Notes on Complex Analysis
Export PDFGitHub

§1.1 Topological Preliminaries

1.1 Topological Preliminaries

The following definitions are subject to the assumption where the topological space is defined to be 𝑋=ℂ𝑛. This is satisfactory to the main purpose of our proceeding passage, but it is noteworthy that it can be generalized to more abstract sets (which will be discussed in later sections, especially in Section 3.2.1Section 3.2.1Section 3.2.1).

Definition 1.1.1 (Open Ball).

Let 𝐵𝑛(𝑎,𝑟)⊂ℂ𝑛 denote the 𝑛-dimensional ball with radius 𝑟 centered at 𝑎=(𝑎1,𝑎2,…,𝑎𝑛)∈ℂ𝑛, defined by

𝐵𝑛(𝑎,𝑟)={(𝑧1,𝑧2,…,𝑧𝑛)∈ℂ𝑛:∑𝑗=1𝑛|𝑧𝑗−𝑎𝑗|2<𝑟2}.

Definition 1.1.2 (Open and Closed Sets).

A set 𝑈⊆ℂ𝑛 is open if for every 𝑧∈𝑈, there exists an open ball centered at 𝑧 that is fully contained in 𝑈. A set 𝐹⊆ℂ𝑛 is closed if its complement, ℂ𝑛∖𝐹, is open.

(Note that these definitions hold only for the standard topology on ℂ𝑛 induced by the Euclidean metric. For now, this is unimportant, but will be explained later in Section 3.2.1Section 3.2.1Section 3.2.1.)

Definition 1.1.3 (Accumulation Point).

A point 𝑧∈ℂ𝑛 is an accumulation point of 𝑋 if for any open set 𝑈 containing 𝑧, (𝑈∖{𝑧})∩𝑋≠⌀

Theorem 1.1.4.

A set 𝑋⊆ℂ𝑛 is closed iff 𝑋 contains all of its accumulation points.
Proof.
  1. We first prove that closedness implies the inclusion of accumulation points. Let 𝑋 be a closed set, and assume that some accumulation point 𝑧0 of 𝑋 satisfies 𝑧0∉𝑋. Since 𝑋 is closed, ℂ𝑛∖𝑋 is open, and there exists an open ball 𝐵𝑛(𝑧0,𝜀) that is fully contained in ℂ𝑛∖𝑋. However, this contradicts the definition of accumulation point, since 𝐵𝑛(𝑧0,𝜀)∖{𝑧0} does not intersect with 𝑋.
  2. Assume 𝑋 is an arbitrary set which includes all its accumulation points. We will show that ℂ𝑛∖𝑋 is open, which implies that 𝑋 is closed. Let 𝑧0∈ℂ𝑛∖𝑋. Since 𝑧0 is not an accumulation point of 𝑋 (as otherwise 𝑧0∈𝑋), there exists an open ball 𝐵𝑛(𝑧0,𝜀) such that (𝐵𝑛(𝑧0,𝜀)∖{𝑧0})∩𝑋=⌀. Then, 𝐵𝑛(𝑧0,𝜀)⊂ℂ𝑛∖𝑋, and hence ℂ𝑛∖𝑋 is open. □

Definition 1.1.5 (Closure).

For a set 𝑋⊆ℂ𝑛, define the closure of 𝑋, or 𝑋‾ to be the intersection of all closed sets containing 𝑋. In other words, it is the union of 𝑋 and its accumulation points.

Because the accumulation points of 𝑋 all lie in 𝑋‾, the closure of 𝑋 is a closed set. Moreover, 𝑋 is closed iff 𝑋=𝑋‾.

Definition 1.1.6 (Interior).

For a set 𝑋⊆ℂ𝑛, the interior of 𝑋, denoted 𝑋⚬, is the union of all open sets contained in 𝑋, or the set of points 𝑧∈ℂ𝑛 such that there exists an open neighborhood of 𝑧 that is fully contained in 𝑋.

Definition 1.1.7 (Set Covering).

A cover 𝒞︀ of a set 𝑋 is a collection of sets {𝑈𝑛} such that

⋃𝑛∈ℕ𝑈𝑛⊇𝑋.
A cover is open if every set in the collection is open.

Definition 1.1.8 (Compact Set).

A set 𝑋⊆ℂ𝑛 is compact iff every open cover has a finite subcover.

Theorem 1.1.9 (Heine–Borel).

A set 𝑋⊆ℂ𝑛 is compact iff it is closed and bounded.
Proof.

We will first show that any compact set is closed and bounded.

First we will show that 𝑋 is bounded. Suppose that ∀𝑅>0, ∃𝑧∈𝑋 where ‖𝑧‖>𝑅. Consider the collection of open sets

𝒰︀={𝐵𝑛(0,𝑘)|𝑘∈ℕ}.

𝒰︀ forms an open cover of 𝑋. Then by the assumption that 𝑋 is compact, there exists a finite subcover in 𝒰︀, namely {𝐵𝑛(0,𝑘1),…,𝐵𝑛(0,𝑘𝑚)} which covers 𝑋. Then,

𝑋⊆⋃𝑖=1𝑚𝐵𝑛(0,𝑘𝑖)=𝐵𝑛(0,max{𝑘1,…,𝑘𝑚}).

By contradiction, 𝑋 must be bounded.

𝑋 must also be a closed set. For the sake of contradiction, assume that there exists a point 𝑧0∈𝑋‾∖𝑋. Since 𝑧0∉𝑋, the following open collection of sets covers 𝑋:

𝒰︀={ℂ𝑛∖𝐵𝑛(𝑧0,1𝑘)‾|∀𝑘∈ℕ}.

