Electronic Journal of Qualitative Theory of Differential Equations 2012, No. 35, 1-27; http://www.math.u-szeged.hu/ejqtde/
On the Lyapunov Funtional of Leslie-Gower
Predator-Prey Models with Time-Delay and Holling's
Funtional Responses
Chao-Pao Ho
1
, Che-Hao Lin
1 ,∗
, Huang-Nan Huang
1 , 2
1
Department ofMathematis, Tunghai University,
Taihung,Taiwan 40704, R.O.C.
2
HarbinInstitute ofTehnology ShenzhenGraduate Shool,
Shenzhen518055, China
Abstrat
The global stability on the dynamial behavior of the Leslie-Gower predator-prey
systemwithdelayedpreyspeigrowthisanalyzedbyonstruting theorresponding
Lyapunovfuntional. ThreedierenttypesoffamousHolling'sfuntionalresponsesare
onsideredinthepresentstudy. Thesuientonditionsfortheglobalstabilityanalysis
oftheunique positiveequilibrium point arederived aordingly. A numerial example
ispresentedtoillustrate theeetofdierentHolling-Type funtionalresponsesonthe
global stabilityof theLeislie-Gower predator-preymodel.
Keywords: Leslie-Gowerpredator-preymodels,globalstability,timedelay,Lyapunovfun-
tional, Holling'sfuntional response
2010 MSC: 34D23, 37N25, 92D25
1 Introdution
Predator-prey models have been studied by many authorsfor a long time. Most of studies
areinterestedintheglobalstabilityoftheuniquepositiveequilibriumpointofthepredator-
preysystemswithorwithoutdelay. Popularmethodsintheglobalstabilityofpredator-prey
∗
Correspondingauthor. E-mail: linhthu.edu.tw
Lyapunov funtion[1,2,6,7,8℄ toemploy theDulaCriterion plus thePoinaré-Bendixson
Theorem[10℄thelimitylestabilityanalysis[10,11,13℄andtheomparisonmethod[11,13℄.
But amore realistimodelshouldinludesome of the past states ofthe populationsystem;
that is, a real system should be modeled with time delay. As disussed in the referenes
[14, 15, 17℄, the global stability analysis for the system with time delay relied mainly on
onstruting aorresponding Lyapunov funtional.
TheglobalstabilityfortheLotka-Volterramodelhasbeenextensivelyaddressed,e.g. see
[24℄. Asoneofthefamousmodelindesribingthedynamibehaviorofpredator-preysystem,
the arryingapaityofthe predatorpopulationintheLotka-Volterramodelisindependent
of the prey population. Atually, the arrying apaity of the predator population should
depend onthe prey populationwhihresults intothe so-alledLeslie-Gowermodelwhih is
a Kolmogorov-Typemodeland is of the form:
˙
x(t) = x(t)
r
1 − x(t) K
− p(x)y(t),
(1.1)˙
y(t) = y(t)
δ − β y(t) x(t)
(1.2)
where
x(t)
andy(t)
denote the density of prey and predator, respetively;r
,β
andδ
arepositive onstants; and
K
is the environment arrying apaity. Also,p(x)
denotes thefuntional response of the predator. This system has an unique positive equilibrium point.
VariousmodiationsofLeslie-Gowermodelsand assoiatedglobalstabilityproblemanbe
referred to[12℄ and the referenes ited therein.
There are many dierent kinds of the predator-prey models with time delay in the lit-
erature, for more details we an refer to [3℄, [5℄ and [20℄. The disussion on the eet of
time delay to the dynami behavior of the system (1.1) and (1.2) are mainly fous on the
Leslie-Gower terms in (1.2). Alternatively, we are onerned with the eet of the single
time delay
τ
onthe logistiterm of the prey,x(t − τ )/K
, in (1.1) whih wasrst proposedand disussed by [18℄. The time delay appearing in the intra-spei interation term of
the prey equation represents a delayed prey growth eet. The stability, bifuration, and
periodisolutions about similar predator-preysystems are extensively studied in literature,
e.g., [4,9, 16, 21, 22, 23,25,26℄.
Inpresentstudy,weestablishglobalstabilityanalysisofLeslie-Gowerpredator-preymod-
els with time delay. Three dierent funtional responses of the predator, i.e.,
p(x)
term in(1.1),are onsidered byonstrutingtheir orrespondingLyapunov funtionalsaordingto
the Korobeinikov approahfor the non-delaymodel[13℄. This paper isorganized as follows.
Insetion2,weintroduesomeusefuldenitionsandtheoremsandtheboundofthedynam-
ialbehavior of the Leslie-Gowerpredator-prey system with a singledelay. In setion3, we
and III funtionalresponses . Finally,weillustrateour resultsbysome numerialexamples.
2 The Model with Time Delay
2.1 Preliminaries
Dene
C ≡ C([−τ, 0], R n )
isthe Banah spae of ontinuous funtions mappingthe interval[−τ, 0]
intoR n with the topology of uniform onvergene; i.e., for φ ∈ C
, the norm of φ
is
denedas
kφk = sup
θ∈[−τ,0]
|φ(θ)|
,where|·|
isanynorminR n. Denex t ∈ C
asx t (θ) = x (t +θ)
,
θ ∈ [−τ, 0]
. Considerthefollowinggeneralnonlinearautonomoussystemofdelaydierential
equation
˙
x (t) = f ( x t ),
(2.1)where
f : Ω → R n and Ω
is a subset of C
. In this paper, we need the following denitions,
theorems and lemmas.
Denition 2.1. [15℄
1. Thesolution
x = 0
ofthesystem(2.1)
issaidtobestableif,foranyσ ∈ R
,ε > 0
,thereis a
δ = δ(ε, σ)
suh thatφ ∈ B(0, δ)
impliesx t (σ, φ) ∈ B ( 0 , ε)
fort ≥ σ
. Otherwise,we say
x = 0
is unstable.2. Thesolution
x = 0
ofthe system(2.1)
issaidtobeasymptotiallystableif itisstable and there is ab 0 = b(σ) > 0
suh thatφ ∈ B (0, b 0 )
impliesx (σ, φ)(t) → 0
ast → ∞
.3. The solution
x = 0
of the system(2.1)
is said tobe uniformlystable if the numberδ
inthe denition of stableis independent of
σ
.4. The solution
x = 0
ofthe system(2.1)
is said tobeuniformlyasymptotiallystableif it is uniformlystable and there is ab 0 > 0
suh that, for everyη > 0
, there is at 0 (η)
suh that
φ ∈ B(0, b 0 )
impliesx t (σ, φ) ∈ B( 0 , η)
fort ≥ σ + t 0 (η)
,for everyσ ∈ R
.Denition2.2. [23℄System
(2.1)
issaidtobeuniformlypersistentifthereexists aompatregion
D ⊂
intR 2 + suhthateverysolutionofthesystem(2.1)
eventuallyentersandremains
in the region
D
.Lemma 2.3. [15℄ Let
u(·)
andw(·)
be nonnegative ontinuous salar funtions suh thatu(0) = w(0) = 0
;w(s) > 0
fors > 0
,lim
s→∞ u(s) = +∞ and that V : C → R
is ontinuous
and satises
V (φ) ≥ u(|φ(0)|), V ˙ (φ) ≤ −w(|φ(0)|).
