Solutions of Hammerstein Equations in The Space BV (I)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Quaestiones Mathematicae 37(2014), 359–370.


c 2014 NISC (Pty) Ltd
http://dx.doi.org/10.2989/16073606.2014.894675

SOLUTIONS OF HAMMERSTEIN EQUATIONS IN


THE SPACE BV (Iab )
W. Aziz
Universidad de Los Andes, Departamento de Fı́sica y Matemáticas, Trujillo-Venezuela.
E-Mail wadie@ula.ve

H. Leiva
Universidad de Los Andes, Departamento de Matemática, Mérida-Venezuela.
E-Mail hleiva@ula.ve

N. Merentes
Universidad Central de Venezuela, Escuela de Matemáticas, Caracas-Venezuela.
E-Mail nmer@ciens.ucv.ve

Abstract. In this paper we study the existence of solutions for Hammerstein integral
equations in the space of two-variables bounded variation functions BV (Iab )

u(x) = v(x) + λ k(x, y)f (y, u(y))dy, (λ ∈ R)
b
Ia

where a = (a1 , a2 ), b = (b1 , b2 ) ∈ R2 , Iab = [a1 , b1 ] × [a2 , b2 ], v : Iab → R, k : Iab × Iab →


R, and f : Iab × R → R. Under some conditions on λ, k and f we prove that this
equation admits only one solution in the space BV (Iab ) for each v ∈ BV (Iab ).

Mathematics Subject Classification (2010): Primary 93B05; Secondary 93C25.


Key words: Hammerstein integral equation, Banach spaces, bounded variation.

1. Introduction. This paper has been motivated by the works done in [8],
[7] and [6] where the authors study the existence of solutions for Hammerstein
equations in different spaces of bounded variation functions in an interval I = [α, β].
Particularly in [8], where the authors study the existence of BV (I)-solutions for
the following Hammerstein equations

(1) x(t) = g(t) + λ K(t, s)f (x(s))ds, t ∈ I, λ ∈ R
I

where g : I → R is a BV (I)-function, f : R → R is a locally Lipschitz function and


K : I × I → R. More precisely, the authors prove, under some conditions on λ,
K and f , that this equation admits only one solution in the space BV (I) for each
g ∈ BV (I).
1 This work has been supported by CDCHT-ULA-C-1796-12-05-AA and BCV.

Quaestiones Mathematicae is co-published by NISC (Pty) Ltd and Taylor & Francis
360 W. Aziz, H. Leiva and N. Merentes

In this paper we study the existence of solutions for Hammerstein integral


equations in the space of two-variables bounded variation functions BV (Iab ) (see
[3])


(2) u(x) = v(x) + λ k(x, y)f (y, u(y))dy, (λ ∈ R)
Iab

where a = (a1 , a2 ), b = (b1 , b2 ) ∈ R2 , Iab = [a1 , b1 ] × [a2 , b2 ], x = (x1 , x2 ), y =


(y1 , y2 ) ∈ Iab , v : Iab → R, k : Iab × Iab → R and f : Iab × R → R. Under some
conditions on λ, k and f we prove that this equation admits only one solution in
the space BV (Iab ) for each v ∈ BV (Iab ).
A special case of equation (2), namely the nonlinear Volterra-Hammerstein
integral equation, corresponds to λ = 1

(3) u(x) = v(x) + k(x, y)f (y, u(y))dy.
Iab

2. Preliminaries. Let Iab = [a1 , b1 ]×[a2 , b2 ] be the basic rectangle (the domain
of functions) with a = (a1 , a2 ), b = (b1 , b2 ) ∈ R2 such that a1 < a2 and b1 < b2 . For
the sake of brevity we will write x = (x1 , x2 ), y = (y1 , y2 ) for x, y ∈ R2 , and x ≤ y
if x1 ≤ y1 and x2 ≤ y2 . Let π1 = {ti }m i=0 and π2 = {sj }j=0 be partitions of [a1 , b1 ]
n

and [a2 , b2 ], respectively (i.e., m, n ∈ N, a1 = t0 < t1 < . . . < tm−1 < tm = b1 and
a2 = s0 < s1 < . . . < sn−1 < sn = b2 ). The (two-dimensional) Hardy-Vitali-Tonelli
variation [10, 13] of a function f : Iab −→ R is defined by
∑ n
m ∑

