• 検索結果がありません。

A NEW ORDERED COMPACTIFICATION

N/A
N/A
Protected

Academic year: 2022

シェア "A NEW ORDERED COMPACTIFICATION"

Copied!
8
0
0

読み込み中.... (全文を見る)

全文

(1)

Internat. J. Math. & Math. Scl.

VOL. 16 NO. (1993) 117-124

A NEW ORDERED COMPACTIFICATION

D.C.KENT

DepartmentofPureandAppliedMathematics Washington State University

Pullman, WA99164

T.A.RICHMOND DepartmentofMathematics WesternKentucky University

BowlingGreen, KY42101

(Received September I0, 1991 and in revised form April II, 1992)

ABSTRACT. A new Wallman-type ordered compactification

%X

is constructed using maximal C

Z-

filters

{which

havefilterbasesobtainedfrom increasing and decreasingzero

sets)

asthe underlyingset. A necessaryand sufficient conditionisgiven for

oX

tocoincidewiththe Nachbin compactification

[oX;

in particular

%X oX

whenever

X

has thediscreteorder. The Wallman ordered compactification equals

oX

whenev.er XisasubspaceofR". Itis shownthat

oX

isalways

T,

butcanfail to be

Tx-ordered

or

T2.

KEY WORDS AND PHRASES.

CZ-set,

maximalCZ-filter,

T-ordered

space,

T2-ordered

space,Nachbin

compactification, Wallman ordered compactification.

1991AMSSUBJECTCLASSIFICATION CODES. 54F05, 54D35,54D10

0. INTRODUCTION.

L.Nachbin

[10]

initiatedthestudy of ordered compactifications whenhe characterizedthe topological ordered spacesthatallow

T2-ordered

compactifications

(we

callthese

Ts.s-ordered spaces),

and constructed the largest such

T-ordered

compactification [oX by embedding X in an

ordered

cube. The Nachbin

(or

Stone-(ech

ordered)

compactifieation

oX

has beenstudiedand applied byvarious authors

(see,

for instance,ouroddnumbered

references).

Asecond ordered

(but

notnecessarily

T2-ordered)

compactiflcation

oX,

cared the Wdlman ordered eompactification,wasintroducedby Choeand Park

[2].

Anecessary and sufficient conditionfor

woX

[oX was givenin

[6],

and in

[8]

the separation propertiesof

woX

were

investigated.

Itis well known

(e.g.,

see

[4])

thattheStone-(echcompactification

X

ofa

Ts.s

topologicalspacecan be described asaWallman-typecompactification usingmaxima]filtersofzerosetsasthe underlyingset for thecompactification. Wehaveextended this construction to

Ts.s-ordered

spaces, and the result isa newordered compactificationwhichwecall

%X.

Thisnewcompactification,like[oXand

woX,

has the universal extensionproperty for increasing, continuous mapsintocompact,

T-ordered

spaces. Withthe helpofthis universalproperty,weobtainnecessary andsufficientconditions for

%X

[oX;inparticular thisequality holds when the order ofXisdiscrete. Asanalternativeapproachtoconstructing

oX, %X

ismoresatisfactorythan

oX,

in thesensethat

oX

and

floX

coincideon alarger classof spacesthan do

oX

and[oX. Althoughwehave not yetcharacterizedthe class of spaces forwhich

%X woX,

wehave

shown thatthisclass includesallsubspacesof

R n.

Thisresultenablesustoshowthat

%X

canexhibit

thesame "pathological behaviorrelative toseparation properties thatwasdemonstrated for

oX

in

[8].

(2)

118 D.C. KENT AND T.A. RICHMOND Forexample,

%X

fails tobe

T-ordered

ifX R for,t

>

3.

Itremainsanopen question whether[oXcanbedescribed viaaWallman-typeordered compactification forall

Ts.s-ordered

spaces

X’.

1. PRELIMINARIES.

Let

(X, <)

bea

poser

and letAbeanon-empty subset ofX. Let

d(A) {z

EX"z

<

aforsome aE

A)

and

i(A) {z

EX"a

<

zforsomeaE

A),

incaseA

{z},

wewrite

d(z)

and

i(z)

rather than

d({z))

and

i({z}).

