B5 Notebook
高等代数与几何
高等微积分
College Mathematics
Complete Handwritten Lecture Notes & Solved Problems
Part I: Algebra, Geometry, and Advanced Calculus I
Vector spaces, linear independence, basis and dimension coordinates, linear subspaces, sums and direct sums.
Page 1
Page 2
No.
Data
山穷水复疑无路
柳暗花明又一村
Page 3
Linear Space
- Axiomatic Definition of Linear Space
- V: Nonempty set
- F: Field
: Scalar multiplication is a unary operation on F
: (λ, v) → λv ∈ V
- ∀v, w ∈ V, ∀λ, μ ∈ F. For F with respect to the unit element, there is:
- 1v = v
- λ(μv) = (λμ)v
- (λ + μ)v = λv + μv
- λ(v + w) = λv + λw
- λ(v + w) = λv + λw
- V forms a linear space under addition and scalar multiplication in the field F.
- V is a linear space over F. V(F).
- Basis and Dimension of Linear Space
- Linear Independence: V(F) is a linear space. v1, v2, ..., vm ∈ V. If all λ1, λ2, ..., λm ∈ F such that (λ1, ..., λm) (v1 v2 vm) = 0
- Then v1, ..., vm are linearly dependent.
- If v1, ..., vm are linearly dependent, then one of them can be expressed as a linear combination of the others in the field F.
- If v1, ..., vm are linearly independent, then none of them can be expressed as a linear combination of the others in the field F.
Theorem: If a set of vectors {v1, ..., vm} is linearly independent, then any subset of it is also linearly independent.
- If a set of vectors {v1, ..., vm} is linearly dependent, then any subset of it is also linearly dependent.
- If a set of vectors {v1, ..., vm} is linearly dependent, then there exists a non-zero λ1, ..., λm such that λ1v1 + ... + λmv = 0. Otherwise, there would be a non-zero λ1, ..., λm such that λ1v1 + ... + λmv = 0.
Page 4
Theorem: If the vector set {B1, ..., Bs} of a vector space V(F) can be linearly expressed by another vector set {α1, ..., αr}, and if s > r, then {B1, ..., Bs} is linearly dependent. If {B1, ..., Bs} is linearly independent, then s ≤ r.
Proof: 1) Suppose B0 = ∑j=1s λij αj (where λij ∈ F, j = 1, 2, ..., s), and suppose x1B1 + x2B2 + ... + xsBs = 0. Then ∑j=1s xjBj = ∑j=1s (∑i=1r λij xj)αj = ∑i=1r (∑j=1s λij xj)αi = 0. Since all xi (i = 1, ..., r) have coefficients of 0, ∑j=1s λij xj = 0 (r equations). Therefore, s > r. Since the system has a non-trivial solution, {B1, ..., Bs} is linearly dependent.
- The converse of 1).
- Dimension and basis of a linear space.
If the linear space V(F) has a finite subset B = {α1, ..., αn} that is linearly independent, and L(B) = V, then B is a basis of V, n is the dimension of V, i.e., dimV = n. Proof: If V(F) has linearly independent subsets B1 = {α1, ..., αn} and B2 = {β1, ..., βm} that can span V, then βj ∈ V(F) = L(α1, ..., αn) (j = 1, ..., m), so m ≤ n. Therefore, n ≤ m. Thus, m = n. Therefore, dimV = n.
- Rank of a vector set.
If S is a subset of V(F), and S has a linearly independent subset B = {α1, ..., αr}, and each element of S can be linearly expressed by B, then the number of elements in B is the rank of S, rankS = r. The method for finding the maximum linearly independent set: row reduction of the matrix to echelon form.
Page 5
The coordinates of (a1, ..., an) ∈ Fn in basis B are:
B1 = {q1, ..., qn}, B2 = {p1, ..., pn} is a basis of V.
Let:
$$\begin{cases} \vec{p}_1 = a_{n1}\vec{q}_1 + ... + a_{n1}\vec{q}_n \\ \vec{p}_2 = a_{n2}\vec{q}_1 + ... + a_{n2}\vec{q}_n \\ \vdots \\ \vec{p}_n = a_{nn}\vec{q}_1 + ... + a_{nn}\vec{q}_n \end{cases}$$
Then:
B2 = AB1
B1 = B2A
If:
$$\vec{x} = B_1\begin{pmatrix} x_1 \\ \vdots \\ x_n \end{pmatrix} = B_1\vec{X}, \quad \vec{y} = B_2\begin{pmatrix} y_1 \\ \vdots \\ y_n \end{pmatrix} = B_2\vec{Y}$$
Then:
B1X = B2Y = B1AY
∴ X = AY, Y = A-1X
- Linear Subspace:
If W is a subspace of V, and W is closed under the operations of V, then W is a subspace of V.
Subspace: rank W = rank V
Non-subspace: rank W < rank V
Theorem: If W is a subspace of V, then W can be extended to a basis of V (i.e., W can be added to V to form a basis of V).
Proof: Let W's basis B = {x1, ..., xm}, and V's basis be {x1, ..., xn}. If there exists x_{m+1} ∈ V such that {x1, ..., xm, x_{m+1}} is linearly independent, then dim V < n, which contradicts the assumption. If m+1 = n, then {x1, ..., xm, x_{m+1}, ..., xn} is linearly independent and forms a basis of V.
Theorem: A non-empty subset W of V is a subspace of V if and only if W is closed under the operations of V.
Proof: Necessity: If not closed, then W is not a subspace of V.
Page 6
Sufficiency: Since W is a subset of V, the rules of scalar multiplication and addition in V hold for W. We need to show that the zero vector 0 ∈ W and the additive inverse -x ∈ W for each x ∈ W. Therefore, W is closed under scalar multiplication. When λ = 0 or λ = 1, 0.x = 0 ∈ W and (-1)x = -x ∈ W. Thus, W is a subspace of V.
- Subspace intersection and sum
Intersection and direct sum:
Intersection: W1 ∩ W2 = {x | x ∈ W1 and x ∈ W2}
Sum: W1 + W2 = {x | x = x1 + x2, x1 ∈ W1, x2 ∈ W2}
W1 ∩ W2 and W1 + W2 are still subspaces of V.
Proof:
- Let x, y ∈ W1 and z ∈ W2. Since W1 is a subspace, λx + μy ∈ W1. Similarly, λx + μy ∈ W2. Therefore, λx + μy ∈ W1 ∩ W2.
- Let x, y ∈ W1 + W2. Then there exist x1, y1 ∈ W1 and x2, y2 ∈ W2 such that x = x1 + x2 and y = y1 + y2. Thus, x + y = (x1 + x2) + (y1 + y2) = (x1 + y1) + (x2 + y2) ∈ W1 + W2. λx = λ(x1 + x2) = λx1 + λx2 ∈ W1 + W2. Therefore, W1 + W2 is a subspace of V.
Direct sum: W1 and W2 are subspaces of V. If W1 ∩ W2 = {0}, then W1 + W2 is called the direct sum of W1 and W2, denoted as W1 ⊕ W2.
Dimension formula: dim(W1 + W2) = dim(W1) + dim(W2) - dim(W1 ∩ W2)
If W1 + W2 is a direct sum, then dim(W1) + dim(W2) = dim(W1 + W2).
Proof: Assume dim(W1) = s, dim(W2) = t, and dim(W1 ∩ W2) = r.
Page 7
Let \( W_1 \cap W_2 = L(\vec{\alpha}_1, \vec{\alpha}_2, \ldots, \vec{\alpha}_r) \)
\( W_1 = L(\vec{\alpha}_1, \vec{\alpha}_2, \ldots, \vec{\alpha}_r, \vec{\beta}_1, \ldots, \vec{\beta}_{s-r}) \)
\( W_2 = L(\vec{\alpha}_1, \ldots, \vec{\alpha}_r, \vec{\beta}_1, \ldots, \vec{\beta}_{t-r}) \)
\(\therefore W_1 + W_2 = L(\vec{\alpha}_1, \ldots, \vec{\alpha}_r, \vec{\beta}_1, \ldots, \vec{\beta}_{s-r}, \vec{\beta}_1, \ldots, \vec{\beta}_{t-r})\). Just need to ensure \(\dim(W_1 + W_2) = s + t - r\).
Let \( a_1 \vec{\alpha}_1 + \ldots + a_r \vec{\alpha}_r + b_1 \vec{\beta}_1 + \ldots + b_{s-r} \vec{\beta}_{s-r} + c_1 \vec{\beta}_1 + \ldots + c_t \vec{\beta}_{t-r} = \vec{0} \)
That is \( a_1 \vec{\alpha}_1 + \ldots + a_r \vec{\alpha}_r + b_1 \vec{\beta}_1 + \ldots + b_{s-r} \vec{\beta}_{s-r} = -c_1 \vec{\beta}_1 - \ldots - c_t \vec{\beta}_{t-r} \)
The two sides of the equation belong to \( W_1 \) and \( W_2 \). Since there is an equal sign, they both belong to \( W_1 \cap W_2 \).
\(\therefore -c_1 \vec{\beta}_1 - \ldots - c_t \vec{\beta}_{t-r} = a_1 \vec{\alpha}_1 + \ldots + a_r \vec{\alpha}_r \).
That is \( c_1 \vec{\beta}_1 + \ldots + c_t \vec{\beta}_{t-r} + a_1 \vec{\alpha}_1 + \ldots + a_r \vec{\alpha}_r = \vec{0} \).
The above equation is a linear combination of the basis of \( W_2 \). Since \( a_1 = \ldots = a_r = c_1 = \ldots = c_t = 0 \).
\(\therefore a_1 \vec{\alpha}_1 + \ldots + a_r \vec{\alpha}_r + b_1 \vec{\beta}_1 + \ldots + b_{s-r} \vec{\beta}_{s-r} = \vec{0} \)
The above equation is a linear combination of the basis of \( W_1 \). Since \( a_1 = \ldots = a_r = b_1 = \ldots = b_{s-r} = 0 \).
\(\therefore \vec{\alpha}_1, \ldots, \vec{\alpha}_r, \vec{\beta}_1, \ldots, \vec{\beta}_{s-r}, \vec{\beta}_1, \ldots, \vec{\beta}_{t-r}\) are linearly independent.
Find \( S_1 + S_2 \): Find the bases of \( S_1 \) and \( S_2 \), then solve the maximal linearly independent set of \( S_1 \) and \( S_2 \).
Find \( S_1 \cap S_2 \): Find the basis of \( S_1 \cap S_2 \).
Find the complement of \( S_1 \): First determine the dimension of the complement space, then use the determinant to determine the basis of the complement space.
Page 8
Section 1: Definition
A linear transformation \( T \) from a vector space \( V \) over a field \( F \) to another vector space \( W \) over the same field \( F \) is a function \( T: V \to W \) such that for all \( u, v \in V \) and all \( \lambda \in F \), the following properties hold:
- \( T(u + v) = T(u) + T(v) \)
- \( T(\lambda u) = \lambda T(u) \)
A linear transformation \( T \) from a vector space \( V \) to itself is called a linear operator on \( V \).
A linear transformation \( f \) from a vector space \( V \) to a field \( F \) is called a linear functional on \( V \).
A mapping \( f \) from \( V \times V \) to \( F \) is linear if it satisfies:
- \( f(\alpha u + \beta v, w) = \alpha f(u, w) + \beta f(v, w) \)
- \( f(u, \alpha v + \beta w) = \alpha f(u, v) + \beta f(u, w) \)
A linear transformation \( T \) from \( V^n \) to \( F \) is called a multilinear function on \( V \).
Section 2: Image and Kernel of a Linear Transformation
Given a linear transformation \( T \) from \( V_1(F) \) to \( V_2(F) \), the image of \( T \) (the range) is defined as:
\[ \text{Im}(T) = \{ T(v) \mid v \in V_1 \} \]
The kernel of \( T \) (the set of all vectors in \( V_1 \) that map to the zero vector in \( V_2 \)) is defined as:
\[ \text{Ker}(T) = \{ v \in V_1 \mid T(v) = 0 \} \]
The kernel of \( T \) is also written as \( T^{-1}(0) \).
Page 9
- Linear Mapping Operations: Multiplication, Addition, Scalar Multiplication.
- The product of linear mappings and the inverse of a linear mapping is a linear mapping.
- V1 → V2, V2 → V3. Then V1 → V3. Let x, y ∈ V1.
Proof: α₂α₁(λx + μy) = α₂(λα₁(x) + μα₁(y))
= λ(α₂(α₁(x))) + μ(α₂(α₁(y)))
= λ(α₂α₁(x) + μα₂α₁(y)).
- V1 → V2, V2 → V1, σ⁻¹σ = I_V1, σσ⁻¹ = I_V2. Let x, y ∈ V2.
Proof: σ⁻¹(λ₁x + λ₂y) = σ⁻¹[λ₁(σ⁻¹(x)) + λ₂(σ⁻¹(y))]
= σ⁻¹[λ₁σ⁻¹(x) + λ₂σ⁻¹(y)]
= λ₁σ⁻¹(x) + λ₂σ⁻¹(y).
- The set of all linear mappings from V1 to V2 forms a linear space L(V1, V2).
Addition: (σ + τ)(x) = σ(x) + τ(x)
Scalar multiplication: λσ(x) = λσ(x)
Proof: (σ + τ)(λx₁ + λ₂x₂) = σ(λx₁ + λ₂x₂) + τ(λx₁ + λ₂x₂)
= λ(σ + τ)(x₁) + λ(σ + τ)(x₂).
(λσ)(λ₁x₁ + λ₂x₂) = λ(λ₁σ(x₁) + λ₂σ(x₂))
= λ(λ₁σ(x₁) + λ₂σ(x₂)).
∴ L(V1, V2) forms a linear space under addition and scalar multiplication.
Furthermore, the linear space of linear mappings is isomorphic to the linear space of linear mappings from V1 to V2. The number of linear mappings from V1 to V2 is the product of the dimension of the linear space of linear mappings from V1 to V2 and the dimension of the linear space of linear mappings from V1 to V2.
That is, kerσ has rank V1 possibilities, Imσ has rank V2 possibilities.
- L(V1, V2) is an abelian group. L(V1, V2) is a vector space. Linear transformation L(V1, V2) is an abelian group.
Page 10
- Rank of Linear Mappings
def: r(σ) = dim(σ(V1))
$$\begin{array}{c|c|c}
\hline
\text{Kero} & \sigma & \text{Imo} \\
\hline
n & m & \text{Rank} \\
\hline
\end{array}$$
$$\therefore \text{rank}(\sigma) + \text{dim}(\text{Kero}) = n$$
Theorem: Let V1, V2, V3 be m, n, s-dimensional linear spaces, σ ∈ L(V1, V2), τ ∈ L(V2, V3). Then rank(σ) + rank(τ) - n ≤ rank(τσ) ≤ min(rank(σ), rank(τ)).
Proof: Right side: Since σ(V1) ⊂ V2, (τσ)(V1) ⊂ τ(V2)
$$\therefore \text{dim}(\tau\sigma)(V_3) \leq \text{dim} \tau(V_2) \therefore \text{rank}(\tau\sigma) \leq \text{rank}(\tau)$$
Also: τσ(V1) = τ(σ(V1)) ∴ dim(τσ)(V1) ≤ dim(σ(V1))
∴ rank(τσ) ≤ rank(σ).
∴ rank(τσ) ≤ min(rank(σ), rank(τ)).
Left side: ∴ rank(σ) + dim ker(σ) = m,
rank(τ) + dim ker(τ) = n,
rank(τσ) + dim ker(τσ) = m.
∴ rank(τσ) + dim ker(τσ) = rank(σ) + dim ker(σ).
∴ rank(τσ) = rank(σ) + dim ker(σ) - dim ker(τσ).
∵ ker(σ) ∩ σ(V1) = {0},
∴ rank(τ(ker(σ))) ≤ dim ker(σ).
Also: rank(τ(ker(σ))) = rank(τ) - rank(τσ)
∴ rank(τσ) = rank(σ) + dim ker(σ) - dim ker(τσ) ≥ rank(σ) + rank(τ) - rank(τσ)
- dim ker(τσ)
= rank(σ) + rank(τ) - n.
Page 11
Theorem: If σ, τ ∈ L(V1, V2), then rank(σ + τ) ≤ rank(σ) + rank(τ).
Proof: Since β ∈ (σ + τ)(V1), there exists x ∈ V1 such that β = (σ + τ)(x) = σ(x) + τ(x) ∈ σ(V1) + τ(V1). Therefore, (σ + τ)(V1) ⊆ σ(V1) + τ(V1). Hence, dim(σ + τ)(V1) ≤ dim[σ(V1) + τ(V1)] ≤ dim σ(V1) + dim τ(V1).
Linear mapping and its basis image relationship:
- Uniqueness: Two different linear mappings acting on the basis have different images.
Proof: If σ(αi) = τ(αi) for i = 1, ..., n, and B = {α1, ..., αn} is a basis of V1, then ∀ξ = x1α1 + ... + xnαn ∈ V1, we have σ(ξ) = x1σ(α1) + ... + xnσ(αn) = x1τ(α1) + ... + xnτ(αn) = τ(x1α1 + ... + xnαn) = τ(ξ). Thus, σ = τ.
- Linearity: If B = {α1, ..., αn} is a basis of V1 and S = {β1, ..., βm} is a basis of V2, then σ(αi) = βi is linear.
Proof: For ξ = x1α1 + ... + xnαn ∈ V1, define σ(ξ) = x1β1 + ... + xnβn. Since ∀ξ1 = b1α1 + ... + bnαn, ξ2 = c1α1 + ... + cnαn and ∀λ1, λ2 ∈ F, we have σ(λ1ξ1 + λ2ξ2) = σ((λ1b1 + λ2c1)α1 + (λ1b2 + λ2c2)α2 + ... + (λ1bn + λ2cn)αn) = (λ1b1 + λ2c1)β1 + ... + (λ1bn + λ2cn)βn = λ1(b1β1 + ... + bnβn) + λ2(c1β1 + ... + cnβn) = λ1σ(ξ1) + λ2σ(ξ2).
- Linear mapping model - solving linear equation systems.
- Homogeneous equation AX = 0.
X ∈ ker(σ) (∼ ker(A)).
dim ker(σ) = n - r.
Page 12
The equation (*) can be expressed as:
$$
x_1\begin{bmatrix}a_{11}\\ \vdots \\ a_{m1}\end{bmatrix} + x_2\begin{bmatrix}a_{12}\\ \vdots \\ a_{m2}\end{bmatrix} + \cdots + x_n\begin{bmatrix}a_{1n}\\ \vdots \\ a_{mn}\end{bmatrix} = \begin{bmatrix}0\\ \vdots \\ 0\end{bmatrix}.
$$
The null space of A is a subspace of R(A), N(A) ⊆ R(A), dim N(A) ≤ dim R(A). In fact, dim N(A) = n - r, dim R(A) = r.
(*) can also be expressed as:
$$(a_{i1}, \ldots, a_{in})\begin{pmatrix}x_1\\ \vdots \\ x_n\end{pmatrix} = 0 \quad j=1,2,\ldots,m, \quad \therefore (a_i, x) = 0 \cdot R(A^T) \subseteq N(A)$$
This is a orthogonal subspace. N(A) = (R(A^T))^T.
Hugo diagram:
$$
\begin{array}{c|c|c}
N(A) & \vdots & R(A) \\
\hline
R(A^T) & \vdots & \vdots \\
\hline
n & \vdots & m \\
\end{array}
$$
Page 13
- Non-homogeneous system AX = B
- Multiple solutions:
$$
\begin{cases}
\overline{x}_0 \\
\vdots \\
\overline{x}_r
\end{cases}
$$
B: V2(mnR) vector, X: A's solution space N(A), homogeneous system's general solution.
X0: Non-homogeneous system's particular solution, r rank (as shown in figure, each rank R(A^T) has a corresponding X0 with B).
- Unique solution:
$$
\overline{x}_0
$$
n = m. A is an invertible matrix, there is a unique solution X0 corresponding to B, X0 = A^(-1)B.
- No solution:
$$
\overline{x}_0
$$
No X0 in V1 corresponds to B in V2.
Note: For mxn matrix A, reduce it to row echelon form (row operations: multiplication, addition, swapping). The r x r part is the rank part of the transformation, the m-r rows of 0 vectors correspond to the kernel part of V1. The n-r columns correspond to the rank part of the transformation in V2 (the part that is not mapped to zero).
Page 14
III. Linear Mapping 2 - Linear Transformation and Similarity Standard Form.
- Similarity Matrix Equivalence Class.
For a linear transformation $\sigma \in L(V, V)$, given a basis of the vector space $V$, $\sigma$ corresponds to a specific matrix $A$. For different bases of $V$, $\sigma$ generally corresponds to different matrices. These matrices corresponding to the same linear transformation $\sigma$ form an equivalence class.
Given $\sigma \in L(V, V)$, $B_1 = \{\vec{x}_1, ..., \vec{x}_n\}$, $B_2 = \{\vec{p}_1, ..., \vec{p}_n\}$ are two bases of $V$.
- If the basis transformation matrix is $C: (\vec{B}_1, ..., \vec{B}_n) = (\vec{x}_1, ..., \vec{x}_n)C$, $C$ is invertible. Then $(\vec{a}_1, ..., \vec{a}_n) = (\vec{B}_1, ..., \vec{B}_n)C^{-1}$.
Let $\sigma$ in $B_1$ correspond to the matrix $A \in M_n(F)$.
Since linear transformation (mapping) can be expressed in the form $\sigma(\vec{x}_1, ..., \vec{x}_n) = (\vec{x}_1, ..., \vec{x}_n)A$,
$\therefore \sigma((\vec{B}_1, ..., \vec{B}_n)C^{-1}) = ((\vec{B}_1, ..., \vec{B}_n)C^{-1})A$,
$\therefore (\sigma(\vec{B}_1, ..., \vec{B}_n))C^{-1} = (\vec{B}_1, ..., \vec{B}_n)(C^{-1}A)$,
$\therefore \sigma(\vec{B}_1, ..., \vec{B}_n) = (\vec{B}_1, ..., \vec{B}_n)(C^{-1}AC)$.
Thus, $\sigma$ in $B_2$ corresponds to the matrix $C^{-1}AC$.
- Conversely, if $\sigma$ in the basis $B_1 = \{\vec{x}_1, ..., \vec{x}_n\}$ corresponds to the matrix $A$, then $C^{-1}AC$ is $\sigma$'s matrix in another basis $\{\vec{p}_1, ..., \vec{p}_n\}$, where the basis $\{\vec{p}_1, ..., \vec{p}_n\}$ is obtained by multiplying the basis $\{\vec{x}_1, ..., \vec{x}_n\}$ by the invertible matrix $C$ from the right.
- Reciprocity: $\forall A, A \in M_n(F), A \sim A: A = E^{-1}AE$.
Symmetry: $\forall A, B \in M_n(F)$, if $A \sim B$, then $B \sim A$; if $A = C^{-1}BC$, then $B = (C^{-1})^{-1}AC^{-1}$.
Transitivity: $\forall A_1, A_2, A_3 \in M_n(F)$. If $A_1 \sim A_2, A_2 \sim A_3$, then $A_1 \sim A_3$:
If $A_1 = C_1^{-1}A_1C_1$ and $A_2 = C_2^{-1}A_2C_2$, then $A_1 = C_1C_2^{-1}A_2C_2C_1 = (C_2C_1)^{-1}A_2(C_2C_1)$.
- Does the similarity matrix equivalence class have a simplest representative (diagonal matrix)? - Conditions and methods.
- First, we determine the goal: if $\sigma$ in the basis $B = \{\vec{x}_1, ..., \vec{x}_n\}$ corresponds to the matrix $A$ (non-diagonal matrix), we hope to find the corresponding $C$ such that $A = C^{-1}AC$, $\Lambda$ is a diagonal matrix $\begin{bmatrix} \lambda_1 & & \\ & \ddots & \\ & & \lambda_n \end{bmatrix}$ ($\Lambda = C^{-1}AC$).
Page 15
- C is a transition matrix of a basis. Given B = {α1, ..., αn}, the corresponding basis for which C can be expressed as a diagonal matrix is (α1, ..., αn)C*. In this basis, we have:
$$\sigma((\alpha_1, ..., \alpha_n)C^*) = ((\alpha_1, ..., \alpha_n)C^*) \cdot \begin{bmatrix} \lambda_1 \\ \vdots \\ \lambda_n \end{bmatrix}$$
Let (α1, ..., αn)C* = (e1, ..., en), then (e1, ..., en)C = (α1, ..., αn).
$$\sigma(e_1, ..., e_n) = (e_1, ..., e_n) \begin{bmatrix} \lambda_1 \\ \vdots \\ \lambda_n \end{bmatrix} = (\lambda_1 e_1, ..., \lambda_n e_n)$$
Therefore, (σ(e1), ..., σ(en)) = (λ1e1, ..., λnen)
Thus, σ(ei) = λiei ⇒ (σ - λiI)ei = 0
Written in matrix form: (A - λiE)X = 0. For each λi, Xi is a solution, so we can write (A - λiE)X = 0
For this to have a solution (the basis for which C can be expressed as a diagonal matrix), ker(A - λiE) ≠ {0}, so |A - λiE| = 0. This is a characteristic equation, known as the characteristic equation, and the solutions are called eigenvalues.
By the fundamental theorem of algebra, λ has n solutions λ1, ..., λn.
- Since λi has real or complex roots, substituting λ into (A - λiE)X = 0 will produce a system of linear equations with real or complex coefficients. Since our goal is to find real eigenvectors (e1, ..., en) (basis of Vn), we need to determine some conditions to achieve this goal.
First, assume the solutions λ1, ..., λn are all real. Substituting λ into (A - λiE)X = 0, since λ1, ..., λn may have repeated roots, we assume there are m distinct λs (m ≤ n): λn, ..., λm. Therefore, we can get m equations (A - λnE)X = 0, ..., (A - λmE)X = 0. These are all linear equations.
Since A ∈ Mn(F), the solutions of each (m) linear equations are all subspaces of Vn. Therefore, the decomposition of W = V1 + ... + Vm is a direct sum.
- Below is the proof that these subspaces {V1, ..., Vm} are a direct sum. We need to show that the zero vector in W is the only vector in the intersection of all Vj. That is,
$$\sum_{j=1}^{m} \overline{V_j} = 0$$
where $$\overline{V_j} \in V_j$$
Page 16
Induction:
- When m=2, assume $\vec{\xi}_1 + \vec{\xi}_2 = \vec{0}$. Then $0(\vec{\xi}_1 + \vec{\xi}_2) = \lambda_1\vec{\xi}_1 + \lambda_2\vec{\xi}_2 = (\lambda_1 - \lambda_2)\vec{\xi}_1 = \vec{0}$. Since $\vec{\xi}_1$ and $\vec{\xi}_2$ are obtained from different eigenvalues of $(A-\lambda E)x = \vec{0}$, $\lambda_1 \neq \lambda_2$. Thus, $\vec{\xi}_1 = \vec{0} \Rightarrow \vec{\xi}_2 = \vec{0}$.
- Assume the statement is true for m-1. Consider the case of m eigenvectors. Assume $\vec{\xi}_1 + \dots + \vec{\xi}_m = \vec{0}$. Then $\vec{\xi}_m = -(\vec{\xi}_1 + \dots + \vec{\xi}_{m-1})$. Thus, $0(\vec{\xi}_1 + \dots + \vec{\xi}_m) = \lambda_1\vec{\xi}_1 + \dots + \lambda_m\vec{\xi}_m = \vec{0}$. Since $(\lambda_1 - \lambda_m)\vec{\xi}_1 + \dots + (\lambda_{m-1} - \lambda_m)\vec{\xi}_{m-1} = \vec{0}$, by the induction hypothesis, $\vec{\xi}_1 = \dots = \vec{\xi}_{m-1} = \vec{0}$. Thus, $\vec{\xi}_m = \vec{0}$.
From the above, the zero vector of W can only be decomposed into the zero vector of $V_{\lambda_i}$ (i=1,...,m). Thus, W is trivial.
- Our goal is to express the solution of $(A-\lambda E)x = \vec{0}$ as a linear combination of $\vec{e}_1, \dots, \vec{e}_n$. Since all vectors are eigenvectors, the sum of their dimensions is n, which means the sum of the dimensions of all $V_{\lambda_i}$ is n. Thus, we need to find the condition such that $\dim(V_{\lambda_1} + \dots + V_{\lambda_m}) = \dim V_{\lambda_1} + \dots + \dim V_{\lambda_m} = n$.
The condition is: when the roots of $|A-\lambda E| = 0$ are not complex, each $\lambda_i$ has a geometric multiplicity equal to the corresponding dimension of the eigenspace $\dim V_{\lambda_i}$.
- Considering the above condition, since the roots of $|A-\lambda E| = 0$ are not complex, according to the fundamental theorem of algebra, the sum of the multiplicities of the m different eigenvalues $\lambda_i$ (i=1,...,m) is n. Let the geometric multiplicity of $\lambda_i$ be $k_i$, i.e., $\sum_{i=1}^m k_i = n$.
If for each $\lambda_i$, its geometric multiplicity $k_i$ is not less than the corresponding dimension of the eigenspace $V_{\lambda_i}$, i.e., $k_i \geq r_i$ for i=1,...,m, then when the condition is met, we have $\sum_{i=1}^m k_i = \sum_{i=1}^m \dim V_{\lambda_i} = n$. Thus, the solutions of $(A-\lambda E)x = \vec{0}$ can form a basis for $V_n$.
Page 17
The text discusses the relationship between the rank of a matrix and its eigenvalues. It proves that for any matrix A, the rank of A is less than or equal to the number of its eigenvalues. The proof involves showing that the rank of A is equal to the number of non-zero eigenvalues. The author uses the concept of a basis and the properties of linear transformations to establish this result. The proof is detailed and involves several steps, including the construction of a basis for the eigenspace corresponding to a given eigenvalue and the use of the rank-nullity theorem.
Page 18
- Summary and Extension
In each eigenspace, the algebraic multiplicity of the corresponding eigenvalue is the number of solutions of the characteristic equation. For a specific matrix A, through C^(-1)AC transformation, it can be converted into a diagonal matrix D. C is the transformation matrix formed by the basis of V1 + V2 + ... + Vm as column vectors. Since each eigenspace V has a non-unique basis, C is not unique. Therefore, the diagonalization of A is not unique in any basis.
- If A can be diagonalized over the complex field, then each eigenvalue of A must have an algebraic multiplicity equal to its geometric multiplicity. According to the fundamental theorem of algebra, the sum of the algebraic multiplicities of all eigenvalues is n.
- The geometric multiplicity of each eigenspace dim ker(A - λiE) is equal to the rank of A - λiE.
- When dim V is not equal to the algebraic multiplicity of the corresponding eigenvalue, A cannot be diagonalized. The simplest form of a similar matrix is the Jordan form matrix J.
$$
J = \begin{bmatrix} J_1 & & & \\ & J_2 & & \\ & & \ddots & \\ & & & J_m \end{bmatrix} \quad \text{where} \quad J_i = \begin{bmatrix} \lambda_i & 1 & & \\ & \lambda_i & 1 & \\ & & \ddots & 1 \\ & & & \lambda_i \end{bmatrix}
$$
Here, $J_i$ is a Jordan block. The rank of $J_i - \lambda_i E$ is $r - 1$, and the geometric multiplicity of each Jordan block is 1. The Jordan form of A is a block diagonal matrix consisting of m Jordan blocks.
Page 19
IV. Linear Transformation + Translation → Affine Transformation
- Definition: A point transformation in the plane, if it can be expressed in an affine coordinate system (affine coordinate frame) as:
$$
\begin{pmatrix}
x' \\
y'
\end{pmatrix}
= A
\begin{pmatrix}
x \\
y
\end{pmatrix}
+
\begin{pmatrix}
x_0 \\
y_0
\end{pmatrix}
$$
where $A = (a_{ij})_{2x2}$ is invertible, then this point transformation is called an affine transformation.