V2 (f, Iab ) = sup f (ti−1 , sj−1 ) + f (ti , sj ) − f (ti−1 , sj ) − f (ti , sj−1 )
π1 ,π2
i=1 j=1

where the supremum is taken over all pairs (m, n) ∈ N2 and (π1 , π2 ) with π1 a
partition of [a1 , b1 ] and π2 a partition of [a2 , b2 ]. If x2 ∈ [a2 , b2 ] is fixed and [x1 , y1 ]
is a subinterval of [a1 , b1 ], the Jordan variation of the one variable function f (·, x2 )
defined by f (·, x2 )(t) = f (t, x2 ), t ∈ [a1 , b1 ], on the interval [x1 , y1 ] is the quantity:
m


Vxy11 (f (·, x2 )) = sup f (ti , x2 ) − f (ti−1 , x2 )
π1
i=1

where the supremum is taken over all partitions π1 = {ti }m i=1 (m ∈ N) of [x1 , y1 ].
A similar definition applies to the Jordan variation Vxy22 (f (x1 , ·)) if x1 ∈ [a1 , b1 ] is
fixed and [x2 , y2 ] is a subinterval of [a2 , b2 ].
We define the total variation of f : Iab −→ R by

(4) T V (f, Iab ) = Vab11 (f (·, a2 )) + Vab22 (f (a1 , ·)) + V2 (f, Iab ),

and the Hardy space of functions with finite total variation by


{ }

BV (Iab ; R) = f : Iab −→ R T V (f, Iab ) < ∞ .
Solutions of Hammerstein equations in the space BV (Iab ) 361

The value V2 (f, Iab ) is usually defined (cf. [1, 2, 10]) with the supremum taken over
all partitions of Iab . It is known ([11, 12]) that BV (Iab ; R) is Banach space with
respect to the norm:


(5) f = |f (a)| + T V (f, Iab ), f ∈ BV (Iab ; R).

In the case of functions f : Iab −→ E, where E is a Banach space, the definition of


BV (Iab ; E) is analogous. We just replace | · | by the norm of E and E by R. That
is to say, by BV (Iab ; E) we will denote the space of all functions f : Iab −→ E such
that T V (f ; Iab ) < +∞ with the norm


f b
= f (a) + T V (f ; Iab ).
BV (Ia ;E) E

Additionally the following concept of a variation of a vector valued function will


be applied. Given a function f : I0a → E. We consider

m [ ] ∑
n [ ]
T V(f, I0a ) = sup εi f (ti , a2 ) + f (ti−1 , a2 ) + sup ηj f (a1 , sj ) + f (a1 , sj−1 )
π1 π2
i=1 j=1

m ∑
n [ ]
+ sup εi ηj f (ti , sj ) + f (ti−1 , sj−1 ) − f (ti , sj−1 ) − f (ti−1 , sj ) ,
π1 ,π2
i=1 j=1

where π1 : 0 = t0 < t1 < · · · < tm = a1 , π2 : 0 = s0 < s1 < · · · < sn = a2 and


|εi | = 1, |ηj | = 1.
The following proposition is obtained by Lemma 1.2 in [5] and Lemma 1 in [4].
Proposition 2.1. Let f : I0a → E be a vector valued function. Then
{ }
T V(f, I0a ) = sup T V(h ◦ f, I0a ) : h ∈ E ∗ , ∥h∥ ≤ 1 ,

where E ∗ denote the dual space of E.

Observation 2.2. In the case of a real function f : I0a → R, we shall obtain

T V (f, I0a ) = T V(f, I0a ).

Also it is satisfied
T V (f, I0a ) ≥ T V(f, I0a ),
where f : I0a → E.

The next result is a direct consequence of the proposition above.