ThesetAissaidtobe decregsin

(respectively, increas/n)

ifA

d(A) (respectively,

A

i(A)).

Aset whichis eitherincreasingordecreasing issaid tobemonotone;if

A d(A)Oi{A),

thenAisconvez.

If

f" {X, <) (Y, <)

isafunction betweentwo

posers,

then

1’

is

increas/n9 (respectively, decrea/ng)

if

z

<

yinXimplies

f(z) < f(I/) (respectively, (I/) < (z)

inY.

A topologicalordered space

(X, <,r)

isatriple consistingofaposet

(X, <)

and aconvex topology

r on

X;

ris conveziftheopen monotone setsformanopensubbase. Theterm space willalwaysmean topological orderedspace,and

(X, <, r)

willbeshortenedto

X

when thereisnoambiguity.Notethat every topological spacecanberegardedas atopological orderedspace relative tothediscreteorder

(equality).

Let Ebe the space

[0,1]

with itsusual order andtopology. For anarbitrary space X we denote by

CI*(X) (respectively, CD*(X))

theset of all increasing

{respectively, decreasing),

continuous maps from

X

intoE. An

increa.sin

zeroset

(respectively, decreasin9

zero

set)

isasetof the form

f-l(0)

where

f CD*(X)

(respectively,

f CI*(X)).

Thesetof all increasingzerosets

(respectively,

decreasingzero

sets)

onXwillbe designatedby

IZ(X) (respectively, DZ(X)).

Usingstandardproceduresdescribedin

[4],

oneeasilyproves the next twopropositions.

PROPOSITION1.1 IfXisaspace,

f _ CI*{X), CD*(X),

andaE

E,

then:

DZ(X); (b) f-([a, 1]) IZ{X); (c) g-([0,a]) IZ(X),

and

(d) 9-([a, 1]) DZ(X).

(ffi) /- ([o, ,,l) e

PROPOSITION 1.2 Foranyspace

X, IZ(X)

and

DZ(X)

areclosed undercountable intersections andfinite unions.

AsubsetAofaspaceXiscalledaC-zeroset

(or C2"-set)

if there isB

I2"(X)

andC

DZ(X)

such

thatA B cC. Let

CZ(X)

bethesetof all CZ-setsonX. One easilyverifiesthe following.

PROPOSITION1.3

(a)

A

CZ(X)

iffthereis g

CI*(X)

andh

OD*(X)

such thatA

]-(0),

where

1/2(g

-t-

h). (b)

Theset

C2"(X)

is closed undercountable intersections.

It isgenerally not true that

C2"(X)

isclosed underfiniteunions; forinstance,

[0,1]

and

[2,3]

are

CZ-setsin

R

whose union is notaC2"-set.

By a

falter :?"

on

X,

wealways mean aproper setfilter

(one

that does not contain

).

Thefilter on X generated by

(z),

forz

X,

will bedenoted by 9. Ifafilter 1"has afilter base ofincreasing zerosets, then }" is called an 12"-filter;

D2"-fdter

and

C2"-filter

aredefinedsimilarly. For an arbitrary

filter }" on

X,

let

I2"(}’) (respectively, D2"(}’), C2’(}’))

be the filter on

X

generated by }"

I2"(X) (respectively,

}"

c DZ(X), Y nC2"(X)).

Notethat

IZ(t)

(respectively,

DZ(I), C2"(1"))

isthefinest12’- filter(respectively, D2"-filter,

CZ-filter)

coarserthan}’. Thenextproposition follows fromZorn’s Lemma.

PROPOSITION1.4 If )"isaC2"-filter

(respectively,

12"-filter,

2"-filter),

thereissmaximalC2"-filter

(respectively,

I2"-filter,

D2"-filter)

finer than}’.

PROPOSITION1.5 Let

X,

Ybespacesand[

X

Yanincreasing,continuous map.

(a)

IfA

IZ(Y)

(respectively, A

DZ(Y),

A

CZ(Y)),

then

I-(A) I2"(X)

(respectively,

I-(A) DZ(X), .f-(A) . CZ(X)).

(b)

If}" isafilter on

X,

then

IZ(I(}’)) < 1(I2"(}’)), D2"(f(}’)) _< I(DZ(Yf)),

and

CZ(I(}’)) < I(CZ(I)).

