Chaos, Solitons and Fractals 32 (2007) 1916–1926 www.elsevier.com/locate/chaos
Periodic solutions for a two-species nonautonomous competition system with diffusion and impulses Lingzhen Dong b
a,*
, Lansun Chen b, Peilin Shi
a
a Department of Mathematics, Taiyuan University of Technology, Taiyuan 030024, China Department of Applied Mathematics, Dalian University of Technology, Dalian 116023, China
Accepted 28 December 2005
Communicated by A. Helal
Abstract Rx By re-estimating the upper bound of 0 eui ðtÞ dt (i = 1, 2), we generalize a result about the existence of a positive periodic solution for a two-species nonautonomous patchy competition system with time delay. Based on that system, we consider the impulsive harvesting and stocking, and establish a two-species nonautonomous competition Lotka–Volterra system with diffusion and impulsive effects. With the continuation theorem of coincidence degree theory, we obtain the existence of a positive periodic solution for such a system. At last, two examples are given to demonstrate our results. 2006 Elsevier Ltd. All rights reserved.
1. Introduction One of the most interesting questions in mathematical biology concerns the existence of positive periodic solutions for population dynamical systems. For the continuous Lotka–Volterra systems, such a problem has been investigated extensively, and many skills and techniques have been developed. The existence of positive periodic solutions for such systems can be obtained by Brouwer fixed point theorem [2], by standard techniques of bifurcation theory [3], or by theory of topological degree [1]. In fact, these methods have been widely applied to various L–V systems [4–6]. However, in population dynamics, many evolutionary processes experience short-time rapid changes after undergoing relatively long smooth variation. For examples, the harvesting and stocking occur at fixed moments, and some species usually immigrate at the same time every year, etc. If we still thought of the population dynamical systems with these phenomena as continuous systems, it would be unreasonable or incorrect. We should establish systems with impulsive effects. Recently, theories for impulsive differential equations have been introduced into population dynamics [7–10,12,13]. To the authors knowledge, the population dynamical systems with diffusion and impulsive effects are seldom discussed. *
Corresponding author. E-mail address:
[email protected] (L. Dong).
0960-0779/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.chaos.2006.01.003
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
In this paper, we mainly study the following impulsive system: 8 9 x_ ðtÞ ¼ x1 ðtÞðr1 ðtÞ a1 ðtÞx1 ðtÞ b1 ðtÞyðtÞÞ þ D1 ðtÞðx2 ðtÞ x1 ðtÞÞ; > > > > > 1 = > > x_ 2 ðtÞ ¼ x2 ðtÞðr2 ðtÞ a2 ðtÞx2 ðtÞÞ þ D2 ðtÞðx1 ðtÞ x2 ðtÞÞ; > > > > > > < y_ ðtÞ ¼ yðtÞ r3 ðtÞ a3 ðtÞx1 ðtÞ b3 ðtÞyðtÞ bðtÞ R 0 KðsÞyðt þ sÞds ; > ; s 9 > > Dx1 ðtk Þ ¼ ck x1 ðtk Þ; > > = > > > > Dx2 ðtk Þ ¼ d k x2 ðtk Þ; > > > > ; : Dyðtk Þ ¼ ek yðtk Þ;
1917
t 6¼ tk ; ð1:1Þ
with period where ri(t), ai(t) (i = 1, 2, 3), Di(t) (i = 1, 2), bi(t) (i = 1, 3), b(t) are all positive periodic continuous functions R0 x > 0, s is a nonnegative constant, K(s) is a continuous nonnegative function on [s, 0] such that s KðsÞds ¼ 1, and there exists a number q > 0 such that ck+q = ck, dk+q = dk, ek+q = ek, tk+q = tk + x (k 2 Z+). From the viewpoint of biology, we assume 1 + ck, 1 + dk, 1 + ek are all positive. System (1.1) shows that species X can diffuse between patch 1 and 2, while species Y is restricted to patch 1, and in patch 1 species X competes with Y for space or food. Furthermore, species X in two patches and species Y are harvested and stocked periodically at fixed moments, which occur when 1 + ck, 1 + dk, 1 + ek are smaller than 1 or larger than 1. These human activities occur frequently and therefore the study of their effects on the population behaviors is of great practical value. This paper R x is arranged as follows. In the next section, Theorem 2.1 in [1] is generalized by re-estimating the upper bound of 0 eui ðtÞ dt (i = 1, 2). And we introduce the continuation theorem of coincidence degree theory, which is used to prove the existence of a positive periodic solution of (1.1). Moreover, some notations are introduced. In Section 3, we study the existence of a positive periodic solution of (1.1). At last, two examples are worked out.
2. Generalization Rx We estimate the upper bound of 0 eui ðtÞ dt (i = 1, 2) in a easier way, and Theorem 2.1 in [1] can be generalized. In fact, let v(t) = max{u1(t), u2(t)}. We know that v(t) is a x-periodic continuous function and (1) If u1(t) > u2(t), or u1(t) = u2(t) but u_ 1 ðtÞ P u_ 2 ðtÞ, then vðtÞ ¼ u1 ðtÞ and
u_ 1 ðtÞ 6 k½r1 ðtÞ a1 ðtÞeu1 ðtÞ 6 k½rm1 al1 eu1 ðtÞ .
