NONLINEAR INTEGRAL INEQUALITIES INVOLVING ... - Ele-Math
NONLINEAR INTEGRAL INEQUALITIES INVOLVING ... - Ele-Math
NONLINEAR INTEGRAL INEQUALITIES INVOLVING ... - Ele-Math
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
M athematical<br />
Inequalities<br />
& Applications<br />
Volume 12, Number 4 (2012), 811–825<br />
<strong>NONLINEAR</strong> <strong>INTEGRAL</strong> <strong>INEQUALITIES</strong> <strong>INVOLVING</strong><br />
MAXIMA OF THE UNKNOWN SCALAR FUNCTIONS<br />
M. BOHNER, S. HRISTOVA AND K. STEFANOVA<br />
(Communicated by J. Pečarić)<br />
Abstract. This paper deals with some nonlinear integral inequalities that involve the maximum<br />
of the unknown scalar function of one variable. The considered inequalities are generalizations<br />
of the classical integral inequality of Gronwall–Bellman. The importance of these integral inequalities<br />
is due to their wide applications in qualitative investigations of differential equations<br />
with “maxima”, and it is illustrated by some direct applications.<br />
1. Introduction<br />
In the past few years, a number of integral inequalities was established by many<br />
scholars, which are motivated by certain applications such as existence, uniqueness,<br />
continuous dependence, comparison, perturbation, boundedness and stability of solutions<br />
of differential and integral equations (see, for example, [4, 7, 12, 13, 15] and the<br />
references cited therein). Among these integral inequalities, we cite the famous Gronwall<br />
inequality and its various generalizations [1, 2, 5, 10, 9, 8].<br />
In the last few decades, great attention has been paid to automatic control systems<br />
and their applications to computational mathematics and modeling. Many problems<br />
in control theory correspond to the maximal deviation of the regulated quantity (see<br />
[14]). Such kind of problems could be adequately modeled by differential equations<br />
that contain the maxima operator. Note that such equations involving “maxima” of the<br />
unknown function are called differential equations with “maxima”, see [3, 6]. In his<br />
survey [11], A. D. Mishkis also points out the necessity to study differential equations<br />
with “maxima”.<br />
The purpose of this paper is to establish some new nonlinear integral inequalities in<br />
the case when the “maxima” of the unknown scalar function is involved in the integral.<br />
Several cases depending on the type of nonlinearity are considered. These inequalities<br />
are mathematical tools in the theory of differential equations with “maxima”. Their<br />
importance is illustrated by some direct applications obtaining bounds of the solutions.<br />
<strong>Math</strong>ematics subject classification (2010): 26D10, 34D40.<br />
Keywords and phrases: Integral inequalities, maxima, scalar functions of one variable, differential<br />
equations with “maxima”.<br />
c○Ð, Zagreb<br />
811<br />
Paper MIA-15-69
812 M. BOHNER, S. HRISTOVA AND K. STEFANOVA<br />
∞.<br />
2. Main results<br />
Let h > 0 be a constant and suppose t0 and T are fixed points with 0 t0 < T <br />
DEFINITION 2.1. The function α ∈ C 1 ([t0,T),R+) is said to be from the class<br />
F if it is nondecreasing and satisfies α(t) t for t ∈ [t0,T).<br />
Let αi,βj ∈ F for i = 1,2,...,n and j = 1,2,...,m. Denote<br />
<br />
J = min min<br />
1in αi(t0), min<br />
1 jm β <br />
j(t0) .<br />
2.1. Constant additive term<br />
In this subsection, we discuss the case of a constant k in the inequality (2.1) below.<br />
For this purpose, we define the following functions.<br />
DEFINITION 2.2. The function ω ∈ C(R+,R+) is said to be from Ω1 if<br />
(i) ω(x) > 0 for x > 0 and ω is a nondecreasing function;<br />
(ii) ∞ dx<br />
ω(x) = ∞.<br />