Corollary 2.3. Let f : I0a → E be a vector valued function. The following items
are equivalent:
1. T V(f, I0a ) < +∞.
2. T V(h ◦ f, I0a ) < +∞ for every continuous functional h ∈ E ∗ .
362 W. Aziz, H. Leiva and N. Merentes

By a BV (Iab )-solution of (2) or (3), defined on some rectangle R ⊂ Iab we will


understand a function f : R → R (or f : R → E) which satisfies (2) or (3) for
every (t, s) ∈ R and such that T V (f, R) < +∞.
Lastly, by C(Iab , R) (C(Iab , E)) we will denote the Banach space of all continuous
functions f : Iab → R (f : Iab → E) with usual norm ∥f ∥C = sup ∥f (t, s)∥.
(t,s)∈Iab

3. Hammerstein integral equation. To study equation (2), we shall assume


the following hypothesis:

1o v : Iab −→ R is a BV (Iab ; R)-function.

2o f : Iab × R −→ R is locally Lipschitz function.

3o K : Iab × Iab −→ R is a function such that

T V (K(·, α)) ≤ M (α), a.e. α = (α1 , α2 ) ∈ Iab ,

where M : Iab −→ R+ is integrable in the Lebesgue sense (shortly: L-


integrable) and K(β, (·, ·)) is L-integrable for every β = (β1 , β2 ) ∈ Iab .

Now, we are ready to prove the following theorem.

Theorem 3.1. Under the above assumptions there exists a number ρ > 0 such
that for every λ satisfying |λ| < ρ, equation (2) has a unique BV (Iab ; R)-solution
defined on Iab .


Proof. Let r > 0 be such that v b
< r, and denote by Lr the Lipschitz
BV( Ia )
constant of f corresponding to the cube Iab × [−r, r]. Choose a number ρ > 0 such
that
∫ [ ]
(6) ∥v∥BV (Iab ) + 3ρ sup |f (y, u(y))| |K(a, y)| + M (y) dy < r,
y∈Iab Iab


(7) ρLr [|K(a, y)| + 3M (y)]dy < 1.
Iab

Denote by B r the closed ball of center zero and radius r in the space BV (Iab ; R).
Fix |λ| < ρ. Define G(u)(x) = v(x) + λF (u)(x), where

(8) F (u)(x) = K(x, y)f (y, u(y))dy, x = (x1 , x2 ) ∈ B r .
Iab

If u ∈ BV (Iab ; R), then the mapping x → f (x, u(x)) is globally Lipschtiz on


Iab . Hence it is measurable in the Lebesgue sense (shortly: L-measurable) and,
obviously, bounded. Thus, the mappings F and G are well defined. First, we
Solutions of Hammerstein equations in the space BV (Iab ) 363

verify that G(B r ) ⊂ B r . Indeed, for any x ∈ B r we have

T V (F ◦ u; R) := Vab11 (F ◦ u) + Vab22 (F ◦ u) + V2 (F ◦ u)
∑m

= sup (F ◦ u)(ti , a2 ) − (F ◦ u)(ti−1 , a2 )
π1
i=1
n


+ sup (F ◦ u)(a1 , sj ) − (F ◦ u)(a1 , sj−1 )
π2
j=1
∑ n
m ∑

+ sup (F ◦ u)(ti−1 , sj−1 ) − (F ◦ u)(ti−1 , sj ) − (F ◦ u)(ti , sj−1 )
π1 ,π2
i=1 j=1

+(F ◦ u)(ti , sj )

m ∫ ∫


= sup K((ti , a2 ), y)f (y, u(y))dy − K((ti−1 , a2 ), y)f (y, u(y))dy
π1 Iab
i=1
I b

n ∫ ∫
a


+ sup K((a1 , sj ), y)f (y, u(y))dy − K((a1 , sj−1 ), y)f (y, u(y))dy
π2 Iab Iab
j=1 ∫
∑ ∑
m n
+ sup K((ti−1 , sj−1 ), y)f (y, u(y))dy
π1 ,π2 Iab
∫ i=1 j=1 ∫
− K((ti−1 , sj ), y)f (y, u(y))dy − K((ti , sj−1 ), y)f (y, u(y))dy

Iab Iab


+ K((ti , sj ), y)f (y, u(y))dy


Iab


m [ ]

