這張圖在說 absolute value 量的是什麼:以實部與虛部為兩股的直角三角形,斜邊長就是 |z|。ℝ² 的 norm 用的是同一個三角形、同一條斜邊——兩套定義寫下來一字不差。
THEOREM · absolute value 的四條性質
For all w, z in ℂ: (i) |z| ≥ 0, and |z| = 0 holds exactly when z = 0; (ii) z z̄ = |z|²; (iii) |wz| = |w| |z|; (iv) | |w| − |z| | ≤ |w ± z| ≤ |w| + |z|.
Proof. (i) The real number c² + d² is a sum of two squares, hence never negative, and its non-negative square root is |z|. Should |z| = 0, then c² + d² = 0; since neither square can be negative, both vanish, so c = d = 0 and z is the zero element. The converse is immediate.
(i) 是定義的直讀。因為 c² 與 d² 都不是負數(§5-2 的 5.5:非零實數的平方嚴格正,零的平方是零),它們的和也不是負數,開根號取的又是非負的那一支,所以 |z| ≥ 0。至於等號:兩個非負數相加得零,只可能各自為零,於是 c = d = 0。這一條是例行核對。
(ii) By Definition 13.1, z z̄ = (c, d)(c, −d) = (c·c − d·(−d), c·(−d) + c·d) = (c² + d², 0). Reading the pair (c² + d², 0) as the real number it stands for, this is exactly |z|².
(iii) Compute wz = (ac − bd, ad + bc), so |wz|² = (ac − bd)² + (ad + bc)². Expanding, the first bracket contributes a²c² − 2abcd + b²d² and the second contributes a²d² + 2abcd + b²c². Since the two middle terms cancel, what remains is a²c² + b²d² + a²d² + b²c² = (a² + b²)(c² + d²) = |w|² |z|². Because both |wz| and |w||z| are non-negative reals with equal squares, they are equal.
(iv) The number |z| is, letter for letter, the norm of (c, d) in ℝ². Theorem 8.7 established the triangle inequality for that norm, and the reversed estimate on the left follows from it in the usual way; both statements transfer to ℂ unchanged.
The absolute value of z = (x, y) in ℂ is the very number Section 8 calls the norm of (x, y) in ℝ². The two systems therefore carry identical balls, hence identical open sets, closed sets, cluster points, compact sets and connected sets. Every theorem of Sections 9 through 12 that is stated for ℝᵖ accordingly holds in ℂ, read with p = 2 — the Bolzano-Weierstrass Theorem 10.6, the Heine-Borel Theorem 11.3 and Theorem 12.7 among them. Results stated for ℝ alone, such as 9.11 and 12.8, describe the line and fall outside this claim.
Proof. For z = (x, y) and w = (x′, y′), the difference z − w is the pair (x − x′, y − y′), so |z − w| = ((x − x′)² + (y − y′)²)^(1/2). That is precisely the distance between (x, y) and (x′, y′) in ℝ². Consequently the ball of radius r about z consists of the same points in either system.
第一步把兩邊的距離對齊。因為 13.1 的減法是逐分量做的,z − w 的兩個分量就是 x − x′ 與 y − y′;代進絕對值的定義,得到的算式與 §8-3 的 8.9(open ball、closed ball 與 sphere)所用的距離一字不差。於是「以 z 為心、r 為半徑的球」在兩邊圈到的是同一批點——不是「對應」,是同一個集合。具體:z = (1, 1)、r = 1 圈到的那塊圓盤,兩邊寫出來的成員清單完全一致。
Now every notion of Sections 9 to 12 is built from those balls. A set is open when each of its points owns a ball inside it (9.1); closed sets are the complements of open sets (9.4); a cluster point is one every ball about which catches another point of the set (10.3); compactness is stated through coverings by open sets (11.1); connectedness through pairs of open sets (12.1). Since the balls agree, each of these classes of sets is the same in ℂ as in ℝ².
