第 15卷第 2期
200】年 6月
模 糊 系 统 与 数 学
Fuzzy Systems and M athematics
Vo1.15,No.2
Jun.,2001
Article ID:1001—7402(200])02-0051一g4
Convergence Theorems of the Choquet Integral
GUo Cai—mei .ZHANG De-IF
(1·Department of Basic Science,Changchun University,Changchun 130022
.China
2 Depa rtment of Computer Science.Jilin University,Changchun 130023㈣Chi )
Abstract:So fa r·the Choquet integral with respect to a fuzzy ~leasu re ha h n tudi d uch e ten vP
by Dr M urofu i’Sugeno and F,ome othe rs
. But the convergence theory is not enough The pape r|㈣ i is
Ju t d scuss1ng this topic·and~hows∞ 0nvergence theorems for the Choquet integra1.These in Lude
generalized R?Gnotone convergence theorem·Fatou。 lerrl[I]8s,etc We conclude that the Choquet inte ra1
has the same convergence theorenls as Sugeno integral o'wns
.
Key words:M easu re Theory;('hcxiuet Integral;Fuzzy Measure:Fuzzy Integral}Convergence The[1re】lL
CLC number:(I1 74.1 2 Document code.A
1 Introduction
Since Sugeno introduced the concept of
f uzzy measures in 1 974.the fuzzy integral with
respect 1o a fuzzy measure is generally developed.
such as Sugeno .Ralescu .Zhao ”一.Suarez!。.
W u .etc.The common property of all the vari
OOS kinds of fuzzy integrals is that they are monc~
tonic functional from the space of nonnegat[ve
measurable functions to E0,。。] Just based on the
property,M urofushi and Sugeno also view the
Choquet integral which x~-as introduced by
Choquet[ ]in 1 953.as a kind of fuzzy,integrals.
and have given a much extensive study in E3 6].
It ls well—known that convergence theorems
a very important in classical integral theory.
But to the Choquet integra1.convergence theory
is not enough.Since so far,only the monotone
convergence theorem ls shown . W hethcr can
other convergence theorems be established? The
answer is JLJS1 the paper s purpose.
In the paper.we will show various kinds of
gcnerali~d convergence theorems of the Choquet
integra1.these include gegeralized monotone COIl—
vergenee theorem ,generalized Fatou s 1emmas.
etc. The rest of the paper is divided into two
parts. Section 2 will give some concepts and
reviews on fuzzy measures and the Choquet inte
gra1 as preparation.Section 3 wil1 show the main
results of this pape r.
2 Fuzzy measures and
the Choquet integral
I et be a nonempty classical set. a
o-algebra formed by the subsets of .( , )
the measurable space.
Definition 2.1 Let : 一E0.。。]be a
set f1】nct】on.Then
:麓 口n da do nin OSZ ;KE I.⋯of SE e of J ilinU uive r sgy ⋯ 。
noil—additireintegraIs,ZHANG DeIIt1 964 ).male.born n Nong’an JI1 p~ofessor and vice—president of Jgin Prov inst。t Educatlon.
f rcn1[恒 on :ugzy analvsts
维普资讯 http://www.cqvip.com
模 糊 系 统 与 数 学 200]正
(1)be is said to be empty null if be(0)一_0; follow
(2)beis said to be monotone ifbe(A)≤be(B)
whenever A[ B;
(3)be is said to be continuous form below if
( )十 ( )whenever A 十A;
(4) is said to be conditionally continuous
form above_f (A ) (A)whenever ,.+A.
and be(A )<。e for a fixed n-≥ l_
Definition 2.2一 A set function be:一e/一
[0,。。]is called a fuzzy'Ineasure if it is empty
null and monotone.
The triplet f。X . ≯ .日j is called a fuzzy
measure space.be is said to be continuous.if it is
both continuous from below and conditionally
continuous from above.
I eI the symbol M ‘ )denote the set of all
fuzzy measures on (X一。 ).
Proposition 2.1~ Let{ ..}be a sequence
of fuzzv measures on(X,。 )1.c.{ } M tX).
and a set function from √ to[0.。。:.If{ ..}
converges to be(setwise convergence,i.e. pr,.(.1j
一 t^ )for every ^∈ 。,simply wrhc as —
J.then be is a fuzzy measure.
Remark 2.1 In this proposition. if
,
gA..
(,?≥1)is continuous.and 一 .then it doesn 【
imply that is continuous. (see a counter
example given by Zhang .)
Proposition 2.2 Let{ ..}be a.sequence
of fuzzy measures on (X. 。,).and be a set—
function from。 to[0.∞].If{ ..}converges
uniformly to ,then ( ≥ 1)is continuous
implies th “is conlinuous.
A function f:X一 [0.。c]is said to be
.measurable if f ( )∈ 。,for each B∈
Borel([0.。。]).
All the measurable function from 二, to
Ln.∞]is denmed hy F(x)
Definition2 3t I et/ ∈ F(JY). ∈
M ( ). Then the Choque1 integral of/ with
respect to .which denoted by‘c)Ifdbe is as
(c)I —l be(f>t)dt
W here the right side integral is Lebesgue
integra1.and(厂>r)stand for{ ∈X:,( )>f}.
