signðx
2
ðtδÞx
n
2
Þ; …;
jx
m
ðtδÞx
n
m
jsignðx
m
ðtδÞx
n
m
ÞÞ
T
;
jyðtτÞy
n
jsignðyðtτÞy
n
Þ
¼ðjy
1
ðtτÞy
n
1
jsignðy
1
ðtτÞy
n
1
Þ; jy
2
ðtτÞ
y
n
2
jsignðy
2
ðtτÞy
n
2
Þ; …;jy
m
ðtτÞy
n
m
jsignðy
m
ðtτÞy
n
m
ÞÞ
T
;
where signðxÞ is defined as when x4 0; signðxÞ¼1; when
xo 0; signðxÞ¼1, when x ¼0; signðxÞ¼0; τ; δ are defined in ðH
3
Þ,
x
n
¼ðx
n
1
; x
n
2
; …; x
n
m
Þ
T
; y
n
¼ðy
n
1
; y
n
2
; …; y
n
m
Þ
T
will be defined in the proof
of Theorem 3.1.
Throughout this paper, we make the following assumptions:
ðH
1
Þ There exist positive constants α
i
; β
i
ði ¼ 1; 2; …; mÞ such that
for ∀x; y∈R
jf
i
ðxÞf
i
ðyÞj≤α
i
jxyj;
jg
i
ðxÞg
i
ðyÞj≤β
i
jxyj;
ðH
2
Þ (Sign conditions). There exist positive constants
B
i
; D
i
ði ¼1; 2; …; mÞ such that for ∀x; y∈R
signðxyÞðb
i
ðxÞb
i
ðyÞÞ≥B
i
jxyj;
signðxyÞðd
i
ðxÞd
i
ðyÞÞ≥D
i
jxyj;
ðH
3
Þ
τ
ji
¼τ; δ
ij
¼δ:
ðH
4
Þ Functions a
i
(r) and c
j
(r) are continuous, 0o a
i
o a
i
o a
i
and
0o
c
j
o c
j
o c
j
, for all r∈R; i; j ¼1; 2; …; m.
In this paper, we need the following definitions and lemmas.
Definition 1. A point ðx
n
; y
n
Þ
T
∈R
m
R
m
is said to be an equilibrium
point of system (1.3) if
a
i
ðx
n
i
Þ b
i
ðx
n
i
Þ ∑
m
i ¼ 1
s
ij
f
j
ðy
n
j
ÞþI
i
"#
¼0; i ¼ 1; 2; …; m;
c
j
ðy
n
i
Þ d
i
ðy
n
i
Þ ∑
m
i ¼ 1
t
ji
g
i
ðx
n
i
ÞþJ
j
"#
¼0; j ¼1; 2; …; m;
where x
n
¼ðx
n
1
; x
n
2
; …; x
n
m
Þ
T
; y
n
¼ðy
n
1
; y
n
2
; …; y
n
m
Þ
T
.
Lemma 1 (Deimling [33]). Let Hðλ; xÞ : ½0; 1
Ω-R
m
be a continuous
homotopic mapping. If Hðλ; xÞ¼y has no solutions in ∂Ω for λ∈½0; 1
and y∈R
m
\Hðλ; ∂ΩÞ, where ∂Ω denotes the boundary of Ω, then the
topological degree degðHðλ; xÞ; Ω; yÞ of Hðλ; xÞ is a constant which is
independent of λ. In this case, degðHð0; xÞ; Ω; yÞ¼degðHð1; xÞ; Ω; yÞ.
Lemma 2 (Deimling [33]). Let HðxÞ :
Ω-R
m
be a continuous map-
ping. If degðHðxÞ; Ω; yÞ≠0, then there exists at least one solution of
HðxÞ¼yinΩ.
Lemma 3 (Deimling [33]). Let Ω⊂R
m
be a nonempty, bounded, open
set and f : R
m
-R
m
be an Ωadmissible map, i.e., fðxÞ≠0 for all x∈∂Ω,
then degðf ; ΩÞ¼∑
x∈f
1
ð0Þ∩Ω
signdetDf ðxÞ; where detDf ðxÞ denotes
Jacobi determinant of f(x) at point x,signdetDf ðxÞ denotes the symbol
of Ja cobi determinant of f(x) at point x.