In the case when the nonlinear functions under the integrals in the inequality (2.1)<br />
below are from the set Ω1 , we obtain the following result.<br />
THEOREM 2.3. Let the following conditions be fulfilled:<br />
(A 1 ) φ ∈ C([J − h,t0],R+).<br />
(A 2 ) k > 0.<br />
(A 3 ) αi,βj ∈ F for i = 1,2,...,n, j = 1,2,...,m.<br />
(A 4 ) fi,gj ∈ C([J,T),R+) for i = 1,2,...,n, j = 1,2,...,m.<br />
(A 5 ) ωi, ˜ω j ∈ Ω1 for i = 1,2,...,n, j = 1,2,...,m.<br />
(A 6 ) ψ ∈ C 1 (R+,R+) is increasing, ψ(0) = 0, and limt→∞ ψ(t) = ∞.<br />
(A 7 ) u ∈ C([J − h,T),R+) satisfies for some p 0 the inequalities<br />
<br />
ψ u(t)<br />
k+<br />
n<br />
∑<br />
αi(t)<br />
i=1 αi(t0)<br />
m β j(t)<br />
+<br />
∑<br />
j=1<br />
β j(t0)<br />
fi(s)u p (s)ωi<br />
g j(s)u p (s) ˜ω j<br />
<br />
u(s) ds (2.1)<br />
<br />
max<br />
ξ ∈[s−h,s] u(ξ)<br />
<br />
ds for t ∈ [t0,T),<br />
u(t) φ(t) for t ∈ [J − h,t0]. (2.2)
Then for t0 t t1 , the inequality<br />
holds, where<br />
t1 = sup<br />
<br />
<strong>NONLINEAR</strong> <strong>INTEGRAL</strong> <strong>INEQUALITIES</strong> 813<br />
u(t) ψ −1<br />
<br />
Ψ −1<br />
<br />
W −1<br />
W Ψ(M) <br />
+ A(t)<br />
<br />
r<br />
Ψ(r) =<br />
r0<br />
r<br />
W(r) =<br />
r1<br />
(2.3)<br />
ds<br />
ψ −1 (s) p, 0 < r0 < k, (2.4)<br />
ds<br />
q ψ−1 (Ψ−1 (s)) , 0 < r1<br />
<br />
< Ψ(M), (2.5)<br />
q(t) = max max<br />
1in ωi(t), max<br />
1 jm ˜ω <br />
j(t) , (2.6)<br />
A(t) =<br />
n<br />
∑<br />
i=1<br />
αi(t)<br />
αi(t0)<br />
M = max<br />
fi(s)ds+<br />
<br />
k, ψ<br />
<br />
m β j(t)<br />
∑<br />
j=1 β j(t0)<br />
max<br />
s∈[J−h,t0] φ(s)<br />
τ ∈ [t0,T) : W Ψ(M) + A(t) ∈ Dom W −1 ,<br />
W −1<br />
W Ψ(M) <br />
+ A(t) ∈ Dom Ψ −1 and<br />
Ψ −1<br />
<br />
W −1<br />
W Ψ(M) <br />
+ A(t)<br />
<br />
∈ Dom ψ −1<br />
Proof. Define a function z : [J − h,T) → R+ by<br />
g j(s)ds, (2.7)<br />
, (2.8)<br />
for t ∈ [t0,τ]<br />
⎧<br />
n αi(t)<br />
M + ∑ fi(s)u<br />
⎪⎨ i=1 αi(t0)<br />
z(t) =<br />
⎪⎩<br />
p <br />
(s)ωi u(s) ds<br />
m β j(t)<br />
+ ∑ g j(s)u<br />
j=1 β j(t0)<br />
p t ∈ [t0,T),<br />
(s) ˜ω j max u(ξ) ds,<br />
ξ ∈[s−h,s]<br />
M, t ∈ [J − h,t0].<br />
The function z is nondecreasing. Since ψ(u(t)) ψ(max s∈[J−h,t0] φ(s)) M = z(t)<br />
for t ∈ [J −h,t0] by (2.2) and (2.8) and ψ(u(t)) z(t) for t ∈ [t0,T) by (2.1) and (2.8),<br />
the inequality<br />
u(t) ψ −1 z(t) holds for t ∈ [J − h,T). (2.9)<br />
<br />
.
814 M. BOHNER, S. HRISTOVA AND K. STEFANOVA<br />
Note that maxξ ∈[s−h,s] ψ−1z(ξ) = ψ−1z(s) for s ∈ [β j(t0),βj(T)), j = 1,2,...,m.<br />
Then from inequality (2.1) and the definition of the function q, we get for t ∈ [t0,T)<br />
n αi(t) <br />
z(t) M + fi(s)<br />
ψ −1 z(s) <br />
ds<br />
+<br />
M +<br />
+<br />
∑<br />
i=1 αi(t0)<br />
m β j(t)<br />
∑<br />
j=1 β j(t0)<br />
n αi(t)<br />
∑<br />
i=1 αi(t0)<br />
m β j(t)<br />
∑<br />
j=1 β j(t0)<br />
ψ −1 z(s) p<br />
ωi<br />
<br />
g j(s) ψ −1 z(s) p <br />
˜ω j ψ −1 z(s) <br />
ds<br />
<br />
fi(s) ψ −1 z(s) p <br />
q ψ −1 z(s) <br />
ds<br />
<br />
g j(s) ψ −1 z(s) p <br />
q ψ −1 z(s) <br />
ds =: K(t),<br />
(2.10)<br />
where the function K : [t0,T) → [M,∞) is nondecreasing and satisfies K(t0) = M . Differentiate<br />
the function K and use its monotonicity (observe Definition 2.1) and (2.10)<br />
to obtain<br />
K ′ (t) =<br />
n<br />
∑<br />
i=1<br />
m<br />
+ ∑<br />
j=1<br />
n<br />
∑ fi<br />
i=1<br />
m<br />
+ ∑<br />
j=1<br />
<br />
<br />
+<br />
<br />
fi αi(t) <br />
ψ −1 z(αi(t)) p <br />
q ψ −1 z(αi(t)) <br />
α ′ i(t)<br />
<br />
g j β j(t) <br />
ψ −1 z(β j(t)) p <br />
q ψ −1 z(β j(t)) <br />
β ′ j (t)<br />
αi(t) <br />
ψ −1 K(αi(t)) p<br />
q<br />
<br />
ψ −1 z(αi(t)) <br />
α ′ i (t)<br />
<br />
g j β j(t) <br />
ψ −1 K(β j(t)) p <br />
q ψ −1 z(β j(t)) <br />
β ′ j (t)<br />
ψ −1 K(t) p n<br />
∑<br />
m<br />
