Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Monatsh Math (2016) 181:975–989

DOI 10.1007/s00605-015-0853-1

The distributional Henstock–Kurzweil integral


and applications

Guoju Ye1 · Wei Liu1

Received: 20 August 2015 / Accepted: 1 December 2015 / Published online: 14 December 2015
© Springer-Verlag Wien 2015

Abstract In this paper, we study the distributional Henstock–Kurzweil integral and


its properties, such as strong and weak convergence theorems. We also establish the
structure of the space of Henstock–Kurzweil integrable distributions, which is proved
to be an ordered Banach space with a regular cone. Furthermore, fixed point theorems
in such space are presented to deal with nonlinear integral equations involving this
integral.

Keywords Distribution · Distributional derivative · Distributional Henstock–


Kurzweil integral · Convergence theorem · Cone · Fixed point theorem · Integral
equation

Mathematics Subject Classification 45G10 · 26A42 · 46F30 · 46F05

1 Introduction

In this paper, we investigate a general integral form named the distributional Henstock–
Kurzweil integral. This integral is closely related to the space of continuous functions

Communicated by G. Teschl.

Supported by the Fundamental Research Funds for the Central Universities (No. 2014B38114).

B Wei Liu
liuw626@hhu.edu.cn
Guoju Ye
yegj@hhu.edu.cn

1 College of Sciences, Hohai University, Nanjing 210098, People’s Republic of China

123
976 G. Ye, W. Liu

and Schwartz distributional derivative. Namely, a distribution (or generalized function)


f is distributionally Henstock–Kurzweil integrable on an interval [a, b] if there exists
a continuous function F such that the distributional derivative of F is f and denote
b
by a f = F(b) − F(a), and F is called the primitive of f . With this definition,
this integral comprises Riemann, Lebesgue, Henstock–Kurzweil, Perron, Denjoy, and
improper integrals as special cases [1–13].
It is well-known that the Henstock–Kurzweil integral has very wide application in
many fields of mathematics and physics and many interesting results appeared, for
instance, differential equations and integral equations [14–19], Fourier analysis [6],
economics [20–22], mechanics [20] and quantum theory [23], and so on. However, a
pity is that the space of Henstock–Kurzweil (shortly HK) integrable functions is not
complete under Alexiewicz norm, i.e., it is not a Banach space. So, many mathemati-
cians try to solve this problem, for instance, Kurzweil [7,8], etc. But, as far as we
know, there is no satisfactory solution.
Denote the spaces of the Henstock–Kurzweil integrable functions and distributions
by HK and DHK , respectively. The space DHK contains the space HK and overcomes
the incompleteness of the space HK [12]. The space DHK is a Banach space with
the Alexiewicz norm and it is isometrically isomorphic to the space of continuous
functions on a closed interval with uniform norm. Some other basic properties are
also introduced in Sect. 2. Section 3 is devoted to the weak and strong convergence
theorems. In Sect. 4, we prove that DHK is an ordered Banach space with a regular cone
under certain ordering and then give a fixed point theorem in DHK . The last section,
using the obtained results, we prove the existence of solutions of nonlinear integral
equations involving the distributional Henstock–Kurzweil integral. At last, we present
two examples to show that our results are more general than the results in the case of
HK integral.

2 The distributional Henstock–Kurzweil integral

The theory of distributions, or generalized functions, was founded by L. Schwartz in


the 1940’s and it extends the notion of function so that all distributions have derivatives
of all orders. Distributions are defined as continuous linear functionals on the space
of the functions

Cc∞ = {φ : R → R | φ ∈ C ∞ and φ has compact support in R},

where the support of a function φ is the closure of the set on which φ does not vanish.
Denote it by

supp(φ) = {x ∈ R : φ(x) = 0}.

A sequence {φn } ⊂ Cc∞ converges to φ ∈ Cc∞ if there exists a compact set K


such that all φn have support in K and for every m ∈ N the sequence of derivatives
(m)
φn converges to φ (m) uniformly on K . Denote Cc∞ endowed with this convergence
property by D. Here φ is called a test f unction if φ ∈ D. The distributions are

123
The distributional Henstock–Kurzweil integral and… 977

defined as continuous linear functionals on D. The space of distributions is denoted


by D , which is the dual space of D. That is, if f ∈ D then f : D → R, and we write
 f, φ ∈ R for φ ∈ D.
For all f ∈ D , we define the distributional derivative f  of f to be a distribution
satisfying  f  , φ = − f, φ  , where φ is a test function. Further, we write distrib-
utional derivative as f  . If a function f is differentiable, then its ordinary pointwise
derivative denotes as f  (x) where x ∈ R. From now on, all derivative in this paper
will be distributional derivatives unless stated otherwise.
Let I be an open interval in R, we define

D(I ) = {φ : I → R | φ ∈ Cc∞ and φ has compact support in I }.