Then
x = 0
is globally asymptotially stable. That is, every solution of the system(2.1)
approahes
x = 0
ast → +∞
.2.2 The Leslie-Gower System
Consider the Leslie-Gowerpredator-prey system with time delay
τ
modeledby˙
x(t) = x(t)
r
1 − x(t − τ) K
− p(x) x(t) y(t)
,
˙
y(t) = y(t)
δ − β y(t) x(t)
(2.2)with the initialonditions
x(θ) = φ(θ) ≥ 0, θ ∈ [−τ, 0], φ ∈ C([−τ, 0], R ),
x(0) > 0, y(0) > 0,
(2.3)where
r, K, c, β, δ
andτ
are positive onstants,x
andy
denote the densities of prey andpredator population, respetively. The biologialpopulationis to bedisussed and we need
onlytoonsiderthe rstquadrantin
xy
-plane. Thefollowingonsistentondition with(2.2)is assumed:
(A)
p ∈ C 1 ([0, ∞), [0, ∞))
;p(0) = 0
andp ′ (x) ≥ 0
forallx ≥ 0
.The popular funtional responses of the predator,
p(x)
, in the literature arep(x) = cx p(x) = c 1+x x , and p(x) = c 1+x x 2 2 of the Holling-type I, II, and III, respetively, for some
positive onstant c
. And itis evident that p(x) ≤ c max{x, x 2 }
for these three responses.
c
. And itis evident thatp(x) ≤ c max{x, x 2 }
for these three responses.Lemma2.4. Everysolution ofthe system
(2.2)
withthe initialonditions(2.3)
existsin theinterval
[0, ∞)
and remains positive for allt ≥ 0
.proof. It istrue beause
x(t) = x(0) exp ˆ t
0
r
1 − x(s − τ ) K
− p(x(s)) x(s) y(s)
ds
, y(t) = y(0) exp
ˆ t
0
δ − β y(s) x(s)
ds
and
x(0) > 0
andy(0) > 0
.Lemma 2.5. Let
(x(t), y(t))
denote the solution of(2.2)
with the initial onditions(2.3)
,then
0 < x(t) ≤ M, 0 < y(t) ≤ L
(2.4)eventually for all large t, where
M = Ke rτ ,
(2.5)L = δ
β M.
(2.6)proof. Now, we want toshow that there exists a
T > 0
suh thatx(t) ≤ M
fort > T
. ByLemma 2.4, we know that solutions of the system (2.2) with the initialonditions (2.3) are
positive, hene by assumption (A),and (2.2) beomes
˙
x(t) = rx(t)
1 − x(t − τ ) K
− p(x(t))y(t)
≤ x(t)
r
1 − x(t − τ) K
.
(2.7)Taking
M ∗ = K(1 + k 1 )
for0 < k 1 < e rτ − 1
. The situation ofx(t)
with respet toM ∗ is
ategorized intotwopossibleases.
Case 1: Suppose
x(t)
isnot osillatoryaboutM ∗. That is, there exists a T 0 > 0
suhthat
either
x(t) ≤ M ∗ fort > T 0 (2.8)
or
x(t) > M ∗ for t > T 0 .
(2.9)
If(2.8) holds, then for
t > T = T 0,
x(t) ≤ M ∗ = K(1 + k 1 ) < Ke rτ = M.
Suppose (2.9) holds, (2.7) impliesthat for
t > T 0 + τ
˙
x(t) ≤ rx(t)
1 − x(t − τ ) K
< −k 1 rx(t).
It follows that
ˆ t
T 0 +τ
˙ x(s) x(s) ds <
ˆ t
T 0 +τ
(−k 1 r) ds = −k 1 r(t − T 0 − τ ),
then
0 < x(t) < x(T 0 + τ ) e −k 1 r(t−T 0 −τ) → 0
ast → ∞
. That is,lim
t→∞ x(t) = 0 by
the Squeeze Theorem. It ontradits to (2.9). Therefore, there exist a
T 1 > T 0 suh
that
x(T 1 ) ≤ M
. Ifx(t) ≤ M
for allt ≥ T 1, and then there exist a T > 0
suh that
x(t) ≤ M
forall t ≥ T
.
Case 2: Suppose
x(t)
is osillatory aboutM ∗, then there must exist a T 2 > T 1 suh that
T 2 be the rst time whih x(T 2 ) > M ∗. Therefore, there exists a T 3 > T 2 suh that
T 3 be the rst time whih x(T 3 ) < M ∗ by above disussion. By above, we know that
x(T 1 ) ≤ M ∗, x(T 2 ) > M ∗ and x(T 3 ) ≤ M ∗ where T 1 < T 2 < T 3. Then, by the
T 3 > T 2 suh that
T 3 be the rst time whih x(T 3 ) < M ∗ by above disussion. By above, we know that
x(T 1 ) ≤ M ∗, x(T 2 ) > M ∗ and x(T 3 ) ≤ M ∗ where T 1 < T 2 < T 3. Then, by the
x(T 3 ) < M ∗ by above disussion. By above, we know that
x(T 1 ) ≤ M ∗, x(T 2 ) > M ∗ and x(T 3 ) ≤ M ∗ where T 1 < T 2 < T 3. Then, by the
x(T 2 ) > M ∗ and x(T 3 ) ≤ M ∗ where T 1 < T 2 < T 3. Then, by the
T 1 < T 2 < T 3. Then, by the
Intermediate Value Theorem, there exists
T 4 and T 5 suh that
x(T 4 ) = M ∗ , T 1 ≤ T 4 < T 2 , x(T 5 ) = M ∗ , T 2 < T 5 ≤ T 3
x(T 4 ) = M ∗ , T 1 ≤ T 4 < T 2 , x(T 5 ) = M ∗ , T 2 < T 5 ≤ T 3
and
x(t) > M ∗ for T 4 < t < T 5. Hene there is a T 6 ∈ (T 4 , T 5 )
suh that x(T 6 )
is a
T 6 ∈ (T 4 , T 5 )
suh thatx(T 6 )
is aloalmaximum and (2.7), we have
0 = ˙ x(T 6 ) ≤ x(T 6 )
r
1 − x(T 6 − τ ) K
and
x(T 6 − τ) ≤ K.
Integrating both sides of (2.7) on the interval
[T 6 − τ, T 6 ]
, we haveln
x(T 6 ) x(T 6 − τ)
= ˆ T 6
T 6 −τ
˙ x(s) x(s) ds ≤
ˆ T 6
T 6 −τ
r
1 − x(s − τ) K
ds ≤ rτ
and
x(T 6 ) ≤ x(T 6 − τ ) e rτ ≤ Ke rτ = M.
Applying the same operation on the amplitude of the trajetory
x(t)
, we an nd asequenes of
T 6 suhthat every x(T 6 )
isa loalmaximumof x(t)
,and itsamplitude is
less then
M
. Hene we an onlude that there exists aT > 0
suh thatx(t) ≤ M
fort ≥ T.