∑<br />
j=1<br />
i=1<br />
From (2.4) and (2.11), we have that<br />
(Ψ ◦ K) ′ (t) = <br />
<br />
fi(αi(t))q ψ −1 z(αi(t)) <br />
α ′ i (t)<br />
<br />
g j(β j(t))q ψ −1 z(β j(t)) <br />
β ′ <br />
j(t) .<br />
K ′ (t)<br />
ψ−1K(t) n<br />
p ∑<br />
i=1<br />
+<br />
<br />
fi αi(t) <br />
q ψ −1 z(αi(t)) <br />
α ′ i(t)<br />
m<br />
∑<br />
j=1<br />
<br />
g j β j(t) <br />
q ψ −1 z(β j(t)) <br />
β ′ j (t).<br />
Integrate (2.12) from t0 to t ∈ [t0,t1] and change the variables to get<br />
<br />
n αi(t) <br />
Ψ K(t) Ψ(M)+ fi(s)q ψ −1 z(s) <br />
ds<br />
+<br />
∑<br />
i=1 αi(t0)<br />
m β j(t)<br />
∑<br />
j=1 β j(t0)<br />
<br />
g j(s)q ψ −1 z(s) <br />
ds =: K1(t),<br />
(2.11)<br />
(2.12)<br />
(2.13)
<strong>NONLINEAR</strong> <strong>INTEGRAL</strong> <strong>INEQUALITIES</strong> 815<br />
where the function K1 is nondecreasing and satisfies K1(t0) = Ψ(M) and, due to (2.10)<br />
and (2.13),<br />
z(t) K(t) Ψ −1 K1(t) holds for t ∈ [t0,t1). (2.14)<br />
Differentiate the function K1 and use its monotonicity (observe Definition 2.1) and<br />
(2.14) to obtain<br />
K ′ <br />
1(t) = fi(αi(t))q ψ −1 z(αi(t)) <br />
α ′ i(t)<br />
n<br />
∑<br />
i=1<br />
+<br />
<br />
q<br />
m<br />
∑<br />
j=1<br />
<br />
g j(β j(t))q ψ −1 z(β j(t)) <br />
β ′ j (t)<br />
ψ −1<br />
Ψ −1 K1(t) n<br />
∑ fi(αi(t))α<br />
i=1<br />
′ i(t)+<br />
From (2.15) and (2.5), we get<br />
(W ◦ K1) ′ (t) =<br />
<br />
K ′ 1 (t)<br />
<br />
q ψ−1 <br />
Ψ−1K1(t) <br />
n<br />
∑<br />
i=1<br />
fi(αi(t))α ′ i(t)+<br />
m<br />
∑<br />
j=1<br />
m<br />
∑<br />
j=1<br />
g j(β j(t))β ′ <br />
j(t) .<br />
g j(β j(t))β ′ j(t).<br />
Integrate (2.16) from t0 to t ∈ [t0,t1] and change the variables to get<br />
(2.15)<br />
(2.16)<br />
W K1(t) = W Ψ(M) + A(t), (2.17)<br />
where the function A is defined by (2.7). Since W −1 is increasing and since, due to<br />
(2.9) and (2.14), u(t) ψ−1 (z(t)) ψ−1 <br />
Ψ−1K1(t) <br />
, (2.17) implies the required<br />
inequality (2.3). <br />
COROLLARY 2.4. Let k = 0 and φ(t) ≡ 0. Suppose (A 3 )–(A 7 ) hold. Then for<br />
t0 t t2 , the inequality<br />
u(t) ψ −1<br />
<br />
Ψ −1<br />
<br />
W −1<br />
<br />
A(t)<br />
<br />
holds, where<br />
t2 = sup<br />
<br />
τ ∈ [t0,T) : A(t) ∈ Dom W −1 , W −1 A(t) ∈ Dom Ψ −1<br />
and Ψ −1<br />
W −1 A(t) <br />
∈ Dom ψ −1 <br />
for t ∈ [t0,τ] .<br />
Proof. This claim follows from Theorem 2.3 by choosing an arbitrary ε > 0, letting<br />
k = φ(t) = ε , and taking the limit as ε → 0. <br />
Note that the inequalities (2.1), (2.2) could have another type of solution as follows,<br />
which is simpler than (2.3) but the used integral function is more complicated.
816 M. BOHNER, S. HRISTOVA AND K. STEFANOVA<br />
THEOREM 2.5. Suppose (A 1 )–(A 7 ) hold. Then for t0 t t3 , the inequality<br />
u(t) ψ −1<br />
Ψ −1<br />
<br />
Ψ1(M)+A(t)<br />
(2.18)<br />
holds, where Ψ −1<br />
1 is the inverse function of<br />
Ψ1(r) =<br />
r<br />
r3<br />
1<br />
ds<br />
ψ −1 (s) pq ψ −1 (s) , 0 < r3 < k, (2.19)<br />
the functions q and A and the constant M are defined by (2.6), (2.7), and (2.8), respectively,<br />
and<br />
<br />
t3 = sup τ ∈ [t0,T) : Ψ1(M)+A(t) ∈ Dom Ψ −1<br />
1 and<br />
Ψ −1<br />
<br />
1 Ψ1(M)+A(t) ∈ Dom ψ −1 <br />
for t ∈ [t0,τ] .<br />
Proof. Following the proof of Theorem 2.3, we obtain the inequalities (2.10) and<br />
(2.11). From (2.11) and Definition 2.1, we conclude<br />
K ′ <br />
(t) ψ −1 K(t) p <br />
q<br />
+<br />
m<br />
∑<br />
j=1<br />
From (2.19) and (2.20), we have that<br />
g j(β j(t))β ′ j (t)<br />
<br />
.<br />
∑<br />
ψ −1 K(t) n<br />
i=1<br />
(Ψ1 ◦ K) ′ K<br />
(t) =<br />
′ (t)<br />
<br />
ψ−1K(t) p <br />
q ψ−1K(t) <br />
<br />
n<br />
∑<br />
i=1<br />
<br />
fi αi(t) α ′ i(t)+<br />
m<br />
∑<br />
j=1<br />
fi(αi(t))α ′ i(t)<br />