(3)

PROOF.

(a)

IfA6

IZ(Y),

thenA

g-’(0),

forg

CD*(Y).

Then

J’-’(A) (9 I)-’(0),

where

g

CD*(X),

d

/-(A) IZ(X).

The ohercs esimilar.

(b)

followseilyfrom

().

A spe X is defined

o

be

T-odeed

if, for eh

X, i(z)

d

d(z)

re cled

se.

Aspe X

T-odeed

if, whenever z Fin

X,

there n increing neighborhood of z nd dreing neighborhood V of Fsuch th V

;

equivalently,

(X, , ) T-ordered

ifthe order is cloud sub ofXx X. Aspe Xis

T.-odeed

if isfiesthe followingconditions:

(1)

Ifx

X,

Aisa

cloud subt of

X,

andz

A,

then thereis

/ CI*(X)

d

CD*(X)

such

that/(z) g(z)

0d

/(F) v g(F)

for

A; (2)

If F in

X,

thereis

/ CI*(X)

such

that/(F)

0

and/(z)

1.

The

Ts.s-ordered

spiese prisely thesubspesof compt,

T-ordered

sp

(see [10]).

Aspe

X

is defined to be

T-ordered

ifit is

T-ordered

d, whenever A d

B

e disjoint cld subts with Adreing d

B

increing, thereedisjointopen ts and

V,

the forr dreing,

t

latter increing, suchthatA HdB V. Notethat: compt d

T-ordered T-ordered Ts.s-ordered T-ordered T-order.

obrvethat

T-order T, T-ordered T,

d

Ts.s-order Ts.s (i.e.,

completelyregular d

T);

it isnottrue,however,that

T-ordered T.

h the reminder of thksection we

exane

some properties of

Ts.s-ordered

spies, th spi emphisontherole played by CZ-sets.

PROPOSITION 1.6 Let

X

be a

T.s-ordered

spe. t z

X,

d let

(z)

be the filter of

nighborhoods ofz.

() () CZ(())

() () o o o m o (X ) (X B), CZ(X) a DZ(X).

() CZ(X) a ,u, o

X.

(d)

ff isafilronXsuchthat z,then

CZ() .

PROOF.

(a)

t Vbe anopenneighborhood ofz. Then thereare

CI*(X)

d g

CD*(X)

such that

() g(z)

0d

I(F) v g(F)

if F X V. Then

-([0, ]) g-[(0, }])

is CZ-set

neighborhood of which subset ofV.

{b)

Let l,g,and V be h theproof

of(a).

IfB

1-’{1)

d A

g-’{1),

then A

DZ{X),

B

IZ(X),

andz

{X- A) {X- B)

V.

(c)

d

{d)

follow iediately from

{b)

d

{a),

respectively.

PROPOSITION1.7

In

a

Ts.s-ordered

spe

X,

thefollongstamenareequivalent:

(a)

z y;

(b) Z() ; (=) OZ() .

PROOF. It obvious that

{a) (b).

Toshow

{b) (a),

suppose y is in eh member of

IZ(X)

contningz, butz y. Then thereis

f CI*(X) su

that

f(y)

0d

f{z)

1. Thusy

f-{1),

but

f-l(1)

isamember of

IZ{X)

contningz. Thestablishesthat

{a) {b),

d

{c) {a)

followsby adu

arment.

Inthe next sectionweshl constructacomptcation bdonmaxalCZ-filters. Thenext two propositionswillbeuseful inthisendeavor.

PROPOSITION1.8 IfXisa

Ts.5-ordered

space andz

X,

the

CZ()

isthe uniquemaximalCZ- filteronXcoarserthan

.

PROOF. Wealready know that

CZ(%)

isthefinest CZ-filtercoarserthan

.

Suppose isaCZ-filter

and

CZ()

<

.

ThenthereisaCZ-set G such thatz X G. By Proposition

1.6(a),

thereisa

CZ-neighborhood Hof zsuch thatH X-G. SinceH

CZ(),

the assumption that

CZ()

<

,

is

(4)

120 D.C. KENT AND T.A. RICHMOND

contradicted,anditfollows that

CZ(%)

isamaximal CZ-filter. Itisobviously the onlymaxims]CZ-filter coarserthan k.