= sup K((ti , a2 ), y) − K((ti−1 , a2 ), y) f (y, u(y))dy
π1 Iab
i=1
∑ n ∫ [ ]

+ sup K((a1 , sj ), y) − K((a1 , sj−1 ), y) f (y, u(y))dy
π2 Iab
j=1 ∫
∑∑
m n [
+ sup K((ti−1 , sj−1 ), y) − K((ti−1 , sj ), y) − K((ti , sj−1 ), y)
π1 ,π2 Iab
i=1 j=1
]

+K((ti , sj ), y) f (y, u(y))dy

∑m ∫

≤ sup |f (y, u(y))| sup K((ti , a2 ), y) − K((ti−1 , a2 ), y) dy
y∈Iab π1 Iab
n ∫
i=1


+ sup |f (y, u(y))| sup K((a1 , sj ), y) − K((a1 , sj−1 ), y) dy
y∈Iab π2 b
j=1 Ia
∑m ∑n ∫

+ sup |f (y, u(y))| sup K((ti−1 , sj−1 ), y)
b
y∈Ia π ,π b
1 2
i=1 j=1 Ia

−K((ti−1 , sj ), y) − K((ti , sj−1 ), y) + K((ti , sj ), y) dy

≤ sup |f (y, u(y))|3 M (y)dy < +∞.
y∈Iab Iab
364 W. Aziz, H. Leiva and N. Merentes

Hence F (u) ∈ BV (Iab ; R) and


∥G(u)∥BV (Iab ) ≤ ∥v∥BV (Iab ) + |λ|∥F (u)∥BV (Iab )
[ ]
= ∥v∥BV (Iab ) + |λ| |F (u)(a)| + T V (F (u))
≤ ∥v∥BV (Iab ) + 3|λ| sup |f (y, u(y))
y∈Iab
∫ [ ]

· |K(a, y)| + M (y) dy < r.
Iab

So G(B r ) ⊂ B r .
Now, we show that G is a contraction. In fact, for any u1 , u2 we obtain
∥G(u1 ) − G(u2 )∥BV (Iab ) = |G(u1 )(a) − G(u2 )(a)| + T V (λ(F (u1 ) − F (u2 )))

= |λ(F (u1 )(a) − F (u2 )(a))| + Vab11 (λ(F (u1 ) − F (u2 ))) + Vab22 (λ(F (u1 ) − F (u2 )))
+V 2 (λ(F (u1 ) − F (u2 )))
∫ [ ]

≤ λ K(a, y) f (y, u1 (y)) − f (y, u2 (y)) dy
Iab
∫ ∑ m

+|λ| sup K((ti , a2 ), y) − K((ti−1 , a2 ), y)||f (y, u1 (y)) − f (y, u2 (y)) dy
π1 Iab i=1
∫ ∑n

+|λ| sup K((a1 , sj ), y) − K((a1 , sj−1 ), y)||f (y, u1 (y)) − f (y, u2 (y)) dy
π2 Iab j=1
∫ ∑m ∑
n

+|λ| sup K((ti−1 , sj−1 ), y) − K((ti−1 , sj ), y) − K((ti , sj−1 ), y)
π1 ,π2 Iab i=1 j=1

+K((ti , sj ), y)| · |f (y, u1 (y)) − f (y, u (y)) dy
{2∫
≤ |λ| sup |f (y, u1 (y)) − f (y, u2 (y))| |K(a, y)|dy
y∈Iab Iab
∫ ∑
m
+ sup |K((ti , a2 ), y) − K((ti−1 , a2 ), y)|dy
π1 I b i=1
∫ a∑ n
+ sup |K((a1 , sj ), y) − K((a1 , sj−1 ), y)|dy
π2 Iab j=1
∫ ∑ m ∑
n
+ sup |K((ti−1 , sj−1 ), y) − K((ti−1 , sj ), y) − K((ti , sj−1 ), y)
π1 ,π2 Iab i=1 j=1
}
+K((ti , sj ), y)|dy
{ ∫ [ ] }
≤ |λ| Lr ∥u1 − u2 ∥BV (Iab ) |K(a, y)| + 3M (y) dy
∫ Iab
= ∥u1 − u2 ∥BV (Iab ) |λ|Lr |K(a, y) + 3M (y)|dy.
Iab

In view of condition (7) and the Banach fixed point theorem, we infer that G
has a unique fixed point in B r which is obviously a BV (Iab )-solution of (2), defined
Solutions of Hammerstein equations in the space BV (Iab ) 365

on Iab . 2

Now, we shall illustrate Theorem 3.1 by the following example.

