240 CHAPTER 10. REGULAR MEASURES

≤ mp

([x : limsup

r→0

1mp (B(x,r))

∫B(x,r)

| f (y)−g(y)|dmp (y)>ε

2

])+mp

([x : limsup

r→0

1mp (B(x,r))

∫B(x,r)

|g(y)−g(x)|dmp (y)>ε

2

])+mp

([x : |g(x)− f (x)|> ε

2

]).

≤ mp

([M ( f −g)>

ε

2

])+mp

([| f −g|> ε

2

])(10.3)

Now∥ f −g∥1 ≥

∫[| f−g|> ε

2 ]| f −g|dmp ≥

ε

2mp

([| f −g|> ε

2

])and so using Claim 1 and 10.3, 10.2 is dominated by

(2ε+ 5p

ε

)∫| f −g|dmp. But by Propo-

sition 10.1.4, g can be chosen to make the above as small as desired. Hence 10.2 is 0.

mp

([limsup

r→0

1mp (B(x,r))

∫B(x,r)

| f (y)− f (x)|dmp (y)> 0])

≤∞

∑k=1

mp

([limsup

r→0

1mp (B(x,r))

∫B(x,r)

| f (y)− f (x)|dmp (y)>1k

])= 0

By completeness of mp this implies[limsup

r→0

1mp (B(x,r))

∫B(x,r)

| f (y)− f (x)|dmp (y)> 0]

is a set of mp measure zero. ■The following corollary is the main result referred to as the Lebesgue Differentiation

theorem.

Definition 10.2.2 f ∈ L1loc (Rp,mp) means f XB is in L1 (Rn,mp) whenever B is a

ball.

Corollary 10.2.3 If f ∈ L1loc (Rp,mp), then for a.e.x,

limr→0

1mp (B(x,r))

∫B(x,r)

| f (y)− f (x)|dmp (y) = 0 . (10.4)

In particular, for a.e.x,

limr→0

1mp (B(x,r))

∫B(x,r)

f (y)dmp (y) = f (x)

Proof: If f is replaced by f XB(0,k) then the conclusion 10.4 holds for all x /∈ Fk whereFk is a set of mp measure 0. Letting k = 1,2, · · · , and F ≡ ∪∞

k=1Fk, it follows that F is aset of measure zero and for any x /∈ F , and k ∈ {1,2, · · ·}, 10.4 holds if f is replaced byf XB(0,k). Picking any such x, and letting k > |x|+1, this shows

limr→0

1mp (B(x,r))

∫B(x,r)

| f (y)− f (x)|dmp (y)

240 CHAPTER 10. REGULAR MEASURESIAmp (|e stimsup ara J. \h(w) elude») > 5) )+a (|2stimsue Tae Into a0@ldmo 0) > 5] )+p ([@ : |g (@) ~ f(@)| > 5]).cma([mr-a>{])-+mo([ir-ai> §]) avalf—sll, = hy-eoe If —gldm, > mp ({lr-2! S |)and so using Claim | and 10.3, 10.2 is dominated by (2 + 2) J \|f —g|dmp. But by Propo-sition 10.1.4, g can be chosen to make the above as small as desired. Hence 10.2 is 0.mp ([limsup ara ay |, IP) Fe) lame (0) > 0} )< Yim (|timsup ao (uF (eam) > ;]) =0k=l r30 MpBy completeness of m, this implies1Himsup rary Jig all wl F(@)ldinp(») > 0r30 Mp (B (x, r)) B(x,r) ( ,is a set of m, measure zero. HlThe following corollary is the main result referred to as the Lebesgue Differentiationtheorem.Definition 10.2.2 /< Lh. (R?,mp) means f 2% is in L' (R",m,) whenever B is aball.Corollary 10.2.3 /f f € L},.(R’,mp), then for a.e.x,loc; 1lim | en fF lamp (9) = 0, (10.4)r+0 Mp (B(a,r))In particular, for a.e.x,lim ——1__ [ en f (y) dmp (y) = f (a)r0 My (B(a,r))Proof: If f is replaced by f 2(9,x) then the conclusion 10.4 holds for all x ¢ F, whereFy is a set of mp, measure 0. Letting k = 1,2,---, and F = Ue_,/, it follows that F is aset of measure zero and for any a ¢ F, and k € {1,2,---}, 10.4 holds if f is replaced byf %p(o,k)- Picking any such z, and letting k > |a|+ 1, this showslim aac Ian fw) F@lemer+0 Mp (B(a,7r))