
˙
ϕ, −1≤θ <0, 0
−τdηtϕt, θ0, 3.14
forϕ∈C1and its adjoint byA∗is defined by
A∗ψ
⎧⎪
⎪⎨
⎪⎪
⎩
−ψ,˙ 0< s≤θ, 0
−τψ−θdηθ, s0,
3.15
forψ∈C1∗. LetFv
j≥21/j!Fjv. Then Sys.3.13becomes
˙
xJx
j≥2
1 j!fj1
x, y ,
˙
yAQ1y
j≥2
1 j!fj2
x, y ,
3.16
where
fj1 x, y
Ψ0Fj
Φx y
, fj2 x, y
I−πX0Fj
Φx y
. 3.17
On the center manifold,3.16can be approximated as
˙
xJx
j≥2
1
j!fj1x,0. 3.18
4. Simple-Zero Singularity
In this section, we assume that the conditionH2holds. From the definition ofτ∗, we know thatτ∗>0 if and only ifq > q0. ThereforeH2is equivalent to
kk∗, q > q0, τ >0, τ /τ∗. 4.1 From ii of Lemma 2.4 and ii of Lemma 2.7, we know that, at 0,0, the characteristic equation of the linear part of Sys. 2.5 has a simple zero root and the rest of roots have negative parts. We treatkas a bifurcation parameter neark∗.
SetC : C−τ,0,R3,C∗ : C0, τ,R3∗. Letμ k−k∗. Then Sys.2.5 can be rewritten as
du1 dt α
βγ
q u10−βu20 μu10 i2u210 i0u31t
Oμ|u|2 |u|4 , du2
dt k∗u1−τ−qu2t μu1−τ−βu2−τ i2u21−τ i3u31−τ Oμ|u|2 |u|4 , dμ
dt 0.
4.2
The linearization of Sys.4.2at0,0,0is du1
dt αβγ
q u10−αβu20, du2
dt k∗u1−τ−qu20−βu2−τ, dμ
dt 0.
4.3
Letηθ Aδθ Bδθ τwhere
A
⎛
⎜⎜
⎜⎝ αβγ
q −αβ 0
0 −q 0
0 0 0
⎞
⎟⎟
⎟⎠, B
⎛
⎜⎜
⎝
0 0 0
k∗ −β 0
0 0 0
⎞
⎟⎟
⎠. 4.4
LetX u1, u2, μTand
FXt
⎛
⎜⎜
⎜⎝
αμu10 αi2u210 αi3u310 Oμ|u|2 |u|4 μu1−τ i2u21−τ i3u31−τ Oμ|u|2 |u|4
0
⎞
⎟⎟
⎟⎠. 4.5
Define
Lϕ 0
−τdηθϕθ, ∀ϕ∈C. 4.6
Then Sys.4.2becomes
Xt ˙ LXt FXt. 4.7
From3.7, the bilinear form can be expressed as ψ, ϕ
ψ0ϕ0
0
−τψξ τBϕξdξ. 4.8
It is not hard to see that the infinitesimal generatorA:C1 → BCis given by
Aϕϕ˙ X0
Lϕ−ϕ0˙
⎧⎨
⎩
˙
ϕ, −τ≤θ <0,
Aϕ0 Bϕ−τ, θ0, 4.9
forϕ∈C1and its adjointA∗by
A∗ψ
⎧⎨
⎩
−ψ,˙ 0< s≤θ,
ψ0A ψτB, s0, 4.10
forψ∈C1∗.
Next we obtain the bases for the center spaceP and its adjoint spaceP∗, respectively.
LetAϕ0 forϕ∈C1, that is,
ϕθ ˙ 0 for −τ ≤θ <0, Aϕ0 Bϕ−τ 0 for θ0. 4.11
then we know thatϕis a constant vectora1, a2, a3T ∈R3\ {0}such that
A Ba1, a2, a3T 0. 4.12
Then we have two linearly independent solutions ϕ1 q, γ,0T,ϕ2 0,0,1T which are bases for the center spaceP. LetΦ ϕ1, ϕ2.
Similarly, letA∗ψ 0 forψ∈C1∗, that is,
−ψs ˙ 0 for 0< s≤τ, ψ0A ψτB0 fors0, 4.13
then we know thatψis a constant vectorb1, b2, b3∈R3∗\ {0}such that
b1, b2, b3A B 0. 4.14
From this we have two linearly independent solutionsψ1 −q β, αβ,0andψ2 0,0,1 which are bases for the center spaceP∗. LetΨ rψ1, ψ2Twithrbeing determined such that ψ1, ϕ11. In fact
r 1
qαβγτ−τ∗. 4.15
Clearlyr is well defined sinceτ−τ∗/0. It is not hard to check that ˙Φ ΦJ, ˙Ψ −JΨand Ψ,ΦI, whereJ0 0
0 0
.
