Scribd
Upload
30 views7 pages

Proofs 20 1 Print

G

Uploaded by

Dilfuza Uzaqbaeva
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
30 views7 pages

Proofs 20 1 Print

G

Uploaded by

Dilfuza Uzaqbaeva
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 7
Chapter 20. The Construction of Particular Measures 20.1. Product Measures: Fubini and Tonelli Theorems—Proofs RE ANALYSIS Lemma 20.1. Let {Ay x By}{ be a countable disjoint collection of measurable rectangles whose union also is a measurable rectangle A x B. Notice that index k ranges over ALL of the rectangles which compose Ax B (so there is no Aj x Byj where i # j). Then H(A) -v(B) = 7 H(Ak) (Bx). ia Proof. Fix a point x € A. For each y € B, the point (x,y) € Ax B and since {Aj x B,}22 is a disjoint collection, then (x,y) is in exactly one Ax x By. So we can write B as the following disjoint union: B= Uralxca,} Bx (here x is a fixed element of A). By the countable additivity of measure v, 1(B) = Syajeca,) ¥(Ba) (here x is a fixed element of A). Proof (continued). A Equivalently, we have v(B) = D, f U((Ak x 81) du(s)) by the Monotone tat Convergence Theorem; the partial sums are increasing = So Anel6) by part (1), ma since Ay x By is a measurable rectangle = («x v)(E) by the definition of ps x v So the result holds for E € Re. Proof (continued). (3) Suppose E is in Rys and (11 x v)(E) < 00. Since R is a semiring (closed under finite intersections), there is a descending sequence {E,}{®; of sets in R, whose intersection is E. Since (11 v)(E) < 00, without loss of generality (j1 x v)(E) < oo (from the definition of j x v in terms of the outer measure induced by the premeasure : x v on R). By the continuity of measure x (Proposition 17.2), im (ux v) Eq) = (x v)(E) (3) Since Ey € Rz and the result holds for R., sets, (1x v)(Ex) = fy ¥((Ev)x) dpu(x), and since (12 x v)(E) < 90 and v((E1)x) is nonnegative, then v((E1)x) < 00 for almost all x € X by Proposition 18.9. For each x € X, Ey is the intersection of the descending sequence {(Ex)x}{21, and so Ey is v-measurable (B is a o-algebra). So by continuity of measure v, for almost all x € E (the x for which v((Ex)x) < 00) we have limy sac “((Ex)x) — (Ex) (Continuity of Measure for descending sequences requires finite measure; see Proposition 17.2) Proof (continued). Furthermore, the function x > v((E;)x) is nonnegative and integrable (since (y x v)(E1) < 00) and for each k € N, dominates a.e. the function x ++ »((Ej)x) since the Ex form a descending sequence. So we have [Xen ais) = [ (im AEx)x) d(x) Bi I aime =I Cl (Eis) 4s) by the Lebesgue Kim, Jy Dominated Convergence Theorem slim, (ex v)(Ex) since the result holds on Re sets Ex = (ux v\(E) from (3) So the result holds for R,,, sets. a Lemma 20.4. Assume the measure v is complete, LetE < X x ¥ be measurable with respect to yi x v. If (1 x v)(E) = 0, then almost al x EX, the x-section of E, Ex, is v-measurable and (Ex) = 0. Therefore 0=(ux(E)= [60 dy(x) =0. Proof. Since (ji x v)(E) < oo it follows from Proposition 17.10 that there is a set A € Rag for which EC Aand (ux v)(A) = (ux v)(E) = 0. Since A is Res, by Lemma 20.3 we have that for all x € X that the xcsection of A, Ax, is v-measurable and (u x v)(A) = fy v(Ax) duu(x). So the integral is 0 for almost all x € X by Problem 18.19. However, for all x © X we have Ey C Ay. By the completeness of v, v(E,) =0 and so Ey is v-measurable. So I (Ex) dyu(x) = 0 = (10 x v)(E) x Proof. Since (j1 x v)(E\ A) = 0, then by Lemma 20.4 (the completeness of v is used here) for almost all x < X, (\E)x is v-measurable and U((A\ E)x) = 0. So »(Ax) = v(Ex) for almost all x € X. So (ux rE) = (ux v)(A) by above f (Az) d(x) by Lemma 20.3 since AE Res Ix = [ (Ex) dyi(x) since v(Ax) = v( Ex) a.e. on X. F a Proposition 20.5. Assume the measure v is complete. Let EX x Y be measurable with respect to yx x v and (2 x v)(E) < oc. The for almost all xX, the x-section of E, Ex, is a v-measurable subset of Y, the function x + v(E,) is semeasurable for all x € X, and (wx y(e) = [ v(E.) aus Proof. Since (jc x v)(E) < oo it follows from Proposition 17.10 that there is a set A € Rag for which Ec Q and (ux v)(A\ E) =0. By the excision property of measure jx x v (Proposition 17.1), we have (ux v)(E) = (ux v)(A). Since Ae Res then by Lemma 20.3, Ay is a v-measurable function. So by the finite additivity of v (Proposition 17.1) (Ax) = v(x (A\ E)x) = o(Ex) + ((A\ E)x), Theorem 20.6. Assume measure v is complete. Let ¢:X x Y > Rbea simple function that is integrable over X x Y with respect to js x v. Then for almost all x € X, the x-section of #, 4(x,-), is integrable over Y with respect to v and [oaiexn= f [[ eorraovr] date Proof. First, if xe is a characteristic function on a subset E of X x Y of finite measure (to get integrability), then fo yxedies hexy (ux v)(E) where y = 1 on E, by the def ion of integral of a characteristic function; page 366 f V((xe)x) du(x) by Proposition 20.5 x Fubini’s Theorem. Let (X,A, 1) and (Y, B,v) be measure spaces and let v be complete, Let Proof (continued). F be integrable over X x Y with respect to the product measure pu xv. Then for almost all x € X, the x-section of f, F(x, )(y). is integrable over fo predexn = [ (f etonavon) aon Y with respect to and heey x \Iy 1 ifyeb I rauxr)= [Lf Ae a | ace see (xe xe =000={ 5 yee, anes [Fauxnd= [| [ evar} duce H((xe)x) = fy ey) dv(y). So the result holds for characteristic functions Proof. Since integration is linear (Theorem 18.12), we assume f is nonnegative (otherwise, we break f into f* and f- and consider these Second, for general simple and integrable y, y is a linear combination of parts individually). By the Simple Approximation Theorem there is an characteristic functions and this result then follows by the linearity of increasing sequence {is} of simple functions that converges pointwise on integration (Theorem 18.12) as applied to the integral with respect to X x Y tof and 0< yu (the integral of f(-, y) over Y with respect to 41) Theorem implies that the x-section of f is v-measurable for almost is jemeasurable, x EX and x fy f(x,y) du(y) is p-measurable Proof (continued). Also, Srey f F(x v) = fy Sy fay) du(y)) du(x), and so F is integrable over X x Y with respect to 4x x by (10). Now applying Fubini's Theorem, since f is integrable over X x Y with respect to jr x v and since 11s complete we have f,faxer= [Cf fosrrenb9 don)

You might also like