跳轉到

Q8:內積空間

考點:內積定義、正交性、Gram-Schmidt 正交化、正規化


核心洞察

內積的本質:測量「方向的相似度」

\[\langle \mathbf{u}, \mathbf{v} \rangle = \|\mathbf{u}\| \|\mathbf{v}\| \cos\theta\]

抽象內積三公理

  1. 對稱性\(\langle \mathbf{u}, \mathbf{v} \rangle = \langle \mathbf{v}, \mathbf{u} \rangle\)
  2. 線性\(\langle a\mathbf{u} + b\mathbf{v}, \mathbf{w} \rangle = a\langle \mathbf{u}, \mathbf{w} \rangle + b\langle \mathbf{v}, \mathbf{w} \rangle\)
  3. 正定性\(\langle \mathbf{v}, \mathbf{v} \rangle \geq 0\),等號 \(\Leftrightarrow \mathbf{v} = \mathbf{0}\)

邏輯鏈

內積 → 範數 ‖v‖ = √⟨v,v⟩ → 正交 ⟨u,v⟩ = 0
                    Gram-Schmidt 正交化
                    正規正交基(QR 分解的核心)

題型分類

類型 1:函數空間 Gram-Schmidt

\(\mathcal{P}_2\) 上內積 \(\langle p, q \rangle = \int_{-1}^1 p(x)q(x)dx\),對 \(\{1, x, x^2\}\) 做 Gram-Schmidt。

Step 1:計算所需內積

\(\langle \cdot, \cdot \rangle\) 計算
\(\langle 1, 1 \rangle\) \(2\) \(\int_{-1}^1 1\,dx\)
\(\langle x, 1 \rangle\) \(0\) 奇函數
\(\langle x, x \rangle\) \(\frac{2}{3}\) \(\int_{-1}^1 x^2\,dx\)
\(\langle x^2, 1 \rangle\) \(\frac{2}{3}\) \(\int_{-1}^1 x^2\,dx\)
\(\langle x^2, x \rangle\) \(0\) 奇函數

Step 2:Gram-Schmidt

\[u_1 = 1\]
\[u_2 = x - \frac{\langle x, 1 \rangle}{\langle 1, 1 \rangle} \cdot 1 = x - 0 = x\]
\[u_3 = x^2 - \frac{\langle x^2, 1 \rangle}{\langle 1, 1 \rangle} \cdot 1 - \frac{\langle x^2, x \rangle}{\langle x, x \rangle} \cdot x = x^2 - \frac{2/3}{2} - 0 = x^2 - \frac{1}{3}\]

正交基底\(\{1, x, x^2 - \frac{1}{3}\}\)(即 Legendre 多項式 \(P_0, P_1, P_2\) 差常數倍)

類型 2:驗證內積

證明 \(\langle p, q \rangle = \int_0^1 p(x)q(x)dx\)\(\mathcal{P}_n\) 上是合法內積。

驗證三公理

  • 對稱性:\(\int_0^1 pq = \int_0^1 qp\)
  • 線性:積分的線性性 ✓
  • 正定性:\(\int_0^1 p(x)^2 dx \geq 0\),等號 \(\Leftrightarrow p \equiv 0\)

類型 3:投影公式

\(W = \text{span}\{1, x\} \subseteq \mathcal{P}_2\),求 \(x^2\)\(W\) 上的正交投影。

公式(用正交基):

\[\text{proj}_W(v) = \sum_{i} \frac{\langle v, u_i \rangle}{\langle u_i, u_i \rangle} u_i\]

\(\{1, x\}\) 已正交(\(\langle 1, x \rangle = 0\)\([-1,1]\) 上)。

\[\text{proj}_W(x^2) = \frac{\langle x^2, 1 \rangle}{\langle 1, 1 \rangle} \cdot 1 + \frac{\langle x^2, x \rangle}{\langle x, x \rangle} \cdot x = \frac{2/3}{2} + 0 = \frac{1}{3}\]

陷阱

陷阱 1:正交 ⟹ 線性獨立?

非零向量才成立!

  • 非零正交向量組必線性獨立 ✓
  • \(\mathbf{0}\) 與任何向量正交,但 \(\{\mathbf{0}\}\) 線性相依 ✗

證明(非零情況):設 \(c_1\mathbf{v}_1 + \cdots + c_k\mathbf{v}_k = \mathbf{0}\)

\(\langle \cdot, \mathbf{v}_i \rangle\)\(c_i \|\mathbf{v}_i\|^2 = 0\)

\(\mathbf{v}_i \neq \mathbf{0} \Rightarrow c_i = 0\)

重要澄清:線性獨立 \(\not\Rightarrow\) 正交!\((1, 0)\)\((1, 1)\) 線性獨立但不正交。

陷阱 2:(U⊥)⊥ = U?

有限維成立,無限維不一定!