- Property ①: If $\varphi$ and $\psi$ are affine transformations, then $\varphi \circ \psi$ is an affine transformation, and $\varphi^{-1}$ is also an affine transformation.
Proof:
$$
\begin{cases}
\varphi\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) = A\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) + \begin{pmatrix}
x_0 \\
y_0
\end{pmatrix} \\
\psi\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) = B\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) + \begin{pmatrix}
x_1 \\
y_1
\end{pmatrix}
\end{cases}
$$
$$
\Rightarrow \varphi\circ\psi\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) = \varphi\left[B\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) + \begin{pmatrix}
x_1 \\
y_1
\end{pmatrix}\right]
$$
$$
= A\left[B\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) + \begin{pmatrix}
x_1 \\
y_1
\end{pmatrix}\right] + \begin{pmatrix}
x_0 \\
y_0
\end{pmatrix}
$$
$$
= AB\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) + \left[A\left(\begin{pmatrix}
x_1 \\
y_1
\end{pmatrix}\right) + \begin{pmatrix}
x_0 \\
y_0
\end{pmatrix}\right]
$$
Since $A$ and $B$ are invertible, $AB$ is also invertible.
Therefore, $\varphi\circ\psi$ is an affine transformation.
And
$$
\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) = A^{-1}\left[\left(\begin{pmatrix}
x' \\
y'
\end{pmatrix} - \begin{pmatrix}
x_0 \\
y_0
\end{pmatrix}\right)\right] = A^{-1}\left(\begin{pmatrix}
x' \\
y'
\end{pmatrix}\right) - A^{-1}\left(\begin{pmatrix}
x_0 \\
y_0
\end{pmatrix}\right)
$$
Since $A^{-1}$ is invertible, $\varphi^{-1}$ is also an affine transformation.
- Property ②: An affine transformation maps a straight line to a straight line, two parallel straight lines to two parallel straight lines, and three non-collinear points to three non-collinear points.
Proof:
Let $\tau$ be an affine transformation:
$$
\tau\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) = \left(\begin{pmatrix}
x' \\
y'
\end{pmatrix}\right) = A\left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) + \begin{pmatrix}
x_0 \\
y_0
\end{pmatrix}
$$
$$
\therefore \left(\begin{pmatrix}
x \\
y
\end{pmatrix}\right) = A^{-1}\left(\begin{pmatrix}
x' - x_0 \\
y' - y_0
\end{pmatrix}\right)
$$
For a line $l: ax + by + c = 0$:
$$
(a, b)A^{-1}\left(\begin{pmatrix}
x - x_0 \\
y - y_0
\end{pmatrix}\right) = 0 \quad \text{, i.e.} \quad (a, b)A^{-1}\left(\begin{pmatrix}
x' \\
y'
\end{pmatrix}\right) - (a, b)A^{-1}\left(\begin{pmatrix}
x_0 \\
y_0
\end{pmatrix}\right) = 0
$$
Therefore, the line $l$ is still a line in the same affine coordinate frame.
If $l_1$ and $l_2$ are parallel lines, then $\tau$ maps $l_1$ and $l_2$ to $l_1'$ and $l_2'$, respectively. Since $l_1$ and $l_2$ are parallel, $l_1'$ and $l_2'$ are also parallel.
If $l_1$ and $l_2$ intersect at point $M$, then $\tau(M) \in l_1 \cap l_2$.
Since $l_1$ and $l_2$ are parallel, $l_1'$ and $l_2'$ are also parallel.
Page 20
Property (3): Let A, B, C be collinear points in the affine coordinate system. Let $\overrightarrow{AB} = \lambda \overrightarrow{BC}$, where $\lambda$ is called the simple ratio of collinear points A, B, C. Then, the affine transformation preserves the simple ratio of collinear points.
Proof: Let $A = \left(\begin{array}{c} x_1 \\ y_1 \end{array}\right)$, $B = \left(\begin{array}{c} x_2 \\ y_2 \end{array}\right)$, $C = \left(\begin{array}{c} x_3 \\ y_3 \end{array}\right)$.
$$\varphi\left(\begin{array}{c} x_0 \\ y_0 \end{array}\right) = P\left(\begin{array}{c} x_0 \\ y_0 \end{array}\right) + \left(\begin{array}{c} x_0 \\ y_0 \end{array}\right)$$
$$\therefore \overrightarrow{AB} = \varphi(\overrightarrow{B}) - \varphi(\overrightarrow{A}) = P\left(\begin{array}{c} \frac{x_2 - x_1}{y_2 - y_1} \\ 1 \end{array}\right), \quad \overrightarrow{BC} = P\left(\begin{array}{c} \frac{x_3 - x_2}{y_3 - y_2} \\ 1 \end{array}\right)$$
If $\overrightarrow{AB} = \lambda \overrightarrow{BC}$, then $\left(\begin{array}{c} \frac{x_2 - x_1}{y_2 - y_1} \\ 1 \end{array}\right) = \lambda \left(\begin{array}{c} \frac{x_3 - x_2}{y_3 - y_2} \\ 1 \end{array}\right) \Rightarrow$
$$\overrightarrow{AB} = P\left(\begin{array}{c} \frac{x_2 - x_1}{y_2 - y_1} \\ 1 \end{array}\right) = \lambda P\left(\begin{array}{c} \frac{x_3 - x_2}{y_3 - y_2} \\ 1 \end{array}\right) = \lambda \overrightarrow{BC}$$
Property (4): Any three non-collinear points in the plane $\{A, B, C\}$ can be transformed into any other three non-collinear points $\{A', B', C'\}$ by a unique affine transformation $\varphi$.
Proof: Take the coordinate system $\{A, \overrightarrow{AB}, \overrightarrow{AC}\}$, then $A' = \left(\begin{array}{c} x_0 \\ y_0 \end{array}\right)$, and $\overrightarrow{AA'} = x_0 \overrightarrow{AB} + y_0 \overrightarrow{AC}$
$$\overrightarrow{AB'} = a_{11} \overrightarrow{AB} + a_{12} \overrightarrow{AC}$$
$$\overrightarrow{AC'} = a_{21} \overrightarrow{AB} + a_{22} \overrightarrow{AC}$$
Since $\overrightarrow{AB}$ and $\overrightarrow{AC}$ are not collinear, $P(a_{ij})$ is a non-singular matrix.
Let $\varphi: \left(\begin{array}{c} x \\ y \end{array}\right) \rightarrow \left(\begin{array}{c} x' \\ y' \end{array}\right) = P\left(\begin{array}{c} x \\ y \end{array}\right) + \left(\begin{array}{c} x_0 \\ y_0 \end{array}\right)$, then $\varphi(\overrightarrow{AB}) = \overrightarrow{AB'}$
Uniqueness: If there are two $\varphi, \varphi'$, $\varphi^{-1} \circ \varphi_1: \{A, B, C\} \rightarrow \{A', B', C'\}$, $\varphi_2^{-1} \circ \varphi_1 = I$, then $\varphi = \varphi'$.
2. Geometric definition: The transformation that transforms collinear points into non-collinear points is a one-to-one correspondence in the plane. It can be defined as an affine transformation of vectors: $\forall A, B \in R^2$, $\varphi(\overrightarrow{AB}) = (\varphi(\overrightarrow{A}), \varphi(\overrightarrow{B}))$.
$$\left\{\begin{array}{l} \varphi(\overrightarrow{v} + \overrightarrow{u}) = \varphi(\overrightarrow{v}) + \varphi(\overrightarrow{u}) \\ \varphi(\lambda \overrightarrow{v}) = \lambda \varphi(\overrightarrow{v}) \end{array}\right.$$
$$\left\{\begin{array}{l} A: x_1, y_1 \\ B: x_2, y_2 \\ \varphi\left(\begin{array}{c} x_1 \\ y_1 \end{array}\right) = P_1\left(\begin{array}{c} x_1 \\ y_1 \end{array}\right) + \left(\begin{array}{c} x_0 \\ y_0 \end{array}\right) \\ \varphi\left(\begin{array}{c} x_2 \\ y_2 \end{array}\right) = P_2\left(\begin{array}{c} x_2 \\ y_2 \end{array}\right) + \left(\begin{array}{c} x_0 \\ y_0 \end{array}\right) \end{array}\right.$$
Page 21
Theorem: Given a transformation T in the affine basis I = {O, e1, e2}, its representation is:
$$\begin{bmatrix} x' \\ y' \end{bmatrix} = \begin{bmatrix} a_{11} & a_{12} \\ a_{21} & a_{22} \end{bmatrix} \begin{bmatrix} x \\ y \end{bmatrix} + \begin{bmatrix} x_0 \\ y_0 \end{bmatrix} \quad \quad \begin{pmatrix} a_{11} & a_{12} \\ a_{21} & a_{22} \end{pmatrix} = A$$
For any non-parallel vectors α and β:
$$\frac{|T(\vec{\alpha}) \times T(\vec{\beta})|}{|\vec{\alpha} \times \vec{\beta}|} = |\det(A)|$$
Proof: Let α = (x1, y1), β = (x2, y2), T(α) = (x1', y1') = Aα, T(β) = (x2', y2') = Aβ. Then:
$$\frac{|T(\vec{\alpha}) \times T(\vec{\beta})|}{|\vec{\alpha} \times \vec{\beta}|} = \frac{|A\vec{\alpha} \times A\vec{\beta}|}{|\vec{\alpha} \times \vec{\beta}|} = |\det A|$$
Corollary: If any area region D in the plane is transformed by a transformation T into region D', then:
$$\frac{S_D'}{S_D} = |\det A|$$
- Geometric definition and algebraic definition of equivalence.
- Geometric definition implies algebraic definition
Given a transformation T(x, y) = A(x, y) + (x0, y0) = (x', y'), it can be shown that T can cause a plane transformation of a non-zero vector transformation T(y) = A(y).
If A, B, C are collinear, i.e., there exists λ such that AB = λBC. Then AB' = T(AB) = A(AB), BC' = A(BC). Still satisfies AB' = A(AB) = A(λBC) = λA(BC) = xBC' (linear transformation).
- Affine transformation maps collinear points to collinear points.
Below we prove T is a transformation:
Since A is invertible, (x, y) = A^(-1)(x' - x0) = A^(-1)(x') - A^(-1)(x0) is also a transformation (algebraic definition).
∴ T is a transformation
∴ Algebraic definition implies geometric definition.
Page 22
- Geometric definition implies algebraic definition.
- Given T satisfies the geometric definition, first prove that T can be deduced from the definition. Affine transformation T will map non-collinear points to non-collinear points:
Proof. Assume T maps A, B, C (non-collinear) to A', B', C' (collinear). Let this line be l. By the geometric definition, the images of A, B, C on l are collinear. For any point D in the plane, there always exists a line DE such that DE intersects AB, BC at E, F. Since E, F and D are collinear, D' must also be collinear with E and F. Thus, D' must be on l. Therefore, T is a mapping from R^2 to R^2, which contradicts the geometric definition. Hence, T maps collinear points to collinear points and non-collinear points to non-collinear points.
- Next, prove that T maps parallel lines to parallel lines:
Proof. Assume T maps parallel lines to intersecting lines. Let A' ∈ l' and B' ∈ l'. Then T(A') ∈ l1 and T(B') ∈ l2. Since l1 and l2 are parallel, T maps parallel lines to parallel lines.
- Point transformation T can induce a vector transformation. Prove that T is a linear transformation:
Let T(x) = x', T(y) = y', T(z) = z'.
Since OA' || BC' and OB' || AC', we should have OA' || BC' and OB' || AC'.
Thus, C' is a vertex of parallelogram O'A'C'B'. Therefore, OC' = x' + y'.
Also, T(x + y) = x' + y'.
Next, prove that ∀x ∈ R, T(λx) = λT(x).
Consider n ∈ N*. By (4), T(nx) = x' + x' + ... + x' = n x' = n T(x).
Consider n ∈ Z. If n < 0, T(nx - nx) = T(0) = 0. Therefore, T(nx) = -n T(x).
Therefore, T is a linear transformation.
Page 23
The note discusses linear transformations in R^2. It starts by considering a linear transformation T: R^2 → R^2. The author proves that T is a linear transformation by showing that it satisfies the properties of linearity. Specifically, for any scalar m and vector x in R^2, T(mx) = mT(x). The author also considers the case when x is a vector in R^2 and shows that T(λx) = λT(x) for any scalar λ in R. The author then defines a basis {e1, e2} for R^2 and shows that T(e1) = e1 and T(e2) = e2. The author then defines a transformation T in terms of a matrix A, where A is the matrix representation of T with respect to the basis {e1, e2}. The author then defines the transformation T(0) as the origin of the coordinate system. The author then defines the transformation T(B) as the vector from the origin to the point B. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x) as the vector from the origin to the point x. The author then defines the transformation T(y) as the vector from the origin to the point y. The author then defines the transformation T(x
Page 24
Five. Projection Geometry.
- Centered Straight Lines and Expanded Planes.
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\pi_1 \rightarrow \pi_2$$
$$\
Page 25
- Any two different lines intersect at a point.
- Original intersecting lines → intersect at ordinary points (two different points)
- Original parallel lines → intersect at infinity points (two identical points)
- Ordinary lines + infinity lines → infinity points
- "Point" and "line" relationships become symmetric.
- On the expanded plane, only the relationship between points and lines and the resulting line of points and lines is maintained. The line of points and lines is the relationship between lines and planes in the usual sense (passing through 0).
- Projective Coordinate System.
- Central Line Projective Coordinate System
- In space, take O as the origin and the affine basis {O; e1, e2, e3} (three non-coplanar vectors) in B(0). Each element in B(0) corresponds to a triple (a1e1 + a2e2 + a3e3) → (a1, a2, a3). This is a one-to-one correspondence. However, since there is no "unit point", the magnitudes of the elements in B(0) are not unique.
- Therefore, define the projective basis:
- Take four points l1, l2, l3, l4 such that any three are not collinear (⇒ general position points), and take the space vector e4. e4 can be decomposed into a linear combination of l1, l2, l3. e4 = e1 + e2 + e3. Thus, when e4 is multiplied by a, e1, e2, e3 all need to be included.
- Therefore, by a1e1 + a2e2 + a3e3 → (a1, a2, a3), the triple completely determines the line.
- Therefore, [l1, l2, l3, l4] is a projective basis, l4 is the unit point.
- l1<1, 0, 0> l2<1, 1, 0> l3<0, 1, 1> l4<1, 1, 1>
B(0) of the line corresponds to a plane in space passing through O. In the above affine basis [O; e1, e2, e3] of the plane, the equation of the line: ax + by + cz = 0. The triple
Therefore, in the projective coordinate system, the set of all lines and points in B(0) is one-to-one corresponding to the set of all triples. Line:
Page 26
- Affine Plane Coordinate System
An affine plane is a plane with a center line. The basis is [e1, e2, e3]. The origin is [0, e1, e2, e3]. To ensure the affine plane coordinates do not depend on the origin, we should have:
For all L in B(0), the coordinates of L in [e1, e2, e3] should be the same as the coordinates of f(L) in [e1', e2', e3'] (where f: E^3 -> E^3 is a space affine transformation: f(0) = 0', f(ek) = ek', k = 1, 2, 3, 4).
Proof: Take a non-zero vector e4' || L such that e4' = e1' + e2' + e3'. The affine frame [0', e1', e2', e3'] is B(0)'. Let e4 = f(e4'). Then e4' || L and e4' = e1' + e2' + e3'. Thus, the coordinates of e4' in [e1', e2', e3'] are the same as the coordinates of f(e4) in [e1', e2', e3']. Therefore, L in the affine coordinate system [e1', e2', e3'] has the same coordinates as B(0) in [e1', e2', e3'].
A1<0, 0, 1> A2<1, 0, 0> A3<0, 1, 0> A4<1, 1, 1> => [e1, e2, e3, e4].
Similarly, lines in the affine plane also have coordinates, which are related to the coordinates of points.
- Point P<\frac{x}{a}, \frac{y}{b}> and line L<\frac{x}{c}, \frac{y}{c}>, their relationship: ax + by + cz = 0.
- A1<\frac{x_1}{a_1}, \frac{y_1}{a_1}> A2<\frac{x_2}{a_2}, \frac{y_2}{a_2}> A3<\frac{x_3}{a_3}, \frac{y_3}{a_3}>. Three points collinear: det\begin{pmatrix} x_1 & x_2 & x_3 \\ y_1 & y_2 & y_3 \\ a_1 & a_2 & a_3 \end{pmatrix} = 0.
- L<\frac{x}{a}, \frac{y}{b}, \frac{z}{c}> L2<\frac{x}{a_2}, \frac{y}{b_2}, \frac{z}{c_2}> L3<\frac{x}{a_3}, \frac{y}{b_3}, \frac{z}{c_3}>. Three lines intersect: det\begin{pmatrix} a_1 & b_1 & c_1 \\ a_2 & b_2 & c_2 \\ a_3 & b_3 & c_3 \end{pmatrix} = 0.
- P<\frac{x_1}{a_1}, \frac{y_1}{b_1}> Q<\frac{x_2}{a_2}, \frac{y_2}{b_2}>. The general form of the line PQ: \lambda\begin{pmatrix} x_1 \\ y_1 \\ z_1 \end{pmatrix} + \mu\begin{pmatrix} x_2 \\ y_2 \\ z_2 \end{pmatrix}.
Thus, P and Q determine the line L<\frac{x}{a}, \frac{y}{b}, \frac{z}{c}>: \lambda\begin{pmatrix} x_1 \\ y_1 \\ z_1 \end{pmatrix} + \mu\begin{pmatrix} x_2 \\ y_2 \\ z_2 \end{pmatrix} = 0. Solving for \lambda and \mu gives:
\begin{pmatrix} x_1 \\ y_1 \\ z_1 \end{pmatrix} \begin{pmatrix} a \\ b \\ c \end{pmatrix} = \begin{pmatrix} 0 \end{pmatrix}.
Page 27
Theorem: Ceva's Theorem: If the lines connecting the vertices of a triangle intersect at a single point, then the lines connecting the vertices to the opposite sides also intersect at a single point.
Proof: Let A, B, C, and P be arbitrary points. Establish a coordinate system. Let A' be the point of intersection of lines AP and BC. Let B' be the point of intersection of lines BP and AC. Let C' be the point of intersection of lines CP and AB. Let P' be the point of intersection of lines A'B', B'C', and C'A'.
The coordinates of A, B, C, and P are:
A(0, 0, 1)
B(0, 0, 0)
C(0, 1, 0)
P(1, 1, 1)
The vectors AB, BC, and AC are:
AB(0, 0, 1)
BC(1, 0, 0)
AC(0, 1, 0)
Since A' lies on the line AP, we can set the coordinates of A' as:
A'(x, y, z)
Similarly, we can set the coordinates of B' and C' as:
B'(x', y', z')
C'(x", y", z")
The vectors A'B', B'C', and C'A' are:
A'B'(y', 1, x')
B'C'(y", 1, x")
C'A'(x", 1, y")
The points P', Q', and R' are the points of intersection of lines A'B', B'C', and C'A' respectively. The coordinates of P', Q', and R' are:
P'(1-x, 0, 0)
Q'(0, 1-y, 0)
R'(0, 0, 1-z)
The determinant of the matrix formed by the coordinates of P', Q', and R' is zero, which implies that P', Q', and R' are collinear.
- Homogeneous Coordinate Transformation:
Given two points P and P' in the same plane, if their homogeneous coordinates are <(x, y, z)^T> and <(x', y', z')^T> respectively, then there exists a transformation matrix H such that:
<(x, y, z)^T> = H <(x', y', z')^T>
H is a 3x3 matrix with a determinant of 1.
Page 28
- Line's Projective Coordinate Transformation Formula: Given H as the transition matrix from J to J', the line in J with coordinates
- Projective Mapping → Projective Transformation: A one-to-one correspondence between two projective planes, mapping lines to lines and points to points. It preserves line structure and point-line incidence.
A self-projective mapping → Projective Transformation: Given H as the transition matrix from J to J', if P in J has coordinates
Line's Projective Transformation: J:
Properties:
- G is a transformation in J, then G^T is the transformation in J'.
- G1, G2 are transformations in J, then G2G1 is the transformation in J.
- G is a transformation in J, H is the transition matrix from J to J', then the transformation in J' is H^TGH.
- Ratio
General Geometry Ratio:
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec{x_0}}{\vec{x_1}} = \frac{s_2}{s_1} \frac{\vec{x_1}}{\vec{x_2}} + t_2$$
$$\frac{\vec{x_3}}{\vec{x_1}} = \frac{s_1}{s_2} \frac{\vec{x_1}}{\vec{x_2}} + t_1$$
$$\frac{\vec
Page 29
- Line intersection: (l1, l2; l3, l4) = (x1, x2; x3, x4) => x3 // l1
- Edge intersection: (A1, A2; A3, A4) = (A1, A2, A3) => A1A3 = λ1A2A2
- Plane intersection: Common plane
π1, π2, π3, π4 and π (plane passing through this line) intersect at the line of intersection
(π1, π2; π3, π4) = (l1, l2; l3, l4). And π's choice is irrelevant.
Extended plane intersection:
P1, P2, P3, P4 collinear, then P3 = λP1 + μP2. P4 = λ'P1 + μ'P2. Intersection (P1, P2; P3, P4) = x' / x. l1, l2, l3, l4 collinear, then l3 = λl1 + μl2. l4 = λ'l1 + μ'l2. Intersection (l1, l2; l3, l4) = x' / x.
- The definition of intersection is independent of the choice of coordinates.
If the coordinates of P1, P2, P3, P4 are respectively p1, p2, p3, p4. Then:
P3P3 = λ(P1P1) + μ(P2P2), P4P4 = λ'(P1P1) + μ'(P2P2)
∴ P3 = (λP1 + μP2) / (λ + μ), P4 = (λ'P1 + μ'P2) / (λ' + μ')
∴ (P1, P2; P3, P4) = (λP1 + μP2) / (λ + μ) = (λ'P1 + μ'P2) / (λ' + μ')
- Line l1, l2, l3, l4 collinear, line l intersects them at points P1, P2, P3, P4. Then:
(l1, l2; l3, l4) = (P1, P2; P3, P4).
Let l's coordinates be u, l1's coordinates be u(1), l2's coordinates be u(2), l3's coordinates be u(3), l4's coordinates be u(4). Then P1, P2's coordinates are u x u(1), u x u(2).
Using l1, l2, l3 collinear, l1, l2, l4 collinear, we can get:
(u(3) = λu(1) + μu(2), u(4) = λ'u(1) + μ'u(2))
∴ u(3) = λu(1) + μu(2), u(4) = λ'u(1) + μ'u(2)
∴ P3 = λP1 + μP2, P4 = λ'P1 + μ'P2
∴ (P1, P2; P3, P4) = (λP1 + μP2) / (λ + μ) = (λ'P1 + μ'P2) / (λ' + μ') = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (P1, P2; P3, P4) = (λ' / λ) * (
Page 30
Six. Space Analytic Geometry
- Introduction: Inner Product Space and Orthogonal Transformation
In linear space, the linear operation cannot describe the vector's magnitude properties. After introducing the inner product, we can measure the length and angle of vectors, thereby transforming the linear space into an inner product space. On the real space V(R), define a binary operation such that V elements α, β correspond to a real number, denoted as (α, β). If α, β ∈ V, λ ∈ R satisfy:
$$
\begin{cases}
(\alpha, \beta) = (\beta, \alpha) \\
(\alpha + \beta, \beta) = (\alpha, \beta) + (\beta, \beta) \\
(\lambda \alpha, \beta) = \lambda (\alpha, \beta) \\
(\alpha, \alpha) \geq 0 \text{ and equal only when } \alpha = 0.
\end{cases}
$$
Then (α, β) is the inner product defined in V(R), which is a real inner product space. n < ∞ → Euclidean space.
Length:
$$
|\alpha| = \sqrt{(\alpha, \alpha)}
$$
Angle:
$$
<\alpha, \beta> = \arccos \frac{(\alpha, \beta)}{|\alpha||\beta|}
$$
- Unitary Orthogonal Basis in Euclidean Space
B = {e1, ..., en} is a basis of V(R). If (ei, ej) = {1 if i = j, 0 if i ≠ j}, then B is a unitary orthogonal basis. We can use the Gram-Schmidt orthogonalization method to orthogonalize a basis in V(R).
Let B = {e1, ..., en} be a basis of V(R).
- Let β1 = e1, β2 = e2 + λ21β1, (β2, β1) = (e2 + λ21β1, e1)
= (e2, e1) + λ21(β1, e1) = 0
∴ λ21 = - (e2, β1) / (β1, β1)
- If we have found two orthogonal non-zero vectors β1, β2, ..., βm-1, let
βm = αm + λm1β1 + λm2β2 + ... + λmmβm + λmmβ1 + ... + λmmβm
such that βm and βk (k = 1, 2, ..., m-1) are orthogonal: (βm, βk) = (αm, βk) + λkm(βk, βk) = 0
∴ λkm = - (αm, βk) / (βk, βk)
- By mathematical induction, we can construct a unitary orthogonal basis from the basis B.
Page 31
$$\vec{\beta}_1 = \vec{a}_1$$
$$\vec{\beta}_2 = \vec{a}_2 - \frac{(\vec{a}_2, \vec{\beta}_1)}{(\vec{\beta}_1, \vec{\beta}_1)} \cdot \vec{\beta}_1$$
$$\vdots$$
$$\vec{\beta}_n = \vec{a}_n - \frac{(\vec{a}_n, \vec{\beta}_1)}{(\vec{\beta}_1, \vec{\beta}_1)} \cdot \vec{\beta}_1 - \cdots - \frac{(\vec{a}_n, \vec{\beta}_n)}{(\vec{\beta}_n, \vec{\beta}_n)} \cdot \vec{\beta}_n - \frac{(\vec{a}_n, \vec{\beta}_1)}{(\vec{\beta}_1, \vec{\beta}_1)} \cdot \vec{\beta}_1$$
Simplify, $\vec{\Sigma}_m = \frac{\vec{\beta}_m}{|\vec{\beta}_m|}$, $m=1,\ldots,n$.
$\therefore$ obtain the orthonormal basis $B^* = \{\vec{\Sigma}_1, \vec{\Sigma}_2, \ldots, \vec{\Sigma}_n\}$
(2) Orthogonal space
$\forall \vec{x} \in W$, $\forall \vec{y} \in W$, $(\vec{x}, \vec{y}) = 0 \Rightarrow \vec{x} \perp W$
$\forall \vec{x} \in W_1$, $\forall \vec{y} \in W_2$, $(\vec{x}, \vec{y}) = 0 \Rightarrow W_1 \perp W_2$. If $W_1 + W_2 = V$, $W_2 = W_1^{\perp}$, $\dim W_1 + \dim W_2 = \dim W_1 + \dim W_1^{\perp} = \dim V$.
(3) Orthogonal transformation, orthogonal matrix
If $\sigma \in L(V, V)$ such that $\forall \vec{x}, \vec{y} \in V$, $(\sigma(\vec{x}), \sigma(\vec{y})) = (\vec{x}, \vec{y})$. $\Leftrightarrow |\sigma(\vec{x})| = |\vec{x}|$.
$\Rightarrow (\sigma(\vec{x}), \sigma(\vec{y})) = (\vec{x}, \vec{y})$. $\sigma$ is an orthogonal transformation. Its orthogonal matrix relative to the orthonormal basis of $V$ is an orthogonal matrix: $\sigma(\vec{e}_1, \ldots, \vec{e}_n) = (\vec{e}_1, \ldots, \vec{e}_n)A$.
- Orthogonal matrix $\Rightarrow$ its column vectors form an orthonormal basis.
Proof: Let $\{\vec{e}_1, \ldots, \vec{e}_n\}$ be an orthonormal basis of $V$. $A$ (orthogonal transformation) is the transformation of $\{\vec{e}_1, \ldots, \vec{e}_n\}$.
$\therefore (\vec{e}_i, \vec{e}_j) = (\sigma(\vec{e}_i), \sigma(\vec{e}_j)) = (\sum_{k=1}^{n} a_{ki} \vec{e}_k, \sum_{l=1}^{n} a_{lj} \vec{e}_l)$
$= \sum_{k=1}^{n} a_{ki} a_{lj} = \vec{e}_i^T A \vec{e}_j$ $(A = (a_{ij})_{n \times n} = (\vec{e}_1, \ldots, \vec{e}_n))$.
$= (\vec{e}_i, \vec{e}_j) = \begin{cases} 1 & j = i \\ 0 & j \neq i \end{cases}$, $i, j = 1, 2, \ldots, n$.
- A column vector forms an orthonormal basis $\Rightarrow$ $A$ is an orthogonal matrix.
Proof: From $\sigma(\vec{e}_1, \ldots, \vec{e}_n) = (\vec{e}_1, \ldots, \vec{e}_n)A$ and $(\vec{e}_i, \vec{e}_j) = \begin{cases} 1 & j = i \\ 0 & j \neq i \end{cases}$, we see that $\{\sigma(\vec{e}_1), \ldots, \sigma(\vec{e}_n)\}$ is also an orthonormal basis of $V$.
Page 32
No.
Date
\[
\begin{aligned}
\sigma(\vec{a}), \sigma(\vec{b}) &= \left(\sum_{i=1}^{n} a_i \sigma(\vec{e}_i), \sum_{j=1}^{n} b_j \sigma(\vec{e}_j)\right) = \sum_{i=1}^{n} a_i b_i \\
&= \left(\sum_{i=1}^{n} a_i \vec{e}_i, \sum_{j=1}^{n} b_j \vec{e}_j\right) = (\vec{a}, \vec{b}).
\end{aligned}
\]
\(\therefore \sigma\) is an orthogonal transformation, \(A\) is an orthogonal matrix.
3) \(A\) has the orthonormal basis \(\{\vec{e}_1, \vec{e}_2, \ldots, \vec{e}_n\}\) of \(\mathbb{R}^n\) \(\Longleftrightarrow A^T A = E\).
Proof: \(A^T A = \begin{pmatrix} \vec{e}_1^T \\ \vdots \\ \vec{e}_n^T \end{pmatrix} (\vec{e}_1, \vec{e}_2, \ldots, \vec{e}_n) = \begin{pmatrix} \vec{e}_1^T \vec{e}_1 & \vec{e}_1^T \vec{e}_2 & \cdots & \vec{e}_1^T \vec{e}_n \\ \vdots & \vdots & \vdots & \vdots \\ \vec{e}_n^T \vec{e}_1 & \vec{e}_n^T \vec{e}_2 & \cdots & \vec{e}_n^T \vec{e}_n \end{pmatrix}\)
\[
= \begin{pmatrix}
(\vec{e}_1, \vec{e}_1) & (\vec{e}_1, \vec{e}_2) & \cdots & (\vec{e}_1, \vec{e}_n) \\
(\vec{e}_2, \vec{e}_1) & (\vec{e}_2, \vec{e}_2) & \cdots & (\vec{e}_2, \vec{e}_n) \\
\vdots & \vdots & \vdots & \vdots \\
(\vec{e}_n, \vec{e}_1) & (\vec{e}_n, \vec{e}_2) & \cdots & (\vec{e}_n, \vec{e}_n)
\end{pmatrix} = E_{n \times n}.
\]
Therefore, \(A\) is an orthogonal matrix \(\Longleftrightarrow A\) has an orthonormal basis \(\Longleftrightarrow A^T A = E \Longleftrightarrow A^{-1} = A^T\)
\(\Rightarrow |A| = 1\) or -1
\(\Rightarrow \langle A \vec{x}, A \vec{y} \rangle = \langle \vec{x}, \vec{y} \rangle\).
\(\Rightarrow\) An orthogonal transformation is a linear transformation.
- Linear Transformations and Space Geometry.
