A_Brief_Introduction_to_Measure_Theory_and_Integration-Richard_Bass.pdf
(
194 KB
)
Pobierz
390103101 UNPDF
ABriefIntroductionto
MeasureTheoryandIntegration
RichardF.Bass
DepartmentofMathematics
UniversityofConnecticut
September18,1998
Thesenotesarec1998byRichardBass.Theymaybeusedforpersonaluseorclassuse,butnotfor
commercialpurposes.
1.Measures.
LetXbeaset.Wewillusethenotation:A
c
={x2X:x/2A}andA−B=A\B
c
.
Definition.AnalgebraorafieldisacollectionAofsubsetsofXsuchthat
(a);,X2A;
(b)ifA2A,thenA
c
2A;
(c)ifA
1
,...,A
n
2A,then[
n
i=1
A
i
and\
n
i=1
A
i
areinA.
Aisa-algebraor-fieldifinaddition
(d)ifA
1
,A
2
,...areinA,then[
1
i=1
A
i
and\
1
i=1
A
i
areinA.
In(d)weallowcountableunionsandintersectionsonly;wedonotallowuncountableunionsandintersections.
Example.LetX=RandAbethecollectionofallsubsetsofR.
Example.LetX=RandletA={AR:AiscountableorA
c
iscountable}.
Definition.Ameasureon(X,A)isafunctionµ:A![0,1]suchthat
(a)µ(A)0forallA2A;
(b)µ(;)=0;
(c)ifA
i
2Aaredisjoint,then
µ([
1
i=1
A
i
)=
1
X
µ(A
i
).
i=1
Example.Xisanyset,Aisthecollectionofallsubsets,andµ(A)isthenumberofelementsinA.
Example.X=R,Athecollectionofallsubsets,x
1
,x
2
,...2R,a
1
,a
2
,...>0,andµ(A)=
P
{i:x
i
2A}
a
i
.
Example.
x
(A)=1ifx2Aand0otherwise.Thismeasureiscalledpointmassatx.
Proposition1.1.Thefollowinghold:
(a)IfA,B2AwithAB,thenµ(A)µ(B).
(b)IfA
i
2AandA=[
1
i=1
A
i
,thenµ(A)
P
1
i=1
µ(A
i
).
(c)IfA
i
2A,A
1
A
2
···,andA=[
1
i=1
A
i
,thenµ(A)=lim
n!1
µ(A
n
).
(d)IfA
i
2A,A
1
A
2
···,µ(A
1
)<1,andA=\
1
i=1
A
i
,thenwehaveµ(A)=lim
n!1
µ(A
n
).
Proof.(a)LetA
1
=A,A
2
=B−A,andA
3
=A
4
=···=;.Nowusepart(c)ofthedefinitionofmeasure.
1
(b)LetB
1
=A
1
,B
2
=A
2
−B
1
,B
3
=A
3
−(B
1
[B
2
),andsoon.TheB
i
aredisjointand
[
1
i=1
B
i
=[
1
i=1
A
i
.Soµ(A)=
P
µ(B
i
)
P
µ(A
i
).
(c)DefinetheB
i
asin(b).Since[
n
i=1
B
i
=[
n
i=1
A
i
,then
µ(A)=µ([
1
i=1
A
i
)=µ([
1
i=1
B
i
)=
1
X
µ(B
i
)
i=1
=lim
n!1
n
X
n!1
µ([
n
i=1
B
i
)=lim
n!1
µ([
n
i=1
A
i
).
i=1
(d)Apply(c)tothesetsA
1
−A
i
,i=1,2,....
Definition.Aprobabilityorprobabilitymeasureisameasuresuchthatµ(X)=1.Inthiscaseweusually
write(,F,P)insteadof(X,A,µ).
2.ConstructionofLebesguemeasure.
Definem((a,b))=b−a.IfGisanopensetandGR,thenG=[
1
i=1
(a
i
,b
i
)withtheintervals
m
(A)=inf{m(G):Gopen,AG}.
Wewillshowthefollowing.
(1)m
isnotameasureonthecollectionofallsubsetsofR.
(2)m
isameasureonthe-algebraconsistingofwhatareknownasm
-measurablesets.
(3)LetA
0
bethealgebra(not-algebra)consistingofallfiniteunionsofsetsoftheform[a
i
,b
i
).IfAis
thesmallest-algebracontainingA
0
,thenm
isameasureon(R,A).
Wewillprovethesethreefacts(andabitmore)inamoment,butlet’sfirstmakesomeremarksabout
theconsequencesof(1)-(3).
Ifyoutakeanycollectionof-algebrasandtaketheirintersection,itiseasytoseethatthiswillagain
bea-algebra.Thesmallest-algebracontainingA
0
willbetheintersectionofall-algebrascontaining
A
0
.
Since(a,b]isinA
0
forallaandb,then(a,b)=[
1
i=i
0
(a,b−1/i]2A,wherewechoosei
0
sothat
1/i
0
<b−a.Thensetsoftheform[
1
i=1
(a
i
,b
i
)willbeinA,henceallopensets.Thereforeallclosedsets
areinAaswell.
