352 CHAPTER 11. REGULAR MEASURES

set E. Show that this implies that∫

f XEdmp = 0 for every measurable E. Explainwhy this requires f = 0 a.e.

10. Suppose f ,g are absolutely continuous on [a,b] . Prove the usual integration by partsformula. Hint: You might try the following:∫ b

af ′ (t)g(t)dm(t) =

∫ b

af ′ (t)

(∫ t

ag′ (s)dm(s)+g(a)

)dm(t)

Now do one integration and then interchange the order of integration.

11. Let x→ F ( f (x)) be absolutely continuous whenever x→ f (x) is absolutely con-tinuous. Then F is Lipschitz. This is due to G. M. Fishtenholz. Hint: Reduce toProposition 11.14.5 using an appropriate Lipschitz continuous function f .

12. Suppose g : [c,d]→ [a,b] and is absolutely continuous and increasing and f : [a,b]→R is Lipschitz continuous. Show that then f ◦g is absolutely continuous and∫ b

af (t)dm(t) =

∫ d

cf ′ (g(s))g′ (s)ds = f (b)− f (a)

13. If f ∈ L1 (Ω,µ), show that limµ(E)→0∫

E | f |dµ = 0. Defining F (x)≡∫ x

a f (t)dm(t)for f ∈ L1 ([a,b] ,m) , verify that F is absolutely continuous with F ′ (x) = f (x) a.e.

14. Show that if f is absolutely continuous, as defined in Definition 11.14.2, then it is ofbounded variation.

15. Let f : [a,b]→R be absolutely continuous. Show that in fact, the total variation of fon [a,b] is

∫ ba | f ′|dm. Hint: One direction is easy, that V [a,b]≤

∫ ba | f ′|dm. To do the

other direction, show there is a sequence of step functions sn (t) ≡ ∑mnk=1 αn

kXInk(t)

which converges to sgn f ′ pointwise a.e., Ink =

(cn

k−1,cnk

). This will involve regularity

notions. Explain why it can be assumed each∣∣αn

k

∣∣≤ 1. Then∣∣∣∣∫ b

af ′sn

∣∣∣∣=∣∣∣∣∣ mn

∑k=1

αnk

∫Ink

f ′∣∣∣∣∣≤ mn

∑k=1

∣∣ f (cnk)− f

(cn

k−1)∣∣≤V ([a,b] , f )

Now pass to a limit using the dominated convergence theorem.

16. Let F (x) =(∫ x

0 e−t2dt)2

. Justify the following:

F ′ (x) = 2(∫ x

0e−t2

dt)

e−x2= 2xe−x2

(∫ 1

0e−x2t2

dt)= 2x

(∫ 1

0e−x2(t2+1)dt

)Now integrate.

F (x) =∫ x

0

∫ 1

02ue−u2(t2+1)dtdu =

∫ 1

0

∫ x

02ue−u2(t2+1)dudt

=∫ 1

0−e−u2(t2+1) 1

1+ t2 |x0dt =

∫ 1

0

(1

1+ t2 − e−x2 11+ t2

)dt

Now let x→ ∞ and conclude F (∞) =(∫

0 e−t2dt)2

=∫ 1

01

1+t2 dt = π

4 .

35210.11.12.13.14.15.16.CHAPTER 11. REGULAR MEASURESset E. Show that this implies that { f2 dm, = 0 for every measurable E. Explainwhy this requires f = 0 a.e.Suppose f,g are absolutely continuous on [a,b]. Prove the usual integration by partsformula. Hint: You might try the following:[roswame= [ro (['eams) +e) amenNow do one integration and then interchange the order of integration.Let x + F (f (x)) be absolutely continuous whenever x — f (x) is absolutely con-tinuous. Then F is Lipschitz. This is due to G. M. Fishtenholz. Hint: Reduce toProposition 11.14.5 using an appropriate Lipschitz continuous function f.Suppose g : [c,d] — [a,b] and is absolutely continuous and increasing and f : [a,b] >R is Lipschitz continuous. Show that then fo g is absolutely continuous andb ‘d[ roam = [Fee as = FO) F(a)If f < L' (Q, w), show that limy(z) +0 Je |f|du = 0. Defining F (x) = f7 f (t)dm/(t)for f € L' ({a,b] ,m) , verify that F is absolutely continuous with F’ (x) = f (x) ae.Show that if f is absolutely continuous, as defined in Definition 11.14.2, then it is ofbounded variation.Let f : [a,b] > R be absolutely continuous. Show that in fact, the total variation of fon [a,b] is [° |f’|dm. Hint: One direction is easy, that V [a,b] < [? | f"|dm. To do theother direction, show there is a sequence of step functions s, (t) = Dy", Ep (t)which converges to sgn f’ pointwise a.e., I!’ = (ch ; ch). This will involve regularitynotions. Explain why it can be assumed each | oc? < 1. Thenb ) mn )/ f'Sn y a | fa k=l qNow pass to a limit using the dominated convergence theorem.<¥ Me) (cha) SV (ab)./)=12Let F (x) = (ie ear) . Justify the following:x 1 1F'(x)=2 (/ ear) e* =2xe* (/ e*"ar) = 2x (/ e* Har)0 0 0Now integrate.x ol 1px[Cf ame # atau = | [ Que" (P+) dudt0 Jo 0 JO1 2/2 1 1 1 2 1_p (P41) dt = | —e* dtI . T4729 o lige © 14h2Now let x > oo and conclude F (ce) = Ue" edt) = fy pdt = ,F (x)