PROPOSITION1.9 Let

f X Y

beacontinuous,increasing map, where

X

is

Ts.s-ordered

and

Y

iscompact and

T2-ordered.

If is amaximalCg-filteron

X,

there isauniquepointI/.u E

Y

such that

f()

IJ inY.

PROOF. Let

Y"

be an ultrs]ilteron

X

such that

,q _< Y’.

Since

Y

is compact andT2, there is aunique pointI/J in

Y

such that

f(Y’)

I/s. Because isamaximal Cg-filter,

Cg(Y’) <_ t,

and

f(),l) >_ f(CZ(Y’)) >_ CZ(f(Y’))

followsby Proposition 1.5.

But f(Y’)

I/ implies

CZ(f())

lttby Proposition

l.e(d),

and therefore

2.

THE

COMPACTIFICATION

Throughoutthissection,we assumethat

X

is

of

X

isapair consisting ofacompactspace

Y

andamap

X Y

such that isbothatopological andanorderembeddingof

X

into

Y

suchthat

tr(X)

isdense in

Y. In

thissection,weshall constructan ordered compactification

(IoX, b)

of

X

and establishsomeofits basicproperties.

Let be thesetofS]Imaxims]Cg-filtersonX. ByProposition 1.8, these includes]Ifiltersof the form

CZ(),

where zEX. Arelation and

DZ() < J.

PROPOSITION2.1

(, <

is a

poser.

PROOF. It isclear that

and

DZ(N) _< .

Since is aCZ-filter,

IZ(.t/) v DZ(N) .q,

and so /

<_ J.

Itis alto true that

IZ(.q) <_

.t/and

IZ(N) _<

PROPOSITION2.2 z

_<

Iin

X

if

CZ(%) < CZ(I)

in

}.

PROOF. Ifz

_< ,

thenbyProposition 1.7,

IZ(%) IZ(CZ(%)) <_ ,

which implies

IZ(CZ(%)) <_

CZ(I).

Likewise,

DZ() <_ ,

which implies

DZ(CZ()) <_ CZ(%).

Thus

CZ(%) < CZ(I).

This reasoning isreversible.

Foranarbitrary,non-emptysubset

A

of

X,

wedefine

{ J }" A J}.

PROPOSITION2.S Let

A,B . CZ(X).

() .n tn’-.

(b) u = ,u".

()

(d) .. Z(x), ,.n i

ni.,..i.

,

in

,.

PROOF. All of the rtions ofthispropitionarroutine, andwe and

,

then

ZZ() <_

and is anincreasing set.

w. nx

d,- ,/,

b ,/,(.) CZ(*),

for

,

X. Sy Propiion

.Z,

,/,i.

.

order

.embedding of

X

inX. Weomitthe routineproofofthenextproposition.

POPOSTO Z.

()

For=y

c_ X, ,/,-() c ,.

(b)

If

, CZ(X),

h.n

,/,-() ,

-d

,/,-(X- ) X ..

(5)

1"21

Let be thetopologyon

)

withclosed subbase

{t

A

CZ(X)}.

Fromthe twopreceding propositions, itfollowsthat hasanopen subbase of monotone opensets;thus

(), < 7)

isatopological orderedspace.

Let

"7oX (, < ).

THEOREM2.5 ForanyT3 s-orderedspaceX,

(7oX, )

isanordered compactification forXwhose topologyis

Tx.

PROOF. First notethat

"

X

%X

isatopological embeddingby Propositions

1.6(c)

and

2.4(b);

isalsoanorderembedding,as weobserved previously.

Toshow that

oX

iscompact,it issufficient to show that any collection

C {i A CZ(X),i I}

of subbasic closed sets in

%X

with the finite intersection property has a non-empty intersection. If

. {A I},

then .4has the finite intersectionproperty by Proposition

2.3(a).

Let 31be any maximal CZ-filtercontaining

q;

then31

Toshow that

7Xo

is

Tx,

let)4,91/be two distinct maximal CZ-filterson X. Then therearedisjoint CZ-setsM 31 andN )/. ItfollowsthatX Nisaneighborhoodof31 not containingI/,andX M isaneighborhood of)/notcontaining 31.