(2) If u2(t) > u1(t), or u2(t) = u1(t) but u_ 2 ðtÞ P u_ 1 ðtÞ, then vðtÞ ¼ u2 ðtÞ and
u_ 2 ðtÞ 6 k½r2 ðtÞ a2 ðtÞeu2 ðtÞ 6 k½rm2 al2 eu2 ðtÞ .
Consequently, Dþ vðtÞ 6 k½r aevðtÞ ; r ¼ maxfrm1 ; rm2 g, Rwhere x vðtÞ e dt 6 rx . Further, 0 a Z 0
x
eui ðtÞ dt 6
rx a
a¼
ð2:1Þ minfal1 ; al2 g.
Integrating (2.1) over [0, x], we have 0 6 k½rx a
Rx 0
vðtÞ
e
dt, which implies that
ði ¼ 1; 2Þ.
Thus, we can replace (H5) of Theorem 2.1 in [1] with ðH 05 Þ : ar3 > am3 r, and obtain Theorem 2.1 0 . If (H3): ri(t) > Di(t) (i = 1, 2), r3 , and (H4): bl3 ðr1 D1 Þ > bm 1 (H 05 ): ar3 > am r, are satisfied, then system (1.1) in [1] has at least one x-periodic solution. 3 In order to obtain the existence of positive x-periodic solutions for (1.1), we shall introduce the continuation theorem of coincidence degree theory [11]. Let X, Y be real Banach spaces, L : Dom L X ! Y be a Fredholm mapping of index zero, and P : X ! X, Q : Y ! Y be continuous projectors such that Im P = Ker L, Ker Q = Im L, so that X = Ker L Ker P, Y = Im L Im Q. Denote by LP the restriction of L to Dom L \ Ker P and by KP : Im L ! Dom L \ Ker P the inverse to LP. Let J : Im Q ! Ker L be an isomorphism of Im Q onto Ker L. Then the continuation theorem can be described as follows.
1918
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
Lemma 2.1. Let X X be an open bounded set and let N : X ! Y be a continuous operator which is L-compact on X (i.e. QN : X ! Y and K P ðI QÞN : X ! Y are compact). Assume that (a) for each k 2 (0, 1), x 2 oX \ Dom L, Lx 5 kNx, (b) for each x 2 oX \ Dom L, QNx 5 0, and (c) deg{JQN, X \ Ker L, 0} 5 0. Then Lx = Nx has at least one solution in X \ Ker L. For convenience, we use the following notations in the rest parts of the paper. (1) (2) (3) (4)
PC(R+, R) = {u(t)ju : R+ ! R, lims!tu(s) = u(t) if t 5 tk, limt!tk uðtÞ ¼ uðtk Þ, limt!tþk uðtÞ exists, k 2 Z+}, PC 0 (R+, R) = {u(t)ju : R+ ! R, u 0 (t) 2 PC(R+, R)}, PCx = {u 2 PC(R+, R)ju(t)R= u(t + x)}, PC 0x ¼ fu 2 PC 0 ðRþ ; RÞjuðtÞ ¼ uðt þ xÞg, x For f 2 PCx, denote f ¼ x1 0 f ðtÞdt, fl = mint2[0,x]jf(t)j, fm = maxt2[0,x]jf(t)j.
3. Positive x-periodic solutions In this section, we demonstrate the existence of a positive x-periodic solution for (1.1). We have Theorem 3.1. If system (1.1) satisfies P lnð1 þ ck Þ > D1 ðtÞ and r2 ðtÞ þ qk¼1 lnð1 þ d k Þ > D2 ðtÞ, P P (2) ððr1 D1 Þx þ qk¼1 lnð1 þ ck ÞÞbl3 > bm1 ðr3 x þ qk¼1 lnð1 þ ek ÞÞ, and Pq P (3) aðr3 x þ k¼1 lnð1 þ ek ÞÞ > am3 ðrx þ qk¼1 lnð1 þ Bk ÞÞ,
(1) r1 ðtÞ þ
Pq
k¼1
where r ¼ maxfrm1 ; rm2 g, Bk = max{ck, dk}, a ¼ minfal1 ; al2 g, then system (1.1) has at least one positive x-periodic solution. Proof. Let x1(t) = exp(u1(t)), x2(t) = exp(u2(t)), y(t) = exp(u3(t)). We obtain 8 9 u_ 1 ðtÞ ¼ r1 ðtÞ D1 ðtÞ a1 ðtÞeu1 ðtÞ b1 ðtÞeu3 ðtÞ þ D1 ðtÞeu2 ðtÞu1 ðtÞ ; > > > = > > > u_ 2 ðtÞ ¼ r2 ðtÞ D2 ðtÞ a2 ðtÞeu2 ðtÞ þ D2 ðtÞeu1 ðtÞu2 ðtÞ ; > t 6¼ tk ; > > > R0 > ; < u_ 3 ðtÞ ¼ r3 ðtÞ a3 ðtÞeu1 ðtÞ b3 ðtÞeu3 ðtÞ bðtÞ s KðsÞeu3 ðtþsÞ ds; 9 > Du1 ðtk Þ ¼ lnð1 þ ck Þ; > > = > > > > Du2 ðtk Þ ¼ lnð1 þ d k Þ; > > > > : ; Du3 ðtk Þ ¼ lnð1 þ ek Þ.