(2.10)Now, wewant toshowthat
y(t)
is bounded above byL
eventually foralllarget
. By(2.10),it follows that for
t > T
˙
y(t) = y(t)
δ − β y(t) x(t)
≤ y(t)
δ − β M y(t)
= δy(t)
"
1 − y(t)
δM β
# .
Therefore,
y(t) ≤ δM/β = L
fort > T
. The proof is omplete.Lemma 2.6. Suppose that the system
(2.2)
satisesr − c max{1, M } L > 0,
(2.11)where
L
dened by(2.6)
. Thenthe system(2.2)
isuniformlypersistent. That is, there existsm
,l
andT ∗ > 0
suh thatm ≤ x(t) ≤ M
andl ≤ y(t) ≤ L
fort ≥ T ∗, i = 1, 2
.
proof. ByLemma 2.5, equation (2.2) follows that for
t ≥ T + τ
˙
x(t) ≥ x(t)
r
1 − M K
− p(x(t)) x(t) L
≥ x(t)
r
1 − M K
− c max{1, M } L
.
(2.12)Integrating both sides of (2.12) on
[t − τ, t]
, wheret ≥ T + τ
, then we havex(t) ≥ x(t − τ ) e (r(1− M K )−c max{1,M} L)τ .
That is,
x(t − τ ) ≤ x(t) e −(r(1− M K )−c max{1,M} L)τ .
(2.13)From (2.2) that for
t ≥ T + τ
˙
x(t) = x(t)r
1 − x(t − τ ) K
− p(x(t))y(t)
≥ x(t) h
r − c max{1, M} L − r
K e −(r(1−M/K)−c max{1,M} L)τ x(t) i
= (r − c max{1, M } L) x(t)
"
1 − x(t)
K(r−c max{1,M} L)
r e (r(1− M K )−c max{1,M} L)τ
# .
It follows that
lim inf
t→∞ x(t) ≥ K r − c max{1, M}L
r e (r−r M K −c max{1,M}L)τ ≡ m ¯
and
m > ¯ 0
. Henex(t) > m ¯ − ε 1 ≡ m > 0
with a positive numberε 1, for large t
. Next,
˙
y(t) ≥ y(t)
δ − β m y(t)
= δy(t)
"
1 − y(t)
δm β
# ,
it implies
lim inf
t→∞ y(t) ≥ δm
β ≡ l.
Therefore,
y(t) > ¯ l − ε 2 ≡ l > 0
with apositivenumberε 2,for large t
. Let
D = {(x, y) | m ≤ x ≤ M, l ≤ y ≤ L}
beaboundedompat regionin
R 2 + thathas positivedistane fromoordinatehyperplanes.
Hene we obtainthat there exists a
T ∗ > 0
suhthat ift ≥ T ∗, then every positivesolution
of the system (2.2) with the initial onditions (2.3) eventually enters and remains in the
region
D
;that is, the system (2.2)is uniformlypersistent.3 The Lyapunov Funtional
The equilibriumpoint
E ∗ = (x ∗ , y ∗ )
of the Leslie-Gower system˙
x(t) = x(t)
r
1 − x(t − τ) K
− p(x) x(t) y(t)
,
˙
y(t) = y(t)
δ − β y(t) x(t)
satises
r
1 − x ∗ K
= p(x ∗ )
x ∗ y ∗ = δ β p(x ∗ ) δx ∗ = βy ∗
orequivalently,is the solution of the systems
x ∗ K + δ
rβ p(x ∗ ) = 1,
βy ∗ − δx ∗ = 0.
One the equilibrium point
E ∗ is found, we an obtain an perturbed system to onstrut
the Lyapunov funtional. Now three dierent types of Holling's funtional responses are
onsidered:
Type I:
p(x) = c x
,Type II:
p(x) = c 1+x x ,
Type III:
p(x) = c 1+x x 2 2.
The Lyapunov funtionals for eah Holling's funtional responses are derived based upon
the formulaproposed by Korobeinikov [13℄ fornon-delay model. Althoughthe Tsai's paper
[22℄ presents a similar result, ours provides larger delay bound for asymptotially stability
whihwillbedemonstrated by various examples innext setion.
3.1 Holling-Type I Funtional Response
When
p(x) = cx
, the equilibriumpointE ∗ (x ∗ , y ∗ )
isthen given byx ∗ = Kβr
βr + δKc , y ∗ = δKr
βr + δKc .
Theorem 3.1. Let
p(x) = c x
be the funtional response of Holling-Type I and the time delayτ
satisesr − c max{1, M } L > 0,
(3.1)M 2 τ < 2 βK
rc ,
(3.2)r δK + c
2β
Mτ < 1
δ ,
(3.3)B 2 − 4AC < 0,
(3.4)where
A = r
δK − rMcτ
2βK − r 2 Mτ
δK 2 ,
(3.5)B = max
− c β + 1
M ,
− c
β + 1 m
,
(3.6)C = 1
M − rMcτ
2βK ,
(3.7)with
M
andm
denedin Lemmas 2.5 and2.6, respetively, thenthe unique positiveequilib- riumE ∗ of the system (2.2)
is globally asymptotially stable.
proof. Dene
z(t) = (z 1 (t), z 2 (t))
byz 1 (t) = x(t) − x ∗
x ∗ , z 2 (t) = y(t) − y ∗ y ∗ .
From (2.2), the perturbed system is given by
˙
z 1 (t) = [1 + z 1 (t)]
−cy ∗ z 2 (t) − rx ∗ z 1 (t − τ ) K
,
(3.8)˙
z 2 (t) = [1 + z 2 (t)]
δx ∗ z 1 (t) − βy ∗ z 2 (t) x ∗ [1 + z 1 (t)]
.
(3.9)Let
V 1 (z(t)) = {z 1 (t) − ln[1 + z 1 (t)]} + {z 2 (t) − ln[1 + z 2 (t)]}
βy ∗ ,
(3.10)then from(3.8) and (3.9),we have
V ˙ (z(t)) =
− c
β + 1
x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t) − 1
x ∗ [1 + z 1 (t)] z 2 2 (t) − rz 1 (t)z 1 (t − τ )
δK .
(3.11)Suppose the inequality
r − c max{1, M } L > 0
hold, then by Lemma 2.6, there exists aT ∗ > 0
suh thatm ≤ x ∗ [1 + z 1 (t)] ≤ M
fort > T ∗. The equation(3.11) implies that
V ˙ 1 (z(t)) ≤
− c
β + 1
x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t) − 1
M z 2 2 (t) − rz 1 (t)z 1 (t − τ )
δK
(3.12)and sine
− rz 1 (t)z 1 (t − τ) δK
= − rz 1 (t) δK
z 1 (t) − ˆ t
t−τ
˙ z 1 (s)ds
= − r
δK z 1 2 (t) + r δK
ˆ t
t−τ
[1 + z 1 (s)]
−cy ∗ z 1 (t)z 2 (s) − rx ∗ z 1 (t)z 1 (s − τ ) K
ds
≤ − r
δK z 1 2 (t) + r δK
ˆ t
t−τ
[1 + z 1 (s)]
cδx ∗
2β (z 2 1 (t) + z 2 2 (s)) + rx ∗
2K (z 2 1 (t) + z 1 2 (s − τ ))
ds
≤
− r
δK + rMcτ
2βK + r 2 Mτ 2δK 2
z 1 2 (t) + rMc 2βK
ˆ t
t−τ
z 2 2 (s)ds + r 2 M 2δK 2
ˆ t
t−τ
z 1 2 (s − τ )ds
then we have
V ˙ 1 (z(t)) ≤
− c
β + 1
x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t) +
− r
δK + rMcτ
2βK + r 2 Mτ 2δK 2
z 1 2 (t) + r 2 M
2δK 2 ˆ t
t−τ
z 2 1 (s − τ)ds − 1
M z 2 2 (t) + rMc 2βK
ˆ t
t−τ
z 2 2 (s)ds.