Thesmallest-algebracontainingtheopensetsiscalledtheBorel-algebra.ItisoftenwrittenB.
AsetNisanullsetifm
(N)=0.LetLbethesmallest-algebracontainingBandallthenullsets.
LiscalledtheLebesgue-algebra,andsetsinLarecalledLebesguemeasurable.
Aspartofourproofsof(2)and(3)wewillshowthatm
isameasureonL.Lebesguemeasureis
themeasurem
onL.(1)showsthatLisstrictlysmallerthanthecollectionofallsubsetsofR.
Proofof(1).Definexyifx−yisrational.Thisisanequivalencerelationshipon[0,1].Foreach
equivalenceclass,pickanelementoutofthatclass(bytheaxiomofchoice)Callthecollectionofsuchpoints
A.GivenasetB,defineB+x={y+x:y2B}.Notem
(A+q)=m
(A)sincethistranslationinvariance
holdsforintervals,henceforopensets,henceforallsets.Moreover,thesetsA+qaredisjointfordierent
rationalsq.
2
µ(B
i
)=lim
disjoint.Definem(G)=
P
1
i=1
(b
i
−a
i
).IfAR,define
Now
[0,1][
q2[−2,2]
(A+q),
wherethesumisonlyoverrationalq,so1
P
q2[−2,2]
m
(A+q),andthereforem
(A)>0.But
[
q2[−2,2]
(A+q)[−6,6],
whereagainthesumisonlyoverrationalq,so12
P
q2[−2,2]
m
(A+q),whichimpliesm
(A)=0,a
contradiction.
Proposition2.1.Thefollowinghold:
(a)m
(;)=0;
(b)ifAB,thenm
(A)m
(B);
(c)m
([
1
i=1
A
i
)
P
1
i=1
m
(A
i
).
Proof.(a)and(b)areobvious.Toprove(c),let">0.ForeachithereexistintervalsI
i1
,I
i2
,...suchthat
A
i
[
1
j=1
I
ij
and
P
j
m(I
ij
)m
(A
i
)+"/2
i
.Then[
1
i=1
A
i
[
i,j
I
ij
and
X
m(I
ij
)
X
i
m
(A
i
)+
X
i
"/2
i
=
X
i
m
(A
i
)+".
i,j
Since"isarbitrary,m
([
1
i=1
A
i
)
P
1
i=1
m
(A
i
).
Afunctiononthecollectionofallsubsetssatisfying(a),(b),and(c)iscalledanoutermeasure.
Definition.Letm
beanoutermeasure.AsetAXism
-measurableif
m
(E)=m
(E\A)+m
(E\A
c
) (2.1)
forallEX.
Theorem2.2.Ifm
isanoutermeasureonX,thenthecollectionAofm
measurablesetsisa-algebra
andtherestrictionofm
toAisameasure.Moreover,Acontainsallthenullsets.
Proof.ByProposition2.1(c),
m
(E)m
(E\A)+m
(E\A
c
)
forallEX.Sotocheck(2.1)itisenoughtoshowm
(E)m
(E\A)+m
(E\A
c
).Thiswillbetrivial
inthecasem
(E)=1.
IfA2A,thenA
c
2AbysymmetryandthedefinitionofA.SupposeA,B2AandEX.Then
m
(E)=m
(E\A)+m
(E\A
c
)
=(m
(E\A\B)+m
(E\A\B
c
))+(m
(E\A
c
\B)+m
(E\A
c
\B
c
)
Thefirstthreetermsontherighthaveasumgreaterthanorequaltom
(E\(A[B))becauseA[B
(A\B)[(A\B
c
)[(A
c
\B).Therefore
m
(E)m
(E\(A[B))+m
(E\(A[B)
c
),
whichshowsA[B2A.ThereforeAisanalgebra.
3
LetA
i
bedisjointsetsinA,letB
n
=[
n
i=1
A
i
,andB=[
1
i=1
A
i
.IfEX,
m
(E\B
n
)=m
(E\B
n
\A
n
)+m
(E\B
n
\A
c
n
)
=m
(E\A
n
)+m
(E\B
n−1
).
Repeatingform
(E\B
n−1
),weobtain
m
(E\B
n
)=
n
X
m
(E\A
i
).
i=1
So
n
X
m
(E)=m
(E\B
n
)+m
(E\B
c
n
)
m
(E\A
i
)+m
(E\B
c
).
i=1
Letn!1.Then
X
m
(E)
m
(E\A
i
)+m
(E\B
c
)
i=1
m
([
1
i=1
(E\A
i
))+m
(E\B
c
)
=m
(E\B)+m(E\B
c
)
m
(E).
ThisshowsB2A.
IfwesetE=Binthislastequation,weobtain
m
(B)=
1
X
m
(A
i
),
i=1
orm
iscountablyadditiveonA.
Ifm
(A)=0andEX,then
m
(E\A)+m
(E\A
c
)=m
(E\A
c
)m
(E),
whichshowsAcontainsallnullsets.