Lemma 4 (Yu et al. [34]). For any vector x; y∈R, and any positive
definite matrix G∈R
mm
, the following matrix inequality holds:
2x
T
y ≤x
T
Gx þ y
T
G
1
y:
Lemma 5. Assume that Q ¼ diagðq
1
; q
2
; …; q
m
Þ is a diagonal matrix,
jzjsignðzÞ¼ðjz
1
jsignðz
1
Þ; jz
2
jsignðz
2
Þ; …; jz
m
jsignðz
m
ÞÞ
T
is a vector,
then ðjzjsignðzÞÞ
T
QjzjsignðzÞ¼jzj
T
Qjzj.
Proof.
ðjzjsignðzÞÞ
T
QjzjsignðzÞ¼ðjz
1
jsignðz
1
Þ; jz
2
jsignðz
2
Þ; …;
jz
m
jsignðz
m
ÞÞ
T
diagðq
1
; q
2
; …; q
m
Þðjz
1
jsignðz
1
Þ;
jz
2
jsignðz
2
Þ; …; jz
m
jsignðz
m
ÞÞ
¼ ∑
m
i ¼ 1
q
i
z
2
i
¼jzj
T
Qjzj: □
Lemma 6. For any x
i
; y
i
ði ¼ 1; 2; …; mÞ∈R
∑
m
i ¼ 1
ðjx
i
jþjy
i
jÞ≤
ffiffiffiffiffiffiffi
2m
p
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
∑
m
i ¼ 1
ðjx
i
j
2
þjy
i
j
2
Þ
s
:
Proof.
∑
m
i ¼ 1
ðjx
i
jþjy
i
jÞ¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
∑
m
i ¼ 1
ðjx
i
jþjy
i
jÞ
"#
2
v
u
u
t
¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
∑
m
i ¼ 1
jx
i
j
!
2
þ ∑
m
i ¼ 1
jy
i
j
!
2
þ 2 ∑
m
i ¼ 1
jx
i
j ∑
m
i ¼ 1
jy
i
j
v
u
u
t
≤
ffiffiffi
2
p
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
∑
m
i ¼ 1
jx
i
j
!
2
þ ∑
m
i ¼ 1
jy
i
j
!
2
v
u
u
t
: ð2:1Þ
Using Lemma 4,wehave
∑
m
i ¼ 1
jx
i
j
!
2
≤∑
m
i ¼ 1
1 ∑
m
i ¼ 1
jx
i
j
2
ð2:2Þ
and
∑
m
i ¼ 1
jy
i
j
!
2
≤∑
m
i ¼ 1
1 ∑
m
i ¼ 1
jy
i
j
2
: ð2:3Þ
Substituting (2.2) and (2.3) into (2.1) gives
∑
m
i ¼ 1
ðjx
i
jþjy
i
jÞ≤
ffiffiffiffiffiffiffi
2m
p
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
∑
m
i ¼ 1
ðjx
i
j
2
þjy
i
j
2
Þ
s
:
This completes the proof of Lemma 6. □
3. Global exponential stability
In this section, we will present our results on global exponential
stability for system (1.3) by applying degree theory, LMI method,
inequality technique and Lyapunov functional.
Theorem 3.1. Under assumptions ðH
1
Þ–ðH
3
Þ, system (1.3) has one
unique equilibrium point which is globally exponentially stable if
there exist m order positive definite diagonal matrices
P; M; P
1
¼ðp
1i
Þ; M
1
¼ðm
1i
Þ; Y
i
ði ¼ 1; 2; 3; 4Þ, and m order matrices
P
2
¼ðp
2ij
Þ; P
3
¼ðp
3ij
Þ; M
2
¼ðm
2ij
Þ; M
3
¼ðm
3ij
Þ and scalar γ 4 0 such
that the following linear matrix inequalities hold:
ðh
1
Þ
Ω
1
¼
a
11
PS 000
S
T
P e
γτ
Y
1
000
nn
a
33
a
34
a
35
nnn
a
44
a
45
nnnn
a
55
0
B
B
B
B
B
B
@
1
C
C
C
C
C
C
A
o 0;
Ω
2
¼
b
11
MT 000
T
T
M e
γδ
Y
2
000
nn
b
33
b
34
b
35
nnn
b
44
b
45
nnnn
b
55
0
B
B
B
B
B
B
@
1
C
C
C
C
C
C
A
o 0;
D. Zhou et al. / Neurocomputing 121 (2013) 512–522514