- The positional relationship between two planes:
The positional relationship between two planes can be described by the following system of linear equations:
\[
\begin{cases}
a_{11} x_1 + a_{12} x_2 + a_{13} x_3 = b_1 \\
a_{21} x_1 + a_{22} x_2 + a_{23} x_3 = b_2
\end{cases}
\]
\[
\begin{pmatrix}
a_{11} & a_{12} & a_{13} \\
a_{21} & a_{22} & a_{23}
\end{pmatrix}
\begin{pmatrix}
x_1 \\
x_2 \\
x_3
\end{pmatrix}
=
\begin{pmatrix}
b_1 \\
b_2
\end{pmatrix}
\]
Page 33
- Two planes parallel → no solution → B in Im0 outside.
$$ r(A) = 1, r(A, B) = 2. $$
2) Two planes coincide → has solution (two-dimensional space) → B in Im0 inside.
$$ \vec{x} = \vec{x}_0 + \vec{x} = \vec{x}_0 + k_1 \vec{x}_1 + k_2 \vec{x}_2. $$
$$ r(A) = 1, r(A, B) = 1. $$
3) Two planes intersect → has one solution (line) → B in Im0 inside.
$$ \vec{x} = \vec{x}_0 + \vec{x}. $$
$$ r(A) = 2, r(A, B) = 2. $$
2) Plane and line positional relationship
Line:
$$ \begin{cases} a_{11} x_1 + a_{12} x_2 + a_{13} x_3 = b_1 \\ a_{21} x_1 + a_{22} x_2 + a_{23} x_3 = b_2 \end{cases} $$
$$ r(A_0) = r(A_0, B) = 2. $$
Plane:
$$ a_{31} x_1 + a_{32} x_2 + a_{33} x_3 = b_3 $$
Joint: A$\vec{x}$ = B, A ∈ M3(F).
- Plane and line parallel → no solution.
$$ \begin{array}{|c|c|c|c|} \hline -1 & -2 & -1 & 2 \\ \hline 2 & 2 & 2 & 2 \\ \hline \end{array} $$
Line:
$$ \vec{x} = \vec{x}_0 + \vec{x}, \vec{x} \text{ and plane normal vector } (a_{31}, a_{32}, a_{33}) \text{ are perpendicular. } $$
Distance of line to O:
$$ \left| \vec{x}_0 - \frac{(\vec{x}_0, \vec{x})}{(\vec{x}, \vec{x})} \vec{x} \right| = \frac{b_3}{\sqrt{a_{31}^2 + a_{32}^2 + a_{33}^2}}. $$
$$ r(A) = 2, r(A, B) = 3. $$
Page 34
- Line in plane -> solution is the original line (-1 dimension) -> B in 2mA range.
$$d = |\vec{x}_0 - (\frac{\vec{x}_2 \cdot \vec{x}}{|\vec{x}|}) \vec{x}| = \frac{b_2}{\sqrt{a_{31}^2 + a_{32}^2 + a_{33}^2}}$$
$$r(A) = r(A, \vec{b}) = 2$$
- Line intersects plane at one point -> solution is a line (a vector) -> A is a line, B is a plane.
$$n(A) = r(A, \vec{b}) = 3$$
3) Line-line spatial relationship.
$$l_1 \left\{ \begin{array}{l} a_{11}x_1 + a_{12}x_2 + a_{13}x_3 = b_1 \\ a_{21}x_1 + a_{22}x_2 + a_{23}x_3 = b_2 \end{array} \right. r(A_1) = r(A_1, \vec{b_1}) = 2$$
$$l_2 \left\{ \begin{array}{l} a_{31}x_1 + a_{32}x_2 + a_{33}x_3 = b_3 \\ a_{41}x_1 + a_{42}x_2 + a_{43}x_3 = b_4 \end{array} \right. r(A_2) = r(A_2, \vec{b_2}) = 2$$
Associated: $$A_{4 \times 3} \vec{x} = \vec{b}$$
1) Intersect -> solution is a point (0 dimension) -> A is a plane, B is a line.
$$n(A) = r(A, \vec{b}) = 3$$
- Coincide -> infinite solutions A is the same, parallel -> no solution.
$$\vec{x} = \frac{1}{2} \vec{x} + \frac{2}{3} \vec{x}$$
$$n(A) = r(A, \vec{b}) = 2$$
- Skew (non-coplanar) -> no solution -> non-coplanar lines in the same plane is skew.
$$n(A) = 3, r(A, \vec{b}) = 4, \vec{b} \notin R(A)$$
Page 35
- Quadratic Forms and Space Analytic Geometry
For a bilinear form f(α, β) on V(F), let B = {e1, ..., en} be a basis, and let α and β be represented in the basis as:
α = ∑xiei, β = ∑yiej.
Then:
f(α, β) = f(∑xiei, ∑yiej) = ∑xjyjf(ei, ej).
Given aij = f(ei, ej) (i, j = 1, ..., n). Then:
f(α, β) = ∑j=1n ∑i=1n aijxjyj = ∑j=1n (a1jx1 + a2jx2 + ... + anjxn)
= (x1, ..., xn) * (a11 a12 ... a1n
a21 a22 ... a2n
...
an1 an2 ... ann) * (y1
y2
...
yn) = α^T A β.
A = (aij)_{n×n} = (f(ei, ej))_{n×n} is the matrix of the bilinear form f(α, β) in the basis B.
Theorem: For a bilinear form f(α, β) in V(F), let B = {e1, ..., en} and B' = {e1', ..., e'n} be two bases. If (e1', ..., e'n) = (e1, ..., en)C, then B' = C^TAC.
Proof: Let A = (aij)_{n×n}, aij = f(ei, ej); B = (bij)_{n×n}, bij = f(ei', ej'). Let C = (cij)_{n×n}, C^T = (cij^T)_{n×n}.
Then:
bij = f(ei', ej') = f(∑k=1n cki ek, ∑m=1n cmj em) = ∑k=1n ∑m=1n cki cmj f(ek, em).
= ∑k=1n ∑m=1n cki cmj akm = ∑m=1n (∑k=1n cki akm) cmj.
= ∑m=1n (C^T A)im cmj = (C^T A C)ij, i, j = 1, ..., n.
∴ B = C^T A C. Thus A is similar to B, A ~ B.
- Similarity of Matrices
Reflexivity: ∀A ∈ Mn(F), A ~ A: A = E^T A E
Symmetry: ∀A, B ∈ Mn(F), if A ~ B then B ~ A: A = C^T B C, B = (C^T)^T A C C^T.
Transitivity: ∀A, B ∈ Mn(F), A = P1^T B P1, B = P2^T C P2. ∴ A = P1^T P2^T C P2 P1 = (P2 P1)^T C (P2 P1).
Page 36
- Real symmetric matrices can be transformed into a simpler form (symmetric form) through the method of orthogonal diagonalization.
(1) Quadratic form - symmetric bilinear function.
$$f(\vec{x},\vec{y}) = f(\vec{y},\vec{x}) \Rightarrow symmetric bilinear function eg. Euclidean space inner product (\vec{x},\vec{y}).$$
$$f(\vec{x},\vec{y}) = -f(\vec{y},\vec{x}) \Rightarrow antisymmetric bilinear function eg. n-th order determinant D(x1...xn) antisymmetric function. f(x1...xn) = ∑aijxixj + ∑2aijxixj = ∑aijxixj = xTAx is symmetric bilinear function. A is a real symmetric matrix. Called n-order quadratic form.
(2) For symmetric bilinear functions (quadratic forms), the matrix can be transformed under different bases:
$$f(\vec{x},\vec{x}) = \vec{x}^T A \vec{x} = (\vec{y}^T)^T A (\vec{y}^T). C is the base transformation matrix.
= \vec{y}^T (C^T A C) \vec{y}.
For a quadratic form, we hope to find a basis such that f(\vec{x},\vec{x}) is simplest in this basis:
If C^T A C = diag(d1, d2...dn) → A's symmetric form.
(3) In fact, any real symmetric matrix (quadratic form) can be transformed into a diagonal matrix with orthogonal matrix Q such that:
Q^T A Q = Q^T A Q = diag(λ1, λ2...λn).
Proof: 1. Real symmetric matrices can be diagonalized first: all eigenvalues are real.
Assume λ = λ, according to λx = λx, (A)^T = A.
x(Ax)^T = (x^T)^T(λx) = (x^T)^T A x = (x^T)^T (A^T x)
= (Ax)^T x = (λx)^T x = x(λx)^T x
∴ x = λ.
2. If A is an n-order real symmetric matrix, then there exists an n-order orthogonal matrix Q such that Q^T A Q = diag(λ1, λ2...λn).
For an n-order real symmetric matrix, if λ is an eigenvalue of A, then A^T x = λx.
Page 37
The note discusses the spectral theorem for symmetric matrices. It starts by stating that if X is a unit vector, it can be extended to an orthonormal basis {X1, X2, ..., Xn}. The matrix A can then be expressed in terms of this basis. The note proceeds to show that A can be written in block form as a diagonal matrix with the eigenvalues on the diagonal and a zero block off the diagonal. It then proves that the zero block can be eliminated by using an orthogonal transformation. The note concludes with a theorem stating that the eigenvectors corresponding to distinct eigenvalues of a symmetric matrix are orthogonal. The proof involves showing that the inner product of any two distinct eigenvectors is zero. The note also mentions that the eigenspaces corresponding to distinct eigenvalues are orthogonal and can be orthogonally diagonalized.
Page 38
- Completing the Square Method.
If the quadratic form has mixed terms, then complete the square term by term, and move all mixed terms into the complete square terms. This will yield an upper triangular matrix with 1s on the diagonal.
If the quadratic form only has mixed terms, first perform a simple transformation that produces a mixed term, then follow the above method to complete the square and eliminate all mixed terms to form a complete square form.
- Elementary Transformation Method.
Using mathematical induction, it can be proven that for any real symmetric matrix A, there exists an elementary matrix (addition, multiplication, permutation) P1P2...Pn such that Pn^T...P1^TAP1P2...Pn = diag(d1, d2, ..., dn).
The transformation matrix C = P1P2...Pn = E. P1P2...Pn.
Therefore, (A) can be transformed into a diagonal matrix through the same type of elementary row and column transformations. Correspondingly, E can be transformed into C.
From the elementary transformation method, it can be seen that the combined standard form is not unique, but can be transformed into the form diag(1, ..., 1, -1, ..., -1, 0, ..., 0). The rank of the matrix is equal to the sum of the positive and negative eigenvalues.
The necessary and sufficient condition for two n-order real symmetric matrices A and B to be equivalent is that their positive and negative eigenvalues are equal.
- Standard Equation and Classification of Quadratic Curves. & Classification of Quadratic Surfaces.
By transforming the quadratic form F(x1, x2, x3) = 0 into the standard form (the normal form) through orthogonal transformation, we can determine the type of the quadratic curve.
The standard form of the ellipsoid: $$\frac{x^2}{a^2} + \frac{y^2}{b^2} + \frac{z^2}{c^2} = 1$$.
The standard form of the single-leaf hyperboloid: $$\frac{x^2}{a^2} + \frac{y^2}{b^2} - \frac{z^2}{c^2} = 1$$.
The standard form of the double-leaf hyperboloid: $$-\frac{x^2}{a^2} + \frac{y^2}{b^2} - \frac{z^2}{c^2} = 1$$.
The standard form of the elliptic paraboloid: $$z = \frac{x^2}{a^2} + \frac{y^2}{b^2}$$.
The standard form of the hyperbolic paraboloid (hyperboloid of one sheet): $$\frac{x^2}{a^2} - \frac{y^2}{b^2} = 2pz$$.
Page 39
- Positive definite quadratic forms and positive definite matrices, other definite quadratic forms.
For any vector x ≠ 0 (x ∈ R^n), if x^T A x > 0, then f(x, x) = f(x_1, x_2, ..., x_n) = x^T A x is a positive definite form. A is a positive definite matrix.
- n-variable quadratic form (standard form) f = a_1 x_1^2 + ... + a_n x_n^2 is positive definite ⇔ a_i > 0 (i = 1, ..., n).
[Proof of necessity: obviously.]
[Proof of sufficiency: Assume a_i ≤ 0. Take x_i = 1, x_j = 0 (j ≠ i). Then f(0, ..., 0, 1, 0, ..., 0) = a_i ≤ 0. Contradiction.]
- If f = x^T A x is positive definite, then f = P^T (C^T A C) P is positive definite, where P = C^T. C is invertible.
- A is positive definite ⇔ A has n positive eigenvalues, i.e., P^T A P = E.
- A is positive definite ⇔ ∃ invertible P, A = P^T P.
∵ A ≈ E. ∵ C invertible, C^T A C = E. ∴ A = (C^T)^{-1} C^T. Let P = C^T, we get.
- A is positive definite ⇔ A has n eigenvalues λ_1, ..., λ_n all greater than zero.
Let λ be an eigenvalue of A. ∴ For any non-zero vector x ≠ 0, A x = λ x. ∴ A = P^T P.
∴ (P^T P) x = λ x. ∴ (P^T x) (P x) = λ (x^T x). ∴ (P x, P x) = λ (x^T x). ∴ (P x, P x) > 0. ∴ λ > 0.
- A is positive definite ⇔ A^T is positive definite.
- A is positive definite ⇔ A has n positive principal minors. i.e., det A_k = det $$\begin{vmatrix} a_{11} & a_{12} & ... & a_{1k} \\ a_{21} & a_{22} & ... & a_{2k} \\ ... & ... & ... & ... \\ a_{k1} & a_{k2} & ... & a_{kk} \end{vmatrix}$$ > 0. k = 1, 2, ..., n.
Theorem: If x^T A x is positive definite, then A has all positive diagonal elements a_{ii} > 0; A has a positive determinant.
Proof: 1) ∵ x^T A x = ∑_{i=1}^n a_{ii} x_i^2 is positive definite. ∴ Take x = (0, ..., 1, 0, ..., 0) and let the 1st element be 1. Then x^T A x = a_{ii} x_i^2 = a_{ii} > 0. 2) ∵ A is positive definite. ∴ ∃ invertible P such that A = P^T P. ∴ |A| = |P| |P| = |P|^2 > 0.
If x^T A x < 0: Negative definite. Characteristic index = n; -A is definite. Odd-order minors < 0, even-order minors > 0. If x^T A x ≥ 0: Semi-definite. Characteristic index = r(A); A has all principal minors ≥ 0. If x^T A x ≤ 0: Negative definite. Characteristic index = r(A).
Page 40
Chapter 7. Differential Geometry
R³ Curves and Surfaces in the Neighborhood of a Point
- Vector Functions
$$\vec{F}(t) = (x(t), y(t), z(t))$$
1.1 Derivative
$$\vec{F}(t_0) = \lim_{\Delta t \to 0} \frac{\vec{F}(t_0 + \Delta t) - \vec{F}(t_0)}{\Delta t} = (\dot{x}(t_0), \dot{y}(t_0), \dot{z}(t_0))$$
If $\|\vec{F}(t_0)\| \neq 0$, then $t_0$ is a regular point, otherwise it is a singular point.
When all points on C are regular points, C is a regular curve. If $\vec{F}(t)$ is continuous and non-zero on $[a, b]$, C is a smooth curve.
- $$\frac{d}{dt} (\vec{F}_1 \cdot \vec{F}_2) = \frac{d\vec{F}_1}{dt} \cdot \vec{F}_2 + \vec{F}_1 \cdot \frac{d\vec{F}_2}{dt}$$
- $$\frac{d}{dt} (\vec{F}_1 \times \vec{F}_2) = \frac{d\vec{F}_1}{dt} \times \vec{F}_2 + \vec{F}_1 \times \frac{d\vec{F}_2}{dt}$$
- $$\frac{d}{dt} (\vec{F}_1, \vec{F}_2, \vec{F}_3) = (\frac{d\vec{F}_1}{dt}, \vec{F}_2, \vec{F}_3) + (\vec{F}_1, \frac{d\vec{F}_2}{dt}, \vec{F}_3) + (\vec{F}_1, \vec{F}_2, \frac{d\vec{F}_3}{dt}) = \frac{d}{dt} (\vec{F}_1, \vec{F}_2, \vec{F}_3)$$
- $$\vec{F}^{(n)}(t) = (x^{(n)}(t), y^{(n)}(t), z^{(n)}(t))$$
- $$\vec{F}(t + \Delta t) = \vec{F}(t) + \vec{F}(t) \Delta t + \cdots + \frac{1}{n!} \vec{F}^{(n)}(t) (\Delta t)^n + \sum (\Delta t)^n$$
1.2 Integral
$$\int \vec{F}(t) dt = \vec{F}(t) + \vec{C}$$
$$\int_a^b \vec{F}(t) dt = \vec{F}(b) - \vec{F}(a)$$
- General Parametric and Arc Length Parametric Descriptions of Frenet Frame (Natural Coordinate System)
Definition
$$\vec{T} = \frac{d\vec{F}}{dt}, \quad \vec{T} = \frac{d\vec{F}}{ds}$$
Arc Length Differentiation
$$ds = \|\vec{F}(t)\| dt = \sqrt{x^2 + y^2 + z^2} dt$$
If $\vec{F} \neq 0$ (regular), then $\frac{ds}{dt} = \|\vec{F}\| = \|\frac{d\vec{F}}{dt}\| > 0$. Thus, $s(t) = \int_0^t \sqrt{x^2 + y^2 + z^2} dt$
$$= \int_0^x \sqrt{1 + [y'(s)]^2} ds = \int_0^x \sqrt{[F^2(s)] + [F'(s)]^2} ds$$
The inverse function $t = t(s)$ exists.
Therefore, there must be a curve C with arc length s as the parameter.
$$\vec{F} = \vec{F}(s) = (x(s), y(s), z(s))$$
Frenet Frame
- Tangent vector: $\vec{T}$
- Normal vector: $\vec{N}$
- Binormal vector: $\vec{B}$
Page 41
- Tangent and Normal Plane
Given $\vec{F} = \vec{F}(s) = (x(s), y(s), z(s))$ for $a \leq s \leq b$.
For $s_0 \in [a, b]$, the tangent vector $\vec{T} = \frac{d\vec{F}}{ds}$ (with $\|\vec{T}\| = \frac{1}{ds} = 1$).
The tangent line equation is $\vec{P} = \vec{F}(s_0) + \lambda \vec{T} = \vec{F}(s_0) + \lambda \vec{T}(s_0)$.
The normal plane equation is $(\vec{P} - \vec{F}(s_0)) \cdot \vec{T}(s_0) = 0$.
Where, $\vec{P} = (x, y, z)$ is a point.
With $t$ as a parameter:
The tangent vector $\vec{T} = \frac{d\vec{F}}{ds} = \frac{d\vec{F}}{dt} \frac{dt}{ds} = \frac{\vec{F}(t)}{\|\frac{d\vec{F}}{dt}\|} = \frac{\vec{F}(t)}{\|\vec{F}(t)\|}$.
At point $\vec{F}(t_0)$,
The tangent line equation is $\vec{P} = \vec{F}(t_0) + \lambda \vec{T}(t_0)$.
The normal plane equation is $\vec{T}(t_0) \cdot (\vec{P} - \vec{F}(t_0)) = 0$.
- Subtangent and Subnormal Plane
The subtangent plane is the plane passing through the point $\vec{F}(s_0)$ and the tangent line and its neighboring point $\vec{F}(s_0 + \Delta s)$ when $\Delta s \to 0$.
The normal vector of the plane is parallel to $\vec{F}(s_0) \times (\vec{F}(s_0 + \Delta s) - \vec{F}(s_0))$.
And $\vec{F}'(s_0) \times (\vec{F}(s_0 + \Delta s) - \vec{F}(s_0)) = \vec{F}'(s_0) \times (\vec{F}'(s_0) \Delta s + \frac{1}{2!} (\vec{F}''(s_0) + \vec{E})(\Delta s)^2)$
$= \frac{1}{2} \vec{F}'(s_0) \times (\vec{F}''(s_0) + \vec{E})(\Delta s)^2$
When $\vec{E} = 0$, $\vec{B} \parallel \vec{F}'(s_0) \times \vec{F}''(s_0)$.
Thus, $\vec{F} \cdot \vec{F}'' = (\frac{\partial x}{\partial s}, \frac{\partial y}{\partial s}, \frac{\partial z}{\partial s}) \cdot (\frac{\partial^2 x}{\partial s^2}, \frac{\partial^2 y}{\partial s^2}, \frac{\partial^2 z}{\partial s^2}) = \frac{\partial x}{\partial s} \frac{\partial}{\partial s} (\frac{\partial x}{\partial s}) + \cdots + \frac{\partial z}{\partial s} \frac{\partial}{\partial s} (\frac{\partial z}{\partial s})$
$= \frac{1}{2} \frac{\partial}{\partial s} \left[ (\frac{\partial x}{\partial s})^2 + \cdots + (\frac{\partial z}{\partial s})^2 \right] = 0$.
$\therefore \vec{F} \perp \vec{F}''$
Page 42
The note discusses the concept of a unit normal vector B in the context of a curve defined by a vector function r(s). It explains how to find B as the cross product of the first and second derivatives of r, normalized by the magnitude of the cross product. The note also covers the tangent vector T and the normal vector N, derived from B and T. It explains that the cross product of T and B gives N, which is a unit vector. The note concludes by stating that T, N, B, and r form a basis for the tangent space at any point on the curve.
Page 43
- Curvature, Torsion, and Frenet's Formula
- Curvature
$$\text{Curvature} = \lim_{\Delta s \to 0} \left\| \frac{\vec{T}(s_0 + \Delta s) - \vec{T}(s_0)}{\Delta s} \right\|$$
$$k(s) = \frac{1}{s} \left\| \frac{\Delta \vec{T}}{\Delta s} \right\| = \left\| \vec{T}'(s_0) \right\| = \left\| \vec{T}'' \right\|$$
$$\text{Radius of curvature} R(s) = \frac{1}{k(s)} = \frac{1}{\left\| \vec{T}'' \right\|}$$
$$\text{Center of curvature} O: \vec{F}(s) + R(s) \vec{N}(s)$$
$$\vec{T}' = k \vec{N}$$
$$\vec{T} = \vec{T} \cdot t'(s) = \frac{\vec{T}}{\left\| \vec{T} \right\|}$$
$$\vec{T}'' = \vec{T} \left[ t(s) \right]^2 + \vec{T} \cdot t''(s)$$
$$\vec{T} \cdot \vec{T}'' = 0$$
$$k = \left\| \vec{T}'' \right\| = \left\| \vec{T} \times \vec{T}'' \right\| = \left\| \vec{T}^2 \cdot t'(s) \times \left[ \vec{T} \left[ t(s) \right]^2 + \vec{T} \cdot t''(s) \right] \right\|$$
$$= \left\| \left[ t'(s) \right]^3 (\vec{T} \times \vec{T}) \right\|$$
$$= \frac{\left\| \vec{T} \times \vec{T}'' \right\|}{\left\| \vec{T} \right\|^3}$$
2. Torsion
$$\text{Torsion} = \frac{d\vec{B}}{ds} = \vec{B}'(s)$$
$$\vec{B} \cdot \vec{T} = 0$$
$$\vec{B}' \cdot \vec{T} = 0$$
$$\vec{B} \cdot \vec{B} = 1$$
$$\vec{B}' \cdot \vec{B} = 0$$
$$\vec{B}'(s) = -\vec{T}(s) \vec{N}(s)$$
$$\left\| \vec{B} \right\| = 1, \vec{N} \parallel \vec{B}$$
$$\therefore \vec{T} = -\vec{B}' \cdot \vec{N}; \left| \vec{T} \right| = \left\| \vec{B}' \right\| = \frac{d\left\| \vec{B} \right\|}{ds}$$
Page 44
Under natural parameters, since $$\frac{1}{k}(\vec{F}''\cdot\vec{B}) = \frac{1}{k}((k\vec{N})'\cdot\vec{B}) = \frac{1}{k}[(\vec{N}\cdot\vec{B})' - (k\vec{N}\cdot\vec{B})]$$
$$= -(\vec{N}\cdot\vec{B}) = \tau$$
And $$\vec{B} = \frac{\vec{F}'\times\vec{F}''}{\|\vec{F}'\|^2} = \frac{1}{k}(\vec{F}'\times\vec{F}'')$$, substituting into the above equation gives
$$\tau = \frac{(\vec{F}',\vec{F}'',\vec{F}''')}{\|\vec{F}'\|^2}$$
Under general parameters, the formula becomes $$\tau = \frac{(\vec{F}',\vec{F}',\vec{F}''')}{(\vec{F}'\times\vec{F}'')^2}$$
(3) Frenet's Formula
In the fixed frame (x, y, z), $$\vec{r}(s) = (x(s), y(s), z(s))$$
In the Frenet frame (T, N, B), $$\vec{r}(s) = \alpha(s)\vec{T} + \beta(s)\vec{N} + \gamma(s)\vec{B}$$
This frame is moving, its first derivative is
$$\frac{d\vec{F}(s)}{ds} = \alpha'T + \beta'N + \gamma'B + \alpha T' + \beta N' + \gamma B'$$
Since $$T' = kN$$, $$B' = -\tau N$$, and for $$N = B \times T$$ we also have:
$$N' = B' \times T + B \times T' = -\tau N \times T + B \times kN = \tau B - kT$$
Therefore, $$\begin{pmatrix} \vec{T}' \\ \vec{N}' \\ \vec{B}' \end{pmatrix} = \begin{pmatrix} 0 & k & 0 \\ -k & 0 & \tau \\ 0 & -\tau & 0 \end{pmatrix} \begin{pmatrix} \vec{T} \\ \vec{N} \\ \vec{B} \end{pmatrix}$$
Theorem: Let $$\vec{F}_1(s)$$ and $$\vec{F}_2(s)$$ be two regular curves in $$\mathbb{R}^3$$, if their k's are everywhere non-zero, but $$k_1(s) = k_2(s)$$, $$\tau_1(s) = \tau_2(s)$$, then $$\vec{F}_1(s)$$ can be transformed through translation and rotation to $$\vec{F}_2(s)$$.
Page 45
- Surface Parametric Equations and Their Normal Vectors
In R^3, the surface S can be represented explicitly as z = f(x, y) or implicitly as F(x, y, z) = 0. Parametric representation is given by r = r(u, v) = (x(u, v), y(u, v), z(u, v)). A point (u, v) is a regular point if r_u × r_v ≠ 0. Otherwise, it is a singular point. If A(u, v) ∈ D and rank $$\begin{vmatrix} x_u & y_u & z_u \\ x_v & y_v & z_v \end{vmatrix}$$ = 2, then r(u, v) is a regular surface.
The u and v curves form a coordinate net:
u coordinate curve: r = r(u, v_0)
v coordinate curve: r = r(u_0, v)
- u curve and v curve tangent vectors:
r_u(u_0, v_0) = (∂x/∂u, ∂y/∂u, ∂z/∂u)(u_0, v_0)
r_v(u_0, v_0) = (∂x/∂v, ∂y/∂v, ∂z/∂v)(u_0, v_0)
- Parameter transformation [r = r(u(û, û), v(û, û))]
u = u(û, û) Surface equation in new parameters: r = r(u(û, û), v(û, û))
v = v(û, û) Tangent vectors of u curve and v curve in new parameters:
r_û = r_u ∂u/∂û + r_v ∂u/∂û
r_û = r_u ∂u/∂û + r_v ∂u/∂û
r_û × r_û = D(û, û) (r_u × r_v)
r_û = r_u ∂u/∂û + r_v ∂u/∂û
r_û = r_u ∂u/∂û + r_v ∂u/∂û
- Tangent plane and normal line equations.
Consider the curve C on the surface S with parametric representation r = r(u(t), v(t)). The tangent vector of C is:
r = r_u ∂u/∂t + r_v ∂v/∂t
r = r_u ∂u/∂t + r_v ∂v/∂t
r = r_u ∂u/∂t + r_v ∂v/∂t
The tangent plane and normal line equations are:
r = r_u ∂u/∂t + r_v ∂v/∂t
r = r_u ∂u/∂t + r_v ∂v/∂t
r = r_u ∂u/∂t + r_v ∂v/∂t
Page 46
The note discusses the tangent plane and normal vector of a surface at a given point. It explains that the tangent plane at point P(t) of a surface S can be described by the cross product of the tangent vectors. The tangent plane is spanned by the tangent vectors. The tangent plane's normal vector is the cross product of the tangent vectors, normalized to be a unit vector. The tangent plane's equation is given by the dot product of the normal vector and the position vector being zero. The parametric form of the tangent plane is given by a linear combination of the tangent vectors. The normal line equation is given by the position vector plus a scalar multiple of the normal vector. The parametric form of the normal line is given by the position vector plus a scalar parameter times the normal vector.
The note then discusses the first fundamental form of a surface. It defines the position vector of a curve C on the surface S as a function of parameters u and v. The tangent vector of the curve at point P is given by the partial derivatives of the position vector with respect to u and v. The arc length element ds is given by the magnitude of the tangent vector. The first fundamental form is defined as the dot product of the tangent vectors. The coefficients E, F, and G are defined as the dot products of the tangent vectors. The first fundamental form is given by the quadratic form in du and dv. The condition for the first fundamental form to be positive definite is given by the determinant of the matrix formed by the coefficients E, F, and G being greater than zero.
Page 47
- Let $\vec{r}_{u} = r_{u} \frac{\partial u}{\partial u} + r_{0} \frac{\partial u}{\partial v}$ and $\vec{r}_{v} = r_{u} \frac{\partial u}{\partial u} + r_{0} \frac{\partial u}{\partial v}$. Then, $\left(\vec{r}_{u}, \vec{r}_{v}\right) = \left(\frac{\partial u}{\partial u}, \frac{\partial u}{\partial v}\right) \left(\vec{r}_{u}, \vec{r}_{v}\right) = \left(\frac{\partial u}{\partial u}, \frac{\partial u}{\partial v}\right) \left(\vec{r}_{u}, \vec{r}_{v}\right)$. The first fundamental form is given by $E = \vec{r}_{u} \cdot \vec{r}_{u}$, $F = \vec{r}_{u} \cdot \vec{r}_{v}$, and $G = \vec{r}_{v} \cdot \vec{r}_{v}$.
- The second fundamental form is given by $\vec{r}_{uu} = r_{uu} u + r_{uv} v$ and $\vec{r}_{uv} = r_{uu} u + r_{uv} v$. The second fundamental form is given by $\vec{r}_{uu} \cdot \vec{n} = r_{uu} u + r_{uv} v$ and $\vec{r}_{uv} \cdot \vec{n} = r_{uu} u + r_{uv} v$.
- The second fundamental form is given by $\vec{r}_{uu} = r_{uu} u + r_{uv} v$ and $\vec{r}_{uv} = r_{uu} u + r_{uv} v$.
