Measure Theory/Riesz' representation theorem

From testwiki
Jump to navigation Jump to search

Theorem (Riesz' representation theorem)

Let X be a locally compact Hausdorff space and let Λ be a positive linear functional on Cc(X). Then, there exists a σ-field Σ containing all Borel sets of X and a unique measure μ such that

  1. Λf=Xfdμ for all fCc(X)
  2. μ(K)< for all compact KΣ
  3. If EΣ, and μ(E)< then μ(E)=inf{μ(V)|EV,V open}
  4. If EΣ, and μ(E)< then μ(E)=sup{μ(K)|EK,K compact}
  5. The measure space (X,Σ,μ) is complete

Proof

Recall the Urysohn's lemma:

If X is a locally compact Hausdorff space and if V is open and K is compact with KV then

there exists f:X[0,1] with fCc(X) satisfying f(K)={1} and suppfV. This is written in

short as KfV

We shall first prove that if such a measure exists, then it is unique. Suppose μ1,μ2 are measures that satisfy (1) through (5)

It suffices to show that μ1(K)=μ2(K) for every compact K

Let K be compact and let ϵ>0 be given.


By (3), there exists open VX with VK such that μ1(V)<μ1(K)+ϵ

Urysohn's lemma implies that there exists fCc(X) such that KfV

(1) implies that μ2(K)Xfdμ2=Λf=Xfdμ1. But μ1(V)<μ1(K)+ϵ, that is μ2(K)<μ1(K)+ϵ. We can similarly show that μ1(K)<μ2(K)+ϵ. Thus, μ2(K)=μ1(K)

Suppose V is open in X, define μ(V)=sup{Λf|fV}

If V1V2 are open , then μ(V1)μ(V2)


If E is a subset of X then define μ(E)=inf{μ(V)|EV,V open}

Define ΣF={EX|μ(E)<,μ(E)=sup{μ(K):KE,K compact}}

Let Σ={EX|EKΣF for all K compact in X}

monotonicity of μ is obvious for all subsets of X


Let EX with μ(E)=0

It is obvious that EΣF which implies that EΣ. Hence, we have that {X,Σ,μ} is complete.

Step 1

Suppose {Ei}i=1 is a sequence of subsets of X. Then, μ(Ei)i=1μ(Ei)

Proof

Let V1,V2 be open subsets of X. We wish to show that μ(V1V2)μ(V1)+μ(V2).

Given ϵ>0, let gCc(X) be such that g(V1V2) (so that supp g=K(V1V2)) and μ(V1V2)ϵΛg. This is possible because μ(V1V2)=sup{Λg|gV1V2}.

Now by Urysohn's lemma we can find hi, i=1,2, such that hiVi and h1+h2=1 on K with hiCc(X).

Thus, higVi and h1g+h2g=g on K.


As Λ is a linear functional ΛfΛg for all fg,

Λg=Λh1g+Λh2gμ(V1)+μ(V2).


Thus, μ(V1V2)μ(V1)+μ(V2)+ϵ for every ϵ>0, i.e. μ(V1V2)μ(V1)+μ(V2).


If {Ei}i=1 is a sequence of members of Σ, there exist open Vi such that given ϵ>0,


μ(Vi)<μ(Ei)+ϵ2i. Define E=EiVi=V, V is open. Let fV. Then μ(V)μ(E) but μ(V)<Λf


Thus, μ(V)μ(Vi)μ(Ei)+ϵ.

Step 2

If K is compact, then KΣF and μ(K)=inf{Λf:Kf}


Proof

It suffices to show that μ(K)< for every compact K

Let 0<α<1 and Kf, define Vα={x:f(x)>α} Then Vα is open and KVα

Then, by Urysohn's lemma, there exists gCc(X) such that KgVα, and hence, αgf on Vα

Since Λgμ(K)[1][2], we have μ(K)1αΛf

As Λf<, we have μ(K)< and hence, KΣF


Let ϵ>0 be

By definition, there exists open VK such that μ(K)>μ(V)ϵ


By Urysohn's lemma, there exists f such that KfV, which implies that μ(V)Λf, that is

μ(K)>Λf+ϵ.

Hence, μ(K)=inf{Λf:Kf}

Step 3

Every open set V satisfies

μ(V)=sup{μ(K):KV,K compact}

If V is open and VX, μ(X)<, then VΣF


Proof

Let V be open. Let α>0 such that α<μ(V). It suffices to show that there exists compact K such that μ(K)>α.

By definition of μ, there exists fCc(X) such that fV and Λf>α

Let K=suppf. Obviously KV.

Let W be open such that KW, then FW and hence Λfμ(W), further Λfμ(K)


Thus, μ(K)>α

Step 4

Suppose μ(Ei) is a sequence of pairwise disjoint sets in ΣF and let E=i=1Ei. Then, μ(E)=i=1μ(Ei)

Proof

If μ(E)=, by step 1, we are done.