Letu Φx y. Then Sys.4.2can be decomposed as
˙ x ΨF
Φx y ,
˙
yAQ1y I−πX0F
Φx y
. 4.16
Writex x1, μ. Note that
Ψ0FΦx
#−rαq2
μx1 qi2x12 q2i3x31 0
$
h.o.t.. 4.17
Here h.o.t. represents higher-order terms. Thus, for sufficiently small μ, on the center manifold, ifi2/0, then Sys.4.2becomes
˙
x1−rαq2μx1−rαq3i2x12 h.o.t.,
˙
μ0. 4.18
Ifi20 andi3/0, then Sys.4.2can be transformed into the following form:
˙
x1−rαq2μx1−rαq4i3x13 h.o.t.,
˙
μ0. 4.19
Thus we have the following results.
Theorem 4.1. Letμbe small. Then consider what follows.
iSuppose thatμ 0. Then ifi2/0, the equilibriumY∗, K∗is unstable, and ifi2 0 andi3/0, then the equilibriumY∗, K∗is asymptotically stable forτ−τ∗i3>0 and unstable ifτ−τ∗i3<0.
iiThe equilibriumY∗, K∗is asymptotically stable ifτ−τ∗μ >0 and unstable ifτ−τ∗μ <
0.
iiiAtY∗, K∗, k∗, Sys.1.4undergoes a transcritical bifurcation ifi2/0 and a pitchfork bifurcation ifi20 andi3/0.
5. Double-Zero Singularity
In this section, we assume that one of the conditionsH3andH6holds andgq2>0, or equivalently, as
kk∗, ττ∗, q > q0, g q2
>0. 5.1
FromSection 2, we can see that, at0,0, the characteristic equation of Sys.2.5has a double root 0 and all other roots have negative real parts ifk k∗ andτ τ∗. We treatk, τas a bifurcation parameter neark∗, τ∗.
By scalingt → t/τ, Sys.2.2can be written as du1t
dt ατ k−γ
u1t−αβτu2t ατi2u21t τi3u31t O
|u1|4 , du2t
dt τku1t−1− q β
τu2t τi2u21t−1 τi3u31t−1 O
|u1|4 .
5.2
LetC:C−1,0,R4, C∗:C0,1,R4∗. Letμ1k−k∗,μ2τ−τ∗. Then onCwe have du1
dt α βγ
qτ∗u10−βτ∗u20 τ∗μ1u10 βγ
qμ2u10−βμ2u20 τ∗i2u210 i2μ2u210 i3τ∗u310 i3μ2u310
Oμ2|u| μ|u|4 , du2
dt βγ
q τ∗u1t−1−qτ∗u2t−βτ∗u2−1 τ∗μ1u1−1 βγ
qμ2u1−1−qμ2u20−βμ2u2−1 i2τ∗u21−1 i2μ2u21−1
i3τ∗u31−1 i3μ2u31−1 Oμ2|u| μ|u|4 , dμ1
dt 0, dμ2 dt 0.
5.3
The linearization of Sys.5.3at0,0,0,0is du1t
dt αβγ
q τ∗u10−αβτ∗u20, du2t
dt k∗τ∗u1−1−qτ∗u20−βτ∗u2−1, dμ1
dt 0, dμ2
dt 0.
5.4
Let
ηθ Aδθ Bδθ 1, 5.5
where
A
⎛
⎜⎜
⎜⎜
⎜⎜
⎝ αβγ
q τ∗ −αβτ∗ 0 0 0 −qτ∗ 0 0
0 0 0 0
0 0 0 0
⎞
⎟⎟
⎟⎟
⎟⎟
⎠
, B
⎛
⎜⎜
⎜⎜
⎜⎝
0 0 0 0
k∗τ∗ −βτ∗ 0 0
0 0 0 0
0 0 0 0
⎞
⎟⎟
⎟⎟
⎟⎠. 5.6
Define
Lϕ 0
−1dηθϕθ, ∀ϕ∈C. 5.7
LetC1C1−1,0,R4. LetX u1, u2, μ1, μ2TandFXt F1, F2,0,0Twhere F1α
τ∗μ1u10 βγ
q μ2u10−βμ2u20 i2τ∗u210 i3τ∗u310
Oμ2|u| μ|u|4 , F2τ∗μ1u1−1 βγ
q μ2u1−1−qμ2u20−βμ2u2−1 i2τ∗u210 i3τ∗u31−1 Oμ2|u| μ|u|4 .