Then the distributions on I are the continuous linear functionals on D(I ). The space
of distributions on I is denoted by D (I ), which is the dual space of D(I ).
Since D(I ) ⊂ D, so D ⊂ D (I ), i.e., if f ∈ D then f ∈ D (I ).
Denote the space of continuous functions on [a, b] by C([a, b]). Let

C0 = {F ∈ C([a, b]) : F(a) = 0}. (2.1)

Then C0 is a Banach space under the norm


F
∞ = sup |F(x)|.
x∈[a,b]

Definition 1 A distribution f in D ((a, b)) is said to be distributionally Henstock–


Kurzweil integrable (shortly DHK ) on an interval [a, b] if there exists a continuous
function F ∈ C0 such that F  = f , i.e., f is the distributional derivative of F. The
b
distributional Henstock–Kurzweil integral of f on [a, b] is denoted by a f (x)dx =
b
F(b) − F(a). For short, a f = F(b) − F(a).
For every f ∈ DHK , φ ∈ D((a, b)), we write
 b
 f, φ = −F, φ  = − F(x)φ  (x)dx.
a

In symbols,
DHK = { f ∈ D ((a, b)) : f = F  , F ∈ C0 }. (2.2)
Obviously, DHK ⊂ D ((a, b)).
Notice that if f ∈ DHK then f has many primitives in C([a, b]), all differing by
a constant, but f has exactly one primitive in C0 . For simplicity of notation, in what
follows we use the letters F, G, . . . for the primitives of f, g, . . . in DHK . Unless
otherwise stated, “ ” denotes the DHK -integral throughout this paper.
The following result is known as the Fundamental Theorem of Calculus.
Lemma 1 ([12, Theorem 4]) (Fundamental Theorem of Calculus)
x
(a) Let f ∈ DHK and F(x) = a f . Then F ∈ C0 and F  = f .

123
978 G. Ye, W. Liu

x
(b) Let F ∈ C([a, b]). Then a F  = F(x) − F(a) for all x ∈ [a, b].

Remark 1 The distributional Henstock–Kurzweil integral is very wide and it includes


the Riemann integral, the Lebesgue integral, the Henstock–Kurzweil integral, the
restricted and wide Denjoy integral. For example, let


 sin n 2 x
F(x) = , x ∈ [0, 1].
n2
n=1

Then F is continuous on [0, 1] and F(0) = 0. Moreover, the function F, apart from
certain exceptional points, indeed is not pointwise differentiable on [0, 1].
Let f = F  . Then f ∈ DHK and

 1 ∞
 sin n 2
f = F(1) = .
0 n2
n=1

However, f is neither Henstock–Kurzweil integrable nor Lebesgue integrable on


[0, 1], because the primitive of Henstock–Kurzweil integrable function is AC G ∗ and
differentiable almost everywhere [1,10].

Remark 2 If take a = −∞ and b = +∞, we obtain distributional Henstock–Kurzweil


integral on R = [−∞, +∞].

For f ∈ DHK , define the Alexiewicz norm in DHK as


f
=
F
∞ .

With the Alexiewicz norm, DHK is a Banach space, see details in [12, Theorem 2].
Moreover, DHK is separable [12, Theorem 3]. 
Let g : [a, b] → R, its variation is V (g) = sup n |g(yn ) − g(xn )| where the
supremum is taken over every sequence {(xn , yn )} of disjoint intervals in [a, b]. A
function g is of bounded variation on [a, b] if V (g) is finite. Denote the space of
functions of bounded variation by BV (see, e.g., [24–26]). The space BV is a Banach
space with norm
g
BV = |g(a)| + V (g).
Recall that C([a, b])∗ = BV by the Riesz Representation Theorem. Since C0 is
the space of continuous functions on [a, b] vanishing at a and DHK is isometrically
isomorphic to C0 due to the definition of the integral, an obvious fact is that the dual
space of DHK is BV (see details in [12]).
Furthermore, integration by parts and Hölder inequality hold.