\(\mathbb{R}^n\) 中:\((U^\perp)^\perp = U\)(對任何子空間 \(U\)

在無限維(如 \(\ell^2\)):\(U \subseteq (U^\perp)^\perp\),但可能嚴格包含

陷阱 3:勾股定理

\[\|\mathbf{u} + \mathbf{v}\|^2 = \|\mathbf{u}\|^2 + \|\mathbf{v}\|^2 \Leftrightarrow \mathbf{u} \perp \mathbf{v}\]

只在正交時成立! 一般情況:

\[\|\mathbf{u} + \mathbf{v}\|^2 = \|\mathbf{u}\|^2 + 2\langle \mathbf{u}, \mathbf{v} \rangle + \|\mathbf{v}\|^2\]

陷阱 4:Cauchy-Schwarz 等號條件

\[|\langle \mathbf{u}, \mathbf{v} \rangle| \leq \|\mathbf{u}\| \|\mathbf{v}\|\]

等號 \(\Leftrightarrow \mathbf{u}, \mathbf{v}\) 線性相依(不只是平行,包含其一為零)


題目

基礎題(大二)

8.1\(\mathcal{P}_2\) 上,內積 \(\langle p, q \rangle = \int_0^1 p(x)q(x)dx\)。對 \(\{1, x, x^2\}\) 做 Gram-Schmidt。

解答

計算內積(在 \([0,1]\) 上):

\(\langle 1, 1 \rangle = 1\)

\(\langle x, 1 \rangle = \frac{1}{2}\)

\(\langle x, x \rangle = \frac{1}{3}\)

\(\langle x^2, 1 \rangle = \frac{1}{3}\)

\(\langle x^2, x \rangle = \frac{1}{4}\)

Gram-Schmidt

\(u_1 = 1\)

\(u_2 = x - \frac{1/2}{1} \cdot 1 = x - \frac{1}{2}\)

\(\langle u_2, u_2 \rangle = \langle x - \frac{1}{2}, x - \frac{1}{2} \rangle = \frac{1}{3} - \frac{1}{2} + \frac{1}{4} = \frac{1}{12}\)

\(\langle x^2, u_1 \rangle = \frac{1}{3}\)\(\langle x^2, u_2 \rangle = \langle x^2, x - \frac{1}{2} \rangle = \frac{1}{4} - \frac{1}{6} = \frac{1}{12}\)

\(u_3 = x^2 - \frac{1/3}{1} - \frac{1/12}{1/12}(x - \frac{1}{2}) = x^2 - \frac{1}{3} - x + \frac{1}{2} = x^2 - x + \frac{1}{6}\)

正交基底\(\{1, x - \frac{1}{2}, x^2 - x + \frac{1}{6}\}\)

8.2\(V = \mathbb{R}^3\)\(W = \text{span}\{(1,1,0), (1,0,1)\}\)。求 \((1,2,3)\)\(W\) 上的正交投影。

解答

先正交化 \(W\) 的基底。

\(\mathbf{u}_1 = (1,1,0)\)

\(\mathbf{u}_2 = (1,0,1) - \frac{\langle (1,0,1), (1,1,0) \rangle}{\|(1,1,0)\|^2}(1,1,0) = (1,0,1) - \frac{1}{2}(1,1,0) = (\frac{1}{2}, -\frac{1}{2}, 1)\)

投影: $\(\text{proj}_W(\mathbf{v}) = \frac{\langle \mathbf{v}, \mathbf{u}_1 \rangle}{\|\mathbf{u}_1\|^2}\mathbf{u}_1 + \frac{\langle \mathbf{v}, \mathbf{u}_2 \rangle}{\|\mathbf{u}_2\|^2}\mathbf{u}_2\)$

\(\langle (1,2,3), (1,1,0) \rangle = 3\)\(\|(1,1,0)\|^2 = 2\)

\(\langle (1,2,3), (\frac{1}{2}, -\frac{1}{2}, 1) \rangle = \frac{1}{2} - 1 + 3 = \frac{5}{2}\)

\(\|(\frac{1}{2}, -\frac{1}{2}, 1)\|^2 = \frac{1}{4} + \frac{1}{4} + 1 = \frac{3}{2}\)

\[\text{proj} = \frac{3}{2}(1,1,0) + \frac{5/2}{3/2}(\frac{1}{2}, -\frac{1}{2}, 1) = (\frac{3}{2}, \frac{3}{2}, 0) + \frac{5}{3}(\frac{1}{2}, -\frac{1}{2}, 1)\]
\[= (\frac{3}{2} + \frac{5}{6}, \frac{3}{2} - \frac{5}{6}, \frac{5}{3}) = (\frac{7}{3}, \frac{2}{3}, \frac{5}{3})\]

8.3 證明:若 \(A\)\(n \times n\) 實對稱矩陣,不同特徵值的特徵向量正交。

解答

\(A\mathbf{v}_1 = \lambda_1\mathbf{v}_1\)\(A\mathbf{v}_2 = \lambda_2\mathbf{v}_2\)\(\lambda_1 \neq \lambda_2\)

\[\lambda_1\langle \mathbf{v}_1, \mathbf{v}_2 \rangle = \langle A\mathbf{v}_1, \mathbf{v}_2 \rangle = \langle \mathbf{v}_1, A^T\mathbf{v}_2 \rangle = \langle \mathbf{v}_1, A\mathbf{v}_2 \rangle = \lambda_2\langle \mathbf{v}_1, \mathbf{v}_2 \rangle\]