<br />
g j β j(t) β ′ j(t).<br />
Integrate inequality (2.21) from t0 to t ∈ [t0,t3] and change the variables to get<br />
Ψ1<br />
<br />
K(t)<br />
Ψ1(M)+<br />
n<br />
∑<br />
i=1<br />
αi(t)<br />
αi(t0)<br />
fi(s)ds+<br />
m β j(t)<br />
∑<br />
j=1 β j(t0)<br />
(2.20)<br />
(2.21)<br />
g j(s)ds = Ψ1(M)+A(t). (2.22)<br />
By (2.9), (2.10), and (2.22), we obtain the required inequality (2.18). <br />
In the case when p = 0, both solutions of inequalities (2.1), (2.2) given in Theorem<br />
2.3 and Theorem 2.5 coincide.<br />
COROLLARY 2.6. Suppose (A 1 )–(A 6 ) hold and assume
<strong>NONLINEAR</strong> <strong>INTEGRAL</strong> <strong>INEQUALITIES</strong> 817<br />
(A ′ 7 ) u ∈ C([J − h,T),R+) satisfies the inequalities<br />
<br />
ψ u(t)<br />
k+<br />
n<br />
∑<br />
αi(t)<br />
i=1 αi(t0)<br />
m β j(t)<br />
+<br />
∑<br />
j=1<br />
β j(t0)<br />
fi(s)ωi<br />
g j(s) ˜ω j<br />
<br />
u(s) ds<br />
u(t) φ(t) for t ∈ [J − h,t0].<br />
Then for t0 t t4 , the inequality<br />
u(t) ψ −1<br />
holds, where W −1 is the inverse function of<br />
r<br />
W(r) =<br />
r4<br />
<br />
max u(ξ)<br />
ξ ∈[s−h,s]<br />
ds for t ∈ [t0,T),<br />
<br />
W −1<br />
<br />
W(M)+A(t)<br />
<br />
ds<br />
q ψ −1 (s) , 0 < r4 < k,<br />
the functions q and A and the constant M are defined by (2.6), (2.7), and (2.8), respectively,<br />
and<br />
<br />
t4 = sup<br />
τ ∈ [t0,T) : W(M)+A(t) ∈ Dom W −1 and<br />
W −1<br />
<br />
W(M)+A(t) ∈ Dom ψ −1 <br />
for t ∈ [t0,τ] .<br />
In the case when the left part of the considered inequality is linear, i.e., ψ(x) = x,<br />
we obtain the following particular case of Theorem 2.3.<br />
COROLLARY 2.7. Suppose (A 1 )–(A 5 ) hold and assume<br />
(A ′′<br />
7 ) u ∈ C([J − h,T),R+) satisfies the inequalities<br />
u(t) k+<br />
n<br />
∑<br />
αi(t)<br />
i=1 αi(t0)<br />
m β j(t)<br />
+<br />
∑<br />
j=1<br />
β j(t0)<br />
fi(s)ωi<br />
g j(s) ˜ω j<br />
<br />
u(s) ds<br />
u(t) φ(t) for t ∈ [J − h,t0].<br />
Then for t0 t t5 , the inequality<br />
u(t) W −1<br />
<br />
<br />
max u(ξ)<br />
ξ ∈[s−h,s]<br />
ds for t ∈ [t0,T),<br />
W(M)+A(t)
818 M. BOHNER, S. HRISTOVA AND K. STEFANOVA<br />
holds, where W −1 is the inverse function of<br />
W(r) =<br />
r<br />
r 5<br />
ds<br />
q(s) , r5 > 0, (2.23)<br />
the functions q and A and the constant M are defined by (2.6), (2.7), and (2.8), respectively,<br />
and<br />
<br />
<br />
t5 = sup<br />
2.2. Monotone additive term<br />
τ ∈ [t0,T) : W(M)+A(t) ∈ Dom W −1 for t ∈ [t0,τ]<br />
In this subsection, we solve an inequality in which the constant k from Subsection<br />
2.1 is replaced by a monotonic function. For this purpose, we introduce the following<br />
set of functions.<br />
DEFINITION 2.8. The function ψ ∈ C 1 (R+,R+) is said to be from Λ if<br />
(i) ψ is an increasing function;<br />
(ii) tψ(x) ψ tx for 0 t 1.<br />
REMARK 2.9. Note that the functions ψ(x) = x and ψ(x) = x n , where n > 1, are<br />
from Λ.<br />
DEFINITION 2.10. The function ω ∈ Ω1 is said to be from Ω2 if<br />
ω(tx) tω(x) for 0 t 1.<br />
THEOREM 2.11. Let the following conditions be fulfilled:<br />
(B1 ) φ ∈ C([J − h,t0],[0, ˜k]), where ˜k = k(t0).<br />
(B2 ) k ∈ C([t0,T),[1,∞)) is nondecreasing.<br />
(B3 ) αi,βj ∈ F for i = 1,2,...,n, j = 1,2,...,m.<br />
(B4 ) fi,gj ∈ C([J,T),R+) for i = 1,2,...,n, j = 1,2,...,m.<br />
(B5 ) ωi, ˜ω j ∈ Ω2 for i = 1,2,...,n, j = 1,2,...,m.<br />
(B6 ) ψ ∈ Λ.<br />
(B7 ) u ∈ C([J − h,T),R+) satisfies for some p 0 the inequalities<br />
<br />
ψ u(t)<br />
k(t)+<br />
n<br />
∑<br />
αi(t)<br />
i=1 αi(t0)<br />
m β j(t)<br />
+<br />
∑<br />
j=1<br />
β j(t0)<br />
fi(s)u p (s)ωi<br />
g j(s)u p (s) ˜ω j<br />
<br />
u(s) ds (2.24)<br />
<br />
max<br />
ξ ∈[s−h,s] u(ξ)<br />
<br />
ds, t ∈ [t0,T),<br />
u(t) φ(t), t ∈ [J − h,t0]. (2.25)<br />
.