ð3:1Þ
T It is obvious that if system (3.1) has an x-periodic solution ðu1 ðtÞ; u2 ðtÞ; u3 ðtÞÞ , then
ðeu1 ðtÞ ; eu2 ðtÞ ; eu3 ðtÞ ÞT ¼ ðx1 ðtÞ; x2 ðtÞ; y ðtÞÞT is a positive x-periodic solution of system (1.1). So in the following we discuss the existence of x-periodic solutions of system (3.1). T To apply P Lemma 2.1, we denote X = {(u1(t), u2(t), u3(t)) jui(t) 2 PCx, i = 1, 2, 3} and denote kðu1 ðtÞ; u2 ðtÞ; u3 ðtÞÞT k ¼ 3i¼1 supt2½0;x jui ðtÞj. Then (X, k Æ k) is a real Banach space. Moreover, let 820 9 1 0 1 0 13 ui ðtÞ 2 PC x ði ¼ 1; 2; 3Þ; ðmk ; nk ; p ÞT ¼ ðDU 1 ðtk ÞDU 2 ðtk Þ; DU 3 ðtk ÞÞT ; > mq m1 u1 ðtÞ > k < = 6B C B C B C7 Y ¼ 4@ u2 ðtÞ A; @ n1 A; . . . ; @ nq A5 a constant vector, k ¼ 1; 2; . . . ; q; > > : ; U ðtÞ; a primitive function of u ðtÞ; i ¼ 1; 2; 3. pq u3 ðtÞ p1 i i . , yq] 2 Y, where y1(t) = (u1(t), u2(t), u3(t))T and yk = (mk, nk, pk)T (k = 1, 2, . . . , q), and define Assume Pthat y = [y1(t), y1, . . P kyk ¼ 3i¼1 supt2½0;x jui ðtÞj þ qk¼1 ky k k. Then (Y, k Æ k) is also a Banach space.
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
1919
Set L : Dom L X ! Y, 2 3 20 1 0 1 0 13 Du1 ðtq Þ u_ 1 ðtÞ Du1 ðt1 Þ u1 ðtÞ 6 7 6B C B C B C7 L4 u2 ðtÞ 5 ¼ 4@ u_ 2 ðtÞ A; @ Du2 ðt1 Þ A; . . . ; @ Du2 ðtq Þ A5; Du3 ðtq Þ u3 ðtÞ u_ 3 ðtÞ Du3 ðt1 Þ where Dom L ¼ fðu1 ðtÞ; u2 ðtÞ; u3 ðtÞÞT 2 X ju_ i ðtÞ 2 PC x g ¼ fðu1 ðtÞ; u2 ðtÞ; u3 ðtÞÞT 2 X jui ðtÞ 2 PC 0x ði ¼ 1; 2; 3Þg. At the same time, we denote f1 ðtÞ ¼ r1 ðtÞ D1 ðtÞ a1 ðtÞeu1 ðtÞ b1 ðtÞeu3 ðtÞ þ D1 ðtÞeu2 ðtÞu1 ðtÞ ; f2 ðtÞ ¼ r2 ðtÞ D2 ðtÞ a2 ðtÞeu2 ðtÞ þ D2 ðtÞeu1 ðtÞu2 ðtÞ ; Z 0 KðsÞeu3 ðtþsÞ ds. f3 ðtÞ ¼ r3 ðtÞ a3 ðtÞeu1 ðtÞ b3 ðtÞeu3 ðtÞ bðtÞ s
Then we may define N : X ! Y, 2
3 20 10 1 0 13 lnð1 þ cq Þ lnð1 þ c1 Þ f1 ðtÞ u1 ðtÞ 6 7 6B CB C B C7 N 4 u2 ðtÞ 5 ¼ 4@ f2 ðtÞ A@ lnð1 þ d 1 Þ A; . . . ; @ lnð1 þ d q Þ A5; lnð1 þ eq Þ u3 ðtÞ f3 ðtÞ lnð1 þ e1 Þ and define two projectors P and Q as P : X ! X, 0Rx 1 1 u1 u1 ðtÞdt 0 C B C 1 BRx P @ u2 A ¼ @ 0 u2 ðtÞdt A. x Rx u3 u3 ðtÞdt 0 0
Q : Y ! Y, 1 2 0R 3 q P x u m 1 ðtÞdt þ k 0 C 0 1 1 0 1 20 0 13 6 B 0 17 k¼1 C 0 6 B mq m1 u1 ðtÞ 0 7 C 61 B R 7 q P x C B C B n C7 6 B u ðtÞdt þ 6B C B C B 7 nk C; @ 0 A; . . . ; @ 0 C Q4@ u2 ðtÞ A; @ n1 A; . . . ; @ q A5 ¼ 6 B 0 2 A7. C 6x B 7 k¼1 C 0 6 B pq u3 ðtÞ p1 0 7 q A 4 @Rx 5 P u3 ðtÞdt þ pk 0 k¼1
3
Obviously, Ker L = R , and 8 9 q P Rx > > 0 u1 ðtÞdt þ > > m ¼ 0 k > > > > 20 1 0 1 0 1 3 > > k¼1 > > mq m1 u1 ðtÞ > > R < = q P 6B C B C B n C7 x u ðtÞdt þ nk ¼ 0 Im L ¼ 4@ u2 ðtÞ A; @ n1 A; . . . ; @ q A5 0 2 > > k¼1 > > > > pq p1 u3 ðtÞ > > q > > R P > > x > > : ; u ðtÞdt þ p ¼ 0 3 k 0 k¼1