Lemma 2 ([12, Definition 6]) (Integration by parts) Let f ∈ DHK and g ∈ BV. Then
f g ∈ DHK and
 b  b
f g = F(b)g(b) − Fdg. (2.3)
a a

123
The distributional Henstock–Kurzweil integral and… 979

Lemma 3 ([12, Theorem 7]) (Hölder inequality) Let f ∈ DHK . If g ∈ BV, then
 
 b 
 f g  ≤ 2
f

g
BV . (2.4)

a

Now we introduce a partial ordering in the space DHK . For f, g ∈ DHK , we say
that
 
f g (or g  f ) if and only if f ≤ g, (2.5)
J J

for any subinterval J in [a, b]. Particularly, for f, g ∈ DHK , by (2.5),


 x  x
f g ⇒ f ≤ g, ∀x ∈ [a, b]. (2.6)
a a

Remark 3 It is known that there is a pointwise ordering in C([a, b]), i.e., F ≤ G in


C([a, b]) if and only if F(x) ≤ G(x) for every x ∈ [a, b]. The ordering in C0 is
similar. In [12], another ordering in DHK is given as

f ≤ g if and only if F ≤ G, (2.7)

where F, G ∈ C0 are the primitives of f, g ∈ DHK , respectively.


It is easy to see that the orderings defined in (2.5) and (2.7) are quite different. By
(2.6), f g in DHK implies F ≤ G in C0 . But the converse is not true. For example,
let F(x) = sin x on [0, π ], so f (x) = cos x, x ∈ [0, π ]. Obviously, F(x) ≥ 0 on
[0, π ], which implies f ≥ 0 under the ordering (2.7). However, for two intervals
J1 = [π/6, π/2] and J2 = [π/2, 5π/6], one has
 
f = 1 − 1/2 = 1/2 > 0, f = 1/2 − 1 = −1/2 < 0.
J1 J2

Therefore, f can not be compared with the zero function under the ordering (2.5).

3 Convergence theorems

In [12,13], many convergence theorems have been introduced, nevertheless, some of


them can be improved. For convenience, we supplement several new convergence
theorems in this section.
Let us start with the definitions of the convergences, see also in [12, Section 7].

Definition 2 (a) A sequence { f n } ⊂ DHK is said to converge strongly to f ∈ DHK if



f n − f
→ 0 as n → ∞.
b
(b) { f n } converges weakly in D((a, b)) if  f n − f, φ = a ( f n − f )φ → 0 (n → ∞)
for each φ ∈ D((a, b)).

123
980 G. Ye, W. Liu

b
(c) { f n } converges weakly in BV if  f n − f, g = a ( f n − f )g → 0 (n → ∞) for
each g ∈ BV.

The following lemma is easy to obtain from their definitions.

Lemma 4 ([12, Theorem 12])


(a) Strong convergence implies weak convergence in D((a, b)) and BV.
(b) Weak convergence in BV implies weak convergence in D((a, b)).
(c) Weak convergence in D((a, b)) does not imply weak convergence in BV.
(d) Weak convergence in BV does not imply strong convergence.

We mean to examine the conditions of a sequence { f n } ⊂ DHK , in order that the


convergence of f n → f (n → ∞) (in some sense) with f ∈ DHK implies that
b b
limn→∞ a f n = a f . We give the following convergence theorems beforehand.

Theorem 1 Let { f n } ⊂ DHK and f ∈ DHK with the primitives Fn and F, respectively.
Then {Fn } is uniformly bounded and Fn (x) → F(x) pointwise on [a, b] as n → ∞
b b
if and only if a f g = limn→∞ a f n g for every g ∈ BV.

Proof Since {Fn } is uniformly bounded in [a, b] and limn→∞ Fn (x) = F(x), x ∈
[a, b], by the dominated convergence theorem of Riemann-Stieltjes integral, one has
 b  b
lim Fn dg = lim Fdg, ∀g ∈ BV.
n→∞ a n→∞ a

Hence, in view of Lemma 2, for every g ∈ BV,


 b  b   b   b
fg = Fg|ab − Fdg = lim Fn g|ab − Fn dg = lim f n g.
a a n→∞ a n→∞ a

The necessity follows.