(3.13)Let
V 2 (z(t)) = rMc 2βK
ˆ t
t−τ
ˆ t
s
z 2 2 (γ)dγds + r 2 M 2δK 2
ˆ t
t−τ
ˆ t
s
z 1 2 (γ − τ)dγds
(3.14)and
V 3 (z(t)) = r 2 Mτ 2δK 2
ˆ t
t−τ
z 1 2 (s)ds,
(3.15)then
V ˙ 2 (z(t)) = r 2 Mτ
2δK 2 z 1 2 (t − τ ) − r 2 M 2δK 2
ˆ t
t−τ
z 1 2 (γ − τ)dγ + rMcτ
2βK z 2 2 (t) − rMc 2βK
ˆ t
t−τ
z 2 2 (γ)dγ
(3.16)and
V ˙ 3 (z(t)) = r 2 Mτ
2δK 2 z 2 1 (t) − r 2 Mτ
2δK 2 z 1 2 (t − τ ).
(3.17)Now we denea Lyapunov funtional
V (z(t))
asV (z(t)) = V 1 (z(t)) + V 2 (z(t)) + V 3 (z(t)),
(3.18)then from(3.13), (3.16) and (3.17) itfollows that for
t ≥ T ∗ V ˙ (z(t)) ≤ −
r
δK − rMcτ
2βK − r 2 Mτ δK 2
z 1 2 (t) +
− c
β + 1
x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t)
− 1
M − rMcτ 2βK
z 2 2 (t).
(3.19)By (3.19),there is
ε > 0
suh thatV ˙ (z(t)) ≤ −ε(z 1 2 (t) + z 2 2 (t))
(3.20)if and only if
A > 0
,C > 0
andB 1 2 − 4AC < 0
whereA
andC
are dened by (3.5) and(3.7), and
B 1 = − c
β + 1
x ∗ [1 + z 1 (t)]
forallpossibletrajetory
(x ∗ [1+z 1 (t)], y ∗ [1+z 2 (t)])
. Sinem ≤ x ∗ [1+ z 1 (t)] ≤ M
fort > T ∗,
i.e.,
− c β + 1
M ≤ − c
β + 1
x ∗ [1 + z 1 (t)] ≤ − c β + 1
m
and by dene
B = max
− c β + 1
M ,
− c
β + 1 m
Then the ondition for
V ˙ (z(t)) ≤ 0
beomesM 2 τ < 2 βK rc , Mτ < 2βK
δKc + 2βr , B 2 − 4AC < 0,
where
A
,B
andC
are given by equations (3.5)-(3.7).Dene
w(s) = εs 2, then w
is nonnegative ontinuous on [0, ∞)
, w(0) = 0
and w(s) > 0
for
s > 0
. It follows that fort ≥ T ∗
V ˙ (z(t)) ≤ −ε[z 1 2 (t) + z 2 2 (t)] = −w(|z(t)|).
Now, we want to nd a funtion
u
suh thatV (z(t)) ≥ u(|z(t)|)
. From (3.10), (3.14) and(3.15) that
V (z(t)) ≥ V 1 (z(t)) = 1
βy ∗ {z 1 (t) − ln[1 + z 1 (t)] + z 2 (t) − ln[1 + z 2 (t)]} .
(3.21)By the Taylor Theorem, we have
z i (t) − ln[1 + z i (t)] = z 2 i (t)
2[1 + θ i (t)] 2 ,
(3.22)where
θ i (t) ∈ (0, z i (t))
or(z i (t), 0)
fori = 1, 2
. Consider the all the possible ases forθ i:
1.
0 < θ i (t) < z i (t)
, fori = 1, 2
,2.
−1 < z i (t) < θ i (t) < 0
,fori = 1, 2
,3.
0 < θ 1 (t) < z 1 (t), −1 < z 2 (t) < θ 2 (t) < 0
,4.
−1 < z 1 (t) < θ 1 (t) < 0, 0 < θ 2 (t) < z 2 (t)
,we an nd a parameter
N e
dened byN e = min
( 1 2βy ∗
x ∗ M
2
, 1 2βy ∗
y ∗ L
2 )
suh that
V (z(t)) ≥ N e |z(t)| 2 .
Dene
u(s) = Ns e 2, then u
is nonnegative ontinuous on [0, ∞)
with u(0) = 0
, u(s) > 0
for
s > 0
and lim
s→∞ u(s) = +∞. Thuswe have
V (z(t)) ≥ u(|z(t)|)
fort ≥ T ∗ .
Hene the equilibrium point
E ∗ of the system (2.2) is globally asymptotially stable by
Lemma 2.3.
Remark 1. In the proof of Theorem 3.1, the orresponding Lyapunov funtional(3.18) for
τ = 0
beomesV (x, y) = 1 βy ∗
x
x ∗ − 1 + ln x ∗ x + y
y ∗ − 1 + ln y ∗ y
(3.23)
whihthe sameas theLyapunov funtionalby Korobeinikov [13℄with anextra onstant
−2
and multipliativeonstant
1/βy ∗ suh that V (x ∗ , y ∗ ) = 0
.
3.2 Holling-Type II Funtional Response
When
p(x) = c 1+x x ,the equilibriumpoint E ∗ (x ∗ , y ∗ )
isobtained by solving
x ∗2 + K
1 K + δc
rβ − 1
x ∗ − 1 = 0, y ∗ = δ
β x ∗ .
Theorem 3.2. Let
p(x) = c 1+x x be the funtional response of Holling-Type II and the time
delay τ
satises
r − c max{1, M } L > 0,
(3.24)M 2
1 + m τ < 2 Kβ rc ,
(3.25)r
δK + c 2β
3x ∗ + 5 (1 + x ∗ )(1 + m)
Mτ + Kc βr
1
1 + m − 1
(1 + x ∗ )(1 + M )
< 1
δ ,
(3.26)B 2 − 4AC < 0,
(3.27)A = r(K − rMτ )
δK 2 + c
β(1 + x ∗ )(1 + M ) − c
β(1 + m) − rMcτ (3x ∗ + 5)
2βK(1 + m)(1 + x ∗ )
(3.28)B = max
− c
β(1 + m) + 1 M
,
− c
β(1 + M) + 1 m
,
(3.29)C = 1
M − rMcτ
2βK(1 + m)
(3.30)with
M
andm
denedin Lemmas 2.5 and2.6, respetively, thenthe unique positiveequilib- riumE ∗ of the system (2.2)
is globally asymptotially stable.
proof. Dene
z(t) = (z 1 (t), z 2 (t))
byz 1 (t) = x(t) − x ∗
x ∗ , z 2 (t) = y(t) − y ∗ y ∗ .