Example 1. Let K : Iab ×Iab → R be defined by the formula: K(t, s) = K1 (t)K2 (s)
for (t, s) ∈ Iab ×Iab , where T V (K1 (t)) < +∞ and K2 is L-integrable on Iab . Obviously
K(t, ·) is L-integrable on Iab for every t ∈ Iab . Further

T V (K(·, s)) = T V (K1 (·)K2 (s)) = K2 (s)T V (K1 ),

and the function s → K2 (s)T V (K1 ) is L-integrable on Iab . Hence the function K
satisfies the assumption 3o .

4. Volterra-Hammerstein integral equation. In this section we shall con-


sider the nonlinear integral equation (3). To this end, we suppose that 1o and 2o
are satisfied. Moreover, we assume the additional hypothesis


4o T = {(t′ , s′ ) : t′ ∈ I0a , s′ ∈ I0t } and K : T → R is a function such that
|K(s′ , s′ )| + T V (K(·, s′ )) ≤ m(s′ ) for a.e. s′ ∈ Iab , where m : Iab → R+ is

L-integrable and K(t′ , ·) is L-integrable on I0t for every t′ ∈ Iab .

Now, we shall prove the following existence theorem for equation (3).

Theorem 4.1. Under the above assumptions there exists a rectangle R ⊂ Iab such
that equation (3) has an unique BV (Iab )-solution, defined on R.

Proof. Let r, Lr denote the numbers defined in the proof of Theorem 3.1. Without
loss of generality, we assume that a = (a1 , a2 ) = (0, 0). Choose a number
∫ d =
(d1 , d2 ) with d1 , d2 > 0 in such a way that ∥v∥BV (I0d ) + sup |f (t′ , u(t′ ))|3 m(s′ )
t′ ∈I0d I0d

ds′ < r and Lr m(s′ )ds′ < 1. Put
I0d

{ ′
e ′ , s′ ) = K(t′ , s′ ), s′ ∈ I0t
K(t
0, s′ ∈ int(It0′ )

′ ′ ′ ′
∫ R = [0, d1 ] × [0, d2 ]. Define G(u)(t ) = v(t ) + F (u)(t ), where F (u(t )) =
and
K(t′ , s′ )f (s′ , u(s′ ))ds′ , u ∈ B r , t′ ∈ R, and B r denotes the closed ball of

I0t
center zero and radius r in the space BV (R). Now, we verify that G maps B r into
366 W. Aziz, H. Leiva and N. Merentes

itself. Indeed, we have

∥G(u)∥BV (I t′ ) ≤ ∥v∥BV (I t′ ) + ∥F (u)∥BV (I t′ )


0 0 0
= ∥v∥BV (I t′ ) + |F (u(0))| + T V (F ◦ u)
0
= ∥v∥BV (I t′ ) + Vab11 (F ◦ u) + Vab22 (F ◦ u) + V2 (F ◦ u)
m
0

= ∥v∥BV (I t′ ) + sup (F ◦ u)(ti , a2 ) − (F ◦ u)(ti−1 , a2 )
0
π1
i=1
n


+ sup (F ◦ u)(a1 , sj ) − (F ◦ u)(a1 , sj−1 )
π2
j=1
∑ n
m ∑

+ sup (F ◦ u)(ti−1 , sj−1 ) − (F ◦ u)(ti−1 , sj ) − (F ◦ u)(ti , sj−1 ) + (F ◦ u)(ti , sj )
π1 ,π2
i=1 j=1
m ∫ ∫