The propositions of the Choquet integral
refered to r1.4].
Definition 2.4一 A set N ∈ ,is calted a
null—set with respect to iff (AUⅣ)一_ ( )·
for all A∈ .
By using the null set .the 'almost ever—
where concept is defined as! J尸‘ )a.e Ine&ns
that there【s a nuIl set N .such that P(』) t rue
for all ∈ .where P 【 ) is a proposition
concerning the points of X.
3 Convergence theorems
rheorem 3.1 I et{ .}[F(X)./∈I-"t j
(1) Let be be cont Jntlous from below If
』,.‘』a.e..then
(f){ dbe十( )Ifdbe
t2)l et he conditionally continuous from
above.If . /a.e.,and iff ≤F a.e.for some
r
g∈F(x)with(c)Igdbe< ∞,then
(C)
.
[fod (C)J fd~
Lemma 3.1 I e1 I [ M (X). ∈ M (X),
/-∈ F(X).
(1)If ..十 ,then
(c)lfd 十(c)Ifdbe
(2)If ,and
(c)ll厂d < 。。for sortie ”≥ 1
lhen
(c)Jfd/J.. (C)Jfd~
Theorem 3.2 I et{ .)[ F(X).f∈
F(x), [ M (X). ∈M (X).If A 十 .
十 .then is continuous from below implies
‘cj 1 .d .十(c)Ifdp,
维普资讯 http://www.cqvip.com
第 2期 GUO Cai—mei,ZHANG De一 Co rgeⅢ Theorems of rhe Choquer Int gH】 53
Theorem 3.3 Let{A }CF(X),,∈F(X).
r
1 cM( ), ∈M( ),and(c)l d <。。
for some ≥l_If .v f a.e., + ,then is
conditionally continuous form above implies tf1at
r
(c)l .d +(c){/d
J J
Theorem 3.4 I et i )[ F(X), .. [
M tx 1.If 堕 is continuous from below.then
r
(c)f(1imf.】d(Iim )≤liIn(c)l/ d
~ ll — l y J
Theorem 3.5 I.et( .}二F‘X J. }[
M(X).and(c)l(sup )d(sup )<。。for s。f【1c
J ^
≥ 1.If lim is conditionally continuous from
above,then
【 ) d,‘≤ t[1_)i(1ira.(.)d‘Tin1 ‘j
。 J ● ·
Theorem 3 6 、.el tj ,=.FtX 3 1∈
Fl ).{ ..}二 (X).,ll∈ ( ),and is bo!h
continuous from below and conditionally contintl
F( ). ∈M( ).If(c)llfo~fld — 0("—
,·then we say that{, is C—mean convergent
to f.It is simply written by fo— _,(f m).
Definition 3.2。 1 Let j \CFtx) {∈
F( ), ∈M ( ).Iflim,a(。 一_厂l≥ e)一0for
ever),£> 0,then we say{j \converges to f in
fuzzy measure . In short.it is denoted by
三 {
Theorem 3.7 Let i [ F(X).f∈
Ftx).If fo一 }c m),then j .
l'heorem 3 8 I.ct i j[ F tX ), ∈
Ff ). ∈ M ( j. ( )= < 。。.If there exists
QF(X).(c)f算d <∞,such that lJ ~ff≤
g for each 1.then j j implies j 一 L m
Proof By the definitlon of the Choquet
integra1.for each ≥¨ 1.we have
“’) _,1.一厂d —j (Jfo—fl>,Jd,
r
(】us from above.If Lc)f(supfDd(sup,%)< but
⋯ 一 一 ⋯
for some ≥1.th /..一厂. 一 impllcs I (Jfo~fl> 1)d
(f)I/ d 一(c)Ijd[x
Corollary 3.1 Let{ }[ F(X ), ,}[
M (X).
(1)If is continuous from below.then
(c)f(1imf,)d ≤lim(C) .d
(2) If is conditionally continuous from
above,{(sup^ )dp< ∞ for some ≥ 1.then
Iim(c)I d ≤(c)l(1iraf,.)d
Coroll~try 3.2 IA f∥[F(X).{ [
M 【 ).Then
【1)((、)I/d(1lm )≤ lim(c)Ifd
(2)If((、,1 (sup,u j<∞ for some ≥1,
th n
¨f 一,≥
([fo—f『≥ )出+f” (1 一厂≥ejd
By making use of{ } e
lim (1 一 ,『≥ £j= 0 for V e> 0
and l 一/I≤g for each ≥ 1,from Lebesgue
dominated convergence theorem of Lebesgue inte:
gra].vee have
Ⅳ
lirai ‘『 一厂 ≥ E)df=0
n o=J
By (1 .一f『≥f)≤ ( )=M
题
快递公司问题件快递公司问题件货款处理关于圆的周长面积重点题型关于解方程组的题及答案关于南海问题
.给出 Choquet积分的一些收敛定理 ,包括广义单
调收敛定理、Fatou引理等。从 中可以看出,Choquet积分与 Sugeno模糊积分具有相同的收敛定理
关键词:和4度论;Choquet积分;模糊测度;模糊积分;收敛定理
维普资讯 http://www.cqvip.com