is closed in Y. Noting that Im P = Ker L, Ker Q = Im L and dimKer L = codimIm L = 3, we know that L is a Fredholm mapping of index zero. Moreover, by computation, the inverse KP of LP has the form KP : Im L ! Ker P \ Dom L, 1 Z Z q q t X X 1 x t 1 X u1 ðsÞds þ mk u1 ðsÞds dt mk þ mk t k C B x 0 0 x k¼1 C tk
1920
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
Since 1 2 0R 3 q P x f1 ðtÞdt þ lnð1 þ ck Þ 0 C 0 1 0 1 6 B 0 17 k¼1 C 0 6 B u1 ðtÞ 0 7 C 61 B R 7 q P B C 6 B x f ðtÞdt þ C B C B 7 lnð1 þ d k Þ C; @ 0 A; . . . ; @ 0 C QN @ u2 ðtÞ A ¼ 6 B 0 2 A7; C 6x B 7 k¼1 C 0 6 B u3 ðtÞ 0 7 q A 4 @Rx 5 P f ðtÞdt þ lnð1 þ ek Þ 0 3 k¼1
and Kp(I Q)N : X ! X,
1 lnð1 þ c Þ k B C 1 0 k¼1 tk
P
f ðsÞds þ 0 1
lnð1 þ ck Þ
0R
1
x f ðsÞds 0 1
þ
q P
k¼1
0 R R 1 q q P P x t 1 1 f ðsÞds dt lnð1 þ c Þ þ lnð1 þ c Þt 1 k k k x Bx 0 0 C k¼1 k¼1 B C B R R C q q P P B 1 x t f ðsÞds dt 1 lnð1 þ d k Þ þ x lnð1 þ d k Þtk C Bx 0 0 2 C; B C k¼1 k¼1 B C q q @ R xR t A P P 1 1 f ðsÞds dt lnð1 þ e Þ þ lnð1 þ e Þt 3 k k k x 0 0 x k¼1
k¼1
by the Lebesque convergence theorem, QN and KP(I Q)N are continuous. Moreover, from the Arzela–Ascoli theorem, QN ðXÞ, K P ðI QÞN ðXÞ are relatively compact for any open bounded set X X. Therefore N is L-compact on X. In the following, the operator equation, Lx = kNx, k 2 (0, 1), is considered, which can be rewritten as: 9 8 u_1 ðtÞ ¼ k½r1 ðtÞ D1 ðtÞ a1 ðtÞeu1 ðtÞ b1 ðtÞeu3 ðtÞ þ D1 ðtÞeu2 ðtÞu1 ðtÞ ; > > > > = > > u2 ðtÞ u1 ðtÞu2 ðtÞ > _ u ðtÞ ¼ k½r ðtÞ D ðtÞ a ðtÞe þ D ðtÞe ; > t 6¼ tk ; 2 2 2 2 2 > > > > > ; < u_ ðtÞ ¼ k½r ðtÞ a ðtÞeu1 ðtÞ b ðtÞeu3 ðtÞ bðtÞ R 0 KðsÞeu3 ðtþsÞ ds > 3 3 3 3 s ð3:2Þ 9 > > Du ðt Þ ¼ k lnð1 þ c Þ; k > 1 k > > = > > > > Du2 ðtk Þ ¼ k lnð1 þ d k Þ; > > > : ; Du3 ðtk Þ ¼ k lnð1 þ ek Þ. Suppose (u1(t), u2(t), u3(t))T 2 X is a solution of system (3.2) for some k 2 (0, 1). Integrating (3.2) over [0, x], it can be obtained that Z x q X lnð1 þ ck Þ ¼ r1 ðtÞ D1 ðtÞ a1 ðtÞeu1 ðtÞ b1 ðtÞeu3 ðtÞ þ D1 ðtÞeu2 ðtÞu1 ðtÞ dt; 0
k¼1
q X
lnð1 þ d k Þ ¼
k¼1
q X
lnð1 þ ek Þ ¼
Z Z
a1 ðtÞeu1 ðtÞ dt þ
0
Z
0
x
r2 ðtÞ D2 ðtÞ a2 ðtÞeu2 ðtÞ þ D2 ðtÞeu1 ðtÞu2 ðtÞ dt;
u1 ðtÞ
r3 ðtÞ a3 ðtÞe
x
a2 ðtÞeu2 ðtÞ dt
Z
x
b1 ðtÞeu3 ðtÞ dt
b3 ðtÞeu3 ðtÞ dt þ
b3 ðtÞe
q X
Z
lnð1 þ d k Þ ¼
lnð1 þ ck Þ ¼
bðtÞ
Z
x
Z
Z
0 u3 ðtþsÞ
KðsÞe
x
ðr1 ðtÞ D1 ðtÞÞdt þ
x
ðr2 ðtÞ D2 ðtÞÞdt þ
0
s
Z 0
0
bðtÞ 0
q X k¼1
k¼1
x
u3 ðtÞ
ds dt.