= ∥v∥BV (I t′ ) + sup K((ti , a2 ), y)f (y, u(y))dy − K((ti−1 , a2 ), y)f (y, u(y))dy
0
π1 I0t′ t′
i=1 I

n ∫ ∫
0


+ sup ′ K((a1 , sj ), y)f (y, u(y))dy − ′ K((a1 , sj−1 ), y)f (y, u(y))dy
π2 I0t t
I0
j=1 ∫ ∫
∑∑
m n
+ sup ′ K((ti−1 , sj−1 ), y)f (y, u(y))dy − K((ti−1 , sj ), y)f (y, u(y))dy
π1 ,π2 I0t Iab

i=1 j=1



− K((ti , sj−1 ), y)f (y, u(y))dy + K((ti , sj ), y)f (y, u(y))dy
′ t′
I0t
m ∫ [
I 0
∑ ]
e i , a2 ), y) − K((t e i−1 , a2 ), y) f (y, u(y))dy
= ∥v∥BV (I0d ) + sup K((t
d
∫ t′ ∈B r i=1 I0
∑ n [ ]
+ sup e 1 , sj ), y) − K((a
K((a e 1 , sj−1 ), y) f (y, u(y))dy

t ∈B r j=1
I d
0

∑∑
m n [
+ sup e i−1 , sj−1 ), y) − K((t
K((t e i−1 , sj ), y) − K((t e i , sj−1 ), y)

t ∈B r i=1 j=1
I d
0

]
e
+K((ti , sj ), y) f (y, u(y))dy

[∫ ∫
≤ ∥v∥BV (I d ) + sup |f (y, u(y))| e s))ds +
V d1 (K(·, e ·))ds
V d2 (K(s,
0 0 0
I0d I0d

y∈I0d
]
+ e s))ds ds
V2 (K(s,
I0d ∫
≤ ∥v∥BV (I0d ) + sup |f (y, u(y))|3 m(s)ds < r,
y∈I0d I0d

for u1 , u2 ∈ B r . Thus G(B r ) ⊂ B r . Further, for any u1 , u2 ∈ B r we obtain

∥G(u1 ) − G(u2 )∥BV (I0d ) = T V (λ(F (u1 ) − F (u2 )))


= Vab11 (λ(F (u1 ) − F (u2 ))) + Vab22 (λ(F (u1 ) − F (u2 )))
+V2 (λ(F (u1 ) − F (u2 )))
Solutions of Hammerstein equations in the space BV (Iab ) 367

∫ ∑
m

≤ sup |K((ti , a2 ), y) − K((ti−1 , a2 ), y)||f (y, u1 (y)) − f (y, u2 (y))|dy
π1 I0t′
∫ ∑ i=1
n
+ sup |K((a1 , sj ), y) − K((a1 , sj−1 ), y)||f (y, u1 (y)) − f (y, u2 (y))|dy

π2 I0t j=1
∫ ∑m ∑
n
+ sup |K((ti−1 , sj−1 ), y) − K((ti−1 , sj ), y) − K((ti , sj−1 ), y)

π1 ,π2 I0t i=1 j=1
+K((ti , sj ), y)| · |f (y, u1 (y)) − f {
(y, u2 (y))|dy
∫ ∑ m
≤ sup |f (y, u1 (y)) − f (y, u2 (y))| sup |K((t e i−1 , a2 ), y)|dy
e i , a2 ), y) − K((t
y∈R t′ ∈B r I0d i=1
∫ ∑
n
+ sup e 1 , sj ), y) − K((a
|K((a e 1 , sj−1 ), y)|dy
t′ ∈B r I0d j=1
∫ ∑ m ∑n
+ sup |K((t e i−1 , sj ), y) − K((t
e i−1 , sj−1 ), y) − K((t e i , sj−1 ), y)
t′ ∈B r I0d i=1 j=1
}
e i , sj ), y)|dy
+K((t
[∫
≤ sup |f (y, u1 (y)) − f (y, u2 (y))| e s))ds
V0d1 (K(·,
I0d
y∈R
∫ ∫ ]
+ e ·)ds +
V0d2 (K(s, e s))ds
V2 (K(s,
I0d ∫ I d
{ 0
}
≤ Lr ∥u1 − u2 ∥R 3m(y)dy
∫ I0d
= Lr ∥u1 − u2 ∥R 3m(y)dy.
I0d

Since Lr m(s)ds < 1, in view of the Banach contraction principle the
I0d
mapping G has a unique fixed point in B r which is obviously a BV (Iab )-solution of
(3), defined on R. The proof is complete. 2

We can obviously illustrate the above result by a similar example as in the


previous section.

