340 CHAPTER 11. REGULAR MEASURES

This is obviously a measure on the Borel sets of Sp−1.

Is this even a good idea? Note mp

(λ−1 ({r}×E)

)= 0 because λ

−1 ({r}×E) is just apart of the sphere of radius r which has mp measure zero. The reason this is so is as follows.Letting α p ≡ mp (B(0,1)) , the sphere of radius r is contained in B(0,r+ ε)\B(0,r− ε)and so the sphere has mp measure no more than α p ((r+ ε)p− (r− ε)p) for every ε > 0.

Lemma 11.11.4 Let G be a Borel set in (0,∞)×Sp−1. Then

mp

(λ−1 (G)

)=∫

0

∫Sp−1

XG (ρ,ω)ρp−1dσdρ (11.16)

and the iterated integrals make sense.

Proof: Let K ≡{

I×E : I is an interval in (0,∞) and E is Borel in Sp−1}. This is a π

system and each set of K is a Borel set. Then if I is one of these intervals, having endpoints a < b,∫

0

∫Sp−1

XI×E (ρ,ω)ρp−1dσdρ =

∫ b

p−1∫

Edσdρ = σ (E)

(bp

p− ap

p

)

= p mp

(λ−1 ((0,1]×E)

)(bp

p− ap

p

)= mp

(λ−1 ((0,1]×E)

)(bp−ap)

= mp

(λ−1 ((a,b)×E)

)= mp

(λ−1 (I×E)

)Let G denote those Borel sets G in (0,∞)×Sp−1 for which, Gn ≡ G∩ (0,n)×Sp−1,

mp

(λ−1 (Gn)

)=∫

0

∫Sp−1

XGn (ρ,ω)ρp−1dσdρ

the iterated integrals making sense. It is routine to verify that G is closed with respect tocomplements and countable disjoint unions. It was also shown above that it contains K .By Dynkin’s lemma, Lemma 9.3.2, G equals the Borel sets in (0,∞)×Sp−1. Now use themonotone convergence theorem. ■

Theorem 11.11.5 Let f be a Borel measurable nonnegative function. Then∫f dmp =

∫∞

0

∫Sp−1

f (ρω)ρp−1dσdρ (11.17)

Proof: From the above lemma, if F is an arbitrary Borel set, it has the same measureas F ∩ (Rp \{0}) so there is no loss of generality in assuming 0 /∈ F .∫

RpXF dmp = mp (F) = mp

(λ−1 (λ (F))

)=∫

0

∫Sp−1

Xλ (F) (ρ,ω)ρp−1dσdρ

=∫

0

∫Sp−1

XF

(λ−1 (ρ,ω)

p−1dσdρ =∫

0

∫Sp−1

XF (ρω)ρp−1dσdρ

Now if f is nonnegative and Borel measurable, one can approximate using Borel simplefunctions increasing pointwise to f and use the monotone convergence theorem to obtain11.17. ■

Note that by Theorem 10.14.9, you can interchange the order of integration in 11.16 ifdesired.

340 CHAPTER 11. REGULAR MEASURESThis is obviously a measure on the Borel sets of S?—'.Is this even a good idea? Note m, (a! ({r} x E)) = because A”! ({r} x E) is just apart of the sphere of radius r which has m, measure zero. The reason this is so is as follows.Letting @, =m, (B(0,1)), the sphere of radius r is contained in B(0,r+e) \B(0,r—e)and so the sphere has m, measure no more than a, ((r+€)” — (r—€)?) for every € > 0.Lemma 11.11.4 Let G be a Borel set in (0,0) x S?~!. Thenmp ( )=[ [% p”dodp (11.16)spoland the iterated integrals make sense.Proof: Let # = {J x E : isan interval in (0,c0) and E is Borel in $?~'}. This is a 2system and each set of .# is a Borel set. Then if / is one of these intervals, having endpoints a <b,2 b bP @P[ | Fire (p.w) p?ldodp = | p’' | dodp =o0(E) (*-<)0 sp-! a E P Ppm, (a! ((0, 1] x E)) (> - “) =m, (a! ((0, 1] x E)) (b? —a”)= m (a ((a,b) x E)) =m, (a (Ix E))Let Y denote those Borel sets G in (0,00) x S?~! for which, G, =GM(0,n) x S?7!,my (A! (Gr) =| I Xz, (p,w) p?dodp0 SPthe iterated integrals making sense. It is routine to verify that ¥ is closed with respect tocomplements and countable disjoint unions. It was also shown above that it contains .%.By Dynkin’s lemma, Lemma 9.3.2, Y equals the Borel sets in (0,00) x §?~!. Now use themonotone convergence theorem.Theorem 11.11.5 Ler f be a Borel measurable nonnegative function. Then[famy= [- [| #(pe)p?'aoap (11.17)Proof: From the above lemma, if F is an arbitrary Borel set, it has the same measureas F 1(R? \ {O}) so there is no loss of generality in assuming 0 ¢ F.— — p-l[,, Zedimy = mp (F) = Imp (ANA et a r) (p,w)p? “dodp-/ [., Rea op ‘dodp = [fo 2X (pw) p? !dodpNow if f is nonnegative e Borel measurable, one can approximate using Borel simplefunctions increasing pointwise to f and use the monotone convergence theorem to obtain11.17.Note that by Theorem 10.14.9, you can interchange the order of integration in 11.16 ifdesired.