<strong>NONLINEAR</strong> <strong>INTEGRAL</strong> <strong>INEQUALITIES</strong> 819<br />
Then for t0 t t6 , the inequality<br />
u(t) k(t)ψ −1<br />
<br />
Ψ −1<br />
<br />
W −1<br />
<br />
W Ψ(1) + A1(t)<br />
(2.26)<br />
holds, where the functions Ψ and W are defined by (2.4) and (2.5), respectively, and<br />
A1(t) =<br />
n αi(t) p m<br />
fi(s) k(s) ds+<br />
β j(t) p g j(s) k(s) ds, (2.27)<br />
∑<br />
i=1<br />
αi(t0)<br />
∑<br />
j=1<br />
β j(t0)<br />
<br />
t6 = sup τ ∈ [t0,T) : W Ψ(1) + A1(t) ∈ Dom W −1 ,<br />
W −1<br />
<br />
W Ψ(1) + A1(t) ∈ Dom Ψ −1 and<br />
Ψ −1<br />
W −1<br />
W Ψ(1) <br />
+ A1(t) ∈ Dom ψ −1 <br />
for t ∈ [t0,τ] .<br />
Proof. From (2.24), (2.25), (B2 ), (B6 ), and 0 1<br />
k(t) 1, we obtain<br />
n <br />
u(t)<br />
αi(t)<br />
ψ 1+<br />
k(t) ∑ fi(s)u<br />
i=1 αi(t0)<br />
p <br />
u(s)<br />
(s)ωi ds (2.28)<br />
k(s)<br />
m β j(t)<br />
+ ∑ g j(s)u p <br />
maxξ∈[s−h,s] u(ξ)<br />
(s) ˜ω j<br />
ds, t ∈ [t0,T),<br />
k(s)<br />
j=1<br />
β j(t0)<br />
u(t) φ(t)<br />
<br />
k(t0) k(t0) 1, t ∈ [J − h,t0]. (2.29)<br />
Let s ∈ [β j(t0),βj(T)), where 1 j m is arbitrary. From the monotonicity of the<br />
function k in [t0,T), we obtain the inequality<br />
max ξ ∈[s−h,s] u(ξ)<br />
k(s)<br />
= u(ξ1)<br />
k(s)<br />
u(ξ1)<br />
max<br />
k(ξ1)<br />
ξ ∈[s−h,s]<br />
where ξ1 ∈ [s − h,s]. Define a function v ∈ C([J − h,T),R+) by<br />
⎧<br />
⎪⎨<br />
u(t)<br />
for t ∈ [t0,T)<br />
k(t)<br />
v(t) =<br />
⎪⎩<br />
u(t)<br />
for t ∈ [J − h,t0].<br />
k(t0)<br />
u(ξ)<br />
k(ξ) ,<br />
Then inequalities (2.28) and (2.29) can be rewritten as<br />
n<br />
ψ v(t) 1+<br />
αi(t)<br />
fi(s)k p (s)v p <br />
(s)ωi v(s) ds (2.30)<br />
∑<br />
i=1 αi(t0)<br />
m β j(t)<br />
+<br />
∑<br />
j=1<br />
β j(t0)<br />
g j(s)k p (s)v p (s) ˜ω j<br />
<br />
max v(ξ)<br />
ξ ∈[s−h,s]<br />
ds, t ∈ [t0,T),<br />
v(t) 1, t ∈ [J − h,t0]. (2.31)
820 M. BOHNER, S. HRISTOVA AND K. STEFANOVA<br />
Theorem 2.3, applied to the inequalities (2.30) and (2.31), implies the validity of inequality<br />
(2.26). <br />
In the case when Theorem 2.5 is applied instead of Theorem 2.3 in the last part of<br />
the proof of Theorem 2.11, we obtain the following result.<br />
THEOREM 2.12. Suppose (B1 )–(B7 ) hold. Then for t0 t t7 , the inequality<br />
u(t) k(t)ψ −1<br />
Ψ −1<br />
<br />
Ψ1(1)+A1(t)<br />
holds, where Ψ1 and A1 are defined by (2.19) and (2.27), respectively, and<br />
t7 = sup<br />
<br />
1<br />
τ ∈ [t0,T) : Ψ1(1)+A1(t) ∈ Dom Ψ −1<br />
1 and<br />
Ψ −1<br />
<br />
1 Ψ1(1)+A(t) ∈ Dom ψ −1 <br />
for t ∈ [t0,τ] .<br />
In the case when p = 0 and the left part of the considered inequality is linear, i.e.,<br />
ψ(x) = x, the results of Theorem 2.11 and Theorem 2.12 coincide, and we obtain the<br />
following result.<br />
COROLLARY 2.13. Suppose (B1 )–(B5 ) hold and assume<br />
(B ′ 7 ) u ∈ C([J − h,T),R+) satisfies the inequalities<br />
u(t) k(t)+<br />
n<br />
∑<br />
αi(t)<br />
i=1 αi(t0)<br />
m β j(t)<br />
+<br />
∑<br />
j=1<br />
β j(t0)<br />
fi(s)ωi<br />
g j(s) ˜ω j<br />
u(t) φ(t), t ∈ [J − h,t0].<br />
Then for t0 t t8 , the inequality<br />
u(t) k(t)W −1<br />
<br />
<br />
u(s) ds<br />
<br />
max<br />
ξ ∈[s−h,s] u(ξ)<br />
<br />
ds, t ∈ [t0,T),<br />
W(1)+A(t)<br />
holds, where the functions W and A are defined by (2.23) and (2.7), respectively, and<br />
t8 = sup<br />
<br />
τ ∈ [t0,T) : W(1)+A(t) ∈ Dom W −1 for t ∈ [t0,τ]<br />
<br />
<br />
.