We now prove the sufficiency. Denote
 b  b
Fn (g) = f n g, F(g) = f g, ∀g ∈ BV.
a a

Thus,
 b  b
lim Fn (g) = lim fn g = f g = F(g), ∀g ∈ BV.
n→∞ n→∞ a a

Hence, Fn as the continuous linear functionals on BV is bounded. By Banach-


Steinhaus theorem, the sequence {Fn } is uniformly bounded on [a, b].
Take g(x) = χ[a,x] , then Fn (x) → F(x) pointwise on [a, b] and the sufficiency is
complete.

123
The distributional Henstock–Kurzweil integral and… 981

Remark 4 In Theorem 1, we obtain a sufficient and necessary condition for the con-
vergence of DHK integrals. A special case was discussed by [27, Corollary 3.3], in
which the author only gave a sufficient condition for HK integrals.
Let O(BV) be the unit ball in BV, i.e.,

O(BV) = {g ∈ BV :
g
BV ≤ 1}. (3.1)

One has
Theorem 2 Let { f n } ⊂ DHK . If f n → f ∈ D ((a, b)) weakly in BV uniformly on
b b
O(BV) as n → ∞, then f ∈ DHK and a f = limn→∞ a f n .
Proof Take g(x) = χ[a,x] , then g ∈ O(BV). By the assumptions of the theorem,
b x
limn→∞ a f n g = limn→∞ a f n exists. Denote
 b  x
F(x) = lim f n g = lim fn ,
n→∞ a n→∞ a

and
 b  x
Fn (x) = fn g = fn .
a a

Then F(x) = limn→∞ Fn (x) for every x ∈ [a, b].


b
Since limn→∞  f n , g = limn→∞ a f n g =  f, g uniformly for g ∈ O(BV), so
limn→∞ f n = f and then limn→∞ Fn (x) = F(x) uniformly for x ∈ [a, b]. Hence,
Fn ∈ C0 yields F ∈ C0 .
Now taking g = φ ∈ D((a, b)), we have

 f, φ = lim  f n , φ = − lim Fn , φ  = −F, φ  = F  , φ .


n→∞ n→∞

b b
So F  = f . Therefore, f ∈ DHK and a f = limn→∞ a fn .
We now discuss the convergence of the product sequence { f gn } for f ∈ DHK and
gn ∈ BV, n = 1, 2, . . ..
Theorem 3 Let f ∈ DHK . If gn , g ∈ BV and gn → g in BV as n → ∞. Then
f g ∈ DHK and
 b  b
f gn → f g (n → ∞).
a a

Proof By Theorem 2 and f ∈ DHK , g ∈ BV, we have f g ∈ DHK . It follows from


Theorem 3 that
 b  b   b 
   
 f g − f g  =  f (g − g)  ≤ 2
f

gn − g
BV → 0 (n → ∞).
 n   n 
a a a

123
982 G. Ye, W. Liu

b b
We have thus proved a f gn → a f g as n → ∞.
Remark 5 This theorem is a reduction of [12, Theorem 19]. The uniform boundedness
condition for {gn } in [12, Theorem 19] is omitted here.
Recall that if X is a separable Banach space and its dual space X ∗ , let E ⊂ X ∗
be bounded, then E is weak* relatively compact. So we can prove the more general
convergence results as follows.
b
Proposition 1 Let f ∈ DHK , {gn } ⊂ BV. If {gn } is bounded in BV and { a f gn }
converges then there exists g ∈ BV such that f g ∈ DHK and
 b  b
f gn → f g (n → ∞).
a a

Proof Since DHK is a separable Banach space and {gn } ⊂ DHK ∗ is bounded in BV,

then {gn } is weak* relatively compact. So, there exist g ∈ BV and a subsequence {gn j }
such that gn j → g ( j → ∞) in BV. By Theorem 3, we have f g ∈ DHK and
 b  b
f gn j → f g ( j → ∞).
a a

b b
b
The convergence of a f gn j implies a f gn → a f g as n → ∞.
b
Proposition 2 Let {gn } ⊂ BV. If a f gn converges for every f ∈ DHK then there
exists a function g ∈ BV such that for every f ∈ DHK , f g ∈ DHK and
 b  b
f gn → f g (n → ∞).
a a
b
Proof Let  f, gn = a f gn . Since for every f ∈ DHK , { f, gn } converges, by
Banach-Steinhaus Theorem, {gn } is bounded in BV. By Proposition 1, there exists
g ∈ BV such that for every f ∈ DHK , f g ∈ DHK and
 b  b
f gn → f g (n → ∞),
a a

which is precisely the assertion of the proposition.