s
0
0
Z
0
0
k¼1
Therefore, Z x
x
KðsÞeu3 ðtþsÞ ds dt þ
Z 0
Z
Z
x
D1 ðtÞeu2 ðtÞu1 ðtÞ dt;
ð3:3Þ
0 x
D2 ðtÞeu1 ðtÞu2 ðtÞ dt;
0 x
a3 ðtÞeu1 ðtÞ dt
q X k¼1
lnð1 þ ek Þ ¼
ð3:4Þ Z
x
r3 ðtÞdt. 0
ð3:5Þ
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
Consequently, Z Z x ju_ 1 ðtÞjdt 6 2 0
Z
a1 ðtÞeu1 ðtÞ dt þ
Z
0 x
ju_ 2 ðtÞjdt 6 2
Z
0
Z
x
ju_ 3 ðtÞjdt 6 2
a2 ðtÞeu2 ðtÞ dt
Z
0
q X
k¼1
lnð1 þ d k Þ;
k¼1 x
r3 ðtÞdt þ
0
q X
lnð1 þ ek Þ ¼ 2r3 x þ
k¼1
q X
D
lnð1 þ ek Þ ¼ d 3 .
k¼1
From Eq. (3.5), we have Z Z x q X b3 ðtÞeu3 ðtÞ dt lnð1 þ ek Þ 6 0
X q b1 ðtÞeu3 ðtÞ dt lnð1 þ ck Þ;
0
x
0 x
x
1921
x
r3 ðtÞdt
0
k¼1
and therefore, P Z x r3 x þ qk¼1 lnð1 þ ek Þ eu3 ðtÞ dt 6 . bl3 0
ð3:6Þ
Let v(t) = max{u1(t), u2(t)}. Then v(t) 2 PCx. Moreover, (1) If u1(t) > u2(t), or u1(t) = u2(t) but u_ 1 ðtÞ P u_ 2 ðtÞ, then vðtÞ ¼ u1 ðtÞ and
u_ 1 ðtÞ 6 k½r1 ðtÞ a1 ðtÞeu1 ðtÞ 6 k½rm1 al1 eu1 ðtÞ .
(2) If u2(t) > u1(t), or u2(t) = u1(t) but u_ 2 ðtÞ P u_ 1 ðtÞ, then vðtÞ ¼ u2 ðtÞ and
u_ 2 ðtÞ 6 k½r2 ðtÞ a2 ðtÞeu2 ðtÞ 6 k½rm2 al2 eu2 ðtÞ .
Denote r ¼ maxfrm1 ; rm2 g and a ¼ minfal1 ; al2 g. Then (
Dþ vðtÞ 6 k½r aevðtÞ ; Dvðtk Þ 6 k lnð1 þ Bk Þ;
ð3:7Þ
where Bk = max{ck, dk} (k 2 Z+). Integrating (3.7) over [0, x], we have Z x q X lnð1 þ Bk Þ 6 rx a evðtÞ dt; 0
k¼1
Rx
Pq
which leads to 0 evðtÞ dt 6 ½rx þ k¼1 lnð1 þ Bk Þ=a. Therefore, P Z x rx þ qk¼1 lnð1 þ Bk Þ ði ¼ 1; 2Þ. eui ðtÞ dt 6 a 0
ð3:8Þ
So, Z
x
ju_ 1 ðtÞjdt 6 2 0
! ! P P am1 r bm1 r3 am q lnð1 þ Bk Þ bm1 qk¼1 lnð1 þ ek Þ þ l x þ 2 1 k¼1 þ a a b3 bl3 q X k¼1
Z
x 0
D
ð3:9Þ
lnð1 þ ck Þ ¼ d 1 > 0;
am r am ju_ 2 ðtÞjdt 6 2 2 x þ 2 2 a
Pq
k¼1
lnð1 þ Bk Þ X D lnð1 þ d k Þ ¼ d 2 > 0. a k¼1
From Eq. (3.4), we know that P Z x ðr2 D2 Þx þ qk¼1 lnð1 þ d k Þ eu2 ðtÞ dt > . am2 0
q
ð3:10Þ
1922
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
From Eq. (3.3), we have Z x Z q X a1 ðtÞeu1 ðtÞ dt lnð1 þ ck Þ P ðr1 D1 Þx 0
x
b1 ðtÞeu3 ðtÞ dt
0
k¼1
and consequently Z
x 0
ðr1 D1 Þbl3 bm1 r3 eu1 ðtÞ dt P x am1 bl3
bm1
Pq
k¼1
q P
lnð1 þ ek Þ bl3
lnð1 þ ck Þ
k¼1
ð3:11Þ
> 0.