2.3. Arbitrary additive term<br />
<strong>NONLINEAR</strong> <strong>INTEGRAL</strong> <strong>INEQUALITIES</strong> 821<br />
In this subsection, we solve a nonlinear inequality in which the constant k of<br />
Subsection 2.1 is replaced by an arbitrary function. In this case, we define the following<br />
set of functions.<br />
DEFINITION 2.14. The function ω ∈ Ω2 is said to be from Ω3 if<br />
ω(x)+ω(y) ω(x+y).<br />
REMARK 2.15. Note that the functions ω(x) = √ x and ω(x) = x are from Ω3 .<br />
THEOREM 2.16. Let the following conditions be fulfilled:<br />
(C2 ) k ∈ C([J − h,T),R+).<br />
(C3 ) αi,βj ∈ F for i = 1,2,...,n, j = 1,2,...,m.<br />
(C4 ) fi,gj ∈ C([J,T),R+) for i = 1,2,...,n, j = 1,2,...,m.<br />
(C5 ) ωi, ˜ω j ∈ Ω3 for i = 1,2,...,n, j = 1,2,...,m.<br />
(C6 ) µ ∈ C([t0,T),[1,∞)) is nondecreasing.<br />
(C7 ) u ∈ C([J − h,T),R+) satisfies the inequalities<br />
u(t) k(t)+ µ(t)<br />
+<br />
<br />
n αi(t)<br />
∑<br />
i=1 αi(t0)<br />
m β j(t)<br />
∑<br />
j=1<br />
β j(t0)<br />
fi(s)ωi<br />
<br />
u(s) ds (2.32)<br />
<br />
g j(s) ˜ω j max u(ξ)<br />
ξ ∈[s−h,s]<br />
ds , t ∈ [t0,T),<br />
u(t) k(t), t ∈ [J − h,t0]. (2.33)<br />
Then for t0 t t9 , the inequality<br />
u(t) k(t)+M(t)e(t)W −1<br />
<br />
W(1)+A2(t)<br />
holds, where the function W is defined by (2.23),<br />
e(t) = 1+<br />
n<br />
∑<br />
αi(t)<br />
i=1 αi(t0)<br />
m β j(t)<br />
+<br />
∑<br />
j=1<br />
β j(t0)<br />
fi(s)ωi<br />
g j(s) ˜ω j<br />
(2.34)<br />
<br />
k(s) ds (2.35)<br />
<br />
max k(ξ)<br />
ξ ∈[s−h,s]<br />
ds, t ∈ [t0,T),
822 M. BOHNER, S. HRISTOVA AND K. STEFANOVA<br />
A2(t) =<br />
n<br />
∑<br />
i=1<br />
αi(t)<br />
αi(t0)<br />
fi(s)M(s)ds+<br />
m<br />
∑<br />
j=1<br />
β j(t)<br />
β j(t0)<br />
g j(s)M(s)ds,<br />
<br />
µ(t) for t ∈ [t0,T),<br />
M(t) =<br />
µ(t0) for t ∈ [J − h,t0],<br />
<br />
t9 = sup τ ∈ [t0,T) : W(1)+A2(t) ∈ Dom W −1 <br />
for t ∈ [t0,τ] .<br />
Proof. Define a function z : [J − h,T) → R+ by<br />
⎧ n αi(t) <br />
∑ fi(s)ωi u(s) ds<br />
⎪⎨ i=1 αi(t0)<br />
m β<br />
z(t) =<br />
j(t) <br />
+ ∑ g j(s) ˜ω j max<br />
⎪⎩<br />
j=1 β j(t0)<br />
ξ ∈[s−h,s] u(ξ)<br />
<br />
ds, t ∈ [t0,T),<br />
0, t ∈ [J − h,t0].<br />
From (2.32) and the definition of the function z, we obtain for s ∈ [β j(t0),βj(T)) and<br />
1 j m<br />
u(t) k(t)+M(t)z(t) for t ∈ [J − h,T]. (2.36)<br />
Since the function M is nondecreasing on [J − h,T), we obtain<br />
max u(ξ) max k(ξ)+M(s) max z(ξ). (2.37)<br />
ξ ∈[s−h,s] ξ ∈[s−h,s]<br />
ξ ∈[s−h,s]<br />
From (2.36), (2.37), (C3 ), and (C5 ), we get for t ∈ [t0,T)<br />
αi(t)<br />
αi(t0)<br />
fi(s)ωi<br />
αi(t)<br />
u(s) ds <br />
<br />
αi(t0)<br />
αi(t)<br />
αi(t0)<br />
fi(s)ωi<br />
fi(s)ωi<br />
<br />
k(s)+M(s)z(s) ds (2.38)<br />
αi(t) <br />
k(s) ds+ fi(s)M(s)ωi z(s) ds<br />
αi(t0)<br />
and<br />
<br />
β j(t)<br />
g j(s) ˜ω j max<br />
β j(t0)<br />
ξ ∈[s−h,s] u(ξ)<br />
<br />
ds<br />
<br />
β j(t)<br />
g j(s) ˜ω j max<br />
β j(t0)<br />
ξ ∈[s−h,s] k(ξ)<br />
<br />
β j(t)<br />
ds+ g j(s)M(s) ˜ω j max<br />
β j(t0)<br />
ξ ∈[s−h,s] z(ξ)<br />
<br />
ds.<br />
(2.39)<br />
From the definition of the function z and (2.36), (2.38), and (2.39), it follows that<br />
n αi(t) <br />
z(t) e(t)+ ∑ fi(s)M(s)ωi z(s) ds (2.40)<br />
i=1 αi(t0)<br />
m <br />
β j(t)<br />
+ ∑ g j(s)M(s) ˜ω j max z(ξ) ds, t ∈ [t0,T),<br />
j=1 β j(t0)<br />
ξ ∈[s−h,s]<br />
z(t) 0, t ∈ [J − h,t0], (2.41)
<strong>NONLINEAR</strong> <strong>INTEGRAL</strong> <strong>INEQUALITIES</strong> 823<br />
where the function e is defined by (2.35). Note that e : [t0,T) → [1,∞) is nondecreasing<br />