On the other hand, the monotone convergence theorem and the dominated conver-
gence theorem in DHK can be modified from [13, Corollaries 4, 5], which are important
tools to deal with differential and integral equations.
Further, we can obtain a variant monotone convergence theorems.
Theorem 4 Let { f n } be a sequence in DHK , h ∈ DHK such that

f 1 f 2 · · · f n · · · h.

123
The distributional Henstock–Kurzweil integral and… 983

b b
Then there exists f ∈ DHK satisfying limn→∞ a f n = a f .
x x
Proof Let Fn (x) = a f n and H (x) = a h. Then Fn , H ∈ C0 for n ∈ N. Since for
n < m, f n f m h, by (2.6),
 x  x  x
Fn (x) = fn ≤ f m = Fm (x) ≤ h = H (x), ∀x ∈ [a, b].
a a a

Thus, {Fn (x)} is also increasing and bounded. So limn→∞ Fn (x) exists. Let

lim Fn (x) = F(x), ∀x ∈ [a, b]. (3.2)


n→∞

Especially, the insertion x = a and x = b in (3.2) yields

lim Fn (a) = F(a) = 0, lim Fn (b) = F(b).


n→∞ n→∞

Moreover, for all m, n ∈ N with n ≤ m, and for each x ∈ [a, b], we have
 x  b
0 ≤ Fm (x) − Fn (x) = ( fm − fn ) ≤ ( f m − f n ) = Fm (b) − Fn (b).
a a

Therefore, {Fn } uniformly converges to F on [a, b]. Consequently, F ∈ C0 .


b b
Put f = F  , then f ∈ DHK and a f = limn→∞ a f n = A, which complete the
proof.

4 Order cone and fixed point theorem

We first present some basic properties of ordered Banach spaces.


A nonempty closed subset X + of a Banach space X is called a order cone if
X + + X + ⊆ X + , X + ∩(−X + ) = {θ }, θ denotes the zero element in X , and cX + ⊆ X +
for each c ≥ 0. It is easy to see that the ordering relation , defined by

x y if and only if y − x ∈ X +

is a partial ordering in X , and that X + = {y ∈ X | θ y}. The space X equipped with


this partial ordering is called an ordered Banach space. The order interval [y, z] =
{x ∈ X | y x z} is a closed subset of X for all y, z ∈ X . A sequence (subset)
of X is called order bounded if it is contained in an order interval [y, z] of X . We say
that a cone X + of Banach space X is normal if there exists such a constant γ ≥ 1 that

0 x y in X implies
x
≤ γ
y
. (4.1)

X + is said to be regular if all increasing and order bounded sequences of X + converge.


If all norm-bounded and increasing sequences of X + converge, we say that X + is fully
regular.

123
984 G. Ye, W. Liu

Since an ordering in the Banach space DHK is given by (2.5), we define

DHK+ = { f ∈ DHK : f  0}. (4.2)

Then DHK+ is an order cone. x x


For f, g ∈ DHK , denote F(x) = a f and G(x) = a g. Since

0 f g ⇒ 0 ≤ F(x) ≤ G(x), ∀x ∈ [a, b]



F
∞ ≤
G


f

g
.

It shows that
Proposition 3 DHK+ is a normal cone in DHK .
Moreover, by Theorem 4 and the definition of regular cone, the following statement
holds.

Theorem 5 The space DHK is an ordered Banach space with the regular cone DHK+ .

Remark 6 For f, g ∈ DHK , define f ∨ g = (F ∨ G) , where F, G ∈ C0 are the


primitives of f, g, respectively, and F ∨ G = sup(F(x), G(x)) = max(F(x), G(x)).
It is easy to see that f ∨ g ∈ DHK . But we do not have f f ∨ g and g f ∨ g.
Hence, DHK is not a lattice and so is not a Banach lattice. In fact, under the ordering
(2.7), DHK is a Banach lattice, see details in [12]. But the monotone convergence
theorem and dominated convergence theorem may not be valid in this situation, and
therefore DHK+ is not a regular cone in DHK . We will discuss the ordering (2.7) in
DHK and the relevant applications in other place.