There then exists a finite subcover 𝒞︀={ℂ𝑛∖𝐵𝑛(𝑧0,1𝑘𝑗)‾|𝑗∈ℕ≤𝑚}. Then,

𝑋⊆ℂ𝑛∖𝐵𝑛(𝑧0,1max{𝑘1,…,𝑘𝑚})‾,

and that 𝑋∩𝐵𝑛(𝑧0,1max{𝑘1,…,𝑘𝑚})‾=⌀. However, by the definition of the accumulation point, every open neighborhood of the accumulation point must intersect 𝑋. Therefore, by contradiction, 𝑋 is closed.

We then prove the converse. By the assumption that 𝑋 is bounded, ∃𝑅>0 such that 𝑋 is contained within the closed cube

𝑄={𝑧∈ℂ𝑛|max𝑗∈ℕ≤𝑛|ℜ𝔢(𝑧𝑗)|≤𝑅∧max𝑗∈ℕ≤𝑛|ℑ𝔪(𝑧𝑗)|≤𝑅}.

Assume that there exists an infinite open cover 𝒰︀ of 𝑋 without finite subcovering.

Bisect 𝑄0=𝑄 into 22𝑛 sub-cubes (for real and complex parts), choose 𝑄1 such that 𝑄1∩𝑋 has no finite subcover of 𝒰︀.

Under the previous assumptions, this is possible since if every sub-cube∩𝑋 had finite subcovering, then 𝑄0∩𝑋=𝑋 would have finite subcovering. Similarly, choose 𝑄2 by bisecting 𝑄1 similarly, and recursively obtain a sequence of cubes:

𝑄0⊃𝑄1⊃𝑄2⊃⋯

Since the side length of each cube tends to 0, ⋂𝑗=0∞𝑄𝑗 consists of a single point 𝑧∞∈ℂ𝑛. Since 𝑋 is closed, 𝑧∞∈𝑋, and since 𝒰︀ covers 𝑋, ∃𝑈∈𝒰︀ such that 𝑧∞∈𝑈. Since 𝑈 is open, ∃𝜀>0 such that 𝐵𝑛(𝑧∞,𝜀)⊆𝑈. ∃𝑁∈ℕ such that ∀𝑘>𝑁, 𝑄𝑘⊂𝐵𝑛(𝑧∞,𝜀). Then taking the intersection with 𝑋 on both sides,

𝑄𝑘∩𝑋⊆𝐵𝑛(𝑧∞,𝜀)∩𝑋⊆𝑈.

This contradicts the assumption that for every 𝑘, 𝑄𝑘∩𝑋 has no finite subcovering, since {𝑈}⊂𝒰︀ clearly covers 𝑄𝑘∩𝑋, as it is a single open set that covers a nonempty subset. Therefore by contradiction, every open cover has finite subcovering. ⁠□

Theorem 1.1.10 (Bolzano–Weierstrass).

Every infinite subset 𝐴 of a compact set 𝑋⊆ℂ𝑛 has an accumulation point in 𝑋.
Proof.

Since 𝑋 is compact, it is bounded by Theorem 1.1.9Theorem 1.1.9Theorem 1.1.9, so there exists a closed cube 𝑄⊂ℂ𝑛 such that 𝐴⊆𝑋⊂𝑄.

Bisect 𝑄0=𝑄 into 22𝑛 congruent sub-cubes. Since 𝐴 is infinite and the sub-cubes are finite in number, at least one of the sub-cubes contains infinitely many points of 𝐴, and choose one to be 𝑄1.

Bisect 𝑄1 into 22𝑛 sub-cubes, and choose a sub-cube 𝑄2⊂𝑄1 that contains infinitely many points of 𝐴. We then obtain the recursive sequence

𝑄0⊃𝑄1⊃𝑄2⊃⋯

Because the side lengths shrink to zero and the cubes are nested, the intersection

⋂𝑘=0∞𝑄𝑘

consists of exactly one point. Call this point 𝑧∞∈ℂ𝑛.

For each 𝑘, 𝑄𝑘 contains infinitely many points of 𝐴. Because the side length of 𝑄𝑘 tends to zero, for any 𝜀>0, ∃𝑁∈ℕ such that ∀𝑘≥𝑁, 𝑄𝑘⊂𝐵𝑛(𝑧∞,𝜀).

Then, 𝐵𝑛(𝑧∞,𝜀) also contains infinitely many points of 𝐴. Therefore, 𝑧∞ is an accumulation point of 𝐴.

We now show that 𝑧∞∈𝑋. Suppose for contradiction that 𝑧∞∉𝑋. Since 𝑋 is compact, it is closed by Theorem 1.1.9Theorem 1.1.9Theorem 1.1.9, thus ℂ𝑛∖𝑋 is open, and ∃𝛿>0 such that

𝐵𝑛(𝑧∞,𝛿)⊂ℂ𝑛∖𝑋.

But then, for sufficiently large 𝑘, we have 𝑄𝑘⊂𝐵𝑛(𝑧∞,𝛿), and hence 𝑄𝑘∩𝑋=⌀. This contradicts the construction of 𝑄𝑘, which ensures that 𝑄𝑘 contains infinitely many points of 𝐴⊂𝑋. ⁠□

Definition 1.1.11 (Support of a Function).

For a set 𝑋 and a function 𝑓:𝑋→ℂ, the support, denoted by supp(𝑓)={𝑧∈𝑋:𝑓(𝑧)≠0}‾, is the closure of the set for which 𝑓 is nonzero.

Remark.

A notable classification of functions comes from the compactness of support—more specifically, its boundedness. Compactly supported functions in 𝐶∞(ℂ) (infinitely differentiable functions in the complex plane) are commonly referred to as bump functions (see Section 3.2.1Section 3.2.1Section 3.2.1).