5. Continuous BV (Iab )-solutions. As we mentioned at the beginning, this


paper has been motivated by the work done in [8], and this section is basically an
imitation of the corresponding section of that work. So, in the same way as in
[8], we shall study the existence of solutions of equation (2) in the space of two-
variables continuous and bounded variation functions. To this end, we assume the
following hypothesis

5o v : Iab → R is a continuous BV (Iab )-function.


368 W. Aziz, H. Leiva and N. Merentes

6o For every ϵ > 0 there exists δ > 0 such that for all t, s, τ ∈ Iab : t = (t′ , t′′ ),
s = (s′ , s′′ ) and τ = (τ ′ , τ ′′ )

∥τ − t∥ < δ ⇒ K(τ, s) − K(t, s) < ϵ.

Theorem 5.1. Suppose that 2o , 3o , 5o and 6o are satisfied. Then there exists a
number ρ > 0 such that for every λ such that |λ| < ρ, equation (2) has an unique
continuous BV (Iab )-solution, defined on Iab .

Proof. Consider the space BVC (Iab ) = BV (Iab ) ∩ C(Iab ) with the norm ∥ · ∥BV (Iab ) .
Let {un }n∈N be a sequence in BVC (Iab ) such that ∥un − u∥BV (Iab ) → 0, for n ∈ N
some u ∈ BV (Iab ). Then it’s clear that ∥un − u∥C(Iab ) → 0, so u ∈ C(Iab ). Hence
BVC (Iab ) considered with the norm ∥ · ∥BVC (Iab ) is a Banach space (cf. [4]). Denote
by B r the closed ball of center zero and radius r in the space BVC (Iab ). Define the
mappings F and G in an analogous way as in the proof of Theorem 3.1. Since


G(u)(t) − G(u)(τ ) ≤ |v(t) − v(s)| + sup |f (u(s))||λ| |K(t, s) − K(τ, s)|ds
b
s∈Ia b
Ia

for u ∈ B r , t, τ ∈ R ⊂ Iab , by 5o and 6o we infer that G(u) is a continuous function.


Further, we argue similarly as in the proof Theorem 3.1.
Now, let us pass on to equation (3). Suppose that

7o for every t ∈ Iab and for every ϵ > 0 there exists δ > 0 such that for all τ ∈ Iab
and s ∈ I0t ∩ I0τ :

∥τ − t∥ < δ ⇒ |K(τ, s) − K(t, s)| < ε.

We assume E is a separable∫ Banach space and BV (Iab )-solutions of equation (1)


and (2), where the sign “ ” stands now for the Lebesgue-Bochner integral. Also
that 1o , 2o , and 3o , in which we replace consequently R by E, are satisfied. 2

Theorem 5.2. Under the assumptions 1o , 2o , and 3o there exists a rectangle R ⊂


Iab such that equation (3) has an unique continuous BV (Iab )-solution defined on R.

Proof. Let r > 0 be such that ∥v∥BV (Iab ) < r and let Lr denote the Lipschitz
constant which corresponds to the ball B r of center zero and radius r in the space
E. Choose a number ρ > 0 such that
(∫ )
( )
∥v∥BV (Iab ) + sup ∥f (z)∥ρ ∥K(0, s)∥ + 3M (s) ds < r
z∈B r I0a


and ρLr er the closed ball of center zero
(∥K(0, s)∥ + 3M (s))ds < 1. Denote by B
Iab
and radius r in the space BV (Iab , E). Define G and F in an analogous way as in the
Solutions of Hammerstein equations in the space BV (Iab ) 369