Since DHK+ is a regular cone in DHK , the next result follows from [28,
Lemma 2.3.1].

Lemma 5 (i) Each totally ordered subset with an upper bound in DHK has the supre-
mum.
(ii) Each totally ordered subset with an lower bound in DHK has the infimum.

Let X be an ordered normed space. The operator T : X → X is increasing if


T u T v, whenever u, v ∈ X and u v.
Using Lemma 5, we give a general fixed point theorem.

Theorem 6 Assume u 0 , v0 ∈ DHK , u 0 v0 , T : [u 0 , v0 ] → DHK is an increasing


operator and u 0 T u 0 , T v0 v0 , then T has least and greatest fixed points u ∗ and
u ∗ on [u 0 , v0 ] and
u ∗ = lim u n , u ∗ = lim vn , (4.3)
n→∞ n→∞
where u n = T u n−1 , vn = T vn−1 (n = 1, 2, . . .) satisfying

u 0 u 1 · · · u n · · · u ∗ u ∗ · · · vn · · · v1 v0 . (4.4)

123
The distributional Henstock–Kurzweil integral and… 985

Proof According to Theorem 5, DHK is an ordered Banach space with the regular
cone DHK+ . By [28, Theorem 3.1.4], the statement holds.
Corollary 1 Assume that the assumptions in Theorem 6 hold. If T has a unique fixed
point x on [u 0 , v0 ], then for every initial value x0 ∈ [u 0 , v0 ], the iterative sequence
xn = T xn−1 (n = 1, 2, . . .) converges to x ∈ DHK , i.e.,
xn − x
→ 0 (n → ∞).
Proof See details in [28, Corollary 3.1.1].

5 Applications

In this section, Theorem 6 is applied to prove existence of solutions for nonlinear


integral equation as follows:
 b
u(t) = f (t, s, u(s))ds, t ∈ [a, b]. (5.1)
a

where f : [a, b] × [a, b] × DHK → DHK , and [a, b] is a compact real interval,
−∞ < a < b < ∞.
Definition 3 We say that u ∈ DHK is a lower solution of (5.1) if
 b
u(.) f (., s, u(s)). (5.2)
a

If the reversed inequality holds in (5.2), we say that u is an upper solution of (5.1). If
equality holds in (5.2), we say that u is a solution of (5.1).
We assume that f satisfies the following hypotheses:
b
( f 1 ) f (t, ., u(.)) and a f (., s, u(s))ds belong to DHK for all t ∈ [a, b] and u ∈ DHK ;
( f 2 ) f (t, s, z) is increasing with respect to z for all (t, s) ∈ [a, b] × [a, b];
b
( f 3 ) there exist u 0 , v0 ∈ DHK , u 0 v0 , such that u 0 a f (., s, u(s))ds v0 for
all u ∈ DHK ;
The main result is given as follows.
Theorem 7 Assume that the hypotheses ( f 1 )–( f 3 ) are satisfied. Then the integral
equation (5.1) has least and greatest solutions, and they are increasing with respect
to f .
Proof Given an operator T : DHK → DHK , such that
 b
T u(t) = f (t, s, u(s))ds, t ∈ [a, b]. (5.3)
a

If u, v ∈ DHK and u v, then by ( f 2 ) and ( f 3 ),


 b  b
u0 f (., s, u(s))ds f (., s, v(s))ds v0 .
a a

123
986 G. Ye, W. Liu

So that T is increasing and T ([u 0 , v0 ]) ⊆ [u 0 , v0 ]. And u 0 T u 0 , T v0 v0 .


Therefore, by Theorem 6, T has least and greatest fixed point u ∗ and u ∗ , and they are
increasing with respect to f .
Next, by successive approximations, we obtain the least and greatest solutions of
the integral equation (5.1).
Proposition 4 Assume that the hypotheses ( f 1 )–( f 3 ) are valid, the operator T satis-
fies Eq. (5.3), then the successive approximations:
(i) u n+1 = T u n converges to the least solution u ∗ of (5.1);
(ii) vn+1 = T vn , converges to the greatest solution u ∗ of (5.1).
Proof From Theorem 7, it follows that

u0 u∗ u∗ u0.