NoneofthisisusefulifAdoesnotcontaintheintervals.Therearetwomainstepsinshowingthis.
LetA
0
bethealgebraconsistingofallfiniteunionsofintervalsoftheform(a,b].Thefirststepis
Proposition2.3.IfA
i
2A
0
aredisjointand[
1
i=1
A
i
2A
0
,thenwehavem([
1
i=1
A
i
)=
P
1
i=1
m(A
i
).
Proof.Since[
1
i=1
A
i
isafiniteunionofintervals(a
k
,b
k
],wemaylookatA
i
\(a
k
,b
k
]foreachk.Sowe
mayassumethatA=[
1
i=1
A
i
=(a,b].
First,
n
X
m(A)=m([
n
i=1
A
i
)+m(A−[
n
i=1
A
i
)m([
n
i=1
A
i
)=
m(A
i
).
i=1
Lettingn!1,
X
m(A)
m(A
i
).
i=1
Letusassumeaandbarefinite,theothercasebeingsimilar.Bylinearity,wemayassumeA
i
=
(a
i
,b
i
].Let">0.Thecollection{(a
i
,b
i
+"/2
i
)}covers[a+",b],andsothereexistsafinitesubcover.
4
1
1
Discardinganyintervalcontainedinanotherone,andrelabeling,wemayassumea
1
<a
2
<···a
N
and
b
i
+"/2
i
2(a
i+1
,b
i+1
+"/2
i+1
).Then
m(A)=b−a=b−(a+")+"
N
X
(b
i
+"/2
i
−a
i
)+"
i=1
1
X
m(A
i
)+2".
i=1
Since"isarbitrary,m(A)
P
1
i=1
m(A
i
).
ThesecondstepistheCarath´eodoryextensiontheorem.Wesaythatameasuremis-finiteifthere
existE
1
,E
2
,...,suchthatm(E
i
)<1foralliandX[
1
i=1
E
i
.
Theorem2.4.SupposeA
0
isanalgebraandmrestrictedtoA
0
isameasure.Define
n
1
X
o
m
(E)=inf
m(A
i
):A
i
2A
0
,E[
1
i=1
A
i
.
i=1
Then
(a)m
(A)=m(A)ifA2A
0
;
(b)everysetinA
0
ism
-measurable;
(c)ifmis-finite,thenthereisauniqueextensiontothesmallest-fieldcontainingA
0
.
Proof.Westartwith(a).SupposeE2A
0
.Weknowm
(E)m(E)sincewecantakeA
1
=Eand
A
2
,A
3
,...emptyinthedefinitionofm
.IfE[
1
i=1
A
i
withA
i
2A
0
,letB
n
=E\(A
n
−[
n−1
i=1
A
i
).The
theB
n
aredisjoint,theyareeachinA
0
,andtheirunionisE.Therefore
m(E)=
1
X
m(B
i
)
1
X
m(A
i
).
i=1
i=1
Thusm(E)m
(E).
Nextwelookat(b).SupposeA2A
0
.Let">0andletEX.PickB
i
2A
0
suchthatE[
1
i=1
B
i
m
(E)+"
1
X
1
X
1
X
m(B
i
\A
c
)
m(B
i
)=
m(B
i
\A)+
i=1
i=1
i=1
m
(E\A)+m
(E\A
c
).
Since"isarbitrary,m
(E)m
(E\A)+m
(E\A
c
).SoAism
-measurable.
Finally,supposewehavetwoextensionstothesmallest-fieldcontainingA
0
;lettheotherextension
becalledn.WewillshowthatifEisinthissmallest-field,thenm
(E)=n(E).
SinceEmustbem
-measurable,m
(E)=inf{
P
1
i=1
m(A
i
):E[
1
i=1
A
i
,A
i
2A
0
}.Butm=non
Let">0andchooseA
i
2A
0
suchthatm
(E)+"
P
i
m(A
i
)andE[
i
A
i
.LetA=[
i
A
i
and
B
k
=[
k
i=1
A
i
.Observem
(E)+"m
(A),hencem
(A−E)<".Wehave
m
(A)=lim
k!1
m
(B
k
)=lim
k!1
n(B
k
)=n(A).
5
and
P
i
m(B
i
)m
(E)+".Then
A
0
,so
P
i
m(A
i
)=
P
i
n(A
i
).Thereforen(E)
P
i
n(A
i
),whichimpliesn(E)m
(E).
Plik z chomika:
Kuya
Inne pliki z tego folderu:
A_Brief_Introduction_to_Measure_Theory_and_Integration-Richard_Bass.pdf
(194 KB)
Distributions-Andras_Vasy.pdf
(76 KB)
Distributions_Generalized_Functions-Ivan_Wilde.pdf
(430 KB)
Introduction_to_Generalized_Functions_with_Applications_in_Aerodynamics_and_Aeroacoustics-Farasat.pdf
(2310 KB)
Introduction_to_Geometric_Measure_Theory-Urs_Lang.pdf
(297 KB)
Inne foldery tego chomika:
algebra
analysis
calculus
complex
computation
Zgłoś jeśli
naruszono regulamin