and e(t0) = 1. Corollary 2.13 applied to inequalities (2.40) and (2.41) implies the<br />
required inequality (2.34). <br />
3. Applications<br />
In this section, we consider the differential equation with “maxima”<br />
with the initial condition<br />
px p−1 x ′ <br />
= F t,x(t), max<br />
s∈[σ(t),τ(t)] x(s)<br />
<br />
for t ∈ [t0,T), (3.1)<br />
x(t) = ϕ(t) for t ∈ [τ(t0) − h,t0], (3.2)<br />
where ϕ : [τ(t0) − h,t0] → R, F : [t0,T)×R×R → R, and p is a natural number.<br />
THEOREM 3.1. (Upper bound) Let the following conditions be fulfilled:<br />
(H 1 ) ϕ ∈ C([τ(t0) − h,t0],R).<br />
(H 2 ) τ,σ ∈ F and there exists a constant h with 0 < τ(t) − σ(t) h for t t0 .<br />
(H 3 ) F ∈ C([t0,T)×R×R,R) satisfies<br />
<br />
F(t,u,v) Q(t) u q + R(t) v q<br />
where Q,R ∈ C([t0,T),R+) and q ∈ (0, p).<br />
for u,v ∈ R and t t0,<br />
(H 4 ) x : [τ(t0) − h,T) → R is a solution of the initial value problem (3.1), (3.2).<br />
Then x satisfies the inequality<br />
<br />
<br />
x(t) p−q M p−q<br />
p +<br />
where M = sup s∈[τ(t0)−h,t0]<br />
p − q<br />
p<br />
p ϕ(s) .<br />
t<br />
Proof. The function x satisfies the integral problem<br />
t0<br />
<br />
Q(s)+R(s) ds for t ∈ [t0,T), (3.3)<br />
x(t) p = ϕ(t0) p +<br />
t<br />
t0<br />
F<br />
<br />
s,x(s), max<br />
ξ ∈[σ(s),τ(s)] x(ξ)<br />
<br />
ds for t ∈ [t0,T),<br />
x(t) = ϕ(t) for t ∈ [τ(t0) − h,t0].
824 M. BOHNER, S. HRISTOVA AND K. STEFANOVA<br />
Then for the norm of the solution x, we obtain<br />
<br />
x(t) p <br />
ϕ(t0) p t <br />
<br />
+ F s,x(s), max<br />
ϕ(t0) p +<br />
t0<br />
t<br />
t0<br />
t<br />
ξ ∈[σ(s),τ(s)]<br />
<br />
ds<br />
x(ξ)<br />
<br />
Q(s) <br />
x(s) q <br />
+ R(s) max<br />
ξ ∈[σ(s),τ(s)] x(ξ)<br />
<br />
<br />
q<br />
ds<br />
ϕ(t0) p + Q(s)<br />
t0<br />
<br />
x(s) q ds<br />
<br />
t<br />
<br />
+ R(s) max x(ξ)<br />
t0 ξ ∈[σ(s),τ(s)]<br />
q ds for t ∈ [t0,T), (3.4)<br />
<br />
x(t) = ϕ(t) for t ∈ [τ(t0) − h,t0]. (3.5)<br />
Change the variable s = τ −1 (η) in the second integral of (3.4), use the inequality<br />
max ξ ∈[σ(s),τ(s)] |x(ξ)| max ξ ∈[τ(s)−h,τ(s)] |x(ξ)| for s ∈ [t0,T) that follows from (H 2 ),<br />
and obtain<br />
<br />
x(t) p <br />
ϕ(t0) p t<br />
+ Q(s)<br />
t0<br />
<br />
x(s) q ds<br />
τ(t)<br />
+ R(τ<br />
τ(t0)<br />
−1 (η))(τ −1 ) ′ <br />
(η)<br />
Note that the conditions of Corollary 2.6 are satisfied for<br />
max<br />
ξ ∈[η−h,η]<br />
<br />
x(ξ) <br />
u(t) = |x(t)|, n = 1, α1(t) = t, m = 1, β1 = τ,<br />
k = |ϕ(t0)| p , f1 = Q, g1 = (R ◦ τ −1 )(τ −1 ) ′ on [τ(t0),T),<br />
ψ(x) = x p , ψ −1 (x) = p√ x, Dom(ψ −1 ) = R+,<br />
ω1(x) = ˜ω1(x) = x q r q<br />
−<br />
, W(r) = s p ds =<br />
0<br />
p p−q<br />
r p ,<br />
p − q<br />
W −1 (r) = p−q<br />
p − q<br />
p r<br />
p , Dom(W −1 ) = R+.<br />
According to Corollary 2.6, from (3.6) and (3.5), we obtain (3.3). <br />
q<br />
dη.<br />
(3.6)<br />
COROLLARY 3.2. Let ϕ(t) ≡ 0. Suppose (H 2 )–(H 4 ) hold. Then the solution x<br />
of the initial value problem (3.1), (3.2) satisfies the inequality<br />
<br />
<br />
x(t) p−q p − q t <br />
Q(s)+R(s) ds for t ∈ [t0,T).<br />
p<br />
t0<br />
REMARK 3.3. Note that in the case of p = q = 1 in (H 3 ), we use the functions<br />
ψ(x) = ψ−1 (x) = x and W(r) = lnr, W −1 (r) = er , apply Corollary 2.6, and obtain<br />
|x(t)| e lnM+ <br />
t<br />
t<br />
t0 Q(s)+R(s) ds t0 Q(s)+R(s) ds<br />
= Me<br />
for t ∈ [t0,T),<br />
which gives us the known result about uniqueness of the solution of a first-order differential<br />
equation with Lipschitz-continuous right-hand side.