Finally,if31

),

then

(31)

converges to 31in%X,andtherefore

(X)

isdensein

oX.

The next theorem shows that

%X

hasthesameuniversal extensionproperty as

woX

and

BoX.

THEOREM2.6 LetXbea

Ta.5-ordered

space,Yacompact,

T:-ordered

space, and

f

X Y be

acontinuous,increasingmap. Thenthereisaunique continuous, increasingmap

]" "loX

Ysuchthat the diagrambelow commutes.

X

PROOF. Let

]’loX Y

be definedby

](3t) ,

where is defined in Proposition 1.9. We firstshow that

]

isincreasing. Let 31

<

) in

2;

then

DZ(.Y) <_

31.

Suppose y) y inY. Then there is g

CI*(Y)

such that

g(y.)

and

g(y)

0. Thus y;

g-([0, ]) DZ(I(M)),

since

f()/)

y inY. But

g-([i,1]) I(31),

since

I(31)

Y, and therefore

)’(31) DZ(I()I)). However, DZ()I)

<_ 31 implies

DZ(f())) <_ f(DZ()I)) _< )’(31)

follows by Proposition 1.5. This contradiction establishesthaty

_<

y;, andso

]

isincreasing.

Wenextshow that

]

is continuous. Let 31

"oX

and letAbeaCZ-neighbgrhood ofy.u inY. Fromthe factthat

)’(31)

y.,wededuce that31

)’-x("A),

andit is easy toseethat

](f-(A))

C_ A. Itremains to

show that

)’- (A)

isaneighborhoodof31in

"loX.

Forthis purpose,weemploy Proposition

1.6(b)

toobtain

C

DZ(Y)

and D

IZ(Y)

such that y

(Y C)

n

(Y D) _

A. Since

(Y C)

n

(Y D)

itfollowsthat31

() f-x(’C)) () f-X(D)).

The latter set is open in

"7oX

andasubsetof

This establishes that

f-x(’-’A)

isaneighborhoodof31which maps intoA,andtheproofiscomplete.

THEOREM2.7 Let Xbe

Ta.5-ordered.

Then

oX BoX

iff thefollowingconditionshold:

(1)

IfM

IZ(X),

N

CZ(X),

andM N 0,thenthere ish

CI*(X)

such that

h(N)

0and

h(M)

1.

(2)

IfM

DZ(X),

N

CZ(X),

andM N

,

thenthere ish

CD*(X)

such that

h(M)

0and

h(N)

1.

PROOF. Since

foX

is the largest

T2-ordered

compactification of

X,

Theorem2.6 implies that

’7oX

,6oXiff

%X

is

T-ordered.

Thusthe proofwillbe achievedby showing that the specifiedconditions arenecessaryandsufficient inorder for

%X

tobe

T2-ordered.

(6)

122 D.C. KENT AND T.A. RICHMOND

Assume that

"loX

is

Tz-ordered

and let M and N beas indicated in

(1).

By Proposition

2.3(a),

f

,

and and are bothclosedsubsets of

7oX.

Furthermore,

.r

isincreasing in

"fox

by

Proposition

2.3(d).

Let

d()

denote thedecreasinghull of in

7oX.

Then

d()

isclosedbyProposition 4, page44,

[10],

and

d(r)

n

2lr

I/t. ByTheorem 1, page30,

[10],

there is g in

CI*{"IoX)

such that

g(l)

0 if4E

d()

and

g()

ift

E/f.

Settingh gok weobtain

(1).

Asimilarargumentestablishes

(2).

Conversely, assumethe twoconditidns, andlet

t,

beelementsof

"loX

such that

t . M.

Then

either

IZ()

/ or

DZ(/) .

If

IZ()

/,then

(because

/ is amaximal

CZ-filter)

thereis

M

IZ()

andaCZ-set N / such thatM n N

.

If his asstated in

(1),

then

h-l([0, ))

and

h-l((1/2, 1])

aredisjointopenneighborhoods of/and respectively, the former decreasing and the latter increasing. If

DZ(/) ,

wecanapply

(2)

to achieve thesameresult.