From (2.2), the perturbed system is given by
˙
z 1 (t) = [1 + z 1 (t)]
cy ∗ z 1 (t)
1 + x ∗ [1 + z 1 (t)] − cy ∗ z 1 (t)
(1 + x ∗ ) (1 + x ∗ [1 + z 1 (t)]) − cy ∗ z 2 (t) 1 + x ∗ [1 + z 1 (t)]
− rx ∗ z 1 (t − τ ) K
,
(3.31)˙
z 2 (t) = [1 + z 2 (t)]
δx ∗ z 1 (t) − βy ∗ z 2 (t) x ∗ [1 + z 1 (t)]
.
(3.32)Let
V 1 (z(t)) = {z 1 (t) − ln[1 + z 1 (t)]} + {z 2 (t) − ln[1 + z 2 (t)]}
βy ∗ ,
(3.33)then from(3.31) and (3.32), we have
V ˙ 1 (z(t)) = 1 βy ∗
z 1 (t) ˙ z 1 (t)
1 + z 1 (t) + z 2 (t) ˙ z 2 (t) 1 + z 2 (t)
=
− c
β (1 + x ∗ [1 + z 1 (t)]) + 1 x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t) + cz 1 2 (t)
β (1 + x ∗ [1 + z 1 (t)])
− cz 1 2 (t)
β(1 + x ∗ ) (1 + x ∗ [1 + z 1 (t)]) − z 2 2 (t)
x ∗ [1 + z 1 (t)] − rz 1 (t)z 1 (t − τ) δK
≤
− c
β (1 + x ∗ [1 + z 1 (t)]) + 1 x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t) +
c
β(1 + m) − c
β(1 + x ∗ )(1 + M)
z 1 (t) 2 − 1
M z 2 2 (t) − r z 1 (t) z 1 (t − τ)
δK .
Suppose the inequality
r − c max{1, M } L > 0
hold, then by Lemma 2.6, there exists aT ∗ > 0
suh thatm ≤ x ∗ [1 + z 1 (t)] ≤ M
fort > T ∗. Sine
− rz 1 (t)z 1 (t − τ)
δK = − rz 1 (t) δK
z 1 (t) − ˆ t
t−τ
˙ z 1 (s)ds
= − r
δK z 1 2 (t) + r δK
ˆ t
t−τ
[1 + z 1 (s)]
cy ∗ z 1 (t)z 1 (s)
1 + x ∗ [1 + z 1 (s)] − cy ∗ z 1 (t)z 1 (s)
(1 + x ∗ ) (1 + x ∗ [1 + z 1 (s)]) − cy ∗ z 1 (t)z 2 (s) 1 + x ∗ [1 + z 1 (s)]
− rx ∗ z 1 (t)z 1 (s − τ ) K
ds
≤ − r
δK z 1 2 (t) + r δK
ˆ t
t−τ
x ∗ [1 + z 1 (s)]
cδ
β (1 + x ∗ [1 + z 1 (s)]) + cδ
β(1 + x ∗ ) (1 + x ∗ [1 + z 1 (s)])
z 1 2 (t) + z 1 2 (s) 2
+ cδ
β (1 + x ∗ [1 + z 1 (s)])
z 1 2 (t) + z 2 2 (s)
2 + r
K
z 2 1 (t) + z 2 1 (s − τ) 2
ds
≤
− r
δK + rMcτ
βK(1 + m) + rMcτ
2βK(1 + x ∗ )(1 + m) + r 2 Mτ 2δK 2
z 1 2 (t) + rM
2K
c
β(1 + m) + c
β(1 + x ∗ )(1 + m) ˆ t
t−τ
z 2 1 (s)ds
+ rMc
2Kβ(1 + m) ˆ t
t−τ
z 2 2 (s)ds + r 2 M 2δK 2
ˆ t
t−τ
z 1 2 (s − τ )ds
then we have
V ˙ 1 (z(t)) ≤
− c
β (1 + x ∗ [1 + z 1 (t)]) + 1 x ∗ (1 + z 1 (t))
z 1 (t)z 2 (t) +
c
β(1 + m) − c
β(1 + x ∗ )(1 + M) − r
δK + rMcτ Kβ(1 + m)
+ rMcτ
2Kβ(1 + x ∗ )(1 + m) + r 2 Mτ 2δK 2
z 1 2 (t)
− 1
M z 2 2 (t) + rMc 2Kβ
1
1 + m + 1
(1 + x ∗ )(1 + m) ˆ t
t−τ
z 1 2 (s)ds
+ rMc
2Kβ(1 + m) ˆ t
t−τ
z 2 2 (s)ds + r 2 M 2δK 2
ˆ t
t−τ
z 1 2 (s − τ )ds.
(3.34)V 2 (z(t)) = rMc 2Kβ
1
1 + m + 1
(1 + x ∗ )(1 + m) ˆ t
t−τ
ˆ t
s
z 1 2 (γ)dγds
+ rMc
2Kβ(1 + m) ˆ t
t−τ
ˆ t
s
z 2 2 (γ)dγds + r 2 M 2δK 2
ˆ t
t−τ
ˆ t
s
z 1 2 (γ − τ )dγds
(3.35)and
V 3 (z(t)) = r 2 Mτ 2δK 2
ˆ t
t−τ
z 1 2 (s)ds,
(3.36)then
V ˙ 2 (z(t))
= rMc
2Kβ 1
1 + m + 1
(1 + x ∗ )(1 + m)
z 2 1 (t) − rMc 2Kβ
1
1 + m + 1
(1 + x ∗ )(1 + m) ˆ t
t−τ
z 1 2 (γ)dγ
+ rMcτ
2Kβ(1 + m) z 2 2 (t) − rMc 2Kβ(1 + m)
ˆ t
t−τ
z 2 2 (γ)dγ + r 2 Mτ
2δK 2 z 1 2 (t − τ) − r 2 M 2δK 2
ˆ t
t−τ
z 2 1 (γ − τ)dγ.
(3.37)and
V ˙ 3 (z(t)) = r 2 Mτ
2δK 2 z 2 1 (t) − r 2 Mτ
2δK 2 z 1 2 (t − τ ).