<strong>NONLINEAR</strong> <strong>INTEGRAL</strong> <strong>INEQUALITIES</strong> 825<br />
R E F E R E N C E S<br />
[1] RAVI P. AGARWAL, SHENGFU DENG, AND WEINIAN ZHANG, Generalization of a retarded<br />
Gronwall-like inequality and its applications, Appl. <strong>Math</strong>. Comput. 165, 3 (2005), 599–612.<br />
[2] ELVAN AKIN-BOHNER, MARTIN BOHNER, AND FAYSAL AKIN, Pachpatte inequalities on time<br />
scales, JIPAM. J. Inequal. Pure Appl. <strong>Math</strong>. 6, 1 (2005), Article 6, 23 pp. (electronic).<br />
[3] VASIL G. ANGELOV AND DRUMI D. BAĬNOV, On the functional-differential equations with “maximums”,<br />
Applicable Anal. 16, 3 (1983), 187–194.<br />
[4] DRUMI D. BAĬNOV AND PAVEL SIMEONOV, Integral inequalities and applications, volume 57 of<br />
<strong>Math</strong>ematics and its Applications (East European Series), Kluwer Academic Publishers Group, Dordrecht,<br />
1992. Translated by R. A. M. Hoksbergen and V. Covachev [V. Khr. Kovachev].<br />
[5] YEOL JE CHO, YOUNG-HO KIM, AND JOSIP PEČARIĆ, New Gronwall-Ou-Iang type integral inequalities<br />
and their applications, ANZIAM J. 50, 1 (2008), 111–127.<br />
[6] SNEZHANA G. HRISTOVA AND LILA F. ROBERTS, Boundedness of the solutions of differential equations<br />
with “maxima”, Int. J. Appl. <strong>Math</strong>. 4, 2 (2000), 231–240.<br />
[7] V. LAKSHMIKANTHAM AND S. LEELA, Differential and integral inequalities: Theory and applications.<br />
Vol. I: Ordinary differential equations, Academic Press, New York, 1969. <strong>Math</strong>ematics in<br />
Science and Engineering, Vol. 55-I.<br />
[8] AILIAN LIU AND MARTIN BOHNER, Gronwall-OuIang-type integral inequalities on time scales, J.<br />
Inequal. Appl., pages Art. ID 275826, 15, 2010.<br />
[9] QING-HUA MA AND JOSIP PEČARIĆ, On certain new nonlinear retarded integral inequalities for<br />
functions in two variables and their applications, J. Korean <strong>Math</strong>. Soc. 45, 1 (2008), 121–136.<br />
[10] QING-HUA MA AND JOSIP PEČARIĆ, Explicit bounds on some new nonlinear retarded integral inequalities<br />
and their applications, Taiwanese J. <strong>Math</strong>. 13, 1 (2009), 287–306.<br />
[11] A. D. MISHKIS, On some problems of the theory of differential equations with deviating argument,<br />
Russian <strong>Math</strong>. Surveys 32, 2 (1977), 181–210.<br />
[12] BABURAO G. PACHPATTE, Inequalities for differential and integral equations, volume 197 of <strong>Math</strong>ematics<br />
in Science and Engineering, Academic Press Inc., San Diego, CA, 1998.<br />
[13] BABURAO G. PACHPATTE, Inequalities for finite difference equations, volume 247 of Monographs<br />
and Textbooks in Pure and Applied <strong>Math</strong>ematics, Marcel Dekker Inc., New York, 2002.<br />
[14] EVGENIĬ PAVLOVICH POPOV, Teoriya lineinykh sistem avtomaticheskogo regulirovaniya i upravleniya,<br />
“Nauka”, Moscow, 1978.<br />
[15] WOLFGANG WALTER, Differential and integral inequalities, Translated from the German by Lisa<br />
Rosenblatt and Lawrence Shampine. Ergebnisse der <strong>Math</strong>ematik und ihrer Grenzgebiete, Band 55.<br />
Springer-Verlag, New York, 1970.<br />
(Received February 22, 2011) M. Bohner<br />
Department of <strong>Math</strong>ematics and Statistics<br />
Missouri University of Science and Technology<br />
Rolla, MO 65409-0020<br />
USA<br />
e-mail: bohner@mst.edu<br />
<strong>Math</strong>ematical Inequalities & Applications<br />
www.ele-math.com<br />
mia@ele-math.com<br />
S. Hristova<br />
Department of Applied <strong>Math</strong>ematics and Modeling<br />
Plovdiv University, Plovdiv 4000<br />
Bulgaria<br />
e-mail: snehri@uni-plovdiv.bg<br />
K. Stefanova<br />
Department of Applied <strong>Math</strong>ematics and Modeling<br />
Plovdiv University, Plovdiv 4000<br />
Bulgaria