5.8
Then Sys.5.3can be transformed into
Xt ˙ LXt FXt. 5.9
Let C∗ C0,1,R4∗. From 3.7, the bilinear inner product between C and C∗ can be expressed by
ψ, ϕ
ψ0ϕ0
0
−1ψξ 1Bϕξdξ, 5.10
forϕ∈ Candψ ∈C∗. As inSection 4, the infinitesimal generatorA :C1 → BCassociated withLis given by
Aϕϕ˙ X0
Lϕ−ϕ0˙
⎧⎨
⎩
˙
ϕ, −1≤θ <0,
Aϕ0 Bϕ−1, θ0, 5.11 forϕ∈C1and its adjoint by
A∗ψ
⎧⎨
⎩
−ψ,˙ 0< s≤1,
ψ0A ψ1B, s0, 5.12 forψ ∈C1∗. FromSection 2, we know that 0 is an eigenvalue ofAandA∗with multiplicity 4. Now we compute eigenvectors ofAandA∗associated with 0, respectively.
Next we obtain the bases for the center spaceP and its adjoint spaceP∗, respectively.
LetAϕ0 forϕ∈C1. This means that
ϕθ ˙ 0 for −1≤θ <0, Aϕ0 Bϕ−1 0 for θ0. 5.13
From this we obtain thatϕθ ϕ0is a constant vector inR4satisfying
A Bϕ0 0. 5.14
This equation has three linearly independent solutions:a1 q, γ,0,0T, a3 0,0,1,0T, a4 0,0,0,1T. Letϕ01be one of those. Suppose thatAa2ϕ01fora2 ∈C1, namely,
˙
a2θ ϕ01 for −1≤θ <0, Aa20 Ba2−1 ϕ01 forθ0. 5.15
This implies that there is a constant vectorϕ02inR4such thata2θ ϕ01θ ϕ02and L
ϕ01θ ϕ02
ϕ01. 5.16
Since
L
ϕ01θ ϕ02 L
ϕ01θ L
ϕ02
−Bϕ01 A Bϕ02, 5.17
we have that
A Bϕ02 I Bϕ01. 5.18
It is easy to see that5.18has no solution ifϕ01 is eithera3 ora4. Forϕ01 a1, settingϕ02 0, l,0,0Tin5.18, we obtain
l− q2γ
q2 qβ−αβγ, 5.19
and hencea2θ θq, l γθ,0,0T. Thus we obtain basesa1,a2,a3,a4for the center spaceP.
LetΦ a1, a2, a3, a4. Then we have that ˙Φ ΦJwhereJ
⎛
⎜⎝
0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
⎞
⎟⎠.
Similarly, letA∗ψ20 forψ2∈C1∗, that is,
−ψ˙2s 0 for 0< s≤1, ψ20A ψ2−1B0 for s0, 5.20
which means thatψ2s ψ20is a constant vectorψ20∈R4∗\ {0}satisfying
ψ02A B 0. 5.21
This equation has three linearly independent solutions: b2 mq β,−mαβ,0,0,b3 0,0,1,0,b4 0,0,0,1. Asserting thatb2, a21 gives
m 2
q2 qβ−αβγ
q4−q2β2 α2β2γ2. 5.22
Letψ20be one ofb2,b3,b4. SupposeA∗b1ψ20, that is,
−b˙1s ψ20 for 0< s≤1, b10A b11Bψ20 fors0, 5.23
which implies that there isψ10∈R4∗such thatb1s −ψ20s ψ10satisfying L∗
−ψ20s ψ10
ψ20. 5.24
Since
L∗
−ψ20s ψ10
−L∗ ψ20s
L∗ ψ10
−ψ02B ψ10A B, 5.25
we have
ψ10A B ψ20I B. 5.26
It is not hard to check that5.26has no solution ifψ02 b3 orb4. Letting ψ20 b2, setting ψ10 n1, n2,0,0in5.26and usingb1, a20, we can getn1andn2:
n1 2
q β
q6−3q5β−3q4β2 q3β3 3q2α2β2γ2−3qα2β3γ2 2α3β3γ3 3
q4−q2β2 α2β2γ22 , n2 2αβ
−q2 2qβ αβγ
q2 qβ−αβγ2
3
q4−q2β2 α2β2γ22 .