( )
proof of Theorem 3.1. Since f (u) ∈ BV (Iab , E), we have T V {f (u(t)) : t ∈ I0a } <
+∞. In view of Corollary 2.3 we infer that (h ◦ f )(u) is scalarly L-measurable for
every h ∈ E ∗ . Since E is separable, by Pettis measurability theorem ([9, page 23])
we deduce that f (u) is strongly L-measurable. Moreover, it’s obviously bounded.
Further, we argue similarly as in the proof of Theorem 3.1. 2

Arguing in a similar way as in the proof of Theorem 4.1 one can obtain the
existence and uniqueness result concerning equation (3).
Now, let us pass on to continuous BV (Iab )-solutions of equations (2) and (3).
In this case we don’t assume that a Banach space E is separable. It’s an obvious
consequence of the fact that the superposition f ◦ u, where u is continuous and f
is locally Lipschitz, is a continuous function. Thus it’s strongly L-measurable and,
obviously, bounded in the considered case, which implies its integrability in the
Lebesgue-Bochner sense. Moreover, note that the space BVC (Iab , E) = BV (Iab , E) ∩
C(Iab , E) considered with the norm ∥ · ∥BV (Iab ) , is a Banach space. Hence arguing
similarly as in the case of real functions one can obtain the following two results.
Theorem 5.3. Suppose that 2o , 3o , 5o and 6o , in which we replace consequently
R by E, are satisfied. Then the conclusion of Theorem 5.1 holds.

Theorem 5.4. Assume that 2o , 4o , 5o and 7o , in which we replace consequently


R by E, are satisfied. Then the conclusion of Theorem 5.2 holds.

In the above two theorems the implication in 6o and 7o has the following form:

∥τ − t∥ < δ ⇒ ∥K(τ, s) − K(t, s)∥ < ϵ,

where ∥ · ∥ denotes the norm of E.

References

1. C.R. Adams and J.A. Clarkson, On the definitions of bounded variation for
functions of two variables, Trans. Amer. Math. 35 (1933), 824–854.
2. , Properties of functions f (x, y) of bounded variation, Trans. Amer.
Math. 36 (1934), 711–730.
3. J. Appell and Chur-Jen Chen, How to solve Hammerstein equations, J. Integral
Equations Appl. 18(3) (2006), 287–296.
4. V.V. Chistyakov, Superposition operators in the algebra of functions of two
variables with finite total variation, Monatshefte Math. 137 (2002), 99–114.
5. , Selections of Bounded Variation, J. Applied Analysis 10(1)
(2004), 1–82.
6. Daria Bugajewska, Dariusz Bugajewski and Grzegorz Lewicki, On Nonlinear
Integral Equations in the space of functions of Bounded Generalized ϕ-Variation,
Journal of Integral Equations and Applications 21(1), (2009), 1–20.
7. Daria Bugajewska and D. O’Regan On Nonlinear Integral Equations and Γ-
Bounded Variation, Acta Math. Hungar. 107(4) (2005), 295–306.
370 W. Aziz, H. Leiva and N. Merentes

8. Dariusz Bugajewski, On BV-Soludtions of some Nonlinear Integral Equations,


Integral Equations and Operator Theory 46 (2003), 387–398.
9. J. Diestel, Sequences and series in Banach spaces, Springer-Verlag, New York/
Berlin/Heidelberg/Tokyo, 1984.
10. G.H. Hardy, On double Fourier series, and especially those which represent the
double zeta–function with real and incommesurable parameters, Quart. J. Math.
Oxford 37 (1905/1906), 53–79.
11. T.H. Hildebrant, Intoduction to the theory of Integration, Academic Press, New
York, 1994.
12. A.S. Leonov, Remarks on the total variation of functions of several variables and
the multidimensional counterpart of the Helly selection principle, Mat. Zametki 63
(1998), 69–80 (Russian); English translation in Math. Notes 63 (1998), 61–71.
13. G. Vitali, Sulle funzione integrali, Atti Accad. Sci. Torino CI Sci. Fis. Mat. Natur.
40 (1904/1905), 1021–1034; and Opere sull’analisi teale, Cremonese (1984), 205–220
(Italian).

Received 10 November, 2012.


Copyright of QM - Quaestiones Mathematicae is the property of Taylor & Francis Ltd and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

You might also like