Page 48
The note discusses the concept of curvature in space curves and surfaces. It defines the curvature of a space curve at a point as the magnitude of the second derivative of the curve's position vector. For a surface, the curvature at a point is defined through the curvatures of all possible curves passing through that point. The note introduces the concept of the normal curvature and the geodesic curvature. The normal curvature is the curvature of a curve on the surface in the direction of the surface's normal. The geodesic curvature is the curvature of a curve on the surface in the direction of the tangent vector to the curve. The note also introduces the concept of the normal vector and the tangent vector at a point on the surface. The normal curvature is given by the dot product of the normal vector and the second derivative of the position vector of the curve. The geodesic curvature is given by the dot product of the tangent vector and the second derivative of the position vector of the curve. The note also introduces the concept of the principal curvatures and the principal directions of a surface. The principal curvatures are the maximum and minimum curvatures of the surface at a point. The principal directions are the directions in which the surface has these curvatures. The note also introduces the concept of the Gaussian curvature and the mean curvature of a surface. The Gaussian curvature is the product of the two principal curvatures. The mean curvature is the average of the two principal curvatures. The note also introduces the concept of the Christoffel symbols and the Riemann curvature tensor. The Christoffel symbols are used to describe the connection between the tangent vectors of the surface. The Riemann curvature tensor is used to describe the curvature of the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is used to describe the deviation of a geodesic from a nearby geodesic. The note also introduces the concept of the geodesic flow and the geodesic flow equation. The geodesic flow is the flow of the geodesics on the surface. The geodesic flow equation is used to describe the flow of the geodesics on the surface. The note also introduces the concept of the geodesic deviation and the geodesic deviation equation. The geodesic deviation is the deviation of a geodesic from a nearby geodesic. The geodesic deviation equation is
Page 49
$$\therefore N = \frac{\vec{r}^{\prime\prime}}{|\vec{r}^{\prime\prime}|} = \frac{\vec{r}^{\prime\prime}}{k}$$
$$\therefore k_n = \vec{r}^{\prime\prime} \cdot \vec{n}^0 = \frac{\Pi(du, dv)}{2(du, dv)} = \frac{2d\mu^2 + 2Mdu dv + N dv^2}{E du^2 + 2F du dv + G dv^2}$$
$$= \frac{L\left(\frac{d\mu}{dv}\right)^2 + 2M\left(\frac{d\mu}{dv}\right) + N}{E\left(\frac{d\mu}{dv}\right)^2 + 2F\left(\frac{d\mu}{dv}\right) + G}$$
$$\therefore k_n \text{ depends only on } \frac{d\mu}{dv}.$$
(2) Principal curvature.
The direction of the principal curvature $k_n$ at a point on the surface is called the principal direction (on the tangent plane). The extremum (maximum and minimum) of this extremum is the principal curvature of the surface at that point.
$\because I(du, dv), II(du, dv)$ are independent of the choice of parameters,
$\therefore$ By choosing appropriate new parameters, $I(du, dv), II(du, dv)$ can be simultaneously diagonalized.
$\because A = \begin{pmatrix} E & F \\ F & G \end{pmatrix}$ is a positive definite matrix (rank 2),
$\therefore$ There exists a real symmetric matrix $B = \begin{pmatrix} L & M \\ M & N \end{pmatrix}$.
$\therefore$ There exists a nonsingular matrix $W$ such that $W^TAW = E_{2 \times 2}$, $W^TBW = \text{diag}(\lambda_1, \lambda_2)$.
Proof: There exists a nonsingular matrix $R$ such that $R^TAR = E$. And $R^TBR$ is still a real symmetric matrix.
$\therefore$ There exists an orthogonal matrix $Q$ such that $Q^T(R^TBR)Q = \text{diag}(\lambda_1, \lambda_2)$, and $Q^T(R^TAR)Q = Q^TEQ = E$.
$\therefore W = RQ$.
$\therefore$ Under the parameter transformation $\left(\frac{d\mu}{dv}\right) = W\left(\frac{d\bar{\mu}}{dv}\right)$,
$\therefore k_n = \frac{\Pi(du, dv)}{2(du, dv)} = \frac{\Pi(d\bar{\mu}, d\bar{v})}{2(d\bar{\mu}, d\bar{v})} = \frac{\lambda_1 d\bar{\mu}^2 + \lambda_2 d\bar{v}^2}{d\bar{\mu}^2 + d\bar{v}^2}$.
When $\lambda_1 \neq \lambda_2$, assume $\lambda_1 < \lambda_2$, $\therefore \frac{\lambda_1(d\bar{\mu}^2 + d\bar{v}^2)}{d\bar{\mu}^2 + d\bar{v}^2} \leq \frac{\lambda_1 d\bar{\mu}^2 + \lambda_2 d\bar{v}^2}{d\bar{\mu}^2 + d\bar{v}^2} \leq \frac{\lambda_2(d\bar{\mu}^2 + d\bar{v}^2)}{d\bar{\mu}^2 + d\bar{v}^2}$
$\therefore \lambda_1, \lambda_2$ are the principal curvatures of the surface $S$ at point $P(u, v)$.
Page 50
Theorem: If the surface S: F=F(u,v) at point P has unequal principal curvatures λ1 and λ2, then λ is a principal curvature if and only if |λA-B|=|λ(EF)-(LM)|=0.
Proof: Since W^T B W = diag(λ1, λ2), i.e., λ1 and λ2 are the eigenvalues of R^T B R. Therefore, λ being a principal curvature is equivalent to λ satisfying R^T B R's characteristic equation: |λE-R^T B R|=0. Since R is invertible, |R|≠0, thus |λA-B|=0.
Theorem: The two principal curvatures correspond to orthogonal principal directions.
Proof: Given km = (λ1d̃u^2 + λ2d̃v^2) / (d̃u^2 + d̃v^2), when d̃v=0, km=λ1; when d̃u=0, km=λ2. Therefore, (d̃u, 0) and (0, d̃v) are the principal directions corresponding to the principal curvatures λ1 and λ2. Thus, the principal directions in the parametric coordinates u and v are given by:
(d̃u/d̃v)_1 = w(d̃u/d̃v) = (a1, a2)(d̃u/d̃v) = d̃u·a1 = d̃r1
(d̃u/d̃v)_2 = w(d̃u/d̃v) = (a1, a2)(d̃u/d̃v) = d̃u·a2 = d̃r2
Here, the principal directions (d̃r1, d̃r2) in the tangent space's basis are the coordinate vectors of the two principal direction vectors e1 and e2 in the basis {e1, e2}. Let ẽj = (r̃u, r̃v) (w1j w2j) = (r̃u, r̃v) aj j=1,2. Therefore, (ẽ1, ẽ2) = (w11r̃u + w12r̃v, w21r̃u + w22r̃v) = (w11, w21) (r̃u r̃v r̃u^2 r̃v^2) (w12) = E. Thus, (ẽ1, ẽ2) = ã1^T A ã2 = ã2^T A ã1 = 0. The two principal directions are orthogonal.
Page 51
- Gaussian curvature (total curvature) and mean curvature.
Gaussian curvature K = k1 * k2
Mean curvature H = 1/2(k1 + k2)
Where k1, k2 are the principal curvatures at point P.
According to |λ(EF) - (LM) | = 0, i.e. (EG - F^2)λ^2 - (LG - 2MF + NE)λ + (LN - M^2) = 0
We get k = |LM| / |EF| * |MN| / |FG|
H = 1/2(|LM| / |EF| + |MN| / |FG|)
Theorem: Let S1, S2 be two regular surfaces defined in D ⊂ R^2. If at each point (u, v) ∈ D, S1 and S2 have the same first and second fundamental forms, then S1 and S2 can be translated and rotated to coincide.
Page 52
Multivariable Calculus and Its Differential Calculus
- n-Dimensional Euclidean Space
$$\mathbb{R}^n = \{ (x_1, x_2, \ldots, x_n) \mid x_i \in \mathbb{R}, i = 1, 2, \ldots, n \}$$
The set $\mathbb{R}^n$ forms a linear space over the real numbers under addition and scalar multiplication.
Euclidean Distance:
$$\|X - Y\|_n = \sqrt{\sum_{i=1}^n (x_i - y_i)^2}$$
Where:
$$X = (x_1, x_2, \ldots, x_n) \in \mathbb{R}^n$$
$$Y = (y_1, y_2, \ldots, y_n) \in \mathbb{R}^n$$
Properties of Euclidean Distance:
- Non-negativity: $\|X - Y\| \geq 0$, and it is zero only when $X = Y$.
- Symmetry: $\|X - Y\| = \|Y - X\|$.
- Triangle Inequality: For all $Z \in \mathbb{R}^n$, $\|X - Y\| \leq \|X - Z\| + \|Z - Y\|$.
δ-Neighborhood of a Point:
$$B(X_0, \delta) = \{ X \in \mathbb{R}^n \mid \|X - X_0\| < \delta \}$$
δ-Neighborhood of a Point Without the Point Itself:
$$B(X_0, \delta) = \{ X \in \mathbb{R}^n \mid 0 < \|X - X_0\| < \delta \}$$
Interior Point:
$S \subset \mathbb{R}^n$, $X_0 \in \mathbb{R}^n$, $\exists X_0 \in B(X_0, \delta) \subset S$, $X_0$ is an interior point of $S$.
Boundary Point:
$S \subset \mathbb{R}^n$, $X_0 \in \mathbb{R}^n$, $\forall \delta > 0$, $B(X_0, \delta) \cap S \neq \emptyset$, $B(X_0, \delta) \nsubseteq S$, $X_0$ is a boundary point of $S$.
Closure of a Set:
$\overline{S} = \mathbb{R}^n \setminus \text{Interior}(S)$ (the interior of $S$ consists of all points that are not boundary points).
Interior of a Set:
The interior of $S$ is the set of all interior points of $S$.
Boundary of a Set:
The boundary of $S$ is the set of all boundary points of $S$.
Closure of a Set:
$\overline{S} = S \cup \partial S$ (the closure of $S$ is the union of $S$ and its boundary).
- Sequence of Points in $\mathbb{R}^n$
$D \subset \mathbb{R}^2$ ($\mathbb{R}^3$) is connected: $\forall \xi, \eta \in D$, $\exists$ line connecting $\xi$ and $\eta$, $line \subset D$.
Open Set:
$D \subset \mathbb{R}^2$ ($\mathbb{R}^3$) is non-empty and connected.
Sequence of Points in $\mathbb{R}^n$:
Let $X_k = (x_1^{(k)}, x_2^{(k)}, \ldots, x_n^{(k)}) \in \mathbb{R}^n$, $k = 1, 2, \ldots$ be a sequence of points in $\mathbb{R}^n$.
Then $\{X_k\}$: $x_1, x_2, \ldots, x_k, \ldots$ is a sequence of points in $\mathbb{R}^n$.
Page 53
A = (a1, a2, ..., an) ∈ R^n, lim Xk = A :
∀ε > 0, ∃N0 ∈ N*, ∀k > N0, ||Xk - A|| < ε
⇔ lim Xk = ai, i = 1, 2, ..., n.
R^n 中的 Cauchy 序列:
∀ε > 0 ∃N0 ∈ N*, ∀m, k > N0, ||Xm - Xk|| < ε. ⇔ R^n 是完备的.
- 函数
n元函数: f: Ω ⊂ R^n → R^l
X → u
显式表示: u = f(X), X = (x1, x2, ..., xn).
隐式表示: F(x1, x2, ..., xn, u) = 0.
向量值函数: f: Ω ⊂ R^n → R^m
X → Y
f(Ω) = {Y ∈ R^m | ∃X ∈ Ω, Y = f(X)}
其中每个分量 yj (j = 1, 2, ..., m) 都是 X = (x1, x2, ..., xn) 的 n元函数
yj = fj(x1, x2, ..., xn), (x1, x2, ..., xn) ∈ Ω
复变函数: f: Ω ⊂ C → C
(x, y) ∈ Ω → (u, v).
{
实部函数 u = u(x, y)
虚部函数 v = v(x, y).
}
Page 54
- Limits and Continuity
Definition of Limit:
f: Ω ⊂ R^n → R^m, x_0 ∈ R^n. For x in the neighborhood B(x_0, δ) of x_0, there exists A ∈ R^m. For all ε > 0, there exists δ > 0, such that for all x ∈ Ω, 0 < ||x - x_0||_n < δ:
||f(x) - A||_m < ε
lim (x→x_0) f(x) = A.
When m = 1, it reduces to the limit of a single function:
∀ε > 0, ∃δ > 0, ∀x ∈ Ω ⊂ R^n, 0 < ||x - x_0||_n < δ: |f(x) - A| < ε.
lim (x→x_0) f(x) = A ⇔ lim (x→x_0) f_j(x) = a_j, j = 1, 2, ..., m. A = (a_1, a_2, ..., a_m).
For vector-valued functions (including n-ary functions), Cauchy convergence criterion and the equivalence between function limits and sequence limits.
Definition of Continuity:
∀ε > 0, ∃δ > 0, ∀x: ||x - x_0||_n < δ, such that ||f(x) - f(x_0)||_m < ε.
When m = 1, it reduces to the continuity of a single function.
⇌ m component functions are continuous.
If x_0 is a boundary point of the domain Ω of the vector-valued function f(x), continuity is defined as:
∀ε > 0, ∃δ > 0, ∀x ∈ B(x_0, δ) ∩ Ω, such that ||f(x) - f(x_0)|| < ε.
Extreme Value Theorem:
Ω ⊂ R^n is a bounded closed set, n-ary function f ∈ C(Ω), then
∃ξ, η ∈ Ω, ∀x ∈ Ω, m = f(ξ) ≤ f(x) ≤ f(η) = M.
f(x_1) = λ, f(x_2) = μ, ∃x ∈ Ω, f(x) = σ ∈ (λ, μ).
kth Order Infinitesimal:
∃β > 0, δ > 0, such that ∀x ∈ B_0(x_0, δ): |f(x)| ≤ βρ^k.
Where ρ^k = ||x - x_0||^k = (∑(x_i - x_0^i)^2)^(k/2).
Page 55
- Multivariable Functions Differentiability and Partial Derivatives
A function u=f(X) is defined in B(X₀, r), where
X = (x₁, x₂, ..., xₙ)
X₀ = (x₁^(0), x₂^(0), ..., xₙ^(0))
For all X ∈ B(X₀, r), Δu = f(X) - f(X₀) = a₁(x₁ - x₁^(0)) + ... + aₙ(xₙ - xₙ^(0)) + o(ρ)
Then u=f(X) is differentiable at X₀.
du = a₁(x₁ - x₁^(0)) + a₂(x₂ - x₂^(0)) + ... + aₙ(xₙ - xₙ^(0))
= a₁dx₁ + a₂dx₂ + ... + aₙdxₙ, called u's total differential at X₀.
Clearly, when ΔX = X - X₀ → 0, f(X) → f(X₀), so differentiable functions must be continuous.
Partial derivatives:
∂u/∂x₁(X₀, x₂^(0), ..., xₙ^(0)) = lim Δx₁→0 Δx₁u/Δx₁, where u is differentiable at X₀.
∂u/∂x₂(X₀), ..., ∂u/∂xₙ(X₀) all exist, and a₁ = ∂u/∂x₁(X₀).
du = ∂u/∂x₁dx₁ + ∂u/∂x₂dx₂ + ... + ∂u/∂xₙdxₙ. Conversely, not true.
Δ according to definition to judge differentiability:
(Δu = f(X) - f(X₀)) = u - ∑ (∂f/∂xᵢ(X₀)dxᵢ) + p higher-order infinitesimal. Thus lim Δu - ∑ (∂f/∂xᵢdxᵢ) / p ≠ 0.
Δ according to partial derivatives' continuity to judge differentiability:
If all partial derivatives ∂u/∂xᵢ(X) at X₀ are continuous, then f(X) is differentiable at X₀.
Conversely, not true.
Page 56
The partial derivative reflects the change rate of a multivariable function along the coordinate axes. The directional derivative reflects the change rate of the function along any direction in space. Let T = (cosα, cosβ, cosγ).
Given P(x, y, z) ∈ B(P₀, δ), the directional derivative at P₀ in the direction of T is defined as:
$$\frac{\partial u}{\partial T}\bigg|_{P_0} = \lim_{P \to P_0} \frac{f(P) - f(P_0)}{\|P - P_0\|}$$
Therefore, Δu = f(P) - f(P₀) = (∂u/∂x)Δx + (∂u/∂y)Δy + (∂u/∂z)Δz + o(δ)
If Δx = ρcosα, Δy = ρcosβ, Δz = ρcosγ.
Then, (∂u/∂T)P₀ = limₚ→ₚ₀ (∂u/∂x)P₀cosα + (∂u/∂y)P₀cosβ + (∂u/∂z)P₀cosγ.
Generally, (∂u/∂T)P₀ = ∑ (∂u/∂xᵢ)P₀cosαᵢ.
= (∂u/∂x₁, ∂u/∂x₂, ..., ∂u/∂xₙ) . (cosα₁, cosα₂, ..., cosαₙ)
= grad u(P₀) . T̂₀ = ∇u(P₀) . T̂₀.
Higher-order partial derivatives: Differentiation of partial derivatives.
Let Ω ⊂ R^n be an open set. If the n-variable function u = f(x) has k-order partial derivatives that are continuous in Ω, then u is said to be k-times continuously differentiable in Ω: u ∈ C^k(Ω). At this time, u's mixed partial derivatives (2 ≤ r ≤ k) are independent of the order of differentiation.
Higher-order differentials: Prove d^n z = (∂/∂x + ∂/∂y)^n z
= (∑ from k=0 to n C_n^k (∂^k z / ∂x^(n-k) ∂y^k) dx^(n-k) dy^k)
= ∑ from k=0 to n C_n^k (∂^k z / ∂x^(n-k) ∂y^k) dx^(n-k) dy^k.
For general n-variable functions, the same can be derived using a similar but more cumbersome method.
Page 57
- Vector-valued function differentiation.
The total differential of f(X) at X0 is:
$$df(X_0) = A\Delta X + o(\|\Delta X\|)$$
where A is the Jacobian matrix.
For an n-variable function, Δf = (a1, a2, ..., an) ΔX. Its Jacobian matrix is a row vector:
$$\therefore \Delta f_j = (\frac{\partial f_1}{\partial x_1}, \frac{\partial f_1}{\partial x_2}, ..., \frac{\partial f_1}{\partial x_n}) \Delta X$$
$$\therefore J(f(X_0)) = \begin{bmatrix} \frac{\partial f_1}{\partial x_1} & \frac{\partial f_1}{\partial x_2} & ... & \frac{\partial f_1}{\partial x_n} \\ \frac{\partial f_2}{\partial x_1} & \frac{\partial f_2}{\partial x_2} & ... & \frac{\partial f_2}{\partial x_n} \\ \vdots & \vdots & \ddots & \vdots \\ \frac{\partial f_m}{\partial x_1} & \frac{\partial f_m}{\partial x_2} & ... & \frac{\partial f_m}{\partial x_n} \end{bmatrix} \bigg|_{X_0}$$
For a composite vector-valued function f(g), we have:
$$d(f\circ g)(X_0) = \frac{\partial (y_1, y_2, ..., y_m)}{\partial (u_1, u_2, ..., u_m)} \bigg|_{X_0} \frac{\partial (u_1, u_2, ..., u_m)}{\partial (x_1, x_2, ..., x_n)} \bigg|_{X_0} dX$$
If k=1, then:
$$\frac{\partial y}{\partial x_i} = \frac{\partial f}{\partial u_1} \frac{\partial u_1}{\partial x_i} + \frac{\partial f}{\partial u_2} \frac{\partial u_2}{\partial x_i} + ... + \frac{\partial f}{\partial u_m} \frac{\partial u_m}{\partial x_i}, \quad i=1, 2, ..., n$$
For a general composite function y = f(u1, u2, ..., um), where ui = gj(x1, x2, ..., xn). If f and gj are in C^2, then:
$$\frac{\partial^2 y}{\partial x_i \partial x_j} = \frac{\partial}{\partial x_i} \left( \frac{\partial y}{\partial x_j} \right) = \frac{\partial}{\partial x_i} \left( \sum_{s=1}^m \frac{\partial y}{\partial u_s} \frac{\partial u_s}{\partial x_j} \right)$$
$$= \sum_{s=1}^m \left[ \frac{\partial}{\partial x_i} \left( \frac{\partial y}{\partial u_s} \right) \frac{\partial u_s}{\partial x_j} + \frac{\partial y}{\partial u_s} \frac{\partial}{\partial x_i} \left( \frac{\partial u_s}{\partial x_j} \right) \right]$$
$$= \sum_{s=1}^m \left[ \left( \sum_{t=1}^m \frac{\partial^2 y}{\partial u_s \partial u_t} \cdot \frac{\partial u_t}{\partial x_i} \right) \cdot \frac{\partial u_s}{\partial x_j} + \frac{\partial y}{\partial u_s} \cdot \frac{\partial^2 u_s}{\partial x_i \partial x_j} \right]$$
Page 58
- Implicit (Vector-valued) Function Existence and Differentiability
Given an n+1 variable function F(x1, x2, ..., xn, y) in a neighborhood B(P0, r) of point P0(x0, y0), if F(x0, y0) = 0 and ∂F/∂y(x0, y0) ≠ 0, then there exists a δ such that B(x0, δ) × (y0 - η, y0 + η) ⊂ B(P0, r) and for all X ∈ B(x0, δ), there exists a unique y ∈ (y0 - η, y0 + η) satisfying F(X, y) = 0. This establishes an n-variable function y = f(X), such that F(X, f(X)) = 0. And
$$\frac{\partial y}{\partial x_i} = -\frac{\frac{\partial F}{\partial x_i}(X, y)}{\frac{\partial F}{\partial y}(X, y)}, \quad i = 1, 2, ..., n.$$
Given m n+m variable functions F1(x1, ..., xn, y1, ..., ym) (i = 1, 2, ..., m) in a neighborhood B(P0, r) of point P0, if Fi(P0) = 0 and ∂(F1, ..., Fm)/∂(y1, ..., ym) is invertible at P0, then there exists a δ and η such that B(x0, δ) × B(y0, η) ⊂ B(P0, r) and for all X ∈ B(x0, δ), there exists a unique Y ∈ B(y0, η) satisfying Fi(X, Y) = 0, i = 1, 2, ..., m. This establishes a vector-valued implicit function Y = f(X). And
$$J(f(X)) = \frac{\partial(y_1, ..., y_m)}{\partial(x_1, ..., x_n)} = -\left(\frac{\partial(F_1, ..., F_m)}{\partial(y_1, ..., y_m)}\right)^{-1} \frac{\partial(F_1, ..., F_m)}{\partial(x_1, ..., x_n)}.$$
If Y = f(X), X = g(Y), Rn → Rn, and J(f(X)) is invertible, then J(g(Y)) = [J(f(X))]^{-1}. Thus, J(f^(-1)(Y)) = [J(f(X))]^{-1}. (For the equation F(X, Y) = 0 = f(X) - Y, the implicit vector-valued function X = g(Y) is determined.)
Page 59
- Surface and Curve Representation, Tangent Plane and Normal Line
- Surface by Implicit Function Method:
$$ S: z = f(x, y) $$
Tangent Plane: $$ z - z_0 = \frac{\partial f}{\partial x}(x - x_0) + \frac{\partial f}{\partial y}(y - y_0) $$
Normal Vector: $$ \vec{n} = \left( \frac{\partial f}{\partial x}, \frac{\partial f}{\partial y}, -1 \right) \Big|_{p_0} $$
Normal Line: $$ \frac{x - x_0}{\frac{\partial f}{\partial x}(x_0, y_0)} = \frac{y - y_0}{\frac{\partial f}{\partial y}(x_0, y_0)} = \frac{z - z_0}{-1} $$
2. Surface by Parametric Representation:
$$ S: \begin{cases} x = x(u, v) \\ y = y(u, v) \\ z = z(u, v) \end{cases} (u, v) \in D_{uv} \subset \mathbb{R}^2 $$
Tangent Plane: $$ \begin{cases} x - x_0 = \frac{\partial x}{\partial u}(u - u_0) + \frac{\partial x}{\partial v}(v - v_0) \\ y - y_0 = \frac{\partial y}{\partial u}(u - u_0) + \frac{\partial y}{\partial v}(v - v_0) \\ z - z_0 = \frac{\partial z}{\partial u}(u - u_0) + \frac{\partial z}{\partial v}(v - v_0) \end{cases} $$
Normal Vector: $$ \vec{n} = \left( \frac{D(y, z)}{D(u, v)}, \frac{D(z, x)}{D(u, v)}, \frac{D(x, y)}{D(u, v)} \right) \Big|_{(u_0, v_0)} $$
Normal Line: $$ \frac{x - x_0}{\frac{D(y, z)}{D(u, v)}} = \frac{y - y_0}{\frac{D(z, x)}{D(u, v)}} = \frac{z - z_0}{\frac{D(x, y)}{D(u, v)}} $$
- Surface by Implicit Function Method:
$$ S: F(x, y, z) = 0 $$
Tangent Plane: $$ \frac{\partial F}{\partial x}(x - x_0) + \frac{\partial F}{\partial y}(y - y_0) + \frac{\partial F}{\partial z}(z - z_0) = 0 $$
Normal Vector: $$ \vec{n} = \left( \frac{\partial F}{\partial x}, \frac{\partial F}{\partial y}, \frac{\partial F}{\partial z} \right) \Big|_{p_0} $$
Normal Line: $$ \frac{x - x_0}{\frac{\partial F}{\partial x}} = \frac{y - y_0}{\frac{\partial F}{\partial y}} = \frac{z - z_0}{\frac{\partial F}{\partial z}} $$
Page 60
- Parametric Representation of Space Curves
$$L: \begin{cases} x = x(t) \\ y = y(t) \\ z = z(t) \end{cases} t \in [a, b]$$
Tangent Line:
$$\begin{cases} x - x_0 = x'(t - t_0) \\ y - y_0 = y'(t - t_0) \\ z - z_0 = z'(t - t_0) \end{cases}$$
Normal Plane:
$$x'(x - x_0) + y'(y - y_0) + z'(z - z_0) = 0$$
5. Parametric Representation of Space Curves Using Surface Intersections:
$$L: \begin{cases} F_1(x, y, z) = 0 \\ F_2(x, y, z) = 0 \end{cases}$$
If F1 and F2 are differentiable at P0(x0, y0, z0), then the tangent planes at P0 are:
$$\begin{cases} \frac{\partial F_1}{\partial x}(x - x_0) + \frac{\partial F_1}{\partial y}(y - y_0) + \frac{\partial F_1}{\partial z}(z - z_0) = 0 \\ \frac{\partial F_2}{\partial x}(x - x_0) + \frac{\partial F_2}{\partial y}(y - y_0) + \frac{\partial F_2}{\partial z}(z - z_0) = 0 \end{cases}$$
The tangent plane at P0 is the plane containing these two tangent planes. The tangent direction is:
$$\vec{T} = (\frac{\partial F_1}{\partial x}, \frac{\partial F_1}{\partial y}, \frac{\partial F_1}{\partial z}) \times (\frac{\partial F_2}{\partial x}, \frac{\partial F_2}{\partial y}, \frac{\partial F_2}{\partial z})$$
- Taylor Formula for Multivariable Functions
$$f(x) = f(x_0) + J(f(x_0)) \Delta x + \frac{1}{2!}(A x)^T H(x^*) A x$$
$$J(f(x_0)) = (\frac{\partial f}{\partial x_1}, \frac{\partial f}{\partial x_2}, \ldots, \frac{\partial f}{\partial x_n})|_{x_0}$$
$$H(x) = \begin{bmatrix} \frac{\partial^2 f}{\partial x_1^2} & \frac{\partial^2 f}{\partial x_1 \partial x_2} & \ldots & \frac{\partial^2 f}{\partial x_1 \partial x_n} \\ \frac{\partial^2 f}{\partial x_2 \partial x_1} & \frac{\partial^2 f}{\partial x_2^2} & \ldots & \frac{\partial^2 f}{\partial x_2 \partial x_n} \\ \vdots & \vdots & \ddots & \vdots \\ \frac{\partial^2 f}{\partial x_n \partial x_1} & \frac{\partial^2 f}{\partial x_n \partial x_2} & \ldots & \frac{\partial^2 f}{\partial x_n^2} \end{bmatrix}$$
Page 61
Proof: Define a function g(t) = f(x_0 + tΔx). Since g(1) = f(x) and g(0) = f(x_0), g(t) at t=0 follows the first-order Taylor formula. Thus, g(t) = g(0) + g'(0)t + 1/2 g''(0)t^2 for t ∈ (0,1). Where g'(t) = ∂f/∂x_1 Δx_1 + ∂f/∂x_2 Δx_2 + ... + ∂f/∂x_n Δx_n = J(f(x_0 + tΔx)) Δx. g'(0) = J(f(x_0)) Δx. g''(t) = ∂/∂t (∂f/∂x_1 Δx_1) = ∑ (∂f/∂x_i) Δx_i = ∑ ∑ (∂^2f/∂x_i ∂x_j) Δx_i Δx_j = (∆x)^T H(x_0 + tΔx) (∆x). g''(0t) = (∆x)^T H(x_0 + otΔx) (∆x).
Take t=1, we get:
If x_0 is a local maximum (minimum) of f, then ∇f(x_0) = 0. The condition is that f is differentiable at x_0. x_0 is a critical point, it is a necessary condition for local extremum. Suppose f is twice continuously differentiable in a neighborhood B(x_0, δ) of x_0. If the Hessian matrix H(x_0) is positive definite, then x_0 is a local minimum; otherwise, x_0 is a local maximum.
Conditional Extremum Problem:
Find min(max) f(x_1, x_2, ..., x_n) subject to φ(x_1, x_2, ..., x_n) = 0.
Given f: Ω ⊂ R^n → R, φ: Ω ⊂ R^n → R are both continuously differentiable functions, and ∂φ/∂x_i is not zero. Then the extremum of the conditional extremum problem in Ω is a Lagrange function:
L(x_1, x_2, ..., x_n, λ) = f(x_1, x_2, ..., x_n) + λφ(x_1, x_2, ..., x_n).
The Lagrange function satisfies:
∂L/∂x_i = ∂f/∂x_i + λ ∂φ/∂x_i = 0 for i = 1, 2, ..., n-1.
∂L/∂λ = φ(x_1, x_2, ..., x_n) = 0.
Thus, ∂f/∂x_i = -λ ∂φ/∂x_i for i = 1, 2, ..., n-1. Therefore, ∂f/∂x_i / ∂φ/∂x_i = -λ.
Page 62
Double Integrals
Double Integral Algorithms:
- Iterated Integration Method:
$$\iint_{D} f(x, y) \, dx dy = \int_{a}^{b} dx \int_{g(x)}^{h(x)} f(x, y) \, dy$$
$$= \int_{c}^{d} dy \int_{x(y)}^{x(y)} f(x, y) \, dx$$
- Change of Variables Method:
$$\iint_{D} f(x, y) \, dx dy = \iint_{D'} f(x(u, v), y(u, v)) \left| \frac{\partial(x, y)}{\partial(u, v)} \right| \, du dv$$
#### Triple Integrals
#### Triple Integral Algorithms:
1. Iterated Integration Method:
$$\iiint_{D} f(x, y, z) \, dv = \int_{a}^{b} dx \int_{g(x)}^{h(x)} dy \int_{z(x, y)}^{z(x, y)} f(x, y, z) \, dz$$
- Change of Variables Method:
$$\iiint_{D} f(x, y, z) \, dv = \iiint_{D'} f(x(u, v, w), y(u, v, w), z(u, v, w)) \left| \frac{\partial(x, y, z)}{\partial(u, v, w)} \right| \, du dv dw$$
#### Cylindrical Coordinates:
$$x = \rho \cos \phi, \quad y = \rho \sin \phi, \quad z = z$$
$$\frac{\partial(x, y, z)}{\partial(\rho, \phi, z)} = \begin{vmatrix} \cos \phi & -\rho \sin \phi & 0 \\ \sin \phi & \rho \cos \phi & 0 \\ 0 & 0 & 1 \end{vmatrix} = \rho$$
$$\therefore \iiint_{D} f(x, y, z) \, dv = \iiint_{D'} f(\rho \cos \phi, \rho \sin \phi, z) \, \rho \, d\rho d\phi dz$$
Spherical Coordinates:
$$x = r \cos \theta \cos \phi, \quad y = r \cos \theta \sin \phi, \quad z = r \sin \theta$$
$$\frac{\partial(x, y, z)}{\partial(r, \theta, \phi)} = \begin{vmatrix} \cos \theta \cos \phi & -r \sin \theta \cos \phi & r \cos \theta \cos \phi \\ \cos \theta \sin \phi & -r \sin \theta \sin \phi & r \cos \theta \sin \phi \\ \sin \theta & 0 & r \cos \theta \end{vmatrix} = r^2 \sin \theta$$
$$\therefore \iiint_{D} f(x, y, z) \, dv = \iiint_{D'} f(r \cos \theta \cos \phi, r \cos \theta \sin \phi, r \sin \theta) r^2 \sin \theta \, dr d\theta d\phi$$
Page 63
Surface Area Problem:
Space surface parametric equations: x = x(u, v), y = y(u, v), z = z(u, v). It determines a vector-valued function F:
F(u, v) = (x, y, z) = (x(u, v), y(u, v), z(u, v))
When ∂(x, y, z) / ∂(u, v) is non-zero, O-uv is O-xyz in the point-to-point correspondence.