(3.38)Now we denea Lyapunov funtional
V (z(t))
asV (z(t)) = V 1 (z(t)) + V 2 (z(t)) + V 3 (z(t)),
(3.39)then from(3.34), (3.37) and (3.38) itfollows that for
t ≥ T ∗ V ˙ (z(t)) = ˙ V 1 (z(t)) + ˙ V 2 (z(t)) + ˙ V 3 (z(t))
≤ −
r(K − rMτ )
δK 2 + c
β(1 + x ∗ )(1 + M) − c
β(1 + m) − rMcτ (3x ∗ + 5) 2Kβ(1 + x ∗ )(1 + m)
z 1 2 (t) +
− c
β (1 + x ∗ [1 + z 1 (t)]) + 1 x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t)
− 1
M − rMcτ
2Kβ(1 + m)
z 2 2 (t)
(3.40)By (3.40),there is
ε > 0
suh thatV ˙ (z(t)) ≤ −ε(z 1 2 (t) + z 2 2 (t))
(3.41)if and only if
A > 0
,C > 0
andB 2 2 − 4AC < 0
whereA
andC
are dened by (3.28) and(3.30), and
B 2 = − c
β (1 + x ∗ [1 + z 1 (t)]) + 1 x ∗ [1 + z 1 (t)]
forallpossibletrajetory
(x ∗ [1+z 1 (t)], y ∗ [1+z 2 (t)])
. Sinem ≤ x ∗ [1+ z 1 (t)] ≤ M
fort > T ∗,
i.e.,
− c
β(1 + m) + 1
M ≤ − c
β (1 + x ∗ [1 + z 1 (t)]) + 1
x ∗ [1 + z 1 (t)] ≤ − c
β(1 + M ) + 1 m
and by dene
B = max
− c
β(1 + m) + 1 M
,
− c
β(1 + M ) + 1 m
Then the ondition for
V ˙ (z(t)) ≤ 0
beomesM 2
1 + m τ < 2 Kβx rc , r
δK + c 2β
3x ∗ + 5 (1 + x ∗ )(1 + m)
Mτ + Kc βr
1
1 + m − 1
(1 + x ∗ )(1 + M )
< 1 δ , B 2 − 4AC < 0,
where
A
,B
andC
are given by (3.28)-(3.30).Dene
w(s) = εs 2, then w
is nonnegative ontinuous on [0, ∞)
, w(0) = 0
and w(s) > 0
for
s > 0
. It follows that fort ≥ T ∗
V ˙ (z(t)) ≤ −ε[z 1 2 (t) + z 2 2 (t)] = −w(|z(t)|).
To nd a funtion
u
suh thatV (z(t)) ≥ u(|z(t)|)
, sine the following relationship is still hold for Holling's Type IIV (z(t)) ≥ V 1 (z(t)) = 1
βy ∗ {z 1 (t) − ln[1 + z 1 (t)] + z 2 (t) − ln[1 + z 2 (t)]} .
andthen bythe sameargumentproposedintheproof ofTheorem 3.1,wean establishthat
this is a nonnegative ontinuous funtion
u
dened on[0, ∞)
withu(0) = 0
,u(s) > 0
fors > 0
andlim
s→∞ u(s) = +∞ and
V (z(t)) ≥ u(|z(t)|)
fort ≥ T ∗ .
Hene the equilibrium point
E ∗ of the system (2.2) is globally asymptotially stable by
Lemma 2.3.
When
p(x) = c 1+x x 2 2, the equilibriumpoint E ∗ (x ∗ , y ∗ )
is obtained by solving
x ∗3 + K
δc rβ − 1
x ∗2 + x ∗ − K = 0, y ∗ = δ
β x ∗ .
Theorem3.3. Let
p(x) = c 1+x x 2 2 be thefuntionalresponse ofHolling-TypeIIIandthetime
delay
τ
satisesr − c max{1, M } L > 0,
(3.42)M 2 (M + 2x ∗ )
1 + m 2 τ < 2 Kβ
rc ,
(3.43)r
δK + c 2β
(5 + 3x ∗2 )(M + 2x ∗ ) + 2x ∗ β(1 + x ∗2 )(1 + m 2 )
Mτ + KC
βr
M + 2x ∗
1 + m 2 + M + x ∗
(1 + x ∗ )(1 + m 2 ) − 2x ∗
(1 + x ∗2 )(1 + M 2 )
< 1
δ ,
(3.44)B 2 − 4AC < 0,
(3.45)where
A = r
δK + 2cx ∗
β(1 + x ∗2 )(1 + M 2 ) − c(M + 2x ∗ )
β(1 + m 2 ) − c(M + x ∗ ) β(1 + x ∗ )(1 + m 2 )
− 3rMc(M + 2x ∗ )τ
2Kβ (1 + m 2 ) − rMc(M + 3x ∗ )τ
Kβ(1 + x ∗2 )(1 + m 2 ) − r 2 Mτ
δK 2 (3.46)
B = max
− cM
β(1 + m 2 ) + 1 M
,
− cm
β(1 + M 2 ) + 1 m
(3.47)
C = 1
M − rMc(M + 2x ∗ )τ
2Kβ(1 + m 2 )
(3.48)where
M
andm
dened in Lemmas 2.5 and 2.6, then the unique positive equilibriumE ∗ of
the system
(2.2)
is globally asymptotially stable.proof. Dene
z(t) = (z 1 (t), z 2 (t))
byz 1 (t) = x(t) − x ∗
x ∗ , z 2 (t) = y(t) − y ∗
y ∗ .
˙
z 1 (t) = [1 + z 1 (t)]
cx ∗ y ∗ z 1 (t)
1 + x ∗2 [1 + z 1 (t)] 2 − 2cx ∗ y ∗ z 1 (t)
(1 + x ∗2 ) (1 + x ∗2 [1 + z 1 (t)] 2 ) + cx ∗ y ∗ z 2 1 (t) 1 + x ∗2 [1 + z 1 (t)] 2
− cx ∗ y ∗ z 1 2 (t)
(1 + x ∗2 ) (1 + x ∗2 [1 + z 1 (t)] 2 ) − cx ∗ y ∗ z 2 (t)
1 + x ∗ 2 [1 + z 1 (t)] 2 − cx ∗ y ∗ z 1 (t)z 2 (t) 1 + x ∗ 2 [1 + z 1 (t)] 2
− rx ∗ z 1 (t − τ) K
,
(3.49)˙
z 2 (t) = [1 + z 2 (t)]
δx ∗ z 1 (t) − βy ∗ z 2 (t) x ∗ [1 + z 1 (t)]
.