If μ(E) is finite then, EΣF and hence, μ is countably additive on ΣF


Suppose, K1,K2 are compact and disjoint then K1,K2ΣF; K1K2ΣF


Claim: μ(K1K2)=μ(K1)+μ(K2)


As X is a locally compact Hausdorff space, there exist disjoint open sets V1, V2 with K1V1, K2V2

Hence, by Urysohn's lemma, there exists gCc(X) such that gK1K2 and Λgμ(K1K2)+ϵ

Now, suppfg=K and Kfg, K(1f)g


Thus, μ(K1)+μ(K2)<Λgμ(K1+K2)+ϵ

Assume μ(E)<. Given EiΣF, there exists compact HiEi such that μ(Hi)>μ(Ei)ϵ2i

Let KN=H1H2HN. Obviously, KN is compact.

Thus, μ(E)μ(KN)=i=1Nμ(Hi)i=1Nμ(Ei)ϵ

and hence, μ(E)i=1μ(Ei) By step 1, we have μ(E)i=1μ(Ei).

Thus, μ(E)=i=1μ(Ei)

Step 5

If EΣF and ϵ>0 then there exists K compact and V open with KEV and μ(VK)<ϵ


Proof

(VK) is open. As EΣF, there exist compact K and open V such that KEV with

μ(V)ϵ2<μ(E)<μ(K)+ϵ2

Now, μ(VK)<μ(V)<μ(E)+ϵ2< (by step 4)

Thus, μ(VK)<ϵ

Step 6

ΣF is a field of subsets of X

Proof

Let A,BΣF and let ϵ>0 be given.

There exist compact K1,K2 and open V1,V2 such that K1AV1, K2AV2 with

μ(V1K1),μ(V2K2)<ϵ


Write AB=(V1K1)(K1V2)(V2K2)

As K1V2 is a closed subsetof K , it is compact


Then, μ(AB)μ(V1K1)μ(K1V2)μ(V2K2)=μ(K1V2)+2ϵ

thus, μ(AB is finite and hence, ABΣF

Now write AB=(AB)B

and AB=A(AB)

Step 7

Σ is a σ-field containing all Borel sets


Proof

Let C be closed

Then, CK is compact for every K compact

Therefore CKΣF and hence CΣ (by definition) and hence, Σ has all closed sets. In particular, XΣ


Let AΣ. Then, AcKK and AcK=K(KA) and hence, AcΣ

Now let A=i=1Ai where AiΣ

We know that AiKΣF for every compact K

Let B1=A1K, Bn=AnK(B1B2Bn1). AK=i=1, but μ(Bi)< and hence, AKΣ

Step 8

ΣF={EΣ:μ(E)<}

Proof

Let EΣF. Then EKΣF for every compact KX

Now, let EΣ, μ(E)<. Given ϵ>0, there exists open V such that EV, μ(V)<, that is, VΣF. Further, there exists compact KV such that μ(VK)<


EΣ implies that EKΣF, that is, there exists compact HEK such that

μ(EK)<μ(H)+ϵ


Therefore, E(EK)(VK) implies that μ(E)μ(H)+2ϵ

As ϵ>0 is arbitrary, we are done.

Step 9

For fCc(X), Λf=Xfdμ


Proof

Without loss of generality, we may assume that f is real valued.

It is obvious from the definition of μ that ΛfXfdμ

Let K=suppf. Hence, as f is continuous, f(K) is compact. and we can write f(K)[a,b] for some a,b. Let ϵ>0. Let 𝒫={a=y0<y1<<yn=b} be an ϵ-fine partition of [a,b]

Let Ei={xX|yi1<f(x)<yi}K. As f is continuous, K is compact, Ei is measurable for every i, and hence, K=i=1nEi

μ(E)=inf{μ(V)|VE,V open}

Hence, we can find open sets ViEi such that μ(Vi)<μ(Ei)+ϵn

f(x)<yi+ϵ for all xVi


We know that if compact KV1V2Vn with Vi open then there exists hiCc(X) wiht hiVi and i=1nhi=1 on K

Hence, there exist functions hiVi such that i=1nhi=1 on K.

Thus, i=1nhi(x)f(x)=f(x) for all xX


By step 2, we have μ(K)i=1nΛhi

hif<(yi+ϵ)hi on each Vi

Thus, Λf=i=1nΛhifi=1nΛhi(yi+ϵ)=(i=1n(yi+ϵ)Λhi+Λi=1n|a|hi)Λi=1n|a|hi

=i=1n(yi+ϵ+|a|)ΛhiΛi=1n|a|hi

i=1n(yi+ϵ+|a|)(μ(Ei)+ϵn)Λi=1n|a|hi

=i=1nyiμ(Ei)+i=1nϵ(μ(Ei)+ϵn+|a|n)i=1n(yiϵ)μ(Ei)+i=1nϵ(2μ(Ei)+ϵn+|a|n)

i=1nf(xi)μ(Ei)+i=1nϵ(2μ(Ei)+ϵn+|a|n)


As ϵ>0 is arbitrary, we have ΛfXfdμ which completes the proof.


Template:BookCat