1

If

X

has thediscreteorder,conditions

(1)

and

(2)

of Theorem2.7reducetothestatementthat disjoint zero setsin X are "completely separated" inthesenseof

[4].

Sincethis is truefor any

Ts.s

space, we

conclude that

%X

[oX

fix

wheneverXisa

Ts.s-ordered

space withthediscreteorder.

Asweshall seein thenext section, therearesimple examples of

Ts.s-ordered

spacesfor which

7oX

is not

T2-ordered.

In thiscase,wemay beinterestedtoknowwhen

%X

satisfiesthe weaker separation properties

"T2

or

"Tl-ordered’.

This sectionconcludeswith twotheorems pertaining tothisproblem.

Examples showing that

%X

neednotsatisfy these latterseparationaxiomsarealso provided in thenext section.

THEOREM2.8 LetXbea

T3.s-ordered

space. Then

%X

is

T2

iff,for each ultrafilter

Y"

on

X,

there isauniquemaximal CZ-filter onXsuch that

CZ(Y’) <_ .

PROOF. Assume

%X

is

T2

and let

"

beanultrafilteronX. Then

(Y’)

converges tosome %X, where isamaximal CZ-filteronX. Itmust betrue that

CZ(Y’) _< ;

otherwise { and

CZ(Y)

would

containdisjoint CZ-setsMandA, andX-A would be aneighborhood of in

%X

notbelongingto

(Y’).

If therewereanothermaximal CZ-filter/finerthan

CZ(Y’),

then

(Y’)

wouldalso converge to / in%X,contradicting the assumption that

%X

is

T2.

Thus isthe uniquemaximal CZ-filtersuch that

cz() <_ ..

Conversely,assumethat

%X

isnot

T;

thenthereisafilter on

%X

convergingto distinctelements and / in

%X.

Let

Y"

beanultralilteronXcontaining thefilterbase

{A

_C X /

}.

One easily

verifiesthat

(Y’)

converges toboth and in

%X.

Thisimplies,asinthe preceding paragraph, that and /arebothmaximal CZ-filters finerthan

CZ(Yf),

whichcontradictsthe uniquenesscondition.

THEOREM2.9 LetXbea

Ts.5-ordered

space suchthat,foreach A

e CZ(X), i(A) e IZ(X)

and

d(A) DZ(X).

Then

%X

is

T-ordered.

PROOF. ForS C_%X,let

iv(S

denote the increasinghullof S and

clS

theclosure of Sin

%X.

We

willshowthatforarbitrary %X,that

cl(i()) i(),

andhence

i()

isclosed in

%X.

The

dual argumentestablishesthat

d{

isalsoclosed.

First,observe thatif /

cl(i()),

then for each A

CZ(X)

such that /

X- A,

there is

i()

such that

X"-

A. Inotherwords,if

//cl(i(t)),

then foreach A

CZ(X)

such that A

,

thereis /%Xsuch that

<

andA

.

Let

cl(i()).

If

/ i({),

then /,andsoeither

IZ()

/or

DZ(.V) .

Asumethe

former;then thereisM

,

f

IZ(X)

such that

M

/. But /

cl(i({))

implies thereis

such that M

.

However

<

implies

IZ() _< ,

acontradiction. Ontheotherhand, suppose

DZ(/)

{. Since )d isamaximal CZ-filter, there isaCZ-setM { and N /

DZ(X)

such

thatMfN

,

andhence N c

i(M) .

Butby assumption,

i(M) IZ(X),

andso

i(M) IZ().

Again, /E

cl(i(())

implies thereis

>

such that

i(M)

f[

t.

However

i(M) IZ({) <_ I

isagain acontradiction. Wetherefore conclude that

i()

isclosedin

%X.

(7)

NEW ORDERED COMPACTIFICATION

3.

%X

AND

cvoX

The Wallman orderedcompactification

(ooX, o)

ofa

Tl-ordered

spaceXwasintroducedbyChoeand Park

[2]

in1979. Inthis sectionwefind conditionsunderwhich

%X cvoX;

thisleadstoexamples showing that

"oX

canfail, in various ways,topreserve theseparation propertiesT2, T2-ordered,and

T1-ordered.