第二步是一次連鎖。§9-1 的 9.1(open set:每個成員都配得到一顆整顆留在集合內的球)只用到球;§9-2 的 9.4(closed set 是 open set 的補集)只用到 open set;§10-2 的 10.3(cluster point:每顆球都撈得到集合的另一個點)只用到球;§11-1 的 11.1(compact:每個 open covering 都裁得出有限子覆蓋)與 §12-1 的 12.1(connected:找不到由兩個 open sets 構成的分割)只用到 open set。既然最底層的球在兩邊完全一致,這一整串定義挑出來的集合族也就完全一致。
A theorem of Sections 9 to 12 stated for ℝᵖ speaks about these classes and nothing else, so each such theorem, together with its proof, reads correctly in ℂ once p is taken to be 2. A theorem stated for ℝ is a different matter: it describes the line, and the line is not what ℂ is.
「全部原封搬過去」講得太痛快了,該補上例外。§5-1 的順序性質(5.1)不是從距離長出來的——它從一個指定的正數集 P 長出來,而那個集合在 ℂ 裡根本配不出來。這件事不是「暫時還沒找到寫法」。
THEOREM · ℂ 排不出相容的大小關係
No subset P of ℂ can play the role that the positive class of 5.1 plays in ℝ: closure under addition, closure under multiplication and trichotomy cannot hold together. Hence the relations > and < of Section 5, and everything resting on them, are unavailable in ℂ.
Proof. Suppose such a P exists, and let z ≠ 0. Trichotomy leaves two possibilities: either z ∈ P, or −z ∈ P. In the first case closure under multiplication gives z · z ∈ P. In the second case it gives (−z)(−z) ∈ P, and the sign rules 4.5 — valid in every field, hence in ℂ — identify (−z)(−z) with z². Either way z² ∈ P.
中途結果:每個非零複數的平方都在 P 裡。所求是一句對所有非零 z 都成立的話,而三分律恰好把 z 的處境限縮成兩種——z ∈ P,或 −z ∈ P(z = 0 已被排除)。第一種情形,因為 P 對乘法封閉,z 乘自己還在 P 裡。第二種情形,封閉性給的是 (−z)(−z) ∈ P;而 §4-3 的 4.5(符號的規則五條,含負負得正)說 (−z)(−z) = z²,這條在任何 field 都成立,ℂ 已驗明是 field,於是照用。兩條路殊途同歸,都把 z² 送進 P。在 ℝ 上這是熟面孔:(−3)² = 9 是正的,走的正是第二條路。
Apply this to two particular elements. Since 1 = (1, 0) is not zero, 1 = 1² ∈ P. Since i = (0, 1) is not zero, i² ∈ P as well; but Definition 13.1 computes i² = (0, 1)(0, 1) = (−1, 0) = −1. Hence 1 ∈ P and −1 ∈ P at the same time.
把中途結果用在兩個具體成員上。取 z = 1 = (1, 0):它不是零,所以它的平方——也就是 1 自己——落在 P 裡。再取 z = i = (0, 1):它也不是零,所以 i² 落在 P 裡;而 13.1 的乘法公式算得 i² = (−1, 0),那正是 −1。於是 1 與 −1 一起在籍。ℝ 上不會發生這種事,因為那裡沒有任何元素的平方等於 −1——i 的存在正是 ℂ 的招牌,也正是它付出的代價。
Trichotomy applied to the element 1 allows exactly one of 1 ∈ P, 1 = 0, −1 ∈ P to hold. Two of them hold. This contradiction shows that no such P exists.
矛盾收口:三分律對元素 1 只准三項中的恰好一項成立,如今第一項與第三項同時為真。假設垮台,這樣的 P 不存在。順帶說明反例欄那個先比實部再比虛部的排法為什麼不算數:它排得出一條線、也守得住三分律,唯獨守不住乘法封閉——上面這場矛盾正是它必然失守的原因。