(3.50)Let
V 1 (z(t)) = {z 1 (t) − ln[1 + z 1 (t)]} + {z 2 (t) − ln[1 + z 2 (t)]}
βy ∗ ,
(3.51)then from(3.49) and (3.50), we have
V ˙ 1 (z(t)) = 1 βy ∗
z 1 (t) ˙ z 1 (t)
1 + z 1 (t) + z 2 (t) ˙ z 2 (t) 1 + z 2 (t)
and after some algebraioperationsimilar tothe proofof Theorem 3.1, itfollows that
V ˙ 1 (z(t)) = − cx ∗ z 1 (t)z 2 (t)
β (1 + x ∗2 [1 + z 1 (t)] 2 ) + z 1 (t)z 2 (t)
x ∗ [1 + z 1 (t)] + cx ∗ z 1 2 (t)
β (1 + x ∗2 [1 + z 1 (t)] 2 )
− 2cx ∗ z 1 2 (t)
β(1 + x ∗2 ) (1 + x ∗2 [1 + z 1 (t)] 2 ) + cx ∗ z 1 3 (t)
β (1 + x ∗2 [1 + z 1 (t)] 2 )
− cx ∗ z 1 3 (t)
β(1 + x ∗2 ) (1 + x ∗2 [1 + z 1 (t)] 2 ) − cx ∗ z 2 1 (t)z 2 (t) β (1 + x ∗2 [1 + z 1 (t)] 2 )
− z 2 2 (t)
x ∗ [1 + z 1 (t)] − rz 1 (t)z 1 (t − τ ) δK
≤
− cx ∗ [1 + z 1 (t)]
β (1 + x ∗2 [1 + z 1 (t)] 2 ) + 1 x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t) + c
β
M + 2x ∗
1 + m 2 − 2x ∗
(1 + x ∗2 )(1 + M 2 ) + M + x ∗ (1 + x ∗2 )(1 + m 2 )
z 1 2 (t)
− z 2 2 (t)
M − rz 1 (t)z 1 (t − τ) δK
Suppose the following inequality
r − c max{1, M } L > 0
hold, then by Lemma 2.6, thereexists a
T ∗ > 0
suh thatm ≤ x ∗ [1 + z 1 (t)] ≤ M
fort > T ∗. Sine
− rz 1 (t)z 1 (t − τ )
δK = − r
δK z 2 1 (t) + r δK
ˆ t
t−τ
z 1 (t) ˙ z 1 (s)ds
≤
− r
δK + rMcτ (M + 3x ∗ )
2βK(1 + m 2 ) + rMcx ∗ τ
Kβ(1 + x ∗2 )(1 + m 2 ) + r 2 Mτ 2δK 2
z 1 2 (t) + rMc
2Kβ
2M + 3x ∗
1 + m 2 + 2(M + 2x ∗ ) (1 + x ∗2 )(1 + m 2 )
ˆ t t−τ
z 1 2 (s)ds + rMc(M + 2x ∗ )
2Kβ(1 + m 2 ) ˆ t
t−τ
z 2 2 (s)ds + r 2 M 2δK 2
ˆ t
t−τ
z 1 2 (s − τ)ds
then we have
V ˙ 1 (z(t)) ≤
− cx ∗ [1 + z 1 (t)]
β (1 + x ∗2 [1 + z 1 (t)] 2 ) + 1 x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t) +
− r
δK − 2cx ∗
β(1 + x ∗2 )(1 + M 2 ) + c(M + 2x ∗ )
β(1 + m 2 ) + rMcτ (M + 3x ∗ ) 2Kβ(1 + m 2 ) rMcτ x ∗
βK(1 + x ∗2 )(1 + m 2 ) + r 2 Mτ 2δK 2
z 2 1 (t)
− 1
M z 2 2 (t) + rMc 2Kβ
2M + 3x ∗
1 + m 2 + 2(M + 2x ∗ ) (1 + x ∗2 )(1 + m 2 )
ˆ t t−τ
z 1 2 (s)ds + rMc(M + 2x ∗ )
2Kβ(1 + m 2 ) ˆ t
t−τ
z 2 2 (s)ds + r 2 M 2δK 2
ˆ t
t−τ
z 2 1 (s − τ)ds
(3.52)Let
V 2 (z(t)) = rMc 2Kβ
2M + 3x ∗
1 + m 2 + 2(M + 2x ∗ ) (1 + x ∗2 )(1 + m 2 )
ˆ t t−τ
ˆ t
s
z 1 2 (γ)dγds + rMc(M + 2x ∗ )
2Kβ (1 + m 2 ) ˆ t
t−τ
ˆ t
s
z 2 2 (γ )dγds + r 2 M 2δK 2
ˆ t
t−τ
ˆ t
s
z 1 2 (γ − τ )dγds
(3.53)and
V 3 (z(t)) = r 2 Mτ 2δK 2
ˆ t
t−τ
z 1 2 (s)ds,
(3.54)then
V ˙ 2 (z(t)) = rMcτ 2Kβ
2M + 3x ∗
1 + m 2 + 2(M + 2x ∗ ) (1 + x ∗2 )(1 + m 2 )
z 1 2 (t)
− rM 2Kβ
2M + 3x ∗
1 + m 2 + 2(M + 2x ∗ ) (1 + x ∗2 )(1 + m 2 )
ˆ t t−τ
z 2 1 (γ )dγ + rMc(M + 2x ∗ )τ
2Kβ(1 + m 2 ) z 2 2 (t) − rMc(M + 2x ∗ ) 2Kβ(1 + m 2 )
ˆ t
t−τ
z 2 2 (γ)dγ + r 2 Mτ
2δK 2 z 1 2 (t − τ ) − r 2 M 2δK 2
ˆ t
t−τ
z 2 1 (γ − τ)dγ.
(3.55)and
V ˙ 3 (z(t)) = r 2 Mτ
2δK 2 z 2 1 (t) − r 2 Mτ
2δK 2 z 1 2 (t − τ ).
(3.56)Now we denea Lyapunov funtional
V (z(t))
asV (z(t)) = V 1 (z(t)) + V 2 (z(t)) + V 3 (z(t)),
(3.57)then from(3.52), (3.55) and (3.56) itfollows that for
t ≥ T ∗ V ˙ (z(t)) = V ˙ 1 (z(t)) + ˙ V 2 (z(t)) + ˙ V 3 (z(t)),
≤ − r
δK + 2cx ∗
β(1 + x ∗2 )(1 + M 2 ) − c(M + 2x ∗ )
β(1 + m 2 ) − c(M + x ∗ ) β(1 + x ∗ )(1 + m 2 )
− 3rMc(M + 2x ∗ )τ
2Kβ (1 + m 2 ) − rMc(M + 3x ∗ )τ
Kβ(1 + x ∗2 )(1 + m 2 ) − r 2 Mτ δK 2
z 2 1 (t) +
− cx ∗ (1 + z 1 (t))
β (1 + x ∗2 [1 + z 1 (t)] 2 ) + 1 x ∗ [1 + z 1 (t)]
z 1 (t)z 2 (t)
− 1
M − rMc(M + 2x ∗ )τ 2Kβ(1 + m 2 )
z 2 2 (t)
(3.58)By (3.58),there is
ε > 0
suh thatV ˙ (z(t)) ≤ −ε(z 1 2 (t) + z 2 2 (t))
(3.59)if and only if
A > 0
,C > 0
andB 3 2 − 4AC < 0
whereA
andC
are dened by (3.46) and(3.48), and
B 3 = − cx ∗ (1 + z 1 (t))
β (1 + x ∗2 [1 + z 1 (t)] 2 ) + 1 x ∗ [1 + z 1 (t)]
forallpossibletrajetory
(x ∗ [1+z 1 (t)], y ∗ [1+z 2 (t)])
. Sinem ≤ x ∗ [1+ z 1 (t)] ≤ M
fort > T ∗,
i.e.,
− cM
β(1 + m 2 ) + 1
M ≤ − cx ∗ (1 + z 1 (t))
β (1 + x ∗2 [1 + z 1 (t)] 2 ) + 1
x ∗ [1 + z 1 (t)] ≤ − cm
β(1 + M 2 ) + 1
m
and by dene
B = max
− cM
β(1 + m 2 ) + 1 M
,
− cm
β(1 + M 2 ) + 1 m
.