(Surface) (Surface)
Given S = ∫∫ √(EG - F²) dudv. Where
E = F_u² = (∂x/∂u)² + (∂y/∂u)² + (∂z/∂u)²
G = F_v² = (∂x/∂v)² + (∂y/∂v)² + (∂z/∂v)²
F = F_u - F_v = (∂x/∂u - ∂x/∂v)² + (∂y/∂u - ∂y/∂v)² + (∂z/∂u - ∂z/∂v)²
Specifically, for x, y parametric equations, F(x, y) = (x, y, f(x, y)), (x, y) ∈ Dxy
Then S = ∬ √(1 + (∂z/∂x)² + (∂z/∂y)²) dxdy. Where
E = 1 + (∂z/∂x)²
G = 1 + (∂z/∂y)²
F = (∂z/∂x)² + (∂z/∂y)²
(∂z/∂x)² = (∂z/∂x)² + (∂z/∂x)² + (∂z/∂x)²
(∂z/∂y)² = (∂z/∂y)² + (∂z/∂y)² + (∂z/∂y)²
(∂z/∂x)² + (∂z/∂y)² = (∂z/∂x)² + (∂z/∂y)² + (∂z/∂x)² + (∂z/∂y)²
Page 64
III. First-Kind Line Integrals and First-Kind Surface Integrals
Integral Domain
Definite Integral in R' interval
Line Integral along a curve
Surface Integral over a surface
- First-Kind Line Integral
$$\int_{L} f(x, y, z) dl = \sum_{i=1}^{n} f(\xi_i, \eta_i, \zeta_i) \Delta l$$
(1) $$\int_{AB} f(x, y, z) dl = \int_{BA} f(x, y, z) dl$$
(2) $$\int_{L} f(x, y, z) dl = \int_{L_1} f(x, y, z) dl + \int_{L_2} f(x, y, z) dl$$
(3) Mean Value Theorem: If the function f(x, y, z) is continuous on L, then there exists a point (ξ, η, ζ) in L such that $$\int_{L} f(x, y, z) dl = f(\xi, \eta, \zeta) L$$
$$dl = \sqrt{[x'(t)]^2 + [y'(t)]^2 + [z'(t)]^2} dt$$
$$\therefore \int_{L} f(x, y, z) dl = \int_{a}^{b} f(x(t), y(t), z(t)) \sqrt{[x'(t)]^2 + [y'(t)]^2 + [z'(t)]^2} dt$$
2. First-Kind Surface Integral
$$\iint_{S} f(x, y, z) ds = \sum_{i=1}^{n} f(\xi_i, \eta_i, \zeta_i) \Delta S_i$$
Assume S is smooth, its parametric equation is F(x, y, z) = F(u, v)
$$\iint_{S} f(x, y, z) ds = \iint_{D_{uv}} f(x(u, v), y(u, v), z(u, v)) \sqrt{EG - F^2} du dv$$
Page 65
Chapter IV. Second Kind Line Integrals and Second Kind Surface Integrals
1. Second Kind Line Integrals
$$\int_{L(A)}^{(B)} F \cdot d\vec{r} = \int_{L(B)}^{(B)} X(x, y, z) dx + Y(x, y, z) dy + Z(x, y, z) dz$$
$$= \lim_{\Delta \to 0} \sum_{i=1}^{n} [X(\xi_i, \eta_i, \zeta_i) \Delta x_i + Y(\xi_i, \eta_i, \zeta_i) \Delta y_i + Z(\xi_i, \eta_i, \zeta_i) \Delta z_i]$$
(1) $$\int_{L(A)}^{(B)} F \cdot d\vec{r} = -\int_{L(B)}^{(A)} F \cdot d\vec{r}$$
(2) $$\int_{L(A)}^{(A)} F \cdot d\vec{r} = \int_{L(A_1)}^{(A_2)} F \cdot d\vec{r} + \int_{L(A_2)}^{(A_1)} F \cdot d\vec{r}$$
(3) When the path L is a closed curve, the counterclockwise direction is the positive direction: $$\oint_{L^+} F \cdot d\vec{r}$$
$$\oint_{L^+} F \cdot d\vec{r} = \oint_{L^+} F \cdot d\vec{r} + \oint_{L^+} F \cdot d\vec{r}$$
Given L is a smooth curve in R^3: $$x = x(t), y = y(t), z = z(t)$$
Vector function $$F(x, y, z) = (P(x, y, z), Q(x, y, z), R(x, y, z))$$
Then $$\int_{L(A)}^{(B)} P(x, y, z) dx = \int_{a}^{b} P(x(t), y(t), z(t)) x'(t) dt$$
$$\int_{L(A)}^{(B)} Q(x, y, z) dy = \int_{a}^{b} Q(x(t), y(t), z(t)) y'(t) dt$$
$$\int_{L(A)}^{(B)} R(x, y, z) dz = \int_{a}^{b} R(x(t), y(t), z(t)) z'(t) dt$$
The sum of these is $$\int_{L(A)}^{(B)} (P, Q, R) \left(\frac{dx}{dt}, \frac{dy}{dt}, \frac{dz}{dt}\right) dt$$
Simultaneously, $$\int_{L(A)}^{(B)} F \cdot d\vec{r} = \int_{L(A)}^{(B)} P dx + Q dy + R dz = \int_{L(A)}^{(B)} (P \cos \alpha + Q \cos \beta + R \cos \gamma) dl$$
This is related to the first kind line integral.
Page 66
- Second Kind Surface Integral
Surface Orientation:
$$\begin{cases} x = x(u, v) \\ y = y(u, v) \\ z = z(u, v) \end{cases} (u, v) \in D_{uv} \subset R^2$$
$$\vec{n}_2 = \pm \left( \frac{\partial x}{\partial u}, \frac{\partial y}{\partial u}, \frac{\partial z}{\partial u} \right) \times \left( \frac{\partial x}{\partial v}, \frac{\partial y}{\partial v}, \frac{\partial z}{\partial v} \right) \neq \vec{0}$$
$$\iint \vec{V}(x, y, z) \cdot d\vec{S} = \lim_{\Delta S \to 0} \sum_{i=1}^n \vec{V}(Q_i) \cdot \Delta \vec{S}_i = \iint \vec{V} \cdot d\vec{S} \cdot \vec{n}^0$$
$$= \iint \left( \begin{array}{c} \cos \alpha \\ \cos \beta \\ \cos \gamma \end{array} \right) \cdot d\vec{S} = \iint (P, Q, R) \left( \begin{array}{c} \cos \alpha \\ \cos \beta \\ \cos \gamma \end{array} \right) \cdot d\vec{S} = \iint (P \cos \alpha + Q \cos \beta + R \cos \gamma) d\vec{S}$$
$$= \iint P dy \wedge dz + Q dz \wedge dx + R dx \wedge dy$$
$$\left\{ \begin{array}{l} \text{Orientation:} \quad \iint \vec{V} \cdot d\vec{S} = -\iint \vec{V} \cdot d\vec{S} \\ \text{Additivity (consistent side):} \quad \iint \vec{V} \cdot d\vec{S} = \iint_{S_1} \vec{V} \cdot d\vec{S} + \iint_{S_2} \vec{V} \cdot d\vec{S} \end{array} \right.$$
$$\because \vec{n}^0 = \pm \frac{\vec{r}_u \times \vec{r}_v}{\|\vec{r}_u \times \vec{r}_v\|} = \pm \frac{(A, B, C)}{\sqrt{A^2 + B^2 + C^2}} = \pm (\cos \alpha, \cos \beta, \cos \gamma)$$
$$\left( \text{where } A = \frac{D(y, z)}{D(u, v)}, B = \frac{D(z, x)}{D(u, v)}, C = \frac{D(x, y)}{D(u, v)} \right)$$
$$d\vec{S} = \sqrt{A^2 + B^2 + C^2} \, du \, dv = \sqrt{EG - F^2} \, du \, dv$$
$$\left\{ \begin{array}{l} dy \wedge dz = \cos \alpha \, d\vec{S} = \pm A \, du \, dv \\ dz \wedge dx = \cos \beta \, d\vec{S} = \pm B \, du \, dv \\ dx \wedge dy = \cos \gamma \, d\vec{S} = \pm C \, du \, dv \end{array} \right.$$
$$\therefore \iint P \, dy \wedge dz + Q \, dz \wedge dx + R \, dx \wedge dy = \pm \iint (P \cos \alpha + Q \cos \beta + R \cos \gamma) \, d\vec{S}$$
$$\therefore \iint \vec{V} \, d\vec{S} = \iint (P \cos \alpha + Q \cos \beta + R \cos \gamma) \, d\vec{S}$$
$$\text{Meanwhile,} \quad \iint \vec{V} \, d\vec{S} = \iint (P \cos \alpha + Q \cos \beta + R \cos \gamma) \, d\vec{S} \text{ is related to the first kind surface integral.}$$
Page 67
- Vector Fields in Plane and Space
- Plane Green's Theorem:
$$\oint_{\partial D} \vec{F} \cdot d\vec{r} = \iint_D P \, dx + Q \, dy = \iint_D \text{rot} \, \vec{F} \cdot d\vec{S} = \iint_D \left( \frac{\partial Q}{\partial x} - \frac{\partial P}{\partial y} \right) \, dx \, dy$$
For a general multiply connected domain, add auxiliary lines and use Green's theorem.
$$\oint_{\partial D} \vec{F} \cdot d\vec{r} = \iint_D P \, dx + Q \, dy$$
In domain D, the integral is path-independent.
$$\Leftrightarrow \oint_{\partial D} \vec{F} \cdot d\vec{r} = 0 \Leftrightarrow \frac{\partial Q}{\partial x} = \frac{\partial P}{\partial y}$$
The integral of $P(x, y) \, dx + Q(x, y) \, dy$ in a simply connected domain D has an antiderivative.
$$\Leftrightarrow \int_{A_1(x_1, y_1)}^{A_2(x_2, y_2)} P \, dx + Q \, dy = u(x_2, y_2) - u(x_1, y_1)$$
- Gauss's Theorem:
$$\iint_{\partial D} \vec{F} \cdot d\vec{r} = \iiint_D \left( \frac{\partial P}{\partial x} + \frac{\partial Q}{\partial y} + \frac{\partial R}{\partial z} \right) \, dx \, dy \, dz = \iiint_D \text{div} \, \vec{F} \cdot dV$$
3. Green's Theorem in Space → Stokes' Theorem:
$$\oint_{\partial S} \vec{F} \cdot d\vec{r} = \iint_S \text{rot} \, \vec{F} \cdot d\vec{S}$$
Where $\text{rot} \, \vec{F} = \begin{vmatrix} \vec{i} & \vec{j} & \vec{k} \\ \frac{\partial}{\partial x} & \frac{\partial}{\partial y} & \frac{\partial}{\partial z} \\ P & Q & R \end{vmatrix} = (Q_y - Q_z) \vec{i} + (P_z - P_x) \vec{j} + (Q_x - P_y) \vec{k}$
Therefore,
$$\oint_{\partial S} \vec{F} \cdot d\vec{r} = \iint_S \left( Q_y - Q_z \right) \, dy \, dz + \left( P_z - P_x \right) \, dz \, dx + \left( Q_x - P_y \right) \, dx \, dy$$
If the region S contains points that do not satisfy the conditions, add auxiliary lines to ensure the path does not include these points.
Page 68
No.
Date
$$\text{rot} \vec{V} = 0 \Leftrightarrow \text{A closed curve } L, \oint \vec{V} \cdot d\vec{r} = 0.$$
$$\Leftrightarrow \text{A, B are points in } \Omega. \int_{(A)}^{(B)} \vec{V} \cdot d\vec{r} \text{ is path independent } (L \subset \Omega).$$
$$\Leftrightarrow \exists u(x, y, z), du = Pdx + Qdy + Rdz.$$
$$\Leftrightarrow \int_{(A)}^{(B)} \vec{V} \cdot d\vec{r} = u(B) - u(A).$$
Summary:
- Gradient of a scalar function: $f \rightarrow \text{grad} f = \nabla f = \frac{\partial f}{\partial x} \hat{i} + \frac{\partial f}{\partial y} \hat{j} + \frac{\partial f}{\partial z} \hat{k}$
- Divergence of a vector field: $\vec{V} \rightarrow \text{div} \vec{V} = \nabla \cdot \vec{V} = \frac{\partial P}{\partial x} + \frac{\partial Q}{\partial y} + \frac{\partial R}{\partial z}$
- Curl of a vector field: $\vec{V} \rightarrow \text{rot} \vec{V} = \nabla \times \vec{V} = \begin{vmatrix} \hat{i} & \hat{j} & \hat{k} \\ \frac{\partial}{\partial x} & \frac{\partial}{\partial y} & \frac{\partial}{\partial z} \\ P & Q & R \end{vmatrix}$
Summary of vector field operations:
- Gradient: direction of steepest ascent and rate of change.
- Conservative field (potential field) / Non-conservative field.
- Potential function $u: du = Pdx + Qdy + Rdz$.
- Divergence: whether the field has a source.
- Source field / Sink field.
- Curl: rotational state of the field.
- Rotational field / Non-rotational field.
Example: Maxwell's equations.
$$\oint \vec{E} \cdot d\vec{s} = 0, \oint \vec{B} \cdot d\vec{s} = 0.$$
$$\nabla \cdot \vec{E} = \rho$$
$$\nabla \times \vec{E} + \frac{\partial \vec{B}}{\partial t} = 0$$
$$\nabla \cdot \vec{B} = 0$$
$$\nabla \times \vec{B} - \frac{\partial \vec{E}}{\partial t} = 0$$
Part II: Algebra, Geometry, and Advanced Calculus II
Complex numbers, complex plane topology, limits, continuity, Cauchy-Riemann equations, differentiation, and analytic functions.
Page 1
Chapter 5. Complex Variable Functions and Calculus
1. Complex Plane
The set of complex numbers \( \mathbb{C} \) forms a field under the operations of addition, subtraction, multiplication, and division. The zero element is \( z = 0 \), and the unit element is \( z = 1 \). The complex plane \( \mathbb{C} \) can be considered as a two-dimensional vector space over the real numbers \( \mathbb{R} \), with a basis of \( \{1, i\} \). Alternatively, it can be viewed as a one-dimensional vector space over the complex numbers \( \mathbb{C} \), with a basis of \( \{1\} \).
If we let \( (0,1) = i \) and \( (1,0) = 1 \), then any complex number \( z = x + iy \in \mathbb{C} \) can be expressed as \( z = x(1,0) + y(0,1) = x + iy \). The real part \( x = \text{Re}(z) \) and the imaginary part \( y = \text{Im}(z) \). Also, \( i^2 = -1 \).
2. Complex Functions
A complex function \( f: \Omega \subset \mathbb{C} \to \mathbb{C} \) maps points in the complex plane to other points in the complex plane. If \( z = x + iy \), then \( w = u + iv \), where \( u = u(x,y) \) and \( v = v(x,y) \). The function can also be written as \( w = f(z) \).
The function \( f \) can be decomposed into two component functions:
- \( u = u(x,y) \) : the real part function
- \( v = v(x,y) \) : the imaginary part function
A complex function can have multiple values: e.g., \( w = \sqrt[n]{z} = (x^2 + y^2)^{\frac{1}{2n}} \left[ \cos \frac{\theta + 2k\pi}{n} + i \sin \frac{\theta + 2k\pi}{n} \right] \), where \( k = 0, 1, \ldots, n-1 \).
3. Limits and Continuity
For a function \( w = f(z) \), if the real part and imaginary part of \( f \) have limits at a point, then the function is continuous at that point. The limit of \( w = f(z) \) at \( z_0 \) is defined as:
\[ \Delta w = f(z_0 + \Delta z) - f(z_0) \]
where \( \Delta z = \Delta x + i \Delta y \), and \( a = \alpha + i \beta \). The function \( f \) is said to be differentiable at \( z_0 \) if:
\[ \Delta w = a \Delta z + o(\Delta z) \]
where \( o(\Delta z) \) represents a term that is higher order than \( \Delta z \).
4. Differentiation of Complex Functions
The derivative of a complex function \( f \) at \( z_0 \) is defined as:
\[ f'(z_0) = \lim_{\Delta z \to 0} \frac{\Delta w}{\Delta z} \]
where \( \Delta w = a \Delta z + o(\Delta z) \). The function \( f \) is said to be differentiable at \( z_0 \) if this limit exists.
Page 2
The derivative of a complex function f(z) at z₀ is defined as the limit of the difference quotient as Δz approaches zero. The function can be expressed as a sum of its real and imaginary parts, u and v, respectively. The change in the function Δw is given by the sum of the real part of the change in u and the imaginary part of the change in v. The change in u and v can be expressed as linear combinations of Δx and Δy, with coefficients α and β. The derivative of the function is then given by the sum of the real and imaginary parts of the derivative of u and v, respectively. The function is differentiable at z₀ if and only if u and v are differentiable at (x₀, y₀) and satisfy the Cauchy-Riemann equations. The integral of a complex function f(z) along a path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral can be expressed as the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of the real and imaginary parts of the integrals of u and v along the path. The integral of f(z) along the path L is given by the sum of
Page 3
- Analytic Functions - Extension of Elementary Real Functions
ω = f(z) is defined in Ω ⊂ C. z₀ is an interior point of Ω. If there exists r > 0 such that f(z) is differentiable in B(z₀, r), then f(z) is analytic at z₀.
Functions analytic in Ω:
- Analytic function ⇔ partial derivatives (∂u/∂x, ∂u/∂y, ∂v/∂x, ∂v/∂y) are continuous in Ω.
- Points in Ω where f(z) is not analytic: singular points.
f(x) is a real-valued function defined on I ⊂ R. F(z) is a complex-valued function defined in Ω ⊂ C. If I ⊂ Ω, x ∈ I, f(x) = F(x), and F(z) is analytic in Ω, then F(z) is an analytic extension of f(x).
- Polynomials: P(z) = aₙzⁿ + aₙ₋₁zⁿ⁻¹ + ... + a₁z + a₀. In C, the analytic extension is:
P(z) = aₙzⁿ + aₙ₋₁zⁿ⁻¹ + ... + a₁z + a₀
- Trigonometric and Hyperbolic Functions:
P(x) = cos(x) = (e^(ix) + e^(-ix))/2, Q(x) = sin(x) = (e^(ix) - e^(-ix))/2i
In C, the analytic extension is:
P(z) = cos(z), Q(z) = sin(z).
sin(z) = (e^(iz) - e^(-iz))/2i = (e^(ix-y) - e^(-ix+y))/2i = (cos(x) + i sin(x) - e^y) / 2i
= -i cos(x) + i sin(x) - e^y / 2i cos(x) - 2i sin(x)
= 1/2e^(ix) (cos(x) - i sin(x)) + e^y / 2 (cos(x) + i sin(x))
= (1/2e^(ix) + e^y / 2) cos(x) - i (e^y / 2 - 1/2e^(ix)) sin(x)
= 1/2e^(ix) (cos(x) + i sin(x)) + e^y / 2 (cos(x) - i sin(x))
= 1/2 (e^(ix) - e^(-ix)) = 1/2 (e^(ix) - e^(-ix)) = cos(z).
∴ d/dz (cos(z)) = (1/2e^(ix) + e^y / 2) cos(x) - i (e^y / 2 - 1/2e^(ix)) sin(x)
= 1/2e^(ix) (cos(x) + i sin(x)) + e^y / 2 (cos(x) - i sin(x))
= 1/2 (e^(ix) - e^(-ix)) = 1/2 (e^(ix) - e^(-ix)) = sin(z).
Similarly, d/dz (cos(z)) = sin(z).
Page 4
The hyperbolic sine and cosine functions are also analytic on the complex plane:
$$\sinh z = \frac{e^z - e^{-z}}{2}, \quad \cosh z = \frac{e^z + e^{-z}}{2}$$
3. Root Functions:
$$\omega_k = n\sqrt{z} = n\sqrt{|z|} \cdot e^{i\frac{\arg z + 2k\pi}{n}}, \quad k = 0, 1, 2, \ldots, n-1$$
If the argument range is appropriately restricted, the multi-valued function can be "single-valued".
- Logarithmic Functions:
$$\omega = \ln z \quad \text{where} \quad z = re^{i\theta}, \quad \omega = u + iv$$
$$e^{u+iv} = re^{i\theta} = e^{u+iv}$$
$$\therefore u = \ln r, v = \theta + 2k\pi$$
$$\therefore \ln z = (\ln r + i(\theta + 2k\pi)) = \ln|z| + i\arg z$$
The principal branch of the logarithm is:
$$\ln z = \ln|z| + i\arg z + i2k\pi, \quad k \text{ is a fixed positive integer}$$
The logarithmic function is multi-valued.
In the cut complex plane, the logarithmic function is locally analytic.
$$\frac{d(\ln z)}{dz} = \frac{d(\ln(x^2+y^2) + i\arctan\frac{y}{x})}{dz} = \frac{x}{(x^2+y^2)} - \frac{iy}{(x^2+y^2)^2} = \frac{1}{z}$$
- Integration of Analytic Functions:
The complex function $f(z) = u(x,y) + iv(x,y)$ in the complex plane can be integrated along a simple closed path $\gamma$ from $A$ to $B$ as the sum of two line integrals:
$$\int_{\gamma(A)}^{(B)} f(z) dz = \int_{\gamma(A)}^{(B)} u dx - v dy + i \int_{\gamma(A)}^{(B)} v dx + u dy$$
Cauchy's Integral Theorem:
If $f(z)$ is analytic in the interior of a simple closed path $\gamma$ and continuous on $\gamma$, then:
$$\oint_{\gamma} f(z) dz = 0$$
This theorem is similar to Green's theorem, where the partial derivatives satisfy:
$$\frac{\partial u}{\partial x} = \frac{\partial v}{\partial y}$$
This is the Cauchy-Riemann condition.
Page 5
Cauchy Integral Formula:
Let the region D have boundary ∂D. If f(ξ) is analytic in D and continuous in D + ∂D, then f(ξ) = 1/(2πi) ∫(D) f(ξ) / (ξ - z) dξ.
Proof: ∀z ∈ D, let F(ξ) = f(ξ) / (ξ - z). Then F(ξ) is analytic in D except at z.
Consider a circle Lp centered at z with radius ρ > 0. By Cauchy's integral theorem in the complex plane, ∫(∂D) f(ξ) / (ξ - z) dξ + ∫(Lp) f(ξ) / (ξ - z) dξ = 0.
∴ ∫(∂D) f(ξ) / (ξ - z) dξ = ∫(Lp) f(ξ) / (ξ - z) dξ.
To prove the original formula, we need to show ∫(Lp) f(ξ) / (ξ - z) dξ = 2πi f(z).
Let ξ - z = ρe^(iθ), θ: 0~2π.
Then ∫(Lp) dξ / (ξ - z) = ∫(0 to 2π) iρe^(iθ) / ρe^(iθ) dθ = i ∫(0 to 2π) dθ = 2πi.
∴ | ∫(Lp) f(ξ) / (ξ - z) dξ - 2πi f(z) | = | ∫(Lp) f(ξ) / (ξ - z) dξ - ∫(Lp) f(z) / (ξ - z) dξ |
= | ∫(Lp) (f(ξ) - f(z)) / (ξ - z) dξ |
∵ ∀ε > 0, ∃δ > 0, such that 0 < ρ < δ, | f(ξ) - f(z) | < ε.
∴ | ∫(Lp) f(ξ) / (ξ - z) dξ - 2πi f(z) | < ε.
Thus, the function f(ξ) in region D can be uniquely determined by its values on the boundary ∂D.
Furthermore, f^(n)(z) = n! / (2πi) ∫(∂D) f(ξ) / (ξ - z)^(n+1) dξ, z ∈ D, n = 1, 2, 3.
By Cauchy's integral formula,
f(z + Δz) - f(z) / Δz = 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ
= 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ
∴ | f(z + Δz) - f(z) / Δz - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ |
= 1/2πi | ∫(∂D) f(ξ) / (ξ - z - Δz)(ξ - z) dξ - 1/2πi
Page 6
The text and formulas in the image are as follows:
---
In ∂D, |f(ξ)| is bounded. Let M be the upper bound, ξ ∈ ∂D. Let d be the shortest distance from z to ∂D. Then when ξ ∈ ∂D, |ξ - z| ≥ d > 0. And when |Δz| < d/2, |ξ - z - Δz| ≥ |ξ - z| - |Δz| ≥ d/2. Thus, for ε > 0, take δ = min{d/2, επd²/M}. Let L be the length of ∂D. When |Δz| < δ, the original expression ≤ δ/2π * M/L² * ∫ |ξ| dξ ≤ ε. Therefore, f'(z) = lim(Δz→0) [f(z + Δz) - f(z)] / Δz = 1/2πi * ∫ f(ξ) / (ξ - z)² dξ.
Assuming the result holds for n, then for n+1 (the result was proven for n=1 in the previous section).
f^(n+1)(z) = [f^(n)(z)]'
= n! / 2πi * ∫ [f(ξ) / (ξ - z - Δz)^(n+1) - f(ξ) / (ξ - z)^(n+1)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z - Δz)^(n+1) - f(ξ) / (ξ - z)^(n+1)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z - Δz)^(n+1) - f(ξ) / (ξ - z)^(n+1)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! / 2πi * ∫ [f(ξ) / (ξ - z)^(n+1)] * [1 - (n+1)Δz / (ξ - z)] dξ
= n! /
Page 7
- Series convergence necessary condition: $\lim_{n\to\infty} u_n = 0$
Series convergence sufficient condition: $S_n \leq M$ (bounded)
- Tests
- Cauchy test: $\forall \epsilon \in \mathbb{N} \, \forall n>N, \sum_{k=n}^{n+p} u_k < \epsilon$, sufficient
- Comparison: $\exists N \, \forall n>N \, u_n \leq u_n$ series diverges, series converges
- D'Alembert's ratio test: $\exists N \, \forall n>N \, \frac{u_{n+1}}{u_n} < 1$ converges, $>1$ diverges
Ratio test limit form: $\lim_{n\to\infty} \frac{u_{n+1}}{u_n} = q$ $q<1$ converges, $q>1$ diverges
- Cauchy root test: $\exists N \, \forall n>N \, \sqrt[n]{u_n} \leq 1$ converges, $>1$ diverges
Root test limit form: $\lim_{n\to\infty} \sqrt[n]{u_n} = l$ $0
- Integral test: $f(x) [1,+\infty)$ non-decreasing, then $\sum f(n) = \sum u_n$ converges ($\sum u_n < \int_1^{\infty} f(x) dx$)
- Raabe test: $\lim_{n\to\infty} n \left(1 - \frac{u_{n+1}}{u_n}\right) = r$ $r>1$ converges, $r<1$ diverges
Essentially compares with $\sum \frac{1}{n^r}$
- Kummer test: Suppose $\sum u_n$ is a positive series, $c_1, c_2, \ldots, c_n$ is a sequence of positive numbers
Let $K_n = c_n - c_{n+1} \frac{u_{n+1}}{u_n}$, if $\exists N, A>0, \forall n>N, K_n \geq A$ then $\sum u_n$ converges, $K_n \leq 0$ then $\sum u_n$ diverges
Proof:
- $K_n = c_n - c_{n+1} \frac{u_{n+1}}{u_n} \geq A > 0$
- $\sum c_n u_n$ is decreasing and has a lower bound, so $c_n u_n$ has a limit
$\sum (c_n u_n - c_{n+1} u_{n+1}) = \sum (c_n u_n - c_{n+1} u_{n+1})$ converges
$\sum A u_n = A \sum u_n$ converges, so $\sum u_n$ converges
- $K_n = c_n - c_{n+1} \frac{u_{n+1}}{u_n} \leq 0$ then $u_n \geq \frac{1}{c_n}$ when $\sum \frac{1}{c_n}$ diverges, $\sum u_n$ diverges
Example: $c_n = n$, $\sum \frac{1}{n}$ diverges, use Raabe test
Example: $c_n = n$, $\sum \frac{1}{n}$ diverges, use Raabe test
Example: $c_n = n$, $\sum \frac{1}{n}$ diverges, use Raabe test
Page 8
- Alternating Series' Leibniz Test Method:
- If (-1)^(n-1)U_n (U_n > 0) and U_n ≥ U_{n+1} and U_n → 0, then ∑(-1)^(n-1)U_n converges.
- (Can also prove convergence of ∑U_n and ∑(-1)^(n-1)U_n by Cauchy's convergence test).
Absolute Convergence: ∑|U_n| converges implies ∑U_n converges (U_n and ∑U_n are alternating series).
Conditional Convergence: ∑U_n converges, ∑|U_n| diverges. (∑U_n and ∑|U_n| both diverge)
- The rearrangement of absolutely convergent series is also absolutely convergent, with the same sum S. Using D'Alembert and Cauchy's direct method, the rearrangement of conditionally convergent series can be equal to any real number (Riemann).
- If U_n = U and V_n = V, then ∑U_n and ∑V_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_nU_n = ∑U_n
Page 9
The series $$\sum_{n=0}^{\infty} \frac{x^{5n}}{(5n)!}$$ can be solved by recognizing that $$I(x) = I(x)$$, solving this differential equation, and substituting the boundary condition $$I(0) = 1$$, $$I^{(n)}(0) = 0$$ for $$n = 1, 2, ..., 5-1$$. This yields the closed-form solution for the series.
The text discusses the properties of a function $$f(x)$$, where $$f(x)$$ is a polynomial of degree $$n$$, and $$f(x)$$ is continuous and differentiable. The function $$f(x)$$ is defined as $$f(x) = x^2 + 1$$. The text also mentions that $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text further discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f(x)$$, such as its derivative and integral. The text also mentions that the function $$f(x)$$ is a quadratic function, and it is continuous and differentiable everywhere. The text also discusses the properties of the function $$f
Page 10
- Analytic Functions and Power Series Expansion
Analytic functions ↔ interior points are differentiable ↔ Cauchy-Riemann conditions / Cauchy integral theorem / power series (Laurent series & Taylor series).
Complex series: ∑∞k=1 ak
A ≤ 0, ∃N, n>N when |Sn - S(= ∑∞k=1 ak)| < ε
A ≤ 0, ∃N, A ≤ p, p>N when |ap - ap-1| < ε
Complex series: If D is a point in the complex plane, fk(z) defined in D, k=1, 2, ...
A ∈ D, series ∑∞k=1 fk(z) = f1(z) + ... * fk(z) converges, then ∑∞k=1 fk(z) in D converges, and the function f(z) = ∑∞k=1 fk(z)
If the series converges in D for z, ∀ε > 0, ∃N(ε), it does not depend on z. When n > N(ε), ∀z ∈ D, then f(z) in D converges.
If each term fk(z) in D is continuous, then the function f(z) is also continuous in D;
If each term fk(z) in D is continuous, then ∫f(z)dz = ∑∞k=1 ∫fk(z)dz;
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(z)
If each term fk(z) in D is continuous, then f(z) is also continuous in D. The derivative f'(z) = ∑∞k=1 f'(k)(
Page 11
Delta Laurent Series / Double Laurent Series
Each complex power series ∑Cn(z-a)^n has a convergence circle |z-a|=R.
In the interior of the convergence circle, the series converges absolutely.
In the exterior, the series diverges.