The construction of

ooX

andadiscussionof itspropertiescan be foundin

[8]. Here,

wereviewonly

afew relevant facts. Although

%X

canbe defined forany

Tl-ordered

space

X,

weshall assume, asin thepreceding section, that XisTa.s-ordered,since it isonly for such spaces that

oX

and

c#oX

canbe compared.

IfAis anynon-empty subsetofX,let

I(A)

denote thesmallestclosed,increasingset containingAand

D(A)

thesmallestclosed,decreasingoversetofA. Ais said tobeac-set ifA

I(A)n D(A).

AspaceX

iscalledac-spaceif, forevery c-setAC_

X, i(A) I(A)

and

d(A) D(A).

AfilteronXwithabase of c-sets iscalled ac-filter. The underlyingsetfor

ooX

isthesetofallmaximal c-filterson X. Indeed,the constructionsof

cvoX

and

7oX

areverysimilar,withthec-setsplaying thesameroleinthe former that the CZ-setsplayinthe latter. Inparticular, if everyc-set in

X

isa

CZ-set,

then

ooX %X.

Thus the

following propositionisobvious.

PROPOSITION3.1 Ifevery increasing closedsetinXis in

IZ(X)

andeverydecreasing closedset inXisin

DZ(X),

then

oX

"/oX.

Another usefulfact,provedin

[6],

isthe following.

PROPOSITION3.2 AspaceXhas the property that

ooX oX

iffXisa

T4-ordered

c-space.

THEOREM3.3 If

X

isa

T4-ordered

spacesuch that,forany sets

F,

Gin

CZ(X), I(F)

G

implies

I(F) D(G)

anddually, then

%X floX.

PROOF. Weshowthat,underthe given assumptions,Xsatisfies conditions

(1)

and

(2)

ofTheorem 2.7. To verify

(1),

let ME

IZ(X)

and N e

C2(A)

be disjoint. Since

I(M) M,

it followsby our assumptionthatM

D(G) .

Thuswe can applyNachbin’sgeneralization ofUrysohn’sLemma

(see

Theorem 1,page30,

[10])

to obtain

e CI*(X)

suchthat

f(M)

1and

/(D(G))

O. This establishes condition

(1);

the proof of

(2)

is similar.

I

COROLLARY3.4 IfXisa

T3.s-ordered

space such that

cooX

=/oX,then

ooX oX.

PROOF. If

ooX oX,

then,by Proposition 3.2,Xisa

T4-ordered c-spa,e.

Everysuchspaceclearly

satisfiesthe requirementsofTheorem 3.3, andsotheconclusionfollows.

1

A

T-ordered

space whose underlyingpartialorder isatotal

(or linear)

order is calledatotallyordered space. Itis shownin

[7]

that

ooX

=/oXforanytotally orderedspaceX.

COROLLARY3.5 IfXisatotally orderedspace, then

ooX %X

[oX.

THEOREM3.6 LetXbeasubspaceofR

PROOF. InviewofProposition 3.1,it is sufficienttoshowthateachclosed,decreasing subsetofX is in

DZ(X)

and eachclosed,increasingsubset ofXis in

IZ(X).

Webegin by defining

(in

the terminology of

[3])

aquasi-pseudo-metricpon

X

defined asbllows: If :r=

(Xl,’--,Xn),

y=

(Yl,’’’,Yn),

then

p(x,y)= (Yl- xl)

V0

+.-.+ (Yn- xn)

V0. IfAisanon-empty, closed, decreasing subset of

X,

wedefine PA X

[0, oo)

as follows:

pA(X)

inf{p(l/,x) y

A}.

Finally, let

hA X

Ebedefinedby

hA

PAA1. Itfollows that

hA CI*(X)

and

hl(0)

A. Thus

A

DZ(X).

The dual argument shows thatanyclosed,increasing subset ofXis in

IZ(X).

ItisshowninTheorem3.4 of

[8]

that

ooR oR

iffn

<_

2;thisyieldsthefollowingconsequence of Theorem3.6.

(8)

124 D.C. KENT AND T.A. RICHMOND

COROLLARY3.7

%R" oR"

iffn

_<

2.

Werecalltwoexamples from

[8]

involving subspacesofR in which

ooX,

and hence also%X,fail to exhibit basicseparation properties. LetS

{(z,I/)

-1

_<

z

_<

1,-1

_<

y

_< 1}

beasubspaceofR

2.

