See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/243064660

Henstock Integrable Functions are Lebesgue Integrable on a Portion

Article  in  Proceedings of the American Mathematical Society · January 1991

DOI: 10.1090/S0002-9939-1991-1034883-6

CITATIONS READS

7 163

1 author:

Zoltán Buczolich
Eötvös Loránd University
138 PUBLICATIONS   539 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Typical properties of functions and measures View project

Translates of continuous functions View project

All content following this page was uploaded by Zoltán Buczolich on 18 December 2014.

The user has requested enhancement of the downloaded file.

 proceedings of the
american mathematical society
Volume 111. Number 1, January 1991

HENSTOCK INTEGRABLE FUNCTIONS

ARE LEBESGUE INTEGRABLE ON A PORTION

ZOLTÁN BUCZOLICH

(Communicated by R. Daniel Mauldin)

Abstract. If a real function / defined on an interval / C Rm is Henstock

integrable, then one can always find a nondegenerate subinterval J C / on
which / is Lebesgue integrable.

Introduction
In his memoir [O; 4.6.4, p. 87] Krzysztof M. Ostaszewski mentioned that "the
question posed by Karták in [K]—whether, for a Perron-integrable function, one
can find a nondegenerate interval on which it is Lebesgue-integrable—remains
unanswered." In this paper we give an affirmative answer to this question.
The one-dimensional case of our theorem is an old and well-known result [S,
Theorem 1.4, Chapter VIII].
Recall that the (classical) Henstock integral is equivalent to the Perron inte-
gral [O, Chapter 2, p. 20]. Since our proof is based on the Henstock integration
process we shall use the name Henstock integral in the statement of our theorem.

Preliminaries
By Rm we denote the w-dimensional Euclidean space. By an interval I we
mean a set of the form x^,[a(, b/\. The w-dimensional Lebesgue measure
of the set £cRffl is denoted by \E\. The Lebesgue integral of a function /
over a set E is denoted by (L) JEf. We say that the intervals 7 and J are
non-overlapping if int(7) n int(/) = 0 . If the intervals Bt. c 7 , i = I, ... , k ,
are non-overlapping and 7 = U{72;.: / = 1, ... , k}, then we say that the set
D = {B¡: i = 1,... , k} is a division of 7 . A partition of an interval 7 is a
collection P = {{B¡, x¡) : i = 1, ... , k}, where D(P) = {B¡: i = 1, ... , k} is
a division of 7 and x¡ G'Bj,, i = 1,..., k.' If a function ô : I —»(0, +00) and
a partition P of I is given we say that P is ¿-fine if Bt c {y: dist(x(, y) <
ô(Xj)} for every i<= 1,,.., k. If / is a function defined on 7 and P is a

Received by the editors September 18, 1989.

1980 Mathematics Subject Classification (1985 Revision). Primary 26A39, 26A42.
This paper was prepared while the author visited the University of California at Davis.

©1991 American Mathematical Society

0002-9939/91 $1.00+ $.25 per page

127

License or copyright restrictions may apply to redistribution; see http://www.ams.org/journal-terms-of-use

 128 zoltan buczolich

partition of 7 , then we put

k
a(f,P) = J2f(xi)\Bl\.
1=1

We also recall the definition of the Henstock integral.

Definition. A function /:/->R is Henstock integrable on 7 if there is a real
number denoted by (77) f} f with the following property: for every e > 0,
there is a S : I —►
(0, +00) such that

a(f,P)-(H)jf <s
for each ¿-fine partition of 7.
We remark that Henstock integrable functions are measurable. This holds
even for generalizations of the Henstock integral, see for example [PI, 4.3 Corol-
lary] or [P2, 4.5 Corollary].
The main result
Theorem. If I c Rm ¿s an interval and f is Henstock integrable on I, then
there exists a nondegenerate interval J c I such that f is Lebesgue integrable
on J.
Proof. Using the existence of (77) /7 /, choose ¿: I —►
(0, +00) such that if P
and P' are ¿-fine partitions of 7 then
\a{f,P)-a{f, P')\ < \I\.
Put An = {jce/: |/(x)| < n, S(x) > l/n}. Since ô(x) is positive and f(x) is
finite we have UneN ^n = J ' aní*>^ Baire's Category Theorem, there exists an
n e N and a portion J c 7 such that An is dense in / . We can also assume
that diam(/) < l/n . We shall prove that (L)/y/ exists. For a contradiction
suppose that this is not true. Then (7_) fy |/| = +00, and we can choose a
measurable set E c J such that
I r I
\(L) f\>3-n-\I\.
I Je I
Put
f f(x), if x e £,
<?(•*)= {
[ 0, otherwise.
Since g(x) is Lebesgue and a fortiori Henstock integrable, we choose ô : / —»
(0, +00) such that ¿'(x) < ¿(x) for x e J , and if T5' is a ¿'-fine partition of
/ then
a{g,P')-{L)j f <n\I\.
Suppose that P' is a fixed ¿'-fine partition of J. Choose P* a ¿-fine
partition of 7 \ / and put Px = P' U P*. Then Px is a ¿-fine partition of 7 .

License or copyright restrictions may apply to redistribution; see http://www.ams.org/journal-terms-of-use

 henstock integrable functions 124

Suppose that P' = {{B¡, x¡): / = 1, ..., k'}. If x¡ $ E then put x" = x(. If
x( e E then, using the density of An in /, choose an x" e An n Bi. Put 7>" =
{{Bi, x'¡): i = 1, ... , k'}. Since ¿(x) > l/n for x e An, diam(7) < l/n,
and ¿' < ¿ the partition P" is ¿-fine on J. Put P2 = P" U P*. Then P2 is
¿-fine on 7. We obtain

\I\>\a(f,P,)-a(f,P2)\ £ /(x,)|5,|+ E /(x;.)|7i.|

{B,x,)€P' (B(,*,)€/''

E MW- E /(*f)i*4
{BI>X¡)€P" (Bl,x';)eP"

E M)W- E /K")i^ *o-

(B^x^eP' (B,,x")Ç.P"

Using the fact that if x( <£ £ then x; = x;" , we have

E /(*mi- E /(*?)W
x,e£, (ß,, x,)gp' x¡eE, (B¡, x" )eP"

>W(g,P)\ E f(x")\Bi = *|-*2-

*je£,(5,,x,")6/>"

Recalling that |(L) /y #| > 3 • n • |/| and |fj(g , 7>')- (L) /y g| < «|7|, we obtain
T, > 2«|7|. To estimate yV2recall that x" G ^„ and hence |/(x")| < n . Thus
*F2<n\J\ < n\I\. Therefore we obtain |7| >y¥x-y¥2 >2n\I\-n\I\ = n\I\ > |7|,
a contradiction proving our theorem. □

References
[K] K. Karták, K leorii vicerozmërného inlegrdlu, Casopis Pest. Mat. 80 (1955), 400-414.
[O] K. M. Ostaszewski, Henstock integration in the plane, Mem. Amer. Math. Soc, no. 63,
Amer. Math. Soc, Providence, RI, 1986.
[PI] W. F. Pfeffer, The multidimensional fundamental theorem of calculus, J. Austral. Math. Soc.
Ser. A 43 (1987), 143-170.
[P2] _, The divergence theorem. Trans. Amer. Math. Soc. 295 (1986), 665-685.
[S] S. Saks, Theory of the integral, Hafner, New York, 1937.

Eötvös Loránd University, Department of Analysis, Budapest, Múzeum krt 6-8, H-

1088, Hungary

License or copyright restrictions may apply to redistribution; see http://www.ams.org/journal-terms-of-use

View publication stats