The given hypotheses imply that

u 0 u 1 · · · u n · · · u ∗ u ∗ · · · vn · · · v1 v0 . (5.4)

Thus (u n )∞ ∞
n=0 is increasing and order bounded, then limit u of (u n )n=0 exists by
Theorem 7, and T u = u. Therefore we obtain that u u ∗ by (5.4). Since u ∗ is the
least fixed point of T , then u ∗ u. Thus u ∗ = u, then (i) holds.
The proof of (ii) is similar.
Now, we give two examples to illustrate our results.
Example 1 Given p ∈ DHK with p  0 on [0, 1], and define a function f : [0, 1] ×
[0, 1] × DHK → DHK by


⎪ M p(t), 0 y M p(t),


⎨ (M + 1) p(t), y  (M + 1) p(t),
f (t, s, y) = (M + 2−m−1 ) p(t), (M + 2−m−1 ) p(t) ≺ (5.5)



⎪ ≺ y ≺ (M + 2−m ) p(t),

− f (t, s, −y), y ≺ 0,

where M ≥ 0, m ∈ N. Thus, the nonlinear integral equation


 1
u(t) = f (t, s, u(s))ds, t ∈ [0, 1] (5.6)
0

has extremal solutions. Moreover, the greatest solution u ∗ = (M + 1) p(t) and the
least solution u ∗ = M p(t).
Proof It is easy to show that the hypotheses ( f 1 )–( f 3 ) hold while u 0 = M p(t) and
v0 = (M + 1) p(t). According to Theorem 7, (5.6) has extremal solutions.
Moreover, by Proposition 4, (5.6) has the least solution u ∗ = M p(t) and the greatest
solution u ∗ = (M + 1) p(t).

123
The distributional Henstock–Kurzweil integral and… 987

Furthermore, in Example 1, if


⎪ M p(t), 0 y M p(t),


⎨ (M + 1) p(t), y  (M + 1) p(t),
f (t, s, y) = (M + 2−m−1 ) p(t), (M + 2−m−1 ) p(t) ≺



⎪ ≺ y (M + 2−m ) p(t),

− f (t, s, −y), y ≺ 0.

Then, Eq. (5.6) has extremal solutions u ∗ = u ∗ = M p(t). Moreover, by Corollary 1,


for every initial value u 0 ∈ [M p(t), (M + 1) p(t)], the iterative sequence u n = T u n−1
(n = 1, 2, . . .) converges to u ∗ ∈ DHK , i.e.,
u n − u ∗
→ 0 (n → ∞).
Here is another example.

Example 2 Consider the distributional integral equation given by


 t
x(t) = x(0) + (x(s) + R  (s))ds, t ∈ [0, 1], (5.7)
0

where x ∈ C([0, 1]), R  is the distributional derivative of Riemann function R, i.e.,



 sin n 2 π t
R(t) = . (5.8)
n2
n=1

Then the Eq. (5.7) has at least one solution.

Proof Obviously, (5.7) is equivalent to

x  = x(t) + R  , t ∈ [0, 1], (5.9)

By (5.9), we have
(e−t x) = e−t R  . (5.10)
Integrating (5.10), we have

  t
−t −t 1
e x(t) − x(0) = e R(t) + e−s sin n 2 π sds. (5.11)
n2 0
n=1

Since
 t n 2 π − n 2 π e−t cos n 2 π t − e−t sin n 2 π t
e−s sin n 2 π sds = , (5.12)
0 n4π 2 + 1

substituting (5.12) into (5.11) yields



 π et − π cos n 2 π t + n 2 π 2 sin n 2 π t
x(t) = et x(0) + . (5.13)
n4π 2 + 1
n=1

123
988 G. Ye, W. Liu

Integrating (5.13), we have


 ∞

t n 2 π et − sin n 2 π t − n 2 π cos n 2 π t
x(s)ds = et x(0) − x(0) + , (5.14)
0 n6π 2 + n2
n=1

which, together with (5.13), implies


 t  t
x(t) − x(0) = x(s)ds + R(t) = (x(s) + R  )ds. (5.15)
0 0

By Definition 3, x(t) given in (5.13) is a solution of Eq. (5.7).

Remark 7 In (5.7), the right integral is distributional Henstock–Kurzweil integral.


Since x + R  is not Henstock–Kurzweil and Lebesgue integrable, so the classical
method is not available to equation given by (5.7).

Acknowledgments The authors are grateful to the referee for the careful reading and helpful comments.