5.27
Hence
b1s ψ10−sψ20
−m q β
s n1, mαβs n2,0,0
. 5.28
Thenb1,b2,b3,b4are bases of the center spaceP∗. LetΨ b1, b2, b3, b4T. ThenΨ,Φ I, Φ ΦJ˙ and ˙Ψ −JΨ.
Letu Φx y, namely,
u1θ qx1 qθx2 y1θ, u2θ γx1
l γθ
x2 y2θ. 5.29
Then Sys.5.9can be decomposed as
˙
xJx Ψ0F
Φx y ,
˙
yAQ1y I−πX0F
Φx y
. 5.30
Writex x1, x2, μ1, μ2. Then, on the center manifold, Sys.5.30becomes
˙ x1
a11μ1 a21μ2 x1
a21μ1 a22μ2 x2 αn1τ∗
i2q2x21 i3q3τ∗x31 n2τ∗
×
i2q2x1−x22 i3q3τ∗x1−x23
h.o.t.,
˙ x2
b11μ1 b21μ2
x1
b12μ1 b22μ2
x2
mα q β
τ∗
i2q2x12 i3q3τ∗x31
mαβτ∗
×
i2q2x1−x22 i3q3τ∗x1−x23
h.o.t.,
˙
μ10, μ˙20,
5.31
where
a11qτ∗αn1 n2, a21 −α q β
γ−qk∗ n1
− 2q β
γ qk∗ n2, a12−qn2τ∗, a22−
lαβn1
q β
l−γ qk∗
n2 , b11 mq2ατ∗, b21mqαβγ, b12 mqαβτ∗, b22 mαβ
− q β
γ qk∗ .
5.32
Next we use techniques of nonlinear transformations in22to transform Sys.5.31 into normal forms. Ifi2/0, then up to the second order, Sys.5.31can be written as
˙ x1
a11μ1 a21μ2
x1
a21μ1 a22μ2
x2
αn1τ∗i2q2x21 n2τ∗i2q2x1−x22 h.o.t.,
˙ x2
b11μ1 b21μ2 x1
b12μ1 b22μ2 x2 mα
q β
τ∗i2q2x21 mαβτ∗i2q2x1−x22 h.o.t.,
˙
μ10, μ˙20.
5.33
This system can be transformed into the following normal form:
˙
x1x2 h.o.t.,
˙
x2ρ1x1 ρ2x2 a1x21 b1x1x2 h.o.t., 5.34
where
ρ1 b11μ1 b21μ2, ρ2 a11 b12μ1 a21 b22μ2, a1mq3ατ∗i2, b12q2
mαβ αn1 n2
τ∗i2.
5.35
Since
∂ρ
∂μ det
⎛
⎜⎜
⎝
∂ρ1
∂μ1
∂ρ1
∂μ2
∂ρ2
∂μ1
∂ρ2
∂μ2
⎞
⎟⎟
⎠
−mq2αγ
mαβ2 αβn1
q β n2
τ∗ −4q3α3β2γ2τ∗
q2 qβ−αβγ q4−q2β2 α2β2γ22 /0,
5.36
we have thatμ1, μ2 → ρ1, ρ2is regular and hence the transversality condition holds.
Ifi20 andi3/0, then up to the third order, Sys.5.31becomes
˙ x1
a11μ1 a21μ2
x1
a21μ1 a22μ2
x2
αn1τ∗i3q3τ∗x31 n2τ∗i3q3τ∗x1−x23 h.o.t.,
˙ x2
b11μ1 b21μ2 x1
b12μ1 b22μ2 x2 mα
q β
τ∗i3q3τ∗x13 mαβτ∗i3q3τ∗x1−x23 h.o.t.,
˙
μ10, μ˙20.
5.37
This system can be transformed into the following normal form:
˙
x1x2 h.o.t.,
˙
x2ρ1x1 ρ2x2 a2x13 b2x21x2 h.o.t., 5.38 wherea2mq4ατ∗i3,b23q3mαβ αn1 n2τ∗i3.
6. Bifurcation Diagrams
In this section, we will use the truncated systems 5.34 and 5.38 to obtain bifurcation diagrams of Sys.5.3.
First, we consider the truncated system of5.34:
˙ x1x2,
˙
x2 ρ1x1 ρ2x2 a1x21 b1x1x2, 6.1