On the boundary, the series may converge or diverge.
∀0<ε If f(z) = ∑Cn(z-a)^n in |z-a| The coefficients satisfy Cp = f^(p)(z) / p! For ∑Cn(z-a)^n, let s = 1/(z-a), then ∑Cn(z-a)^n = C1s + C2s^2 + ... + Cn-1s^(n-1) + ... |s| = |1/(z-a)| < 1/(0 < 1/(z-a) < ∞). For |z-a| > r, ∑Cn(z-a)^n converges. Therefore, ∑Cn(z-a)^n = ∑Cn(z-a)^n + ∑Cn(z-a)^n converges, the convergence domain is r < |z-a| < R. Laurent's Theorem: In the annulus H: r < |z-a| < R (0 ≤ r < R < ∞) of a function f(z) that can be expanded into a double Laurent series f(z) = ∑Cn(z-a)^n, where Cn = ∫(f(z) / (z-a)^(n+1)) dz, n=0, ±1, ... L+ is the circle |z-a|=r (r Proof: ∀z ∈ H: r < |z-a| < R, there exists p1, p2, r < p1 < p2 < R. Let L+ be the circle |z-a|=r (r
Page 12
By Cauchy's integral formula,
$$f(z) = \frac{1}{2\pi i} \oint_{\Gamma_2} \frac{f(\xi)}{\xi - z} d\xi - \frac{1}{2\pi i} \oint_{\Gamma_1} \frac{f(\xi)}{\xi - z} d\xi$$
1) For the first integral:
$$\frac{f(\xi)}{\xi - z} - \frac{f(\xi)}{(\xi - a) - (z - a)} = \frac{f(\xi)}{\xi - a} \cdot \frac{1}{1 - \frac{z - a}{\xi - a}}$$
When $\xi \in \Gamma_2$, $|\xi - a| = r_2 > |z - a|$, so $|\frac{z - a}{\xi - a}| < 1$
$$\therefore \frac{f(\xi)}{\xi - z} = \frac{f(\xi)}{\xi - a} \left(1 - \frac{z - a}{\xi - a}\right)^{-1} = \frac{f(\xi)}{\xi - a} \sum_{n=0}^{\infty} \left(\frac{z - a}{\xi - a}\right)^n$$
The Laurent series in $|\xi - a| = r_2$ converges uniformly, so we can integrate term by term,
$$\therefore \frac{1}{2\pi i} \oint_{\Gamma_2} \frac{f(\xi)}{\xi - z} d\xi = \frac{1}{2\pi i} \oint_{\Gamma_2} \sum_{n=0}^{\infty} \left(\xi - a\right)^n \frac{f(\xi)}{(\xi - a)^{n+1}} d\xi$$
$$= \sum_{n=0}^{\infty} \left(\xi - a\right)^n \cdot \frac{1}{2\pi i} \oint_{\Gamma_2} \frac{f(\xi)}{(\xi - a)^{n+1}} d\xi \quad (C_n = \frac{1}{2\pi i} \oint_{\Gamma_2} \frac{f(\xi)}{(\xi - a)^{n+1}} d\xi)$$
$$= \sum_{n=0}^{\infty} C_n (z - a)^n$$
- For the second integral:
$$\frac{f(\xi)}{\xi - z} = -\frac{f(\xi)}{(\xi - a) - (z - a)} = -\frac{f(\xi)}{z - a} \cdot \frac{1}{1 - \frac{\xi - a}{z - a}}$$
When $\xi \in \Gamma_1$, $|\xi - a| = r_1 < |z - a|$, so $|\frac{\xi - a}{z - a}| < 1$
$$\therefore \frac{f(\xi)}{\xi - z} = -\frac{f(\xi)}{z - a} \left(1 - \frac{\xi - a}{z - a}\right)^{-1} = -\frac{f(\xi)}{z - a} \sum_{n=1}^{\infty} \left(\frac{\xi - a}{z - a}\right)^n$$
The Laurent series in $|\xi - a| = r_1$ converges uniformly, so we can integrate term by term,
$$\therefore \frac{1}{2\pi i} \oint_{\Gamma_1} \frac{f(\xi)}{\xi - z} d\xi = -\frac{1}{2\pi i} \oint_{\Gamma_1} \sum_{n=1}^{\infty} \frac{1}{(z - a)^n} f(\xi) (\xi - a)^n d\xi = -\sum_{n=1}^{\infty} \frac{C_n}{(z - a)^n}$$
($C_n = \frac{1}{2\pi i} \oint_{\Gamma_1} \frac{f(\xi)}{(\xi - a)^{n+1}} d\xi$)
Page 13
The Laurent series of a function f(z) is given by:
$$f(z) = \sum_{n=0}^{\infty} C_n (z-a)^n + \sum_{n=1}^{\infty} \frac{C_n}{(z-a)^n} = \sum_{n=-\infty}^{\infty} C_n (z-a)^n$$
For a function f(z) to be analytic in the annulus |z-a| $$f(z) = \sum_{n=0}^{\infty} C_n (z-a)^n$$ where $$C_n = \frac{1}{2\pi i} \oint_{\gamma} \frac{f(s)}{(s-a)^{n+1}} ds = \frac{f^{(n)}(a)}{n!}, n=0,1,2,...$$ This is the Taylor series of f(z) in the annulus 0<|z-a| Examples: $$e^z = \sum_{n=0}^{\infty} \frac{z^n}{n!}, |z|<+\infty$$ $$\sin z = \sum_{n=0}^{\infty} \frac{(-1)^n z^{2n+1}}{(2n+1)!}, |z|<+\infty$$ $$\cos z = \sum_{n=0}^{\infty} \frac{(-1)^n z^{2n}}{(2n)!}, |z|<+\infty$$ $$(1+z)^{-1} = \sum_{n=0}^{\infty} (-1)^n z^n, |z|<1$$ The Laurent series of a function f(z) has a convergence radius r < |z-a| < R. In the annulus |z-a|=r, |z-a|=R, there is at least one singularity of f(z). Specifically, if f(z) can be Taylor expanded, the convergence radius is |z-a| Based on the above theorem, we can determine the Laurent (Taylor) series convergence radius by analyzing the singularities of f(z). m-order zero point: If f(z) is analytic in the annulus |z-a| $$f(z) = C_m (z-a)^m + C_{m+1} (z-a)^{m+1} + ... = (z-a)^m [C_m + C_{m+1} (z-a) + ...]$$ $$= (z-a)^m \varphi(z), \varphi(z) is analytic at z=a, C_m \neq 0, a is a zero point of f(z) of order m. Then a is a zero point of f(z) of order m.
Page 14
Laurent Series:
- Removable singularity: In 0 < |z - a| < R, f(z) = ∑∞ n=1 Cn(z - a)n. When n = 1, 2, 3, ..., Cn = 0.
- Example: sin(z)/z
- Pole of order m: When n = m+1, m+2, ..., Cn = 0, Cm ≠ 0.
- Example: cos(z) - pole of order 1
- Principal part: f(z) = Cm(z - a)^{-m} + ... + C0 + C1(z - a) + ... + Cn(z - a)^n + ..., a is the principal part of the pole.
- Essential singularity: There are infinitely many Cm not equal to zero.
- Example: e^(1/z)
Residue:
- Suppose a is f(z) removable singularity. Then there exists R > 0, in 0 < |z - a| < R, f(z) is analytic.
- f(z) = ∑∞ n=0 a(z - a)^n
- Where C1 = ∫_L+ f(z) dz. L+: |z - a| = ρ, (0 < ρ < R).
- C1 is the residue of f(z) at z = a. By Cauchy's integral theorem, Res(f(z), z = a) is related to the integral.
Residue Theorem:
- L+ is a closed path, f(z) is analytic in the region D except for finite singularities a1, ..., an.
- If f(z) is continuous in the region {a1, ..., an}, then ∫_L+ f(z) dz = ∑_k=1^n ∫_L+ f(z) dz = 2πi ∑_k=1^n ∫_L+ f(z) dz = 2πi ∑_k=1^n Res(f(z), z = ak).
If a is f(z) pole of order m, then f(z) = φ(z) / (z - a)^m. Where φ(z) is analytic at z = a and φ(a) ≠ 0. Then Res(f(z), z = a) = C1 = ∫_L+ f(z) dz = 1/(2πi) ∫_L+ φ(z) / (z - a)^m dz = φ^(m-1)(a) / (m-1)!.
Logarithmic Residue:
- ∫_L+ f(z) dz / f(z) dz is the logarithmic residue of f(z).
- If a is f(z) n-th zero point, then Res(f(z), z = a) = n.
- If b is f(z) m-th pole point, then Res(f(z), z = b) = -m.
Page 15
Rouche's Theorem: Let L be a closed path, and f(z) and φ(z) be analytic functions inside and on L. If |f(z)| > |φ(z)| on L, then inside L, f(z) and f(z) + φ(z) have the same number of zeros.
Proof:
$$f(z) + \varphi(z) = f(z)\left[1 + \frac{\varphi(z)}{f(z)}\right]$$
$$\Delta_{L} \text{Arg}[f(z) + \varphi(z)] = \Delta_{L} \text{Arg}[f(z)] + \Delta_{L} \text{Arg}\left[1 + \frac{\varphi(z)}{f(z)}\right]$$
$$\Delta_{L} \text{Arg}\left[1 + \frac{\varphi(z)}{f(z)}\right] = 0$$
Thus,
$$\Delta_{L} \text{Arg}[f(z) + \varphi(z)] = \Delta_{L} \text{Arg}[f(z)] = N(f, L) = \frac{1}{2\pi i} \oint_{L} \frac{f'(z)}{f(z)} dz$$
If f(z) has no poles inside L, then
$$\frac{1}{2\pi i} \oint_{L} \frac{f'(z)}{f(z)} dz = N(f, L) = n$$
$$= \text{Res}\left[\frac{f'(z)}{f(z)}\right] = \frac{1}{2\pi i} \oint_{L} \frac{d}{dz}[\ln f(z)] dz$$
The argument of the logarithm reflects the change in the argument of f(z) when z traverses L. At point z, the real and imaginary parts of ln f(z) are respectively (ln |f(z)|, φ). After one full traversal, the real part remains ln |f(z)| and the imaginary part is φ. Thus,
$$\frac{1}{2\pi i} \oint_{L} \frac{f'(z)}{f(z)} dz = \frac{1}{2\pi i} \left\{\left[\ln |f(z)| + i\varphi\right] - \left[\ln |f(z_0)| + i\varphi_0\right]\right\}$$
$$= \frac{\varphi - \varphi_0}{2\pi}$$
$$= \frac{\Delta_{L} \text{Arg}[f(z)]}{2\pi}$$
$$= N(f, L) - P(f, L)$$
If f(z) has no poles inside L, then
$$\frac{\Delta_{L} \text{Arg}[f(z)]}{2\pi} = N(f, L)$$
Rouche's Theorem: Let L be a closed path, and f(z) and φ(z) be analytic functions inside and on L. If:
1. They are both analytic inside and on L.
2. On L, |f(z)| > |φ(z)|.
Then inside L, f(z) and f(z) + φ(z) have the same number of zeros.
Page 16
Chapter 6. Ordinary Differential Equations
1. Second-Order Linear Ordinary Differential Equations
1.1 Structure of Second-Order Linear Homogeneous Equations
Given the second-order linear homogeneous equation \( y'' + p(x) y' + q(x) y = 0 \), if \( y_1 \) and \( y_2 \) are two solutions, then \( \{y_1, y_2\} \) forms a basis of the solution space of the linear space (rank = 2). Thus, \( y(x) = c_1 y_1(x) + c_2 y_2(x) \), where \( y_1(x) \) and \( y_2(x) \) are linearly independent. Otherwise, \( W(y_1, y_2) = W(x) = \begin{vmatrix} y_1(x) & y_2(x) \\ y_1'(x) & y_2'(x) \end{vmatrix} = 0 \).
1.2 Structure of Second-Order Linear Non-Homogeneous Equations
Given the second-order linear non-homogeneous equation \( y'' + p(x) y' + q(x) y = f(x) \), if \( z_1(x) \) and \( z_2(x) \) are two solutions, then \( z_1(x) + p(x) z_1(x) + q(x) z_1(x) = f(x) \) and \( z_2(x) + p(x) z_2(x) + q(x) z_2(x) = f(x) \). Thus, \( [z_1(x) - z_2(x)]'' + p(x) [z_1(x) - z_2(x)]' + q(x) [z_1(x) - z_2(x)] = 0 \). Hence, \( z_1(x) - z_2(x) \) is a solution to the homogeneous equation \( y'' + p(x) y' + q(x) y = 0 \). Thus, \( z(x) = c_1 y_1(x) + c_2 y_2(x) + z_0(x) \), where \( z_0(x) \) is a particular solution of the non-homogeneous equation.
1.3 Constructing Linearly Independent Solutions
Given \( y_1 \) is a non-zero solution of the homogeneous equation \( y'' + p(x) y' + q(x) y = 0 \), and \( y_2 \) is a linearly independent solution. Since \( y_2 \neq c_1 y_1 \), we have \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Therefore, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \). Thus, \( y_2 = c_1 y_1 + c_2 y_1' \
Page 17
The original equation becomes: $$(2y_1 + p(x)y_1)C'(x) + y_1C''(x) = 0$$
Solving for $C'(x)$ gives: $$C'(x) = \frac{1}{y_1^2}e^{-\int p(x)dx}$$
Thus: $$C(x) = \int \frac{e^{-\int p(x)dx}}{y_1^2}dx$$
Therefore: $$y_2 = y_1\int \frac{e^{-\int p(x)dx}}{y_1^2}dx$$
- Using the two linearly independent solutions of the homogeneous equation, construct a particular solution of the non-homogeneous equation. Let the particular solution be $z_0 = C_1(x)y_1(x) + C_2(x)y_2(x)$, where $C_1(x), C_2(x)$ are linearly independent. Then: $$z_0' = C_1(x)y_1(x) + C_2(x)y_2(x) + C_1'(x)y_1(x) + C_2'(x)y_2(x)$$
Given $z_0' = 0$, we have: $$z_0'' = C_1(x)y_1''(x) + C_2(x)y_2''(x) + C_1'(x)y_1''(x) + C_2'(x)y_2''(x)$$
Substituting $z_0, z_0', z_0''$ into the non-homogeneous equation gives: $$C_1y_1 + C_2y_2 = f(x)$$
Since: $$C_1(x)y_1(x) + C_2(x)y_2(x) = 0$$
Thus: $$C_1'(x) = -\frac{y_2(x)f(x)}{W(x)}$$
$$C_2'(x) = \frac{y_1(x)f(x)}{W(x)}$$
Therefore: $$z_0 = C_1(x)y_1(x) + C_2(x)y_2(x) = \int_{x_0}^{x} \frac{y_2(t)f(t)}{W(t)}dt$$
3) Solving the homogeneous equation: $$y'' + py' + qy = 0$$
Assume $y = e^{\lambda x}$, then: $$(\lambda^2 + p\lambda + q)e^{\lambda x} = 0$$
1) If $\Delta > 0$, $y = C_1e^{\lambda_1x} + C_2e^{\lambda_2x}$
2) If $\Delta < 0$, $\lambda_{1,2} = \alpha \pm \beta i$. Thus: $$y = e^{\alpha x}(C_1\cos(\beta x) + C_2\sin(\beta x))$$
3) If $\Delta = 0$, $y_1 = e^{-\frac{p}{2}x}$, using the method of undetermined coefficients, we get: $$y_2 = xe^{-\frac{p}{2}x}$$
Page 18
$$\therefore y = a e^{-\frac{b}{2}x} + c_2 xe^{-\frac{b}{2}x}$$
Solve higher-order equations in the same way.
(4) Solve non-homogeneous equations.
- Generally, $z_0 = \int_{x_0}^{x} \frac{y_1(t) y_1(x) + y_1(t) y_2(x)}{W(t)} \cdot f(t) \cdot dt$.
- $z = a y_1 + c_2 y_2 + z_0$.
Specifically, if the non-homogeneous term $f(x) = P_n(x) e^{ax}$ (a is a complex number, $P_n$ is a polynomial).
1) If $\mu$ is not a characteristic root of the homogeneous equation, then the non-homogeneous differential equation has the form:
$$z_0 = Q_n(x) e^{ax}$$
- If $\mu$ is a single characteristic root of the homogeneous equation, then:
$$z_0 = x Q_n(x) e^{ax}$$
3) If $\mu$ is a double characteristic root of the homogeneous equation, then:
$$z_0 = x^2 Q_n(x) e^{ax}$$
Substitute the special solution $z_0(x)$ into the non-homogeneous equation. Through comparison of coefficients, determine $Q_n(x)$.
For $y'' + py' + qy = f(x)$, $f(x) = Re[f_1(x) + f_2(x)]$ (where $f_1(x) e^{ax}$ also satisfies $P_n(x) e^{ax}$).
We need to solve the two equations:
$$f_1(x) = y'' + py' + qy$$
$$f_2(x) = y'' + py' + qy$$
The solution $z_1(x) + z_2(x)$ is the special solution of the non-homogeneous equation.
For the Euler equation $\sum_{i=0}^{n} a_i x^i y^{(i)} = 0$, by substituting $t = \ln x$, we can obtain a homogeneous equation with constant coefficients.
Page 19
- First-order linear constant coefficient differential equations.
$$\dot{y} = Ay + f(x)$$
$$\dot{y} = (y_1, y_2, ..., y_n)^T, A(x) = (a_{ij}(x))_{n \times n}, f(x) = (f_1(x), f_2(x), ..., f_n(x))^T$$
Second-order linear constant coefficient differential equations are a special case of first-order linear differential equations.
- Structure of the solution of first-order linear differential equations.
The solution set of a first-order linear constant coefficient differential equation is a linear space of dimension n. The general solution of a non-homogeneous equation can be expressed as the general solution of the corresponding homogeneous equation plus a particular solution of the non-homogeneous equation.
- Solution of constant coefficient linear systems.
Given $$\dot{y} = Ay + f(x)$$, the corresponding homogeneous equation $$\dot{y} = Ay$$ has the solution $$y = Pe^{Ax}$$. Substituting into $$(A - \lambda E)P = 0$$, we get $$\lambda$$ is an eigenvalue of A, and P is an eigenvector of A.
- If the eigenvalues $$\lambda_1$$ and $$\lambda_2$$ are not equal, the corresponding eigenvectors $$\vec{p}_1$$ and $$\vec{p}_2$$ are linearly independent. The general solution of $$\dot{y} = Ay$$ is $$\vec{p}_1 e^{\lambda_1 x}$$ and $$\vec{p}_2 e^{\lambda_2 x}$$.
The general solution of $$\dot{y} = Ay + f(x)$$ is $$y = C_1 \vec{p}_1 e^{\lambda_1 x} + C_2 \vec{p}_2 e^{\lambda_2 x}$$.
If the eigenvalues are complex conjugates, the corresponding eigenvectors are linearly dependent. The general solution of $$\dot{y} = Ay$$ is $$\vec{p}_1 e^{\lambda_1 x}$$ and $$\vec{p}_2 e^{\lambda_2 x}$$.
- If the eigenvalues $$\lambda_1 = \lambda_2$$, the general solution of $$\dot{y} = Ay$$ is $$\vec{p}_1 e^{\lambda_1 x} + \vec{p}_2 xe^{\lambda_1 x}$$.
The general solution of $$\dot{y} = Ay + f(x)$$ is $$y = C_1 \vec{p}_1 e^{\lambda_1 x} + C_2 \vec{p}_2 xe^{\lambda_1 x}$$.
For systems of more than two equations, the rules are similar.
Page 20
Delta Constant Substitution Method:
For $$\vec{y}' = A\vec{y} + f(x)$$, let $$\vec{y} = C(x)\vec{\phi}(x)$$, substitute into the original equation to get:
$$\vec{\phi}'(x)C(x) + \vec{\phi}(x)C'(x) = A\vec{\phi}(x)C(x) + f(x)$$.
$$\therefore \vec{\phi}'(x) = A\vec{\phi}(x)$$
$$\therefore \vec{\phi}'(x)C(x) = A\vec{\phi}(x)C(x)$$
$$\therefore \vec{\phi}(x)C'(x) = f(x)$$.
$$\therefore C'(x) = \vec{\phi}^{-1}(x) \cdot f(x)$$.
$$\therefore C(x) = \int_{x_0}^{x} \vec{\phi}^{-1}(x) \cdot f(x) \, dx + C$$.
(Translation of the rest of the text is not provided as it is not in English and appears to be a continuation of the mathematical derivation.)
Page 21
Fourier Analysis and Laplace Transform.
- Orthogonal Basis of Functions {1, sin(nπx/L), cos(nπx/L), sin(2nπx/L), cos(2nπx/L), ...}
Basis Norm |e_n| = √(1/L) ∫[-L, L] cos²(nπx/L) dx = √(1/2) ∫[-L, L] (1 - cos(2nπx/L)) dx = √(1/2) ∫[-L, L] 1 dx = √(1/2) L
∴ f(x) = (1/2) ∫[-L, L] f(x) dx + (1/2) ∫[-L, L] f(x) cos(nπx/L) dx + ... + (1/2) ∫[-L, L] f(x) cos(nπx/L) dx + ...
= (a_0/2) + ∑[n=1,∞] (a_n cos(nπx/L) + b_n sin(nπx/L)) (at the continuity points).
Where a_n = (1/L) ∫[-L, L] f(x) cos(nπx/L) dx, n=0,1,2,...
b_n = (1/L) ∫[-L, L] f(x) sin(nπx/L) dx, n=1,2,...
- Fourier Integral Formula:
f(x) = (a_0/2) + ∑[n=1,∞] (a_n cos(nπx/L) + b_n sin(nπx/L))
= (1/2L) ∫[-L, L] f(t) dt + ∑[n=1,∞] (1/L) ∫[-L, L] f(t) (cos(nπx/L) cos(nπx/L) + ∑[m=1,∞] (cos(nπx/L) cos(mπx/L)) dt
= (1/2L) ∫[-L, L] f(t) dt + ∑[n=1,∞] (1/L) ∫[-L, L] f(t) cos(nπx/L) (cos(t) - t) dt
Let u_n = nπ/L, ∴ u_n = u_n - u_{n-1} = π/L.
∴ ∑[n=1,∞] (1/L) ∫[-L, L] f(t) cos(nπx/L) (cos(t) - t) dt = ∑[n=1,∞] (1/L) ∫[-L, L] f(t) cos(u_n (x-t)) dt
= (1/π) ∑[n=1,∞] (1/L) ∫[-L, L] f(t) cos(u_n (x-t)) dt ∆u_n
When L → +∞, ∆u_n → 0, so the above formula becomes ∑[n=1,∞] (1/π) ∫[-L, L] f(t) cos(u_n (x-t)) dt ∆u_n = ∫[0, +∞] φ(u) du
∴ f(x) = (1/π) ∫[0, +∞] du ∫[-∞, +∞] f(t) cos(u (x-t)) dt -∞ < x < +∞.
= ∫[0, +∞] [a(u) cos(ux) + b(u) sin(ux)] du.
Page 22
- Fourier Transform
$$f(x) = \frac{1}{2\pi} \int_{-\infty}^{+\infty} du \int_{-\infty}^{+\infty} f(t) \cos{u(x-t)} dt$$
$$= \frac{1}{2\pi} \int_{-\infty}^{+\infty} du \int_{-\infty}^{+\infty} f(t) \cos{u(x-t)} dt \quad (\cos{u(x-t)} is u's even function)$$
$$= \frac{1}{2\pi} \int_{-\infty}^{+\infty} du \int_{-\infty}^{+\infty} f(t) [\cos{u(x-t)} + i\sin{u(x-t)}] dt \quad (\sin{u(x-t)} is u's odd function)$$
$$= \frac{1}{2\pi} \int_{-\infty}^{+\infty} du \int_{-\infty}^{+\infty} f(t) e^{iux} dt$$
$$= \frac{1}{2\pi} \int_{-\infty}^{+\infty} e^{iux} du \int_{-\infty}^{+\infty} f(t) e^{-itu} dt$$
$$= \frac{1}{2\pi} \int_{-\infty}^{+\infty} F(u) e^{iux} du$$
$$F(u) is the Fourier transform of f(x), F(u) = \int_{-\infty}^{+\infty} f(t) e^{-itu} dt, written as F(u) = F[f(x)]$$
$$f(x) = \frac{1}{2\pi} \int_{-\infty}^{+\infty} F(u) e^{iux} du is the Fourier inverse transform, written as f(x) = F^{-1}[F(u)]$$
Differentiation property: If f(x) is absolutely integrable and continuous on (-∞, +∞), and its first derivative also satisfies the Dirichlet condition, then
$$F[f'(x)] = iuF[f(x)]$$
$$F[f'(x)] = \int_{-\infty}^{+\infty} f'(x) e^{-itu} dx = f(x) e^{-itu} \Big|_{-\infty}^{+\infty} + iu \int_{-\infty}^{+\infty} f(x) e^{-itu} dx$$
$$= iuF[f(x)]$$
$$\therefore F[f^{(n)}(x)] = (iu)^n F[f(x)]$$
$$F[\int_{-\infty}^{+\infty} f(x) dx] = \frac{1}{iu} F[f(x)]$$
$$\because \frac{d^n F(u)}{du^n} = F[-ix^n f(x)]$$
Convolution property: If f1(x) and f2(x) are absolutely integrable and satisfy the Dirichlet condition on (-∞, +∞), then
$$F[f_1(x) * f_2(x)] = F_1(u) F_2(u)$$
f1 and f2's convolution: f1(x) * f2(x) = ∫_{-∞}^{+∞} f1(y) f2(x-y) dy.
Page 23
No.
Date
$$F[f_1(x) * f_2(x)] = \int_{-\infty}^{+\infty} [f_1(x) * f_2(x)] e^{-iux} dx$$
$$= \int_{-\infty}^{+\infty} \left[ \int_{-\infty}^{+\infty} f_1(y) f_2(x-y) dy \right] e^{-iux} dx$$
$$= \int_{-\infty}^{+\infty} \int_{-\infty}^{+\infty} f_1(y) e^{-iuy} f_2(x-y) e^{-iux} dy dx$$
$$= \int_{-\infty}^{+\infty} f_1(y) e^{-iuy} dy \int_{-\infty}^{+\infty} f_2(x-y) e^{-iux} dx$$
$$= F_1(u) F_2(u)$$
$$\therefore f_1(x) * f_2(x) = F^{-1}[F_1(u) F_2(u)]$$
- Laplace Transform
Let f(t) be defined in [0, +∞) and have a Laplace transform in the region of convergence. Then the integral of f(t) e^(-pt) dt in the region of convergence determines a function F(p) of p.
$$F(p) = L[f(t)]$$
$$f(t) = L^{-1}[F(p)]$$
Differentiation Property: If f(t) and f'(t) both have Laplace transforms, and L[f(t)] = F(p), then
$$L[f'(t)] = pF(p) - f(0)$$
$$L[f'(t)] = \int_0^{+\infty} f'(t) e^{-pt} dt = f(t) e^{-pt} \Big|_0^{+\infty} + p \int_0^{+\infty} f(t) e^{-pt} dt$$
$$= -f(0) + pF(p), \quad \text{Re}(p) > c$$
Generally, L[f^(n)(t)] = p^n F(p) - p^(n-1) f(0) - p^(n-2) f'(0) - ... - f^(n-1)(0)
If f(0) = f'(0) = ... = f^(n-1)(0) = 0, then
$$L[\int_0^t f(x) dx] = \frac{1}{p} F(p)$$
Page 24
Let λ(t) = ∫₀^t f(τ) dτ, then λ(0) = 0, λ'(t) = f(t)
By the properties of integration, ∫[f(t)] = ∫[λ'(t)] = p ∫[∫₀^t f(τ) dτ]
∴ ∫[∫₀^t f(τ) dτ] = 1/p ∫[f(t)] = 1/p F(p).
- Inverse Laplace Transform
If the Laplace transform of f(t) is F(p), i.e., ∫[f(t)] = F(p), then
f(t) = 1/2πi ∫[β+i∞]^(β-i∞) F(p) e^(pt) dp, t > 0.
If p₁, ..., pₙ are all the poles of F(p), and when p → ∞, F(p) → 0, then
f(t) = 1/2πi ∫[β+i∞]^(β-i∞) F(p) e^(pt) dp = ∑(n=1 to n) Res[F(p) e^(pt)] at p = pₙ
∫[f₁(t) * f₂(t)] = ∫₀^∞ [f₁(t) * f₂(t)] e^(-pt) dt
= ∫₀^∞ [∫₀^t f₁(τ) f₂(t-τ) dτ] e^(-pt) dt
= ∫₀^∞ f₁(τ) [∫₀^∞ f₂(t-τ) e^(-pt) dt] dτ
= ∫₀^∞ f₁(τ) ∫₀^∞ f₂(s) e^(-p(τ+s)) ds dτ
= ∫₀^∞ f₁(τ) e^(-pt) dτ - ∫₀^∞ f₂(s) e^(-ps) ds
= F₁(p) F₂(p).
Page 25
Chapter 8. About the Completeness of the Real Number System
- Bolzano's Theorem (Bolzano)
A bounded sequence has a convergent subsequence.
- Monotone Bounded Theorem
A bounded monotone sequence converges.
- Interval Nesting Theorem
If the closed intervals $\{[a_n, b_n]\}$ satisfy the conditions:
(i) $[a_{n+1}, b_{n+1}] \subset [a_n, b_n]$, $n=1, 2, \ldots$
(ii) $\lim_{n\to\infty}(b_n - a_n) = 0$, then there is a unique point $x$ that belongs to all the closed intervals $[a_n, b_n]$ $(n=1, 2, \ldots)$.
- Finite Covering Theorem
From any open covering of $[a, b]$, we can extract a finite open covering of $[a, b]$.
- Accumulation Point Theorem
A bounded infinite set has at least one accumulation point.
- Least Upper Bound Principle (Least Lower Bound Principle)
Any bounded (non-bounded) set has an upper bound (lower bound).
- Completeness Theorem (Cauchy)
A sequence $\{a_n\}$ converges if and only if: $\forall \epsilon > 0$, $\exists N \in \mathbb{N}^+$, $\forall n, m > N$, $|a_n - a_m| < \epsilon$.
6 $\Rightarrow$ 2: Least Upper Bound $\Rightarrow$ Monotone Bounded
Proof: Consider the decreasing case.
Given the sequence $\{a_n\}$ is decreasing and bounded, so $\{a_n\}$ has a lower bound, denoted $A = \inf\{a_n\}$.
$\therefore \forall n, a_n \geq A$; $\forall \epsilon > 0$, $\exists N \in \mathbb{N}^+$, $a_N < A + \epsilon$.
Since $\{a_n\}$ is decreasing, $\therefore \forall n > N$, $a_n \leq a_N < A + \epsilon$.
$\therefore |a_n - A| < \epsilon$.
$\therefore \lim_{n\to\infty} a_n = A$.
Page 26
2→3: Monotone bounded → Interval套
Proof: Given {an} monotone increasing and {bn} monotone decreasing, and b1 is an upper bound of {an}, a1 is a lower bound of {bn}. By the monotone bounded theorem, {an} and {bn} both converge. Let lim an = A, lim bn = B. Therefore, lim (bn - an) = 0, so A = B. Thus, ∀n, an ≤ A; ∀ε > 0, ∃N, an > A - ε. Therefore, A = sup an, similarly B = inf bn. And ∀m, n, am ≤ bn. Thus, ∀n, bn is an upper bound of {an}, and an is a lower bound of {bn}. Therefore, ∀n, A ≤ bn, an ≤ B. Thus, ∀n, an ∈ [A, B]. Therefore, A ∈ [a1, b1].
Assume μ ∈ [a1, b1], then ∀n, an ≤ μ ≤ bn. By the property of limits, A ≤ μ ≤ A, so μ = A.