References
1. Lee, P.Y.: Lanzhou Lectures on Henstock Integration. World Scientific, Singapore (1989)
2. Lee, P.Y., Výborný, R.: Integral: An Easy Approach After Kurzweil and Henstock. Cambridge Uni-
versity Press, Cambridge (2000)
3. Bullen, P.S., Lee, P.Y., Mawhin, J., Muldowney, P., Pfeffer, W.F.: New Integrals. Springer, New York
(1990)
4. Henstock, R.: The general theory of integration. The Clarendon Press, New York (1991)
5. Swartz, C.: Introduction to gauge integrals. World Scientific, Singapore (2001)
6. Lee, T.Y.: Henstock–Kurzweil Integration on Euclidean Spaces. World Scientific, Singapore (2011)
7. Kurzweil, J.: Henstock–Kurzweil Integration: Its Relation to Topological Vector Spaces. World Sci-
entific, Singapore (2000)
8. Kurzweil, J.: Integration Between the Lebesgue Integral and the Henstock–Kurzweil Integral: Its
Relation to Local Convex Vector Spaces. World Scientific, Singapore (2001)
9. Stroock, D.W.: Essentials of Integration Theory for Analysis. Springer, New York (2011)
10. Schwabik, Š., Ye, G.: Topics in Banach Space Integration. World Scientific, Singapore (2005)
11. Ye, G., Schwabik, Š.: The McShane and the Pettis integral of Banach space-valued functions defined
on Rm . Ill. J. Math. 46(4), 1125–1144 (2002)
12. Talvila, E.: The distributional Denjoy integral. Real Anal. Exch. 33, 51–82 (2008)
13. Ang, D.D., Schmitt, K., Vy, L.K.: A multidimensional analogue of the Denjoy–Perron–Henstock–
Kurzweil integral. Bull. Belg. Math. Soc. Simon Stevin 4, 355–371 (1997)
14. Schwabik, Š.: Generalized Ordinary Differential Equations. World Scientific, Singapore (1992)
15. Schwabik, Š.: The Perron integral in ordinary differential equations. Differ. Integral Equ. 6(4), 863–882
(1993)
16. Chew, T.S., Van-Brunt, B., Wake, G.C.: On retarded functional-differential equations and Henstock–
Kurzweil integrals. Differ. Integral Equ. 9(3), 569–580 (1996)
17. Chew, T.S., Van-Brunt, B., Wake, G.C.: First-order partial differential equations and Henstock–
Kurzweil integrals. Differ. Integral Equ. 10(5), 947–960 (1997)
18. Chew, T.S., Flordeliza, F.: On x  = f (t, x) and Henstock–Kurzweil integrals. Differ. Integral Equ.
4(4), 861–868 (1991)
19. Chew, T.S.: On Kurzweil generalized ordinary differential equations. J. Differ. Equ. 76(2), 286–293
(1988)
20. Krejčí, P., Lamba, H., Melnik, S., Rachinskii, D.: Analytical solution for a class of network dynamics
with mechanical and financial applications. Phys. Rev. E 90(3), 032822 (2014)

123
The distributional Henstock–Kurzweil integral and… 989

21. Muldowney, P.: A modern theory of random variation. Wiley, Hoboken, NJ (2012)
22. Muldowney, P.: The Henstock integral and the Black-Scholes theory of derivative asset pricing. Real
Anal. Exch. 26 (1), 117–131 (2000/2001)
23. Gill, T.L., Zachary, W.W.: Foundations for relativistic quantum theory I: Feynman’s operator calculus
and the Dyson conjectures. J. Math. Phys. 43(1), 69–93 (2002)
24. Aye, K.K., Lee, P.Y.: The dual of the space of functions of bounded varriation. Math. Bohem. 131(1),
1–9 (2006)
25. Xue, X., Zhang, B.: Properties of set-valued functions of bounded variation in a Banach space. J.
Harbin Inst. Tech. 3, 102–105, 107 (1991). (Chinese)
26. Wu, C., Xue, X.: Abstract-valued functions of bounded variation in locally convex spaces. Acta Math.
Sin. 33(1), 107–112 (1990). (Chinese)
27. Talvila, E.: Limits and Henstock integrals of products. Real Anal. Exch. 25, 907–918 (1999–2000)
28. Guo, D., Cho, Y.J., Zhu, J.: Partial ordering methods in nonlinear problems. Nova Science Publishers
Inc, New York (2004)

123

You might also like