am1 bl3
With equality (3.5), we know that Z 0 b3 þ b KðsÞds exp max u3 ðtÞ s
P r3 x
Z
t2½0;x
x
a3 ðtÞeu1 ðtÞ dt þ
0
P
q X
lnð1 þ ek Þ
k¼1
P P ar3 am3 r a qk¼1 lnð1 þ ek Þ am3 qk¼1 lnð1 þ Bk Þ xþ > 0. a a
ð3:12Þ
Inequalities (3.6) and (3.8) show that there exist three points si ði ¼ 1; 2; 3Þ such that P rx þ qk¼1 lnð1 þ Bk Þ ði ¼ 1; 2Þ; ui ðsi Þ 6 ln Pqax r3 x þ k¼1 lnð1 þ ek Þ u3 ðs3 Þ 6 ln . bl3 x
ð3:13Þ ð3:14Þ
Consequently, there are three positive numbers di (i = 1, 2, 3) such that ui ðsi Þ < di
ði ¼ 1; 2; 3Þ.
ð3:15Þ
Similarly, inequalities (3.10)–(3.12) show that there exist three points (i = 1, 2, 3) such that ui ðni Þ > qi
ni
ði ¼ 1; 2; 3Þ and three positive numbers qi
ði ¼ 1; 2; 3Þ.
ð3:16Þ
Since for an arbitrary t 2 [0, x], Z t Z X u_ 1 ðsÞds þ k lnð1 þ ck Þ 6 u1 ðs1 Þ þ u1 ðtÞ ¼ u1 ðs1 Þ þ s1
u2 ðtÞ ¼ u2 ðs2 Þ þ u3 ðtÞ ¼ u3 ðs3 Þ þ
Z
s2
Z
u1 ðtÞ P u1 ðn1 Þ
X
u_ 2 ðsÞds þ k
X
u_ 3 ðsÞds þ k
s3
Z Z
x
ju_ 1 ðsÞjds
Z
0
X
Z Z
j lnð1 þ ck Þj;
0
ju_ 2 ðsÞjds þ 0
0
X X
j lnð1 þ d k Þj;
0
ju_ 3 ðsÞjds þ
X
j lnð1 þ ek Þj;
0
j lnð1 þ ck Þj;
0
ju_ 2 ðsÞjds
0
u3 ðtÞ P u3 ðn3 Þ
lnð1 þ ek Þ 6 u3 ðs3 Þ þ
s3
0
u2 ðtÞ P u2 ðn2 Þ
lnð1 þ d k Þ 6 u2 ðs2 Þ þ
s2
t
ju_ 1 ðsÞjds þ
0
s1
t
x
X
j lnð1 þ d k Þj;
0
ju_ 3 ðsÞjds
X
j lnð1 þ ek Þj
0
and inequalities (3.15) and (3.16), we have ðqi þ d i þ ri Þ 6 ui ðtÞ 6 di þ d i þ ri ði ¼ 1; 2; 3Þ; P P P where r1 ¼ 0
jui ðtÞj 6 maxfqi þ d i þ ri ; di þ d i þ ri g ¼ Ri
ði ¼ 1; 2; 3Þ.
Obviously, Ri (i = 1, 2, 3) are independent of k. Now, we are in the position of proving all conditions of Lemma 2.1 are satisfied.
ð3:17Þ
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
1923
Denote M = R1 + R2 + R3 + R0, where R0 is sufficiently large such that solution (a*, b*, c*)T of the system 8 q P > x yx z > r a ¼ lnð1 þ ck Þ; 1 D1 1 e b1 e þ D1 e > > > k¼1 > > < q P r2 D2 a2 ey þ D2 exy ¼ lnð1 þ d k Þ; ð3:18Þ > k¼1 > > > q > R > > 0 KðsÞds ez ¼ P lnð1 þ ek Þ; : r3 a3 ex b3 þ b s k¼1
k(a*, b*, c*)Tk
ja*j
jb*j
jc*j
= + + < M, provided that system (3.18) has a solution or a number of solutions. Let satisfies X = {(u1(t), u2(t), u3(t))T 2 X, k(u1(t), u2(t), u3(t))Tk < M}. Then condition (a) of Lemma 2.1 is satisfied. T 3 T 3 P3For each (u1, u2, u3) 2 oX \ Ker L = oX \ R , we know that (u1(t), u2(t), u3(t)) is a constant vector in R with i¼1 jui j ¼ M. If system (3.18) has a solution or a number of solutions, then 20 1 3 q P u1 u2 u1 u3 r a þ lnð1 þ c 1 D1 1 e b1 e þ D1 e kÞ C0 1 0 1 6B 0 17 20 1 0 1 0 13 k¼1 C 0 6B u1 ðtÞ 0 7 0 0 0 C 6B 7 q P B C 6B CB C B C7 6B C B C B C7 a2 eu2 þ D2 eu1 u2 þ lnð1 þ d k Þ QN @ u2 ðtÞ A ¼ 6B r2 D2 C; @ 0 A; . .. ; @ 0 A7 6¼ 4@ 0 A; @ 0 A;. . .; @ 0 A5. C 6B 7 k¼1 C 0 6B u3 ðtÞ 0 7 0 0 0 q 4@ A 5 R P 0 KðsÞds eu3 þ lnð1 þ ek Þ r3 a 3 eu 1 b3 þ b s k¼1