3→1: Interval套 → Bolzano (Bolzano)
Proof: Given {xn} bounded. ∴ ∃a1, b1, such that ∀n, xn ∈ [a1, b1]. Take any x1 ∈ [a1, b1], divide the interval [a1, b1] into [a1, (a1 + b1)/2] and [(a1 + b1)/2, b1]. Then among these, there must be a subsequence of {xn} that is unbounded. Take one of these intervals, [a2, b2], b2 - a2 = (b1 - a1)/2. Take any x2 ∈ [a2, b2], n2 > n1, ...
Thus, we generate the interval套 [a1, b1] (where [a1, b1] is half of [a1, b1]) and a subsequence {xn} that satisfies xn ∈ [a2, b2]. By the Bolzano定理, ∃! c ∈ [a1, b1], such that {xn} converges to c.
Page 27
1→7: Compactness (Bolzano) → Completeness (Cauchy)
Proof: Necessity.
Assume {x_n} converges. Then ∀ε > 0 ∃N ∈ N ∀n > N such that |x_n - A| < ε/2.
When m, n > N, then |x_m - A| < ε/2, |x_n - A| < ε/2, thus |x_m - x_n| ≤ |x_m - A| + |x_n - A| < ε/2 + ε/2 = ε.
Sufficiency.
Assume {x_n} is a Cauchy sequence. Then for any ε > 0, ∃N_1 > 0, ∀m, n > N_1, such that |x_m - x_n| < 1.
∴ ∀n > N_1, there is |x_n - x_{n+1}| < 1, |x_n| < |x_{n+1}| + 1.
So the sequence {x_n} is bounded, and there exists a subsequence {x_{n_k}}, such that lim x_{n_k} = A.
∀ε > 0, ∃N_2 > 0 ∀k > N_2, such that |x_{n_k} - A| < ε/2. ∃N_3 > 0, ∀m, n > N_3, such that |x_m - x_n| < ε/2.
Take N = max(N_2 + 1, N_3 + 1).
When n > N, there is n > N > N_3.
∴ |x_n - A| ≤ |x_n - x_{n_N}| + |x_{n_N} - A| < ε/2 + ε/2 = ε.
∴ lim x_n = A.
3→4: Interval Covering → Finite Covering.
Proof: If not, then [a, b] cannot be covered by a finite number of open intervals.
Divide the interval [a, b] into two halves, at least one half cannot be covered by a finite number of open intervals.
Repeat this process on [a_1, b_1] (if both halves are covered, take either one).
Repeat this process on [a_1, b_1] into two halves, at least one half cannot be covered by a finite number of open intervals.
This process will yield a sequence of intervals [a_k, b_k], such that ∃c ∈ [a_k, b_k] for any c ∈ [a_k, b_k].
∴ lim (b_k - a_k) = 0, c ∈ [a_k, b_k] for any c ∈ [a_k, b_k]. ∴ ∃k such that [a_k, b_k] ⊂ c, which contradicts the assumption of finite covering.
∴ lim (b_k - a_k) = 0, c ∈ [a_k, b_k] for any c ∈ [a_k, b_k]. ∴ ∃k such that [a_k, b_k] ⊂ c, which contradicts the assumption of finite covering.
Page 28
6→6: Monotone bounded → Convergent
Proof: Let A be a nonempty real set, bounded above.
Suppose A has an element that is not its own upper bound. Let this element be a1.
A has an upper bound b1. Let c1 = (a1 + b1) / 2.
If c1 is an upper bound, let a2 = a1, b2 = c1; otherwise, let a2 = a1, b2 = b1.
Let c2 = (a2 + b2) / 2, ...
Thus, we get a sequence {bn} where each term is an upper bound of A. It is monotone decreasing, bounded below by a1. By the Monotone Convergence Theorem, {bn} converges. Let lim b_n = s.
∴ lim a_n = s.
∀x ∈ A ∀n x ≤ b_n. By monotonicity ∀x ∈ A x ≤ s.
∀ε > 0, ∴ lim a_n = s. ∃N ∈ N+, ∀n > N |a_n - s| < ε, a_n > s - ε.
Because a_{n+1} is not an upper bound, ∴ ∃x ∈ A x > a_{n+1} > s - ε.
∴ s is an upper bound.
4→1: Compactness (Bolzano-Weierstrass)
Proof: Suppose sequence {x_n} is bounded, a as the lower bound, b as the upper bound, but has no convergent subsequence.
∴ ∀A ∈ [a, b] ∃ε_A > 0, (A - ε_A, A + ε_A) contains only finite terms of {x_n}.
[a, b] ⊆ ∪ (A - ε_A, A + ε_A).
By the Compactness Theorem, there exist finite A1, ..., Am such that [a, b] ⊆ ∪ (A_i - ε_{A_i}, A_i + ε_{A_i}).
∴ Each open interval (A_i - ε_{A_i}, A_i + ε_{A_i}) contains only finite terms of {x_n}.
∴ ∪ (A_i - ε_{A_i}, A_i + ε_{A_i}) contains only finite terms of {x_n}.
∴ [a, b] contains only finite terms of {x_n}.
1→4: Compactness (Bolzano-Weierstrass) → Finite Covering
Proof: With 3→4's proof method, by forming a sequence of intervals a_n, b_n, we can derive lim a_n = lim b_n = c.
Page 29
7→2: Completeness (Cauchy) → Monotonic and Bounded
Proof: Assume the sequence {a_n} is monotonic increasing and bounded but divergent. By the Cauchy convergence criterion, ∃ε>0, ∀N, ∃m,n>N, |a_m - a_n| > ε. For N=1, ∃m_1 > n_1, |a_{m_1} - a_{n_1}| > ε. For N=m_1, ∃m_2 > n_2 > m_1, |a_{m_2} - a_{n_2}| > ε. For N=m_2, ∃m_3 > n_3 > m_2, |a_{m_3} - a_{n_3}| > ε. ... ∴ {a_n} is monotonic increasing. ∴ ∀k∈N*, a_{m_k} > a_{n_k} + ε ≥ a_{m_{k-1}} + ε > (a_{n_{k-1}} + ε) + ε = a_{n_{k-1}} + 2ε ≥ a_{m_{k-2}} + 2ε > ... > a_{m_1} + (k-1)ε. ∵ ε is given, k can be greater than ∞. ∴ {a_{m_k}} is unbounded, which contradicts the assumption.
1→5: Convergence (Bolzano) → Convergent
Proof: Let A be a bounded infinite set, ∀x∈A. A\{x_1} is a bounded infinite set, ∀x_1∈A\{x_1}. A\{x_1, x_2\} is a bounded infinite set, ∀x_2∈A\{x_1, x_2\}.
5→1: Convergent → Convergent (Bolzano)
Proof: Assume {x_n} is a bounded sequence, thus has a convergent subsequence, let lim_{k→∞} x_{n_k} = a. ∀ε>0, ∃K, ∀k>K, |x_{n_k} - a| < ε. Thus, a's ε neighborhood contains infinitely many points x_{n_{k+1}}, x_{n_{k+2}}, ...
- A is a finite set, at this time {x_n} has infinitely many equal terms, these terms form a convergent constant sequence.
- A is an infinite set, at this time A has a convergent point.
Page 30
The text and formulas in the image are as follows:
---
Let \(a\) be a point in set \(A\). Take any term of the sequence, denoted as \(x_n\).
Let \(\delta_2 = \min\left(\frac{1}{2}, |a - x_1|\right)\), in the \(\delta_2\)-neighborhood of \(a\), take the term of the sequence with index greater than \(n\), denoted as \(x_{n_2}\). Let \(\delta_3 = \min\left(\frac{1}{3}, |a - x_{n_2}|\right)\), in the \(\delta_3\)-neighborhood of \(a\), take the term of the sequence with index greater than \(n_2\), denoted as \(x_{n_3}\)...
This way, we obtain the subsequence \(\{x_{n_k}\}\) converging to \(a\). Because
\(\forall \epsilon > 0\), \(\exists N\), \(\frac{1}{N} < \epsilon\)
\(\therefore \forall k > N\), \(|x_{n_k} - a| < \frac{1}{k} < \epsilon\)
4 → 5: Finite cover → point.
Proof: Suppose there is an unbounded infinite set \(S \subset [-m, m]\). Since \(S\) has a limit point, it must be contained in \([-m, m]\).
Now, assume that \([-m, m]\) does not contain any points of \(S\).
Then, \(\forall x \in [-m, m]\), \(\exists \delta_x > 0\), such that \(U(x, \delta_x) \cap S\) is a finite set. Let
\(H = \{U(x, \delta_x) \mid x \in [-m, m]\}\),
Then \(H\) is a cover of \([-m, m]\). By the finite cover theorem, \(\exists n\)
\(S \subset [-m, m] \subset \bigcup_{i=1}^n U(x_i, \delta_i)\)
\(\therefore U(x_i, \delta_i) \cap S\) is a finite set (\(i = 1, 2, \ldots, n\))
\(\therefore\) The above statement contradicts the assumption that \(S\) is an infinite set.
---
The formulas are:
\(\delta_2 = \min\left(\frac{1}{2}, |a - x_1|\right)\)
\(\delta_3 = \min\left(\frac{1}{3}, |a - x_{n_2}|\right)\)
\(\forall k > N\), \(|x_{n_k} - a| < \frac{1}{k} < \epsilon\)
\(\forall x \in [-m, m]\), \(\exists \delta_x > 0\), such that \(U(x, \delta_x) \cap S\) is a finite set
\(H = \{U(x, \delta_x) \mid x \in [-m, m]\}\)
\(S \subset [-m, m] \subset \bigcup_{i=1}^n U(x_i, \delta_i)\)
\(\therefore U(x_i, \delta_i) \cap S\) is a finite set (\(i = 1, 2, \ldots, n\))
\(\therefore\) The above statement contradicts the assumption that \(S\) is an infinite set.
Page 31
Chapter 9: Properties of Continuous Functions on Closed Intervals
- Boundedness - Extreme Value Theorem
If a function \( f(x) \) is continuous on the closed interval \([a, b]\), then \( f(x) \) has a maximum and minimum value. That is, there exist points \( \xi \) and \( \eta \) in \([a, b]\) such that for all \( x \in [a, b] \), \( f(\xi) \leq f(x) \leq f(\eta) \).
Proof:
For any \( \varepsilon > 0 \), since \( f(x) \in C[a, b] \), we have:
\[
\forall x_0 \in [a, b], \exists \delta_{x_0} > 0, \forall x \in (x_0 - \delta_{x_0}, x_0 + \delta_{x_0}) \cap [a, b].
\]
\[
|f(x) - f(x_0)| < 1, |f(x)| < |f(x_0)| + 1.
\]
\[
\bigcup_{x_0 \in [a, b]} (x_0 - \delta_{x_0}, x_0 + \delta_{x_0}) \text{ forms an open cover of } [a, b].
\]
\[
\text{There exist finitely many } x_0_1, x_0_2, \ldots, x_0_n \text{ such that }
\]
\[
[a, b] \subseteq \bigcup_{k=1}^n (x_0_k - \delta_{x_0_k}, x_0_k + \delta_{x_0_k}).
\]
\[
\therefore |f(x)| \leq \max_{1 \leq k \leq n} \{ |f(x_0_k)| + 1 \} \text{ is bounded.}
\]
Next, we prove that \( f(x) \) has a minimum value.
By the completeness principle, \( f(x) \) has a lower bound, let \( m = \inf_{x \in [a, b]} f(x) \).
For any \( n \in \mathbb{N}^+ \), there exists \( x_n \in [a, b] \) such that \( f(x_n) < m + \frac{1}{n} \).
Since \( \{x_n\} \subseteq [a, b] \) is bounded, there exists a convergent subsequence \( \{x_{n_k}\} \) with limit \( \xi \in [a, b] \).
Since \( f(x) \) is continuous on \([a, b]\),
\[
\lim_{k \to \infty} f(x_{n_k}) = f(\xi).
\]
\[
\therefore f(\xi) < m + \frac{1}{n}.
\]
\[
\lim_{k \to \infty} f(x_{n_k}) \leq \lim_{k \to \infty} (m + \frac{1}{n_k}) = m.
\]
\[
\therefore f(\xi) \leq m.
\]
Since \( m \) is the lower bound, \( m \leq f(\xi) \).
\(\therefore f(\xi) = m\). Thus, \( f(x) \) has a minimum value.
- Intermediate Value Theorem
If \( f(x) \) is continuous on the closed interval \([a, b]\) and \( f(a) \neq f(b) \), then for any \( \mu \in (f(a), f(b)) \) or \( (f(b), f(a)) \), there exists \( \xi \in (a, b) \) such that \( f(\xi) = \mu \).
\(\exists \xi \in (a, b), f(\xi) = \mu\).
Page 32
Proof: Let \(a_1 = a\), \(b_1 = b\), then \(c_1 = \frac{a_1 + b_1}{2}\).
If \(f(c_1) = 0\), take \(c_1\), and we are done.
If \(f(c_1) < 0\), then \(a_2 = c_1\), \(b_2 = b_1\).
If \(f(c_1) > 0\), then \(a_2 = a_1\), \(b_2 = c_1\), and \(c_2 = \frac{a_2 + b_2}{2}\).
Or at some step, we find \(f(c_2) = 0\), or we generate an interval \([a_n, b_n]\) satisfying \(f(a_n) < 0\), \(f(b_n) > 0\), by the interval套定理.
\(\exists \xi = \lim_{n \to \infty} a_n = \lim_{n \to \infty} b_n\), \(f(\xi) = 0\), so \(f(a_n) \leq 0\), \(f(b_n) \geq 0\).
\(\therefore f(\xi) = 0\).
The proof of the zero point existence theorem, and the intermediate value theorem.
\(\because f(x) \in C[a, b]\), \(\therefore g(x) = f(x) - \mu \in C[a, b]\).
\(\therefore g(a) g(b) < 0\).
By the zero point existence theorem, there exists \(\xi \in (a, b)\), \(g(\xi) = 0\), so \(f(\xi) = \mu\).
- Uniform Continuity Theorem (Heine-Borel)
If \(f(x)\) is continuous on \([a, b]\), then \(f(x)\) is uniformly continuous on \([a, b]\).
Proof: For \(\epsilon > 0\), \(\forall x_0 \in [a, b]\), \(\exists \delta_{x_0} > 0\), \(x \in (x_0 - \delta_{x_0}, x_0 + \delta_{x_0}) \cap [a, b]\) such that \(|f(x) - f(x_0)| < \frac{\epsilon}{3}\).
Then \([a, b] \subseteq \bigcup_{x_0 \in [a, b]} (x_0 - \frac{\delta_{x_0}}{3}, x_0 + \frac{\delta_{x_0}}{3})\).
\(\therefore \exists x_1, x_2, \ldots, x_n\), \([a, b] \subseteq \bigcup_{k=1}^n (x_{0k} - \frac{\delta_{x_k}}{3}, x_{0k} + \frac{\delta_{x_k}}{3})\).
Let \(\delta = \min_{1 \leq k \leq n} (\frac{\delta_{x_k}}{3})\), when \(|x - t| < \delta\), \(\exists l, m \in \{1, 2, \ldots, n\}\).
\(x \in (x_{0l} - \frac{\delta_{x_l}}{3}, x_{0l} + \frac{\delta_{x_l}}{3})\), \(t \in (x_{0m} - \frac{\delta_{x_m}}{3}, x_{0m} + \frac{\delta_{x_m}}{3})\).
Assume \(\delta_{x_l} \leq \delta_{x_m}\), then \(x_l \in (x_{0m} - \delta_{x_m}, x_{0m} + \delta_{x_m})\).
Thus, \(|f(x) - f(t)| \leq |f(x) - f(x_l)| + |f(t) - f(x_m)| + |f(x_l) - f(x_m)|\).
\(\therefore f(x)\) is uniformly continuous on \([a, b]\).
Therefore, \(|f(x) - f(t)| \leq |f(x) - f(x_l)| + |f(t) - f(x_m)| + |f(x_l) - f(x_m)|\).
\(\therefore f(x)\) is uniformly continuous on \([a, b]\).
\(\therefore f(x)\) is uniformly continuous on \([a, b]\).
Page 33
Riemann Integration
- Given a finite partition T = {x_0, x_1, ..., x_n} ⊂ [a, b], satisfying a = x_0 < x_1 < x_2 < ... < x_n = b. Then partition T is a partition of the interval [a, b]. Each interval [x_{k-1}, x_k] (k = 1, ..., n) is a subinterval of [a, b]. Δx_k = x_k - x_{k-1} is the length of the subinterval. λ = max{Δx_k} is the diameter of the partition T.
If f(x) is defined on [a, b], for any partition T = {x_0, x_1, ..., x_n}, and any point ξ_k ∈ [x_{k-1}, x_k] (k = 1, ..., n), if lim(ξ_k) Δx_k exists, then f(x) ∈ R[a, b].
∀I ∈ R, ∀ε > 0, ∃δ > 0, ∀T[a, b] = {x_0, x_1, ..., x_n}, ∀ξ_k ∈ [x_{k-1}, x_k], if λ < δ, then |Σ(ξ_k) Δx_k - I| < ε. (Definition)
∴ Not integrable as:
∀I ∈ R, ∃ε > 0, ∀δ > 0, ∃T[a, b] = {x_0, x_1, ..., x_n}, ∃ξ_k ∈ [x_{k-1}, x_k], if λ < δ, then |Σ(ξ_k) Δx_k - I| < ε.
- Necessary condition for integrable function:
If f(x) ∈ R[a, b], then f(x) is bounded on [a, b].
Proof: Since f(x) ∈ R[a, b], for any partition T and any point ξ_k ∈ [x_{k-1}, x_k], if the diameter λ < δ, then
|Σ(ξ_k) Δx_k - ∫_a^b f(x) dx| < 1.
∴ When λ < δ, all integral sums are bounded.
Suppose f(x) is unbounded on [a, b]. Then for any partition T of [a, b] with any diameter less than δ, the partition f(x) must be unbounded in at least one subinterval. Suppose f(x) is unbounded in [x_0, x_1]. ∴ ∀M > 0, ∃ξ_k ∈ [x_{k-1}, x_k] (k = 2, 3, ..., n), such that |f(ξ_k)| > |Σ(ξ_k) Δx_k| + M.
∴ The integral sum and the integral in λ < δ are unbounded. ∴ The integral sum and the integral in λ < δ are unbounded.
Page 34
- Let f(x) be bounded on [a, b]. T is a partition of [a, b]. Define:
$$M_k = \sup_{x \in [x_{k-1}, x_k]} \{f(x)\}, \quad m_k = \inf_{x \in [x_{k-1}, x_k]} \{f(x)\}, \quad k = 1, 2, \ldots, n$$
The upper sum of f(x) corresponding to partition T is:
$$S(f, T) = \sum_{k=1}^{n} M_k \Delta x_k$$
The lower sum of f(x) corresponding to partition T is:
$$s(f, T) = \sum_{k=1}^{n} m_k \Delta x_k$$
Since $m_k \leq f(\xi_k) \leq M_k$ (for partition T),
$$\sum_{k=1}^{n} m_k \Delta x_k \leq \sum_{k=1}^{n} f(\xi_k) \Delta x_k \leq \sum_{k=1}^{n} M_k \Delta x_k, \quad \xi_k \in [x_{k-1}, x_k]$$
Thus,
$$s(f, T) \leq \sum_{k=1}^{n} f(\xi_k) \Delta x_k \leq S(f, T)$$
For the upper sum $S(f, T)$, we have:
$$\sum_{k=1}^{n} f(\xi_k) \Delta x_k \leq S(f, T)$$
For any $\epsilon > 0$, since $M_k = \sup_{x \in [x_{k-1}, x_k]} \{f(x)\}$, there exists $\xi_k \in [x_{k-1}, x_k]$ such that $M_k - \frac{\epsilon}{2^n} < f(\xi_k)$. Thus,
$$\sum_{k=1}^{n} (M_k - \frac{\epsilon}{2^n}) \Delta x_k < \sum_{k=1}^{n} f(\xi_k) \Delta x_k$$
This implies:
$$S(f, T) - \epsilon < \sum_{k=1}^{n} f(\xi_k) \Delta x_k$$
From (*) and (**), we have:
$$S(f, T) = \sup_{\xi_k \in [x_{k-1}, x_k]} \sum_{k=1}^{n} f(\xi_k) \Delta x_k$$
Similarly,
$$s(f, T) = \inf_{\xi_k \in [x_{k-1}, x_k]} \sum_{k=1}^{n} f(\xi_k) \Delta x_k$$
Let $T_1$ and $T_2$ be two partitions of [a, b] such that $T_1 \subseteq T_2$. Then $s(f, T_1) \leq s(f, T_2)$ and $S(f, T_2) \leq S(f, T_1)$.
Proof: Consider $T_1 = \{x_0, x_1, \ldots, x_n\}$ and let $t \in (x_0, x_1)$. Then $T_2$ is $a = x_0 < t < x_1 < x_2 < \ldots < x_n = b$.
Let $M_k = \sup_{x \in [x_{k-1}, x_k]} \{f(x)\}$, $k = 1, 2, \ldots, n$, $M' = \sup_{x \in [x_0, t]} \{f(x)\}$, $M'' = \sup_{x \in (t, x_1]} \{f(x)\}$.
Then $M_1 \geq M'$ and $M_1 \geq M''$.
Thus,
$$M'(t - x_0) + M''(x_1 - t) \leq M_1(t - x_0) + M_1(x_1 - t) = M_1(x_1 - x_0)$$
Therefore,
$$S(f, T_2) - S(f, T_1) = M'(t - x_0) + M''(x_1 - t) - M_1(x_1 - x_0) \leq 0$$
Hence,
$$S(f, T_2) \leq S(f, T_1)$$
Therefore,
$$s(f, T_1) \leq s(f, T_2)$$
Page 35
The text and formulas in the image are as follows:
---
If T1 and T2 are any two partitions of [a, b], then S(f, T1) ≤ S(f, T2).
Proof: Let T = T1 ∪ T2, then T1 ⊆ T, T2 ⊆ T. Therefore, S(f, T1) ≤ S(f, T), S(f, T) ≤ S(f, T2). Hence, S(f, T1) ≤ S(f, T2).
Combining the above two propositions, for any T1 and T2, T = T1 + T2, then S(T1) ≤ S(T) ≤ S(T) ≤ S(T2). Thus, the lower sum S has an upper bound, and the upper sum S has a lower bound. Therefore, there exists a definite integral I = ∫ab f(x)dx, which is the definite integral of f(x) over [a, b].
Since the upper sum is the upper bound of the lower sum, for any ε > 0, there exists T1 such that S(f, T1) < I + ε/2. Let δ > 0 be the smallest interval length in T, and T is a partition of [a, b]. Let T2 = T ∪ T1. Then S(f, T2) ≤ S(f, T1). Therefore, S(f, T2) ≤ S(f, T).
Thus, 0 ≤ S(f, T) - I = S(f, T) - S(f, T2) + S(f, T2) - I ≤ S(f, T) - S(f, T2) + S(f, T2) - I < S(f, T) - S(f, T2) + ε/2.
- When T2 = T, i.e., all the points of T are also points of T. S(f, T) - S(f, T2) = 0.
- When T2 has only one point t that is not a point of T. Suppose t ∈ (x_{i-1}, x_i).
Therefore, S(f, T) - S(f, T2) = M(x_i - x_{i-1}) - [M(t - x_{i-1}) + M(x_i - t)] ≤ 2M(x_i - x_{i-1}) ≤ 2Mδ. Where M = sup_{x∈[a,b]} |f(x)|.
Given δ = min{ε/4n, M, δ}, then when T has a diameter λ < δ, there is:
0 ≤ S(f, T) - S(f, T2) < ε/2.
Therefore, 0 ≤ S(f, T) - I < S(f, T) - S(f, T2) + ε/2 < ε/2 + ε/2 = ε. Thus, lim S(f, T) = I.
---
The formulas are:
- S(f, T1) ≤ S(f, T)
- S(f, T) ≤ S(f, T2)
- S(f, T1) ≤ S(f, T2)
- S(f, T1) ≤ S(f, T2)
- S(f, T) - S(f, T2) = 0
- S(f, T) - S(f, T2) = M(x_i - x_{i-1}) - [M(t - x_{i-1}) + M(x_i - t)]
- S(f, T) - S(f, T2) ≤ 2M(x_i - x_{i-1}) ≤ 2Mδ
- S(f, T) - S(f, T2) ≤ ε/2
- S(f, T) - S(f, T2) < ε/2
- S(f, T) - S(f, T2) + ε/2 < ε/2 + ε/2 = ε
- lim S(f, T) = I
---
Page 36
- Necessary and sufficient conditions for integrability: I = I̅
Necessary condition: Suppose f(x) is integrable on [a, b]. lim (Σ f(ξ_k)Δx_k) = I
For any ε > 0, there exists δ > 0. When the diameter of T < δ, |Σ f(ξ_k)Δx_k - I| < ε/2.
For any partition T, there exists (ξ_k) (k=1, 2, ..., n). Because the upper sum is the upper bound and the lower sum is the lower bound
Σ f(ξ_k)Δx_k > S(f, T) - ε/2
|S(f, T) - I| ≤ |S(f, T) - Σ f(ξ_k)Δx_k| + |Σ f(ξ_k)Δx_k - I| < ε
lim S(f, T) = I
Similarly, lim S(f, T) = I. Therefore, I = I̅
Sufficient condition: Suppose lim S(f, T) = lim S(f, T), then I = I̅
Since S(f, T) ≤ Σ f(ξ_k)Δx_k ≤ S(f, T).
I = lim S(f, T) ≤ lim Σ f(ξ_k)Δx_k ≤ lim S(f, T) = I̅.
Since I = I̅, lim Σ f(ξ_k)Δx_k = I = I̅. Thus, f(x) is integrable on [a, b].
Since I = I̅, lim (S(f, T) - S(f, T)) = 0. Therefore, lim Σ (M_k - m_k)Δx_k = 0. Define w_k = M_k - m_k.
lim Σ w_kΔx_k = 0.
- Commonly Integrable Functions
① f(x) is continuous on [a, b] implies f(x) is integrable on [a, b].
Proof: Since f(x) is continuous on [a, b], it is uniformly continuous on [a, b].
For any ε > 0, there exists δ > 0. When x' and x'' ∈ [a, b] and |x' - x''| < δ, |f(x') - f(x'')| < ε/2.
Therefore, for any partition T, the sum of w_kΔx_k is zero.
ω_k = M_k - m_k = f(ξ_k) - f(η_k) < ε/2 - a.
Thus, for any partition T, the sum of w_kΔx_k is zero.
Page 37
- If f(x) is bounded and has only a finite number of discontinuities in [a, b], then f(x) is integrable in [a, b].
Proof: Assume there is only one discontinuity point c such that a < c < b.
Let ε > 0. Suppose |f(c)| ≤ M, ∀x ∈ [a, b].
Take ε1 = ε / (1 + 8M), C1 = c - ε1, C2 = c + ε1, A = (C1, f(C1)), B = (C2, f(C2)).
If F(x) = {f(x), x ∈ [a, c1] ∪ [c2, b]; g(x), x ∈ (c1, c2)}, g(x) is the line segment AB.
Then F(x) is continuous in [a, b]. By the previous proposition, F(x) is integrable in [a, b].
For any partition T, as long as λ < δ, there is a sum of the form ∑(w_k * Δx_k) < ε.
- In T = {x_0, ..., x_n} of the various intervals [x_{k-1}, x_k], one class completely includes [a, c1] ∪ [c2, b].
Using the subscript 'k' to indicate those with intersections, and 'k'' to indicate those without, the total length is less than 4ε.
∴ ∑(w_k * Δx_k) = ∑(w_k' * Δx_k') + ∑(w_k'' * Δx_k'') ≤ ∑(w_k' * Δx_k') + ∑(w_k'' * Δx_k'') ≤ ε + 2M * 4ε = ε + 8M * 4ε < 2ε.
Thus, lim(ε→0) ∑(w_k * Δx_k) = 0. The proof is similar for multiple discontinuities.
- If f(x) is monotonic in [a, b], then f(x) is integrable in [a, b].
Proof: Assume f(x) is monotonic and non-constant in [a, b].
For any ε > 0, take δ = ε / (f(b) - f(a)) > 0. For any partition T of [a, b], as long as λ < δ, we have:
0 < ∑(w_k * Δx_k) = ∑(f(x_k) - f(x_{k-1})) * Δx_k < ε.
= ε / (f(b) - f(a)) * ∑(f(x_k) - f(x_{k-1})) = ε.
Thus, lim(ε→0) ∑(w_k * Δx_k) = 0.
Page 38
- Weighted Function Properties
- Linearity: $$\int_{a}^{b} (\alpha f(x) + \beta g(x)) dx = \alpha \int_{a}^{b} f(x) dx + \beta \int_{a}^{b} g(x) dx$$
Proof: For any $\alpha \in \mathbb{R}$, any partition $T$ of $[a, b]$, let $\omega_k^{\alpha}$, $\omega_k^{\beta}$ represent $\alpha f(x)$, $f(x)$ in the subinterval $[x_{k-1}, x_k]$. Then $\omega_k^{\alpha} = |\alpha| \omega_k^{\beta}$.
$\therefore \sum \omega_k^{\alpha} \Delta x_k = 0$
$\therefore \sum \omega_k^{\alpha} \Delta x_k = 0 \therefore \alpha f(x)$ is integrable.
Using $\omega_k$, $\omega_k^{\beta}$, $\omega_k^{\alpha}$ to represent $f(x) + g(x)$, $f(x)$, $g(x)$ in the subinterval $[x_{k-1}, x_k]$. Then $\omega_k \leq \omega_k^{\beta} + \omega_k^{\alpha}$
$\therefore \sum \omega_k^{\alpha} \Delta x_k = \sum \omega_k^{\beta} \Delta x_k = 0$
$\therefore \sum \omega_k^{\alpha} \Delta x_k = 0 \therefore f(x) + g(x)$ is integrable.
- Interval Additivity: $$\int_{a}^{b} f(x) dx = \int_{a}^{c} f(x) dx + \int_{c}^{b} f(x) dx$$
Proof: 1) $\Rightarrow$:
Let $T_1$ be any partition of $[a, c]$, $T_2$ be any partition of $[c, b]$. $T = T_1 \cup T_2$ is a partition of $[a, b]$.
$\sum_{k=1}^{n} \omega_k \Delta x_k + \sum_{k=n+1}^{m} \omega_k \Delta x_k = \sum_{k=1}^{m} \omega_k \Delta x_k$
$\sum_{k=1}^{n} f(x_k) \Delta x_k + \sum_{k=n+1}^{m} f(x_k) \Delta x_k = \sum_{k=1}^{m} f(x_k) \Delta x_k$
$\therefore \lambda(T) = \max \{\lambda(T_1), \lambda(T_2)\}$, and $f(x)$ is integrable on $[a, b]$.
$\therefore$ when $\lambda(T_1)$ and $\lambda(T_2)$ both tend to 0, $\lim_{\lambda(T) \to 0} \sum_{k=1}^{m} \omega_k \Delta x_k = 0$
$\therefore 0 \leq \sum_{k=1}^{n} \omega_k \Delta x_k \leq \sum_{k=1}^{m} \omega_k \Delta x_k, 0 \leq \sum_{k=n+1}^{m} \omega_k \Delta x_k \leq \sum_{k=1}^{m} \omega_k \Delta x_k$
$\therefore \lim_{\lambda(T) \to 0} \sum_{k=1}^{n} \omega_k \Delta x_k = \lim_{\lambda(T) \to 0} \sum_{k=1}^{m} \omega_k \Delta x_k = 0$
$\therefore \int_{a}^{c} f(x) dx + \int_{c}^{b} f(x) dx = \int_{a}^{b} f(x) dx$
- $\Leftarrow$:
Let $T = \{x_0, x_1, \ldots, x_n\}$ be any partition of $[a, b]$. Suppose $c \in [x_{i-1}, x_i]$.