\((\lambda_1 - \lambda_2)\langle \mathbf{v}_1, \mathbf{v}_2 \rangle = 0\)

\(\lambda_1 \neq \lambda_2 \Rightarrow \langle \mathbf{v}_1, \mathbf{v}_2 \rangle = 0\) \(\square\)


進階題(大三/碩一)

8.4\(\langle p, q \rangle = p(0)q(0) + p(1)q(1) + p(2)q(2)\)\(\mathcal{P}_2\) 上。 (a) 證明這是合法內積 (b) 求 \(\{1, x, x^2\}\) 的正交基底

解答

(a) 對稱性、線性性顯然。

正定性:\(\langle p, p \rangle = p(0)^2 + p(1)^2 + p(2)^2 \geq 0\)

等號 \(\Leftrightarrow p(0) = p(1) = p(2) = 0\)

\(p \in \mathcal{P}_2\)(次數 \(\leq 2\))有三個根 \(\Rightarrow p \equiv 0\)

(b) \(\langle 1, 1 \rangle = 1 + 1 + 1 = 3\)

\(\langle x, 1 \rangle = 0 + 1 + 2 = 3\)

\(\langle x, x \rangle = 0 + 1 + 4 = 5\)

\(\langle x^2, 1 \rangle = 0 + 1 + 4 = 5\)

\(\langle x^2, x \rangle = 0 + 1 + 8 = 9\)

\(u_1 = 1\)

\(u_2 = x - \frac{3}{3} \cdot 1 = x - 1\)

\(\langle u_2, u_2 \rangle = 1 + 0 + 1 = 2\)

\(\langle x^2, u_2 \rangle = 0 + 0 + 4 = 4\)

\(u_3 = x^2 - \frac{5}{3} - \frac{4}{2}(x-1) = x^2 - 2x + \frac{1}{3}\)

8.5 證明 Cauchy-Schwarz 不等式:\(|\langle \mathbf{u}, \mathbf{v} \rangle| \leq \|\mathbf{u}\| \|\mathbf{v}\|\)

解答

\(\mathbf{v} = \mathbf{0}\),不等式顯然。

\(\mathbf{v} \neq \mathbf{0}\)。對任意 \(t \in \mathbb{R}\)

\[0 \leq \|\mathbf{u} - t\mathbf{v}\|^2 = \|\mathbf{u}\|^2 - 2t\langle \mathbf{u}, \mathbf{v} \rangle + t^2\|\mathbf{v}\|^2\]

\(t = \frac{\langle \mathbf{u}, \mathbf{v} \rangle}{\|\mathbf{v}\|^2}\)

\[0 \leq \|\mathbf{u}\|^2 - \frac{\langle \mathbf{u}, \mathbf{v} \rangle^2}{\|\mathbf{v}\|^2}\]

\(\Rightarrow \langle \mathbf{u}, \mathbf{v} \rangle^2 \leq \|\mathbf{u}\|^2 \|\mathbf{v}\|^2\) \(\square\)

8.6\(Q\)\(n \times n\) 正交矩陣。證明 \(\langle Q\mathbf{u}, Q\mathbf{v} \rangle = \langle \mathbf{u}, \mathbf{v} \rangle\)(正交變換保內積)。

解答

$\(\langle Q\mathbf{u}, Q\mathbf{v} \rangle = (Q\mathbf{u})^T(Q\mathbf{v}) = \mathbf{u}^T Q^T Q \mathbf{v} = \mathbf{u}^T I \mathbf{v} = \langle \mathbf{u}, \mathbf{v} \rangle\)$ \(\square\)


概念關聯

內積空間
    ├──→ 範數:‖v‖ = √⟨v,v⟩
    ├──→ 正交:⟨u,v⟩ = 0
    │        │
    │        └──→ 正交基底 → Gram-Schmidt
    │                        │
    │                        └──→ QR 分解:A = QR
    └──→ 投影:proj_W(v) = Σ ⟨v,uᵢ⟩/‖uᵢ‖² · uᵢ
             └──→ 最小二乘法:A^T A x̂ = A^T b

速查:Gram-Schmidt 公式

\[\mathbf{u}_1 = \mathbf{v}_1\]
\[\mathbf{u}_k = \mathbf{v}_k - \sum_{j=1}^{k-1} \frac{\langle \mathbf{v}_k, \mathbf{u}_j \rangle}{\langle \mathbf{u}_j, \mathbf{u}_j \rangle} \mathbf{u}_j\]
\[\mathbf{e}_k = \frac{\mathbf{u}_k}{\|\mathbf{u}_k\|}\]

← 上一題:四個子空間 | 返回目錄 | 下一題:對角化 →