If system (3.18) does not have any solution, naturally, 0 1 20 1 0 1 0 13 0 0 0 u1 ðtÞ B C 6B C B C B C7 QN @ u2 ðtÞ A 6¼ 4@ 0 A; @ 0 A; . . . ; @ 0 A5; u3 ðtÞ 0 0 0 which shows that condition (b) in Lemma 2.1 is satisfied. Finally, we prove that condition (c) in Lemma 2.1 is satisfied. Define U : Dom L · [0, 1] ! X, 2 3 q P u1 u3 r D a e b e þ lnð1 þ c Þ 1 1 1 k 6 1 7 2 3 k¼1 6 7 D1 eu2 u1 6 7 q P 6 7 6 7 lnð1 þ d k Þ Uðu1 ; u2 ; u3 ; lÞ ¼ 6 r2 D2 a2 eu2 þ 7 þ l4 D2 eu1 u2 5; 6 7 k¼1 6 7 0 q 4 5 R0 P u1 u3 r3 a3 e b3 þ b s KðsÞds e þ lnð1 þ ek Þ k¼1
0 for (u1, u2, u3)T 2 oX \ Ker L. Assuming where l 2 [0, 1] is a parameter. With the mapping U, we have U(u1, u2, u3, l) P5 3 T that this is not true, then there exists a constant vector (u1, u2, u3) with i¼1 jui j ¼ M satisfying U(u1, u2, u3, l) = 0, i.e. r1 D1 a1 eu1 b1 eu3 þ lD1 eu2 u1 þ
q X
lnð1 þ ck Þ ¼ 0;
k¼1
r2 D2 a2 eu2 þ lD2 eu1 u2 þ Z r3 a3 eu1 b3 þ b
q X
lnð1 þ d k Þ ¼ 0;
k¼1
0
s
q X KðsÞds eu3 þ lnð1 þ ek Þ ¼ 0. k¼1
Following the discussion of (3.13)–(3.18), we know that juij < max{di, qi}. Therefore P 3 i¼1 jui j ¼ M. Using the property of topological degree and taking J : Im Q ! Ker L by 3 2 0R 1 1 0R q q P P x x u m u m 1 ðtÞdt þ k 1 ðtÞdt þ k 0 0 6 B C 0 1 C B 0 17 k¼1 k¼1 6 B C 0 C B 0 7 7 1 BR 61 B R C C q q P P x 7 6 B x u ðtÞdt þ C B C C B B nk C; @ 0 A; . . . ; @ 0 A7 ¼ B 0 u2 ðtÞdt þ nk C J6 B 0 2 C; 7 xB 6x B C C k¼1 k¼1 7 6 B C C B 0 0 q q 5 4 @Rx A A @Rx P P u ðtÞdt þ p u ðtÞdt þ p 3 3 k k 0 0 k¼1
k¼1
P3
i¼1 jui j
< M, which contradicts
1924
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
we have degðJQN ðu1 ; u2 ; u3 ÞT ; X \ Ker L; ð0; 0; 0ÞT Þ ¼ degðUðu1 ; u2 ; u3 ; 1Þ; X \ Ker L; ð0; 0; 0ÞT Þ ¼ degðUðu1 ; u2 ; u3 ; 0Þ; X \ Ker L; ð0; 0; 0ÞT Þ ¼ deg
r1 D1 a1 eu1 b1 eu3 þ
q X
lnð1 þ ck Þ; r2 D2 a2 eu2 þ
k¼1
Z r3 a3 e b3 þ b
0
u1
q X
lnð1 þ d k Þ;
k¼1
1 !T q X TA u3 KðsÞds e þ lnð1 þ ek Þ ; X \ Ker L; ð0; 0; 0Þ .
s
k¼1
By the hypotheses of the theorem, the algebraic system 8 q P > > lnð1 þ ck Þ ¼ 0; > r1 D1 a1 x1 b1 y þ > > k¼1 > > > < q P r2 D2 a2 x2 þ lnð1 þ d k Þ ¼ 0; > k¼1 > > > > q > R > 0 KðsÞds y þ P lnð1 þ ek Þ ¼ 0; > : r3 a3 x1 b3 þ b s
k¼1
T has a unique solution ðx1 ; x2 ; y Þ , where P P a1 r3 þ qk¼1 lnð1 þ ek Þ a3 r1 D1 þ qk¼1 lnð1 þ ck Þ > 0; x1 ¼ R 0 KðsÞds a3 b1 a1 b3 þ b s P r2 D2 þ qk¼1 lnð1 þ d k Þ > 0; x2 ¼ a2 R P P 0 KðsÞds r1 D1 þ qk¼1 lnð1 þ ck Þ b3 þ b b1 r3 þ qk¼1 lnð1 þ ek Þ s > 0. y ¼ R 0 KðsÞds a1 b3 þ b b1 a3 s
Consequently, we have degðJQN ðu1 ; u2 ; u3 ÞT ; X \ Ker L; ð0; 0; 0ÞT Þ ¼ sign
Z b1 b3 þ b a2 a2 a3 a1
0
KðsÞds
s
x1 x2 y ¼ 1.