Then $x_0 < x_1 < \ldots < x_{i-1} < c$ is a partition of $[a, c]$, and $c < x_i < \ldots < x_n$ is a partition of $[c, b]$.
Page 39
No.
Date
$\exists M, \forall x \in [a, b], |f(x)| \leq M$, assume $M$ is the diameter.
$\therefore 0 \leq \sum_{k=1}^{n} w_k \Delta x_k = \sum_{k=1}^{n} w_k \Delta x_k + w_i \Delta x_i + \sum_{k=i+1}^{n} w_k \Delta x_k$
$\leq \left[ \sum_{k=1}^{i} (w_k \Delta x_k + w'(c - x_{i-1})) \right] + \left[ \sum_{k=i+1}^{n} (w_k \Delta x_k + w''(x_i - c)) \right] + 2M \Delta$
$\because f(x)$ is in $[a, c]$ and $[c, b]$ on the interval
$\therefore \int_{a}^{b} f(x) dx = \int_{a}^{c} f(x) dx + \int_{c}^{b} f(x) dx$
(3) Absolute value function integrability: $|\int_{a}^{b} f(x) dx| \leq \int_{a}^{b} |f(x)| dx$
Proof: Let $T = \{x_0, x_1, \ldots, x_n\}$ be any partition of $[a, b]$, then
$w_k = \sup_{x \in [x_{k-1}, x_k]} \{ |f(x)| - |f(x)| \}$,
$\bar{w}_k = \sup_{x \in [x_{k-1}, x_k]} \{ |f(x)| - |f(x)| \}$
Then $\bar{w}_k \leq w_k (k = 1, 2, \ldots, n)$
$\therefore 0 \leq \sum_{k=1}^{n} \bar{w}_k \Delta x_k \leq \sum_{k=1}^{n} w_k \Delta x_k$
$\therefore \lim_{\Delta \to 0} \sum_{k=1}^{n} \bar{w}_k \Delta x_k = 0$
$\therefore |f(x)|$ is integrable on $[a, b]$.
According to $-|f(x)| \leq f(x) \leq |f(x)|$ and the integral comparison principle, we have
$-\int_{a}^{b} |f(x)| dx \leq \int_{a}^{b} f(x) dx \leq \int_{a}^{b} |f(x)| dx$
(4) Product function integrability: $f(x), g(x)$ are integrable on $[a, b]$, then $f(x)g(x)$ is also integrable on $[a, b]$.
Proof: Let $T$ be any partition of $[a, b]$, then $w_k = \sup_{x \in [x_{k-1}, x_k]} \{ |f(x)g(x) - f(x)g(x)| \}$,
$M_k, m_k, w_k$ are the upper bound, lower bound, and range of $f(x)$ on $[x_{k-1}, x_k]$,
$M_k, m_k, w_k$ are the upper bound, lower bound, and range of $g(x)$ on $[x_{k-1}, x_k]$.
$\therefore f(x), g(x) \geq 0$
Page 40
$$\therefore w_k \leq M_k \overline{M}_k - m_k \overline{m}_k$$
$$= M_k (\overline{M}_k - \overline{m}_k) + \overline{m}_k (M_k - m_k)$$
$$\leq M_k w_k^f + \overline{M}_k w_k^f \quad \forall x \in [a, b], \text{ if } |w_k| \leq M, |g(x)| \leq \overline{M}.$$
$$\therefore 0 \leq \sum_{k=1}^n w_k \Delta x_k \leq M \sum_{k=1}^n w_k^f \Delta x_k + \overline{M} \sum_{k=1}^n w_k^f \Delta x_k$$
$$\therefore \lim_{n \to \infty} \sum_{k=1}^n w_k^f \Delta x_k = \lim_{n \to \infty} \sum_{k=1}^n w_k^f \Delta x_k = 0$$
$$\therefore \lim_{n \to \infty} \sum_{k=1}^n w_k \Delta x_k = 0.$$
When $f(x)$ and $g(x)$ are of opposite signs, let $f^+(x) = \frac{1 + f(x)}{2}$, $f^-(x) = \frac{1 - f(x)}{2}$, $g^+(x) = \frac{1 + g(x)}{2}$, $g^-(x) = \frac{1 - g(x)}{2}$. Then $f^+(x), f^-(x), g^+(x), g^-(x) \geq 0$.
$\because f(x), g(x)$ are integrable, $\therefore f^+(x), f^-(x), g^+(x), g^-(x)$ are all integrable.
$\therefore f(x)g(x) = [f^+(x) - f^-(x)][g^+(x) - g^-(x)]$
$= f^+(x)g^+(x) - f^+(x)g^-(x) + f^-(x)g^+(x) + f^-(x)g^-(x)$
$\therefore$ each term is integrable, $\therefore f(x)g(x)$ is integrable.
(5) First Mean Value Theorem: If $f(x)$ is continuous on $[a, b]$, $g(x)$ is integrable and non-zero on $[a, b]$, then there exists $\xi \in (a, b)$ such that $\int_a^b f(x)g(x)dx = f(\xi) \int_a^b g(x)dx$.
Proof: Since $f(x)$ is continuous on $[a, b]$, there exists $x_1, x_2 \in [a, b]$, $\forall x \in [a, b]$ such that $m = f(x_1) \leq f(x) \leq f(x_2) = M$.
$\because g(x)$ is non-zero, $\therefore$ it is not zero. Then $mg(x) \leq f(x)g(x) \leq Mg(x)$.
$\therefore m \int_a^b g(x)dx \leq \int_a^b f(x)g(x)dx \leq M \int_a^b g(x)dx$.
When $\int_a^b g(x)dx = 0$, by the above formula, $\int_a^b f(x)g(x)dx = f(\xi) \int_a^b g(x)dx = 0$.
$\therefore$ when $\int_a^b g(x)dx > 0$, let $\mu = \frac{\int_a^b f(x)g(x)dx}{\int_a^b g(x)dx}$, then $m \leq \mu \leq M$.
Page 41
The text and formulas in the image are as follows:
---
If \( m < \mu < M \), according to the Intermediate Value Theorem, there exists \( \xi \in (x_1, x_2) \subset (a, b) \) such that \( f(\xi) = \mu \). That is, \( \int_a^b f(x) g(x) dx = f(\xi) \int_a^b g(x) dx \).
If \( \mu = m \) or \( \mu = M \), it contradicts \( \mu = m \). Since \( \int_a^b g(x) dx > 0 \), there exists \( h > 0 \) such that \( \int_{a+h}^{b-h} g(x) dx > 0 \). If \( f(x) - \mu = f(x) - m \) in \( (a, b) \) is always greater than 0, then there exists \( m_0 > 0 \). For \( x \in [a+h, b-h] \), we have \( f(x) - \mu \geq m_0 > 0 \).
\[
\therefore \int_a^b f(x) g(x) dx - \mu \int_a^b g(x) dx = \int_a^b (f(x) - \mu) g(x) dx = \int_a^b (f(x) - m) g(x) dx
\]
\[
\geq \int_{a+h}^{b-h} (f(x) - \mu) g(x) dx \geq m_0 \int_{a+h}^{b-h} g(x) dx > 0
\]
This contradicts \( \int_a^b f(x) g(x) dx = \mu \int_a^b g(x) dx \). Therefore, \( \exists \xi \in (a, b), f(\xi) = \mu \).
The Second Integral Mean Value Theorem: If \( f(x) \) is monotonic in \( [a, b] \), then there exists \( \xi \in [a, b] \) such that
\[
\int_a^b f(x) g(x) dx = f(a) \int_a^b g(x) dx + f(b) \int_a^b g(x) dx.
\]
---
The text discusses the Intermediate Value Theorem and the Second Integral Mean Value Theorem in the context of definite integrals. It proves the existence of a point \( \xi \) within the interval \( (a, b) \) where the function \( f(x) \) equals a given value \( \mu \), and it also discusses the conditions under which the integral of a product of two functions can be expressed in terms of the function values at the endpoints of the interval.
Page 42
Chapter 11. Uniform Convergence of Function Series
1. Definition of Uniform Convergence:
A sequence of functions {f_n(x)} defined on I converges uniformly to f(x) if for all ε > 0, there exists an N such that for all n > N and all x in I, |f_n(x) - f(x)| < ε.
2. Necessary and Sufficient Conditions:
Necessary condition: For all ε > 0, there exists an N such that for all n > N and all x in I, |f_n(x) - f(x)| < ε. Cauchy condition.
Sufficient condition: lim sup |S(x) - S_n(x)| = 0. Proof:
Necessary condition: If {f_n(x)} converges uniformly to f(x), then for all ε > 0, there exists an N such that for all n > N and all x in I, |f_n(x) - f(x)| < ε. Let p > 0, then n + p > N. Thus, |S_{n+p}(x) - S_n(x)| < ε/2. Since |S_{n+p}(x) - S_n(x)| ≤ |S_{n+p}(x) - S(x)| + |S(x) - S_n(x)| < ε/2 + ε/2 = ε. Sufficient condition: If {S_n(x)} converges to S(x), then for all ε > 0, there exists an N such that for all n > N and all x in I, |S(x) - S_n(x)| < ε. Let p → ∞, then |S(x) - S_n(x)| < ε. Thus, {f_n(x)} converges uniformly to f(x). Proof by Cauchy convergence criterion.
3. Weierstrass Test:
The Weierstrass test states that if {f_n(x)} is a sequence of functions defined on I such that for all x in I, there exists a constant M_n such that |f_n(x)| ≤ M_n and the series ∑M_n converges, then the series ∑f_n(x) converges uniformly on I.
Page 43
The note discusses the convergence of series and functions. It includes theorems and proofs related to uniform convergence, monotonicity, and the integral test. The text also covers the properties of uniformly convergent series and functions, including continuity. The proofs involve epsilon-delta arguments and inequalities. The mathematical content is dense and technical, focusing on advanced calculus concepts.
Page 44
The note discusses the convergence of series and the properties of functions. It states that for any x in the interval I, if the absolute value of the difference between f(x) and f_n(x) is less than delta, then the series converges. It also mentions that the series of functions is uniformly convergent on I. The note further explains that the partial sums of the series are continuous on I. It concludes by stating that the series is uniformly convergent on I. The note also discusses the concept of improper integrals and the conditions under which they can be evaluated. It states that if a function is continuous on the interval [a, b], then the improper integral of the function can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the concept of the integral of a series and the conditions under which it can be evaluated. It states that if the series of functions is uniformly convergent on the interval [a, b], then the integral of the series can be evaluated. The note also discusses the
Page 45
The function f(x) is differentiable on [a, b]. And $$\frac{d}{dx}[\lim_{n\to\infty}f_n(x)]=\lim_{n\to\infty}\frac{d}{dx}f_n(x)$$.
Let u_n(x) ∈ C^1[a, b]. If u_n(x) converges uniformly on [a, b], and if the sequence u_n(x) converges uniformly on [a, b], then u_n(x) is differentiable on [a, b]. And $$\frac{d}{dx}[\lim_{n\to\infty}u_n(x)]=\lim_{n\to\infty}\frac{d}{dx}u_n(x)$$.
Proof: Suppose {f_n(x)} converges uniformly to g(x), and g(x) is continuous. Then g(x) is continuous on [a, x] for x ∈ (a, b). Therefore, f_n(x) = f_n(a) + ∫_a^x f_n'(t) dt, x ∈ [a, b].
Both sides of the equation are continuous. By the theorem of continuous functions, f(x) = f(a) + ∫_a^x g(t) dt. Since g(x) is continuous, f'(x) = g(x).
Thus, $$\frac{d}{dx}[\lim_{n\to\infty}f_n(x)]=\lim_{n\to\infty}\frac{d}{dx}f_n(x)$$.
Since the partial sum of the series {S_n(x)} converges uniformly on [a, b], and {S_n(x)} converges uniformly on [a, b], then S(x) is differentiable. Therefore, S'(x) = ∑_{n=1}^∞ S_n'(x).
Thus, $$\frac{d}{dx}[\lim_{n\to\infty}S_n(x)]=\lim_{n\to\infty}[\frac{d}{dx}S_n(x)]=\lim_{n\to\infty}[\sum_{k=1}^n u_k(x)]=\sum_{k=1}^{\infty}\frac{d}{dx}u_k(x)$$.
(To be continued)
Part III: Multivariable Calculus
Functions of multiple variables, partial derivatives, total differentials, multiple integration, and vector calculus field equations.
Page 1
B5 Notebook
8mm
"BEIFA" is awarded the titles of National Celebrated Export Commodity To Be Given Special Support, National Celebrated Product in Writing Instrument Manufacturing Industry, Provincial Celebrated Product in Zhejiang Province, Provincial
多元微积分
Page 2
Chapter 1: Multivariable Functions and Their Calculus
n-dimensional Euclidean space: R^n = { (x_1, x_2, ..., x_n) | x_i ∈ R, i = 1, 2, ..., n }
For addition and scalar multiplication, R^n is a linear space over the real numbers.
Euclidean distance: ||X - Y||_n = √(Σ^(n)_(i=1) (x_i - y_i)^2) : positivity, symmetry, triangle inequality
n-variable function: Let Ω ⊂ R^n, if for any point X ∈ Ω, there exists a unique value u ∈ R^1 corresponding to it, the corresponding rule is an n-variable function: f: Ω ⊂ R^n → R^1, X → u.
Example: u = √(1 - x^2 - y^2) (x, y) ∈ D ⊂ R^2
Example: Implicit representation: F(x_1, x_2, ..., x_n, u) = 0. (under certain conditions)
R^n → R^m vector-valued function: Let Ω ⊂ R^n, if for any point X ∈ Ω, there exists a vector Y ∈ R^m corresponding to it, the corresponding rule is a vector-valued function from Ω to R^m. Each component y_j (j = 1, 2, ..., m) of Y = (y_1, y_2, ..., y_m) is an n-variable function: y_j = f_j(x_1, x_2, ..., x_n), (x_1, ..., x_n) ∈ Ω.
Example: { x = R sin u cos v u ∈ [0, π] y = R sin u sin v v ∈ [0, 2π) z = R cos u R^2 → R^3
Vector-valued function limit: lim_(X→X_0) f(X) = A (A ∈ R^m):
∀ε > 0, ∃δ > 0, ∀X ∈ Ω and 0 < ||X - X_0||_n < δ → ||f(X) - A||_m < ε.
lim_(X→X_0) f_j(X) = a_j, j = 1, 2, ..., m.
Proof of limit non-existence: counterexample of different limits.
Proof of limit existence: use ε-δ through inequality proof.
Vector-valued function continuity: f(X) is continuous at X_0 point. ∀ε > 0, ∃δ > 0, ∀X: ||X - X_0||_n < δ, → ||f(X) - f(X_0)||_m < ε.
Page 3
Theorem: Continuous vector-valued functions are closed under addition, subtraction, multiplication, and composition. Continuous vector-valued functions have the extreme value theorem and the intermediate value theorem.
kth order infinitesimal function: $p^k = ||x - x_0||^k$. If $\frac{f(x)}{p^k} = constant$ as $x \to x_0$, then $f(x)$ is of kth order.
e.g., $f(x) = \sum_{i,j=1}^n a_{ij} x_i x_j = O(p^2)$.
n-ary function's total differential: $u = f(x)$ is defined in the ball $B(x_0, r)$. For $x \in B(x_0, r)$, $\Delta u = f(x) - f(x_0) = a_1(x_1 - x_1^0) + \ldots + a_n(x_n - x_n^0) + o(p)$. Then $u$ is differentiable at $x_0$, where $o(p)$ is the higher-order infinitesimal differential.
As $x - x_0 \to 0$, $f(x) \to f(x_0)$, so differentiable functions are continuous at $x_0$.
By the definition of total differential, if the n-ary function $u = f(x)$ is differentiable at $x_0$, then its partial derivatives $\frac{\partial u}{\partial x_1}(x_0), \ldots, \frac{\partial u}{\partial x_n}(x_0)$ all exist, and $a_i = \frac{\partial u}{\partial x_i}(x_0)$.
The converse of the above proposition is not true.
By the definition of total differential, if $u$ has all partial derivatives at $x_0$, and $\Delta u - \frac{\partial u}{\partial x_1} \Delta x_1 - \ldots - \frac{\partial u}{\partial x_n} \Delta x_n$ is of higher-order infinitesimal $p$, then $u$ is differentiable at $x_0$. Otherwise, it is not differentiable.
To determine differentiability, a sufficient condition is: if $u$ has all partial derivatives at $x_0$ and they are continuous, then $u$ is differentiable at $x_0$.
Where $x^0 = (cos\alpha_1, cos\alpha_2, \ldots, cos\alpha_n)$, if $u$ is differentiable at $x_0$, then
$\frac{\partial u}{\partial t}|_{p^0} = \frac{\partial u}{\partial x_1}|_{p^0} cos\alpha_1 + \ldots + \frac{\partial u}{\partial x_n}|_{p^0} cos\alpha_n$ (differentiable implies directional derivative exists)
$\frac{\partial u}{\partial t}|_{p^0} = \frac{\partial u}{\partial x_1}|_{p^0} cos\alpha_1 + \ldots + \frac{\partial u}{\partial x_n}|_{p^0} cos\alpha_n$ (differentiable implies directional derivative exists)
Page 4
The gradient of an n-variable function u at point X0 is the vector of partial derivatives. The direction of the maximum directional derivative is the unit vector in that direction. (If differentiable, it exists and is non-zero).
grad u(X0) = (∂u/∂x1, ∂u/∂x2, ..., ∂u/∂xn) → (x1, ..., xn): R^n → R^n
Theorem: u = f(x) in Ω has k-order partial derivatives (u is k-order differentiable in Ω) then u has r-order mixed partial derivatives (2 ≤ r ≤ k) and the order of differentiation is irrelevant.
The higher-order differential of an n-variable function: d^k u = (∂u/∂x1 dx1 + ... + ∂u/∂xn dxn)^k = D^k u
D: linear operator
Vector-valued function differentiation: ∆f = f(x0 + ∆x) - f(x0) = A ∆x + o(∥∆x∥). A ∆x: f(x) has a differential d f(x0); o(∥∆x∥) is higher-order infinitesimal as ∥x - x0∥.
Vector-valued function is differentiable at x0 if and only if each component (n-variable) function is differentiable at x0. Among them
A = J(f(x0)) = (∂f1/∂x1, ..., ∂fm/∂xn) = (∇f1) |x0, ..., (∇fm) |x0.
= (∂(f1, ..., fm)/∂(x1, ..., xn)) |x0.
Chain rule for vector-valued functions: If u = g(x) is differentiable at x0, y = f(u) at u0 = g(x0) is differentiable, then f ∘ g is differentiable at x0. d(f ∘ g)(x0) = J[f(u0)] ∙ J[g(x0)]. That is
J[f ∘ g(x0)] = J[f(u0)] ∙ J[g(x0)].
i = 1, ..., n.
Example: y = f(u1, u2, ..., um), uj = gj(x1, ..., xn), j = 1, ..., m.
Page 5
The Existence of Implicit Functions (Multivariate or Vector-Valued Functions):
- For an n+1 variable function F(x1, x2, ..., xn, y) in a neighborhood B(x0, y0, r) of point (x0, y0), if F is k-times continuously differentiable and F(x0, y0) = 0, ∂F/∂y(x0, y0) ≠ 0, then there exists an n-variable implicit function y = f(x), such that F(x, f(x)) = 0. The function f is k-times continuously differentiable.
$$\frac{\partial y}{\partial x_i} = -\frac{\partial F}{\partial x_i}(x, y) \quad i = 1, 2, ..., n$$
$$\frac{\partial^2 y}{\partial x_i \partial x_j} = -\frac{\partial}{\partial x_j}\left[\frac{\partial F}{\partial x_i}(x, y)\right]$$
Where ∂F/∂xi(x, y(x)) is the composite function of ∂F/∂xi(x, y) and y = y(x).
Example: F(x, y, z) = 0, the implicit function z = z(x, y) is 2-times continuously differentiable. Then,
$$\frac{\partial^2 z}{\partial y \partial x} = \frac{\partial}{\partial y}\left(\frac{\partial z}{\partial x}\right) = -\frac{\partial}{\partial y}\left[\frac{\partial F}{\partial x}(x, y, z(x, y))\right]$$
$$= -\frac{\partial}{\partial y}\left[\frac{\partial F}{\partial x}(x, y, z(x, y))\right] = -\frac{\partial}{\partial y}\left[\frac{\partial F}{\partial x}(x, y, z(x, y))\right]$$
$$\frac{\partial^2 z}{\partial y \partial x} = -\frac{\partial}{\partial y}\left[\frac{\partial F}{\partial x}(x, y, z(x, y))\right] = -\frac{\partial}{\partial y}\left[\frac{\partial F}{\partial x}(x, y, z(x, y))\right]$$
Where ∂y[∂F/∂x(x, y, z(x, y))] = ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F/∂y∂x(x, y) + ∂²F/∂z∂x(x, y) + ∂²F/∂x²(x, y) + ∂²F
Page 6
The function is continuously differentiable.
J(f(x)) = ∂(y1, ..., ym) / ∂(x1, ..., xn) = - (∂(E1, ..., Em) / ∂(y1, ..., ym))⁻¹ ∂(F1, ..., Fm) / ∂(x1, ..., xn).
Let Y = f(X) be a R^n → R^m continuously differentiable vector-valued function. If the Jacobian matrix J(f(X)) is invertible at X ∈ Ω, then f is invertible in B(X, δ). The inverse vector-valued function X = g(Y) is also continuously differentiable. J(g(Y)) = [J(f(X))]⁻¹
Example: A system of linear equations formed by full-rank matrices.
Surface's explicit representation: S = z = f(x, y), (x, y) ∈ Dxy ⊂ R^2.
Tangent plane: z - z0 = (∂f/∂x)(x0, y0)(x - x0) + (∂f/∂y)(x0, y0)(y - y0).
Normal vector: n = (∂f/∂x, ∂f/∂y, -1) |_p0.
Tangent plane equation: (∂f/∂x)(x0, y0) = (∂f/∂y)(x0, y0) = -1.
Surface's parametric representation: S = {x = x(u, v), y = y(u, v), z = z(u, v)}, (u, v) ∈ Duv ⊂ R^2.
Tangent plane: {x - x0 = (∂x/∂u)(u0, v0)(u - u0) + (∂x/∂v)(u0, v0)(v - v0),
y - y0 = (∂y/∂u)(u0, v0)(u - u0) + (∂y/∂v)(u0, v0)(v - v0),
z - z0 = (∂z/∂u)(u0, v0)(u - u0) + (∂z/∂v)(u0, v0)(v - v0)}.
Normal vector (all tangent lines are perpendicular): n = (∂(y, z) / ∂(u, v), ∂(z, x) / ∂(u, v), ∂(x, y) / ∂(u, v)) |_(u0, v0).
Tangent plane equation: (∂(y, z) / ∂(u, v)) |_(u0, v0) = (∂(z, x) / ∂(u, v)) |_(u0, v0) = (∂(x, y) / ∂(u, v)) |_(u0, v0).
Page 7
The surface's implicit function representation: S: F(x, y, z) = 0.
The tangent plane: $$\left(\frac{\partial F}{\partial x}\right)_{P_0}(x - x_0) + \left(\frac{\partial F}{\partial y}\right)_{P_0}(y - y_0) + \left(\frac{\partial F}{\partial z}\right)_{P_0}(z - z_0) = 0$$
The normal vector: n = ($\frac{\partial F}{\partial x}$, $\frac{\partial F}{\partial y}$, $\frac{\partial F}{\partial z}$)${}_{P_0}$
The normal equation: $$\frac{x - x_0}{\frac{\partial F}{\partial x}} = \frac{y - y_0}{\frac{\partial F}{\partial y}} = \frac{z - z_0}{\frac{\partial F}{\partial z}}$$
The parametric representation of a space curve: L: $$\begin{cases} x = x(t) \\ y = y(t) \\ z = z(t) \end{cases}$$
The tangent line: $$\begin{cases} x - x_0 = x'(t_0)(t - t_0) \\ y - y_0 = y'(t_0)(t - t_0) \\ z - z_0 = z'(t_0)(t - t_0) \end{cases}$$
The tangent plane: $$(x'(t_0), y'(t_0), z'(t_0)) \cdot ((x - x_0), (y - y_0), (z - z_0))^T = 0$$
The implicit function representation of the space curve: $$\begin{cases} F_1(x, y, z) = 0 \\ F_2(x, y, z) = 0 \end{cases}$$
The tangent plane at P_0: $$\left(\frac{\partial F_1}{\partial x}\right)_{P_0}(x - x_0) + \left(\frac{\partial F_1}{\partial y}\right)_{P_0}(y - y_0) + \left(\frac{\partial F_1}{\partial z}\right)_{P_0}(z - z_0) = 0$$
$$\left(\frac{\partial F_2}{\partial x}\right)_{P_0}(x - x_0) + \left(\frac{\partial F_2}{\partial y}\right)_{P_0}(y - y_0) + \left(\frac{\partial F_2}{\partial z}\right)_{P_0}(z - z_0) = 0$$
The intersection of the two tangent planes is the tangent line: T = ($\frac{\partial F_1}{\partial x}$, $\frac{\partial F_1}{\partial y}$, $\frac{\partial F_1}{\partial z}$)${}_{P_0} \times$ ($\frac{\partial F_2}{\partial x}$, $\frac{\partial F_2}{\partial y}$, $\frac{\partial F_2}{\partial z}$)${}_{P_0}$
Using the tangent direction, the parametric equation of the tangent line can be written.
Page 8
The second-order Taylor formula with Peano remainder for an n-variable function:
$$f(\mathbf{x}) = f(\mathbf{x}_0) + J(f(\mathbf{x}_0))\Delta\mathbf{x} + \frac{1}{2!}(\Delta\mathbf{x})^T H(\mathbf{x}_0)\Delta\mathbf{x} + o(\|\Delta\mathbf{x}\|^2)$$
n-variable function
Necessary condition for a critical point: f is differentiable at x0 (in all variables), if x0 is a local maximum (minimum) of f, then $$\frac{\partial f}{\partial x_i}(x_0) = 0 \Leftrightarrow \nabla f(x_0) = \vec{0}$$
If f is twice continuously differentiable in a neighborhood of x0, x0 is a critical point:
{If H(x0) is positive definite, x0 is a local minimum}
{If H(x0) is negative definite, x0 is a local maximum}
Example: Let u = f(x, y, z) = x^3 + y^2 + z^2 + 6xy + 2z. Find the critical points.
By $$u_x = u_y = u_z = 0$$ we get two critical points P1(6, -18, -1), P2(0, 0, -1)
The Hessian at P1 is:
$$H(P_1) = \begin{bmatrix} 36 & 6 & 0 \\ 6 & 2 & 0 \\ 0 & 0 & 2 \end{bmatrix}$$
The Hessian at P2 is:
$$H(P_2) = \begin{bmatrix} 0 & 6 & 0 \\ 6 & 2 & 0 \\ 0 & 0 & 2 \end{bmatrix}$$
∴ P1 is a local minimum, P2 is not a local minimum.
Conditional extremum problem:
$$\begin{cases} \min(\max) f(x_1, x_2, \ldots, x_n) \\ g(x_1, x_2, \ldots, x_n) = 0 \end{cases}$$
Objective function (n variables), constraint.
Theorem: The conditional extremum problem in Ω has an extremum point at Lagrange function
$$\mathcal{L}(x_1, x_2, \ldots, x_n, \lambda) = f(x_1, \ldots, x_n) + \lambda g(x_1, \ldots, x_n)$$
The extremum point of the Lagrange function is the extremum point of the original function.
Therefore, by solving the Lagrange function to find the extremum points and then substituting them into the objective function to compare the values, we can obtain the solution.
Page 9
Example
Space Ellipsoid
\[
\left\{
\begin{array}{l}
\frac{x^2}{a^2} + \frac{y^2}{b^2} + \frac{z^2}{c^2} = 1 \\
lx + my + nz = 0
\end{array}
\right.
\]
Objective Function
\[
\min(\max) \quad x^2 + y^2 + z^2
\]
Constraints:
\[
\begin{aligned}
&\text{Lagrange Function:} \quad L = x^2 + y^2 + z^2 + \lambda_1 \left( \frac{x^2}{a^2} + \frac{y^2}{b^2} + \frac{z^2}{c^2} - 1 \right) + \lambda_2 (lx + my + nz) \\
&\text{First-order conditions:} \\
&\quad L_x = 2x + \lambda_1 \frac{2x}{a^2} + \lambda_2 l = 0 \\
&\quad L_y = 2y + \lambda_1 \frac{2y}{b^2} + \lambda_2 m = 0 \\
&\quad L_z = 2z + \lambda_1 \frac{2z}{c^2} + \lambda_2 n = 0 \\
&\text{Substitute the constraints:} \\
&\quad a^* = \left( \frac{B + \sqrt{B^2 - AC}}{A} \right)^{\frac{1}{2}}, \quad b^* = \left( \frac{B - \sqrt{B^2 - AC}}{A} \right)^{\frac{1}{2}} \\
&\text{Where:} \\
&\quad A = a^2 l^2 + b^2 m^2 + c^2 n^2, \quad B = \frac{1}{2} \left[ l^2 a^2 (b^2 + c^2) + m^2 b^2 (c^2 + a^2) + n^2 c^2 (a^2 + b^2) \right], \quad C = a^2 b^2 c^2
\end{aligned}
\]
Page 10
Chapter 2: Variable Parameter Integrals
Definition: I(y) = ∫[a,b] f(x,y) dx, y ∈ [c,d].
Continuity: f(x,y) is continuous in D = [a,b] x [c,d], then I(y) is continuous in [c,d].
Ordering: f(x,y) and ∂f/∂y(x,y) are continuous in D = [a,b] x [c,d], then ∀y₀ ∈ (c,d)
[∂/∂y ∫[a,b] f(x,y) dx]y=y₀ = ∫[a,b] [∂/∂y f(x,y)]y=y₀ dx.
Leibniz's Formula: f(x,y) and ∂f/∂y(x,y) are continuous in D = [a,b] x [c,d], α(y) and β(y) are in [c,d] and bounded, then
∂/∂y ∫[α(y),β(y)] f(x,y) dx = ∫[α(y),β(y)] ∂f/∂y(x,y) dx + f(β(y),y) β'(y) - f(α(y),y) α'(y).
Ordering*: f(x,y) is continuous in D = [a,b] x [c,d], then
∫[c,d] dy ∫[a,b] f(x,y) dx = ∫[a,b] dx ∫[c,d] f(x,y) dy.
For improper integrals, add a uniform convergence condition to use the above theorems.
Example: Calculate ∫[0,∞] (e^(-ax) - e^(-bx)) / x dx (b ≥ a > 0).
By ∫[a,b] e^(-xy) dy = (e^(-ax) - e^(-bx)) / x, then the original integral = ∫[0,∞] dx ∫[a,b] e^(-xy) dy.
When a ≤ y ≤ b, |e^(-xy)| ≤ e^(-ax), thus ∫[a,b] e^(-xy) dx in y ∈ [a,b] is uniformly convergent, so
the original integral = ∫[a,b] dy ∫[0,∞] e^(-xy) dx = ln(b/a).
Page 11
Chapter 3: Double Integrals
The double integral of a bounded function f(x1, ..., xn) over the bounded region Ω in Rn is:
$$\int_{\Omega} \ldots \int_{\Omega} f(x_1, ..., x_n) \, dx_1 \ldots dx_n$$
Theorem: If f is a continuous function on the closed region Ω, then f has a Riemann integral on Ω.
Double integral variable substitution:
$$\iint_{D} f(x, y) \, dx \, dy = \iint_{Duv} f(x(u, v), y(u, v)) \left| