11.12. SYMMETRIC DERIVATIVE FOR RADON MEASURES 341

Example 11.11.6 For what values of s is the integral∫

B(0,R)

(1+ |x|2

)sdy bounded inde-

pendent of R? Here B(0,R) is the ball, {x ∈ Rp : |x| ≤ R} .

I think you can see immediately that s must be negative but exactly how negative? Itturns out it depends on p and using polar coordinates, you can find just exactly what isneeded. From the polar coordinates formula above,∫

B(0,R)

(1+ |x|2

)sdy =

∫ R

0

∫Sp−1

(1+ρ

2)sρ

p−1dσdρ =Cp

∫ R

0

(1+ρ

2)sρ

p−1dρ

Now the very hard problem has been reduced to considering an easy one variable prob-lem of finding when

∫ R0 ρ p−1

(1+ρ2

)s dρ is bounded independent of R. You need 2s+(p−1)<−1 so you need s <−p/2.

11.12 Symmetric Derivative for Radon MeasuresHere we have two Radon measures µ,λ defined on a σ algebra of sets F which are subsetsof an open subset U of Rp, possibly all of Rp. They are complete and Borel and inner andouter regular, and finite on compact sets. Thus both of these measures are σ finite.

In this section is the symmetric derivative Dµ (λ ). In what follows, B(x,r) will denotea closed ball with center x and radius r. Also, let λ and µ be Radon measures and as above,Z will denote a µ measure zero set off of which µ (B(x,r))> 0 for all r > 0. Generalizingthe notion of limsup and liminf,

limsupr→0

f (r)≡ limr→0

(sup{ f (t) : t < r}) , lim infr→0

f (r)≡ limr→0

(inf{ f (t) : t < r})

Then directly from this definition, the limr→0 exists if and only if these two are equal.

Definition 11.12.1 For x /∈ Z, define the upper and lower symmetric derivativesas

Dµ λ (x)≡ limsupr→0

λ (B(x,r))µ (B(x,r))

, Dµ λ (x)≡ lim infr→0

λ (B(x,r))µ (B(x,r))

.

respectively. Also define Dµ λ (x)≡ Dµ λ (x) = Dµ λ (x) in the case when both the upperand lower derivatives are equal. Recall that Z ≡ {x : µ (B(x,r)) = 0 for some r > 0} andthat this set has measure zero.

Lemma 11.12.2 Let λ and µ be Radon measures on Fλ and Fµ respectively and leta,b > 0. If A is a subset of

{x /∈ Z : Dµ λ (x)≥ b

}then λ (A)≥ bµ (A) and if A is a subset

of{x /∈ Z : Dµ λ (x)≤ a

}, then λ (A)≤ aµ (A) .

Proof: Let λ be the outer measure determined by λ , similar for µ and µ . Suppose firstthat A is a subset of

{x /∈ Z : Dµ λ (x)≥ b

}so µ (B(x,r))> 0 for all r > 0 and λ (A)< ∞.

Let small ε > 0, and let V be a bounded open set with V ⊇ A and λ (V )− ε < λ (A) . Thenfor each x ∈ A, λ (B(x,r))

µ(B(x,r)) > b− ε, B(x,r) ⊆ V,for infinitely many values of r which arearbitrarily small. Thus the collection of such closed balls constitutes a Vitali cover forA. By Corollary 9.12.3 there is a disjoint sequence of these closed balls {Bi} such thatµ (A\∪∞

i=1Bi) = 0,

µ (A)≤ µ (A\∪∞i=1Bi)+µ (∪∞

i=1Bi∩A)≤∞

∑i=1

µ (Bi∩A) (11.18)

11.12. SYMMETRIC DERIVATIVE FOR RADON MEASURES 341SsExample 11.11.6 For what values of s is the integral Jy(,p) (1 + le") dy bounded inde-pendent of R? Here B(0,R) is the ball, {a € R? : |x| < R}.I think you can see immediately that s must be negative but exactly how negative? Itturns out it depends on p and using polar coordinates, you can find just exactly what isneeded. From the polar coordinates formula above,2 2\5 _ “| 2\5 .p—l _ [ 25 ptDrow tl ) dy= [ spl (1+p ) p’‘dodp =C, F (I+p ) p?-!dpNow the very hard problem has been reduced to considering an easy one variable prob-lem of finding when fo pP-! (1 + p’)'dp is bounded independent of R. You need 2s +(p—1) <—1 so you need s < —p/2.11.12 Symmetric Derivative for Radon MeasuresHere we have two Radon measures LL, A defined on a o algebra of sets ¥ which are subsetsof an open subset U of R?, possibly all of R?. They are complete and Borel and inner andouter regular, and finite on compact sets. Thus both of these measures are o finite.In this section is the symmetric derivative D, (A). In what follows, B(a,r) will denotea closed ball with center x and radius r. Also, let A and be Radon measures and as above,Z will denote a u measure zero set off of which pu (B(a,r)) > 0 for all r > 0. Generalizingthe notion of lim sup and lim inf,lim sup f (r) = lim (sup {f (¢) :t < r}), lim inf f (r) = lim (inf {f (t) : t < r})r>0 r>0 r>0r>0Then directly from this definition, the lim,_,9 exists if and only if these two are equal.Definition 11.12.1 For ¢ Z, define the upper and lower symmetric derivativesas_ A(B A (BDyA (a) = limsup A(B (wr) D,,A (x) = lim inf A(B(@,r))30 HL (B(a,r)) r>0 U(B(a,r))respectively. Also define DyA (x) = DyA (x) = D,,A (a) in the case when both the upperand lower derivatives are equal. Recall that Z = {x : u(B(a,r)) =0 for some r > 0} andthat this set has measure zero.Lemma 11.12.2 Let 4 and wu be Radon measures on Fj, and Fy, respectively and leta,b >0. IfA is a subset of {x ¢ Z: DyA (x) > b} then 2 (A) > bE (A) and if A is a subsetof {a ¢Z: DA (x) < a}, then A (A) < ap(A).Proof: Let 4 be the outer measure determined by A, similar for ff and w. Suppose firstthat A is a subset of {a ¢ Z: DyA (x) > b} so u(B(ax,r)) > 0 for all r > 0 and A (A) <o.Let small ¢ > 0, and let V be a bounded open set with V D A and A (V) —€ < A (A). Thenforeach x EA, ae > b—e, B(a,r) CV, for infinitely many values of r which arearbitrarily small. Thus the collection of such closed balls constitutes a Vitali cover forA. By Corollary 9.12.3 there is a disjoint sequence of these closed balls {B;} such thatH(A\Uz,Bi) =0,8H(A) < #(A\ U2,B;) + (U2, BNA) < VHB) A) (11.18)1