-
It suffices to prove the second claim, since it implies that for each compact K, ∣F(z)∣<∞ almost everywhere on K, i.e., F(z) converges almost everywhere on K, so F(z) converges almost everywhere on C.
We may assume without loss of generality that K=B(0,R). Indeed, any compact set is contained within such a ball if R is large enough, and we get the result by monotonicity.
By applying Fubini-Tonelli, we may switch the order of integration in the following calculation:
∫B(0,R)∣F(z)∣dA≤∫∣z∣≤R∫w∈C∣z−w∣1dμ(w)dA(z)=∫w∈C∫∣z∣≤R∣z−w∣1dA(z)dμ(w)=∫∣w∣≤2R∫∣z∣≤R∣z−w∣1dA(z)dμ(w)+∫∣w∣>2R∫∣z∣≤R∣z−w∣1dA(z)dμ(w)≤∫∣w∣≤2R∫∣z∣≤R∣z−w∣1dA(z)dμ(w)+∫∣w∣>2R∫∣z∣≤RR1dA(z)dμ(w)≤∫∣w∣≤2R∫∣z−w∣≤3R∣z−w∣1dA(z)dμ(w)+∫∣w∣>2R∫∣z∣≤RR1dA(z)dμ(w)=∫∣w∣≤2R∫03R∫02πrrdrdθdμ(w)+∫∣w∣>2RπRdμ(w)=∫∣w∣≤2R6πRdμ(w)+∫∣w∣>2RπRdμ(w)=6πRμ(∣w∣≤2R)+πRμ(∣w∣>2R)<∞.
-
As before, it suffices to prove this for closed balls [−R,R]⊆L.
Apply (1) and Fubini-Tonelli to K=[−M,M]×[n,n+1], where n∈Z, M∈N, which gives ∫nn+1∫−MM∣F(x+iy)∣dydx<∞. Thus, for almost every y∈[n,n+1], we have ∫−MM∣F(x+iy)∣dx<∞. Call this full measure set YnR and set
Yn=M∈N⋂YnM.
Notice that
m(Ync∩[k,k+1])≤k∈Z∑m((YnM)c∩[k,k+1])=0,
so taking a union over k∈Z, we see that Ync has Lebesgue measure zero. Moreover, if y∈Yn, then ∫−MM∣F(x+iy)∣dx<∞. Thus, if we set
Y=n∈Z⋃Yn,
then Yc also has Lebesgue measure zero as a countable intersection of measure zero sets.
Finally, let R>0 and y∈Y. If M∈N is such that R≤M, then by definition of Y,
∫−RR∣F(x+iy)∣dx≤∫−MM∣F(x+iy)∣dx<∞,
which proves the claim.
-
Notice that in (1), we proved that ∫K∣z−w∣1dA<∞ for any compact set K. Thus, by replacing ∣F(z)∣ with ∫C∣z−w∣1dμ(w) in (2), we see that for almost every horizontal line L in C that
∫K∫w∈C∣(x+iy)−w∣1dμ(w)dx<∞
for any compact K⊆L.
If we run the same argument in (2) but with x and y switched, we obtain the analogous result for vertical lines with conull sets X for vertical lines and Y for horizontal ones.
Let S be the set of squares in C whose vertices in X×Y, so S contains almost every square in C. Then in particular, we may integrate ∣F(z)∣ along ∂S. Combining this with our remark at the beginning, we may apply Fubini's theorem:
2πi1∫∂SF(z)dz=2πi1∫∂S∫w∈Cz−w1dμ(w)dz=∫w∈C(2πi1∫∂Sz−w1dz)dμ(w)=∫w∈CχSdμ=μ(S),
where the second-to-last equality comes from Cauchy's integral formula, and this gives us the desired result.