And the proof is completed. h In fact, let ck = dk = ek = 0, then the above system (1.1) is the system (1.1) in [1], and conditions (1), (2) and (3) of Theorem 3.1 are identical with (H3), (H4) and ðH 05 Þ of Theorem 2.1 0 . Example 1. Consider the system 8 x_ ðtÞ ¼ x1 ðtÞðð2 þ sin tÞ ð2 þ sin tÞx1 ðtÞ yðtÞÞ þ 54 þ sin t ðx2 ðtÞ x1 ðtÞÞ; > > < 1 x_2 ðtÞ ¼ x2 ðtÞðð2 þ cos tÞ ð2 þ sin tÞx2 ðtÞÞ þ 54 þ cos t ðx1 ðtÞ x2 ðtÞÞ; > > : yðtÞ _ ¼ yðtÞ 4þsin t 1 x ðtÞ ð2 þ sin tÞyðtÞ . 2p
ð3:19Þ
2p 1
It is easily proved that this system satisfies (H3), (H4) and ðH 05 Þ, and has at least one positive 2p-periodic solution (Fig. 1). Example 2. For the system 8 9 > x_1 ðtÞ ¼ x1 ðtÞðð2 þ sin tÞ ð2 þ sin tÞx1 ðtÞ yðtÞÞ þ 54 þ sin t ðx2 ðtÞ x1 ðtÞÞ; > > > > = > > < x_2 ðtÞ ¼ x2 ðtÞðð2 þ cos tÞ ð2 þ sin tÞx2 ðtÞÞ þ 5 þ cos tðx1 ðtÞ x2 ðtÞÞ; 4 > > ; t 1 > 2p x1 ðtÞ ð2 þ sin tÞyðtÞ ; > y_ ðtÞ ¼ yðtÞ 4þsin > 2p > > : Dyðtk Þ ¼ 0:05yðtk Þ;
t 6¼ tk ;
ð3:20Þ
where tk = 2kp, k 2 Z+. It can be verified that all the conditions of Theorem 3.1 hold, so there exists at least one positive 2p-periodic solution for this impulsive differential system (Fig. 2).
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
1925
Fig. 1. Positive periodic solution for continuous system (3.19). (a) Phase portrait. (b) and (c) Time-series of species X in patch 1, 2, respectively. (d) Time-series of species Y in patch 1.
Fig. 2. Positive periodic solution for impulsive differential system (3.20). (a) Phase portrait. (b) and (c) Time-series of species X in patch 1, 2 respectively. (d) Time-series of species Y in patch 1.
4. Conclusion In this paper, we study a two-species periodic competition system with diffusion and impulses.RIn fact, system (1.1) x with ck = dk = ek = 0 (k 2 N) was discussed in paper [1]. Here, we re-estimate the upper bounds of 0 eui ðtÞ dt (i = 1, 2) by constructing the special function v(t) = max{u1(t), u2(t)}, then the conditions of Theorem 2.1 in [1], which guarantee the existence of a positive periodic solution, are simplified. Further, the impulsive effects are considered besides the diffusion of species X between two patches, and a periodic competition system with diffusion and impulses is established. With the theory of topological degree, the existence of a positive periodic solution for such an impulsive differential system is
1926
L. Dong et al. / Chaos, Solitons and Fractals 32 (2007) 1916–1926
obtained. Moreover, some examples are given to illustrate our results. How to prove the stability of positive periodic solutions will be our future work.
References [1] Zhang Z, Wang Z. Periodic solution for a two-species competition Lotka–Volterra patch system with time delay. J Math Anal Appl 2002;265:38–48. [2] Song X, Chen L. Persistence and global stability for nonautonomous predator–prey system with diffusion and time-delay. Comput Math Appl 1998;35(6):33–40. [3] Cushing JM. Periodic time-dependent predator–prey system. SIAM J Appl Math 1977;32:82–95. [4] Cushing JM. Periodic Kolmogrov system. SIAM J Math Anal 1982;13:811–27. [5] Li Y. On a periodic neutral delay Lotka–Volterra system. Nonlinear Anal 2000;39(6):767–78. [6] Song X, Chen L. Persistence and periodic orbits for two-species predator–prey system with diffusion. Can. Appl. Math. Quart. 1998;6(3). [7] Panetta JC. A mathematical model of periodically pulsed chemotherapy: tumor recurrence and metastasis in a competition environment. Bull Math Biol 1996;58:425–47. [8] Ballinger G, Liu X. Permanence of population growth models with impulsive effects. Math Comput Modell 1997;26:59–72. [9] Roberts MG, Kao RR. The dynamics of an infectious disease in a population with birth pulse. Math Biosci 1998;149:23–6. [10] Tang Sanyi, Chen Lansun. Density-dependent birth rate, birth pulse and their population dynamic consequences. J Math Biol 2002;44(2):185–99. [11] Gains RE, Mawhin JL. Coincidence degree and nonlinear differential equations. Berlin: Springer-Verlag; 1977. [12] Zhen J, Ma Z, Han M. The existence of periodic solutions of the n-species Lotka–Volterra competition systems with impulsive. Chaos, Solitons & Fractals 2004;22:181–8. [13] Zhang S, Tan D, Chen L. The periodic n-species Gilpin–Ayala competition system with impulsive effect. Chaos, Solitons & Fractals 2005;26:507–17.