In

Example3.6of

[8],

thesubspace

Xl

S

{(0, 0))

of

R

has the property that

%Xl

isneither

T-ordered

nor

T2.

InExample3.7of

[8],

thesubspace

X

S

((0, I/)

-1

_<

I/-< 1and t/

0)

has the property that

%X2

is

T

butnot

T-ordered.

Wedonotknow ofaspaceXfor which

%X

isT]-ordered butnot

T.

Asafinalexample, recall thatifXisa

Ts.s-ordered

space withthe discreteorder,then

%X

[$oX. If,in ddition,

X

ischosennottobeT4,then

ooX (which

in thiscaseisthe ordinary Wallman

compactification)

fails tobeT2,and consequently

ooX %X.

4. UNSOLVEDPROBLEMS.

Find necessaryandsufficient conditionson aspaceXfor

%X

tobe

T-ordered.

(2)

Find conditionsona spaceXwhicharenecessary and sufficient for

%X woX.

(3)

Determinewhether

%R

sis

T.

(4)

Finda

Ts.s-ordered

spaceXforwhich

ooX, oX,

and

%X

aremutuallynon-equivalent.

,(5)

Determinewhether/$oXcanberepresentedas aWallman-typeordered compactification.

REFERENCES

[1]

J.Blatter, "Order Compactification of Totally OrderedSpaces," J. Approz. Theory13

(1975)

56-65.

[2]

T. H. Choe and S. Park, Wallman’s

Type

Order Compact’fication,"

Pacific

J. Math. 82

(1979)

339-347.

[3]

P. Fletcher and W. Lindgren, Q,

asi-Uniform

Spaces, Lecture NotesinPureand Applied Mathemat- ics, Vol. 77,MarcelDekker,

Inc.,

NewYork

(1982).

[4]

L.GillmanandM.Jerison, Rings

of

ContinaoesFnctions,Van Nostrand,Princeton

(1960).

[5]

G.Hommel,"IncreasingRadonMeasuresonLocally Compact OrderedSpaces," Rendiconti Mathe- matica9

(1976)

85-117.

[6]

D.C.

Kent,

Onthe Wallman OrderCompactification,"

Pacific

J. Math. 118

(1985)

159-163.

[7]

D. Kent and T. Richmond, "Ordered Compactlflcation ofTotally Ordered Spaces," lnternat. J.

Math. l Math. $ci. 11

(1988)683-694.

[8]

,"Separation Properties of the Wallman Ordered Compactification," lnternat. J.

Math. Math. Sci. 13

(1990)

209-222.

[9]

T. McCallion, CompactificationsofOrdered Topological

Spaces,"

Proc. Camb. Phil. Soc. 71

(1972)

463-473.

[I0]

L. Nachbin, TopologyandOrder,VanNostrand, NewYorkMath. Studies, 4, Princeton, N.J.

(1905).

参照

関連したドキュメント

´ Ciri´c, “Coupled fixed point theorems for nonlinear contractions in partially ordered metric spaces,” Nonlinear Analysis: Theory, Methods &amp; Applications. Lee, “On ´

For a poset to be a continuous domain, it needs to have sups of only directed sets in addition to the following two conditions: i each element is the sup of elements way below it,

Aydi, “Fixed point theorems for generalized weakly contractive condition in ordered partial metric spaces,” Journal of Nonlinear Analysis and Optimization: Theory and

This finding has relevance to the behavior of the total negation operation, defined for topological spaces by Bankston [1], when it is applied to partially-ordered topological

In this paper, we consider the concept of Ω-distance on a complete, partially ordered G-metric space and prove a fixed point theorem for (ψ, φ)-Weak contraction.. Then, we present

In this paper, we give three main theorems which are new generalizations of Banach fixed point theorem, Kannan fixed point theorem and Chatterjea fixed point theorem in the context

In this paper, we extend some recent fixed point theorems by using this concept and prove various fixed point theorems in generalized partially ordered G-metric spaces.. At first

Therefore, we give a common fixed point theorem for two weakly increasing operators in partially ordered metric spaces using implicit relation technique.. Also we did not use