412 CHAPTER 16. MAPPING THEOREMS

point to which fn (z) converges. Then f is a continuous function defined on Ω. Is f ana-lytic? Yes it is by Lemma 14.8.5. Alternatively, you could let T ⊆ Ω be a triangle. Then∫

∂T f (z)dz = limn→∞

∫∂T fn (z)dz = 0. Therefore, by Morera’s theorem, f is analytic.

As for the uniform convergence of the derivatives of f , recall Theorem 15.4.2 about theexistence of a cycle. Let K be a compact subset of Ω. Then for some n, K is a compactsubset of int(Kn) and let {γk}

mk=1 be closed oriented curves contained γ∗k ⊆ int(Kn)\K such

that ∑mk=1 n(γk,z) = 1 for every z ∈ K. Also let η denote the distance between ∪ jγ

∗j and

K,η ≡ inf{|z−w| : z ∈ K,w ∈ ∪ jγ

∗j

}It follows that η > 0. (Why? In general, two disjoint

compact sets are at a positive distance from each other. ) Then for z ∈ K,

∣∣∣ f (k) (z)− f (k)n (z)∣∣∣ =

∣∣∣∣∣ k!2πi

m

∑j=1

∫γ j

f (w)− fn (w)

(w− z)k+1 dw

∣∣∣∣∣≤ k!

2π∥ fk− f∥Kn

m

∑j=1

(length of γk)1

ηk+1 .

where here ∥ fk− f∥Kn≡max{| fk (z)− f (z)| : z ∈ Kn} . Thus you get uniform convergence

of the derivatives on each compact subset of Ω. ■Another surprising consequence of this theorem is that the property of being one to one

is preserved if the target function is known to not be a constant.

Lemma 16.6.2 Suppose hn is one to one, analytic on Ω, a connected open set (region)and converges uniformly to h on compact subsets of Ω along with all derivatives. Then if his not a constant, it follows that h is also one to one.

Proof: Pick z1 ∈Ω and suppose z2 is another point of Ω. As shown above, h is analytic.Thus, if the zeros of h− h(z1) have a limit point in Ω, then h(z) is a constant which isassumed to not be the case. Since the zeros of h−h(z1) have no limit point, there exists acircular contour bounding a circle which has z2 on the inside of this circle but not z1 suchthat γ∗ contains no zeros of h−h(z1). Taking a subsequence if necessary, it can be assumedγ∗ contains no zeros of hn−hn (z1) either.

•z1

γ•z2

Using the theorem on counting zeros, Theorem 16.3.1, and the fact that hn is one toone, we know that hn−hn (z1) has no zeros inside this circle and so

0 = limn→∞

12πi

∫γ

h′n (w)hn (w)−hn (z1)

dw =1

2πi

∫γ

h′ (w)h(w)−h(z1)

dw,

which shows that h− h(z1) has no zeros in B(z2,r) . In particular z2 is not a zero of h−h(z1) . This shows that h is one to one since z2 ̸= z1 was arbitrary. ■

Theorem 16.6.1 is an example of a normal family of functions.

Definition 16.6.3 Let F denote a collection of functions which are analytic on Ω,a region (open and connected). Then F is normal if every sequence contained in F has asubsequence which converges uniformly on compact subsets of Ω.

412 CHAPTER 16. MAPPING THEOREMSpoint to which f, (z) converges. Then f is a continuous function defined on Q. Is f ana-lytic? Yes it is by Lemma 14.8.5. Alternatively, you could let T C Q be a triangle. ThenJar Ff (z) dz = limn+0 f57 fn (z) dz = 0. Therefore, by Morera’s theorem, f is analytic.As for the uniform convergence of the derivatives of f, recall Theorem 15.4.2 about theexistence of a cycle. Let K be a compact subset of Q. Then for some n, K is a compactsubset of int (K,) and let {7 }/_, be closed oriented curves contained ¥; C int (K,) \ K suchthat P71 (%,z) = 1 for every z € K. Also let 7 denote the distance between U iV; andK,n= int { Jz—w| ize K,weuj7; hie follows that n > 0. (Why? In general, two disjointcompact sets are at a positive distance from each other. ) Then for z € K,f (w) _ Sn (w) dw2mi Jy) (wz)mk} 1mm Ife —fllx, Yo (length of 7;) Wej=llAwhere here || fx — f || x, = max {| fx (z) — f (z)| : z © Kn}. Thus you get uniform convergenceof the derivatives on each compact subset of Q.Another surprising consequence of this theorem is that the property of being one to oneis preserved if the target function is known to not be a constant.Lemma 16.6.2 Suppose hy is one to one, analytic on Q, a connected open set (region)and converges uniformly to h on compact subsets of Q along with all derivatives. Then if his not a constant, it follows that h is also one to one.Proof: Pick z; € Q and suppose zz is another point of Q. As shown above, h is analytic.Thus, if the zeros of h—h(z,) have a limit point in Q, then /(z) is a constant which isassumed to not be the case. Since the zeros of h —h(z;) have no limit point, there exists acircular contour bounding a circle which has zz on the inside of this circle but not z; suchthat y* contains no zeros of h—h(z,). Taking a subsequence if necessary, it can be assumedy* contains no zeros of h, — hy (z1) either.YeZIUsing the theorem on counting zeros, Theorem 16.3.1, and the fact that h, is one toone, we know that h, — h, (z1) has no zeros inside this circle and sohi, (w 1 h' (w)lim — dw = — | —————d0= lim 55 igen) ae ” sai |, oy ha "which shows that / —h(z;,) has no zeros in B(z),r). In particular z is not a zero of h—h(z,). This shows that / is one to one since z2 4 z; was arbitrary. HiTheorem 16.6.1 is an example of a normal family of functions.Definition 16.6.3 Let F denote a collection of functions which are analytic on Q,a region (open and connected). Then ¥ is normal if every sequence contained in ¥ has asubsequence which converges uniformly on compact subsets of Q.