Then the ondition for
V ˙ (z(t)) ≤ 0
beomesM 2 (M + 2x ∗ )
1 + m 2 τ < 2 Kβ rc , r
δK + c 2β
(5 + 3x ∗2 )(M + 2x ∗ ) + 2x ∗ β(1 + x ∗2 )(1 + m 2 )
Mτ + KC
βr
M + 2x ∗
1 + m 2 + M + x ∗
(1 + x ∗ )(1 + m 2 ) − 2x ∗
(1 + x ∗2 )(1 + M 2 )
< 1 δ , B 2 − 4AC < 0,
where
A
,B
andC
are given by (3.46)-(3.48).Dene
w(s) = εs 2, then w
is nonnegative ontinuous on [0, ∞)
, w(0) = 0
and w(s) > 0
for
s > 0
. It follows that fort ≥ T ∗
V ˙ (z(t)) ≤ −ε[z 1 2 (t) + z 2 2 (t)] = −w(|z(t)|).
To nd a funtion
u
suh thatV (z(t)) ≥ u(|z(t)|)
, sine the following relationship is still hold for Holling's Type IIIV (z(t)) ≥ V 1 (z(t)) = 1
βy ∗ {z 1 (t) − ln[1 + z 1 (t)] + z 2 (t) − ln[1 + z 2 (t)]} .
andthen bythe sameargumentproposedintheproof ofTheorem 3.1,wean establishthat
this is a nonnegative ontinuous funtion
u
dened on[0, ∞)
withu(0) = 0
,u(s) > 0
fors > 0
andlim
s→∞ u(s) = +∞ and
V (z(t)) ≥ u(|z(t)|)
fort ≥ T ∗ .
Hene the equilibrium point
E ∗ of the system (2.2) is globally asymptotially stable by
Lemma 2.3
4 Numerial Example
The following numerialexample isused toillustratethe proedures of applying our results
toLeslie-Gowermodelwithout and with time delay. Consider the system
˙
x(t) = x(t)[3 − 10x(t − τ)] − p(x)y(t),
˙
y(t) = y(t)
1 − 6 y(t) x(t)
,
(4.1)x 1 (θ) = x 1 (0), θ ∈ [−τ, 0],
x 1 (0) > 0, x 2 (0) > 0.
(4.2)The orresponding parameter values are
r = 3, K = 0.3, c = 15, δ = 1, β = 6.
The unique positive equilibrium point
E ∗ = (x ∗ , y ∗ )
for three-dierent types of Holling's funtionalresponse are listed below.Table1: Theunique positiveequilibriumpoint
E ∗ oftheLeslie-Gowersystem (4.1)forthree typesof Holling's funtional responses
Holling'sfuntional response
p(x) E ∗ = (x ∗ , y ∗ )
Type I
15x
6 25 , 1
25
TypeII
15 x
1 + x
1 4 , 1
24
Type III
15 x 2
1 + x 2 ≈ (0.2816, 0.0469)
When
τ = 0
, i.e., the modelwithout delay, the unique positive equilibrium pointE ∗ of
the system (4.1) is globally asymptotiallystable by using the Lyapunov funtional (3.23).
The orresponding trajetories of the system are depited inFigure 1.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
x(t)
y(t)
Type I without delay
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
x(t)
y(t)
Type II without delay
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
x(t)
y(t)
Type III without delay
(a)Holling-TypeI (b) Holling-TypeII () Holling-TypeIII
Figure1: Thetrajetoriesofthesystem(4.1)forthreetypesofHolling'sfuntionalresponses
withoutdelay
Whenever
τ = 0.05
,we obtainM = 0.34855
,L = 0.05809
,r − c max{1, M }L = 2.12862
,and hoose
m = 0.19691
. The orresponding suient onditions of Theorems 3.1-3.3 for three dierent types of Holling'sfuntional responses are veried by the following tableTable 2: The parameters inthe suient onditions of Theorems 3.1-3.3.
Holling's funtionalresponse
A B C B 2 − AC
Type I
8.0394 2.5784 2.6512 −78.6074
Type II
6.8076 3.2246 2.6870 −62.7711
Type III
5.0054 4.6395 2.6778 −32.0887
Consequently, by Theorems 3.1-3.3, weonlude that the unique positiveequilibriumpoint
E ∗ of the system (4.1) with initialonditions (4.2) is globally asymptotially stable. The trajetories ofthe delayed system aredepited inFigure2. Butitisindistinguishableinthe
phase portraitsgiven by Figures1and 2 whihorrespond to non-delay and delay systems.
Toobservetheeetoftimedelayondynamialbehavior,wehoosethesystemwithHolling-
Type III funtional response under initialonditions
x(θ) = 0.4
forθ ∈ [−τ, 0]
,x(0) = 0.4
,and
y(0) = 0.05
. Figure 3 shows the time history of the system trajetories for both ases and thetrajetoriesfor thedelay systemis movingavery littlehigherand faster thanthoseof the non-delay one.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
x(t)
y(t)
Type I delay τ =0.05
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
x(t)
y(t)
Type II delay τ =0.05
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
x(t)
y(t)
Type III delay τ =0.05
(a)Holling-TypeI (b) Holling-TypeII () Holling-TypeIII
Figure2: Thetrajetoriesofthesystem(4.1)forthreetypesofHolling'sfuntionalresponses
with delay time
τ = 0.05
0 1 2 3 4 5 6 7 8 0.3
0.32 0.34 0.36 0.38 0.4
Trajectory for the Leslie-Gower system
time t
solution x(t)
nondelay τ =0.05
0 1 2 3 4 5 6 7 8
0.046 0.047 0.048 0.049 0.05 0.051 0.052
Trajectory for the Leslie-Gower system
time t
solution y(t)
nondelay τ =0.05
(a)
x(t)
vs.t
(b)y(t)
vs.t
Figure3: Comparisonof time history of the trajetory ofthe system (4.1)for Holling-Type
III funtional response with delay time
τ = 0.05
under initial onditionsx(θ) = 0.4
forθ ∈ [−τ, 0]
,x(0) = 0.4
,andy(0) = 0.05
BasedonthesuientonditionsofTheorems3.1-3.3,itanbenumeriallyveriedthat
theuniquepositiveequilibriumpoint
E ∗ isgloballyasymptotiallystablewheneverthedelay time is less than the upper bound dened in Table 3. It is evident that the upper bound
on delay time of Holling-Type I from Theorem 3.1 is muh larger than those provided by
Tsai's paper [22℄ and Tsai's paper an't provide the information on the time-delay bounds
of Holling-TypeII & III.
Table 3: The upperbound ondelaytime of theLeslie-Gowersystem (4.1)for threetypesof
Holling's funtionalresponses
Methods Holling-TypeI Holling-TypeII Holling-TypeIII
Present study
0.127607 0.105255 0.0767035
Tsai's paper[22℄
0.006333 − −
Aknowledgments
This work was inpart supported by the National Siene Counil,TAIWAN, under the
NSC grant: NSC 98-2115-M-029-006-MY2.