6.9. EXERCISES 157

Cantor set of Problem 24. Let [0,1] \P = ∪∞j=1I j where I j is open and I j ∩ Ik = /0

if j ̸= k. These intervals are the connected components of the complement of theCantor set. Show m(g(I j)) = m(I j) so

m(g(∪∞j=1I j)) =

∑j=1

m(g(I j)) =∞

∑j=1

m(I j) = 1.

Thus m(g(P)) = 1 because g([0,1]) = [0,2]. By Problem 23 there exists a set,A⊆ g(P) which is non measurable. Define φ(x) = XA(g(x)). Thus φ(x) = 0 unlessx ∈ P. Tell why φ is measurable. (Recall m(P) = 0 and Lebesgue measure is com-plete.) Now show that XA(y) = φ(g−1(y)) for y ∈ [0,2]. Tell why g−1 is continuousbut φ ◦ g−1 is not measurable. (This is an example of measurable ◦ continuous ̸=measurable.) Show there exist Lebesgue measurable sets which are not Borel mea-surable. Hint: The function, φ is Lebesgue measurable. Now recall that Borel ◦measurable = measurable.

27. For x ∈ Rp to be in ∏pi=1 Ai, it means that the ith component of x, xi is in Ai for each

i. Now for ∏pi=1 (ai,bi)≡ R, let V (R) = ∏

pi=1 (bi−ai) . Next, for A ∈P (Rp) let

µ (A)≡ inf

{∑k

V(

Rk)

: A⊆ ∪kRk

}This is just like one dimensional Lebesgue measure except that instead of open in-tervals, we are using open boxes Rk. Show the following.

(a) µ is an outer measure.(b) µ

(∏

pi=1 [ai,bi]

)= ∏

pi=1 (bi−ai) = µ

(∏

pi=1 (ai,bi)

).

(c) If dist(A,B)> 0, then µ (A)+µ (B) = µ (A∪B) so B (Rp)⊆F the set of setsmeasurable with respect to this outer measure µ .

This is Lebesgue measure on Rp. Hint: Suppose for some j,b j−a j < ε. Show thatµ(∏

pi=1 (ai,bi)

)≤ ε ∏i ̸= j (bi−ai). Now use this to show that if you have a covering

by finitely many open boxes, such that the sum of their volumes is less than somenumber, you can replace with a covering of open boxes which also has the sum oftheir volumes less than that number but which has each box with sides less than δ .To do this, you might consider replacing each box in the covering with 2mp openboxes obtained by bisecting each side m times where m is small enough that eachlittle box has sides smaller than δ/2 in each of the finitely many boxes in the coverand then fatten each of these just a little to cover up what got left out and retain thesum of the volumes of the little boxes to still be less than the number you had.

28. ↑Show that Lebesgue measure defined in the above problem is both inner and outerregular and is translation invariant.

29. Let (Ω,F ,µ) be a measure space and let s(ω) = ∑ni=0 ciXEi (ω) where the Ei are

distinct measurable sets but the ci might not be. Thus the ci are the finitely manyvalues of s. Say each ci ≥ 0 and c0 = 0. Define

∫sdµ as ∑i ciµ (Ei). Show that this is

well defined and that if you have s(ω) = ∑ni=1 ciXEi (ω) , t (ω) = ∑

mj=1 d jXFj (ω) ,

then for a,b nonnegative numbers, as(ω)+ bt (ω) can be written also in this formand that

∫(as+bt)dµ = a

∫sdµ + b

∫tdµ . Hint: s(ω) = ∑i ∑ j ciXEi∩Fj (ω) =

∑ j ∑i ciXEi∩Fj (ω) and (as+bt)(ω) = ∑ j ∑i (aci +bd j)XEi∩Fj (ω).

6.9. EXERCISES 15727.28.29.Cantor set of Problem 24. Let [0,1] \ P = U%_)/; where J; is open and 1; i, = 0if j #k. These intervals are the connected components of the complement of theCantor set. Show m(g(J;)) = m(J;) soTs3——ogaIelIM&oN“~Im(g(URal)) =}jThus m(g(P)) = 1 because g([0,1]) = [0,2]. By Problem 23 there exists a set,A C g(P) which is non measurable. Define ¢(x) = 24(g(x)). Thus ¢(x) = 0 unlessx € P. Tell why @ is measurable. (Recall m(P) = 0 and Lebesgue measure is com-plete.) Now show that .24(y) = 6(g7!(y)) for y € [0,2]. Tell why g~! is continuousbut @0g~! is not measurable. (This is an example of measurable o continuous 4measurable.) Show there exist Lebesgue measurable sets which are not Borel mea-surable. Hint: The function, @ is Lebesgue measurable. Now recall that Borel omeasurable = measurable.For x € R? to be in Me, Aj, it means that the i” component of x, x; is in A; for eachi. Now for []?_, (a;,b;) = R, let V (R) = TTL, (bi — ai). Next, for A € A(R?) letu(A) =n Ev (R‘) “AC ut |kThis is just like one dimensional Lebesgue measure except that instead of open in-tervals, we are using open boxes R*. Show the following.(a) pL is an outer measure.(b) w (TT; (ai, bi) = Ty (i — ai) = Be (TP (aii) -(c) If dist (A,B) > 0, then u (A)+u(B) = (AUB) so B(R?) C F the set of setsmeasurable with respect to this outer measure LU.This is Lebesgue measure on R?. Hint: Suppose for some j,b; — aj < €. Show that(TT, (ai,bi)) < €Tiz; (bi — ai). Now use this to show that if you have a coveringby finitely many open boxes, such that the sum of their volumes is less than somenumber, you can replace with a covering of open boxes which also has the sum oftheir volumes less than that number but which has each box with sides less than 6.To do this, you might consider replacing each box in the covering with 2”? openboxes obtained by bisecting each side m times where m is small enough that eachlittle box has sides smaller than 6/2 in each of the finitely many boxes in the coverand then fatten each of these just a little to cover up what got left out and retain thesum of the volumes of the little boxes to still be less than the number you had.+Show that Lebesgue measure defined in the above problem is both inner and outerregular and is translation invariant.Let (Q,.¥,) be a measure space and let s(@) = Y_)c; %z, (@) where the E; aredistinct measurable sets but the c; might not be. Thus the c; are the finitely manyvalues of s. Say each c; > 0 and co = 0. Define f sdu as Y;c;U (E;). Show that this iswell defined and that if you have s(@) = Y7., ci: 2x; (@),t(@) = Li dj Zr, (@),then for a,b nonnegative numbers, as(@) + bt(@) can be written also in this formand that f(as+bt)du =afsdu+bftdu. Hint: s(@) = YL jc; Pe nr, (@) =yy, Yi ci ZEK; (@) and (as + bt) (@) = yi yi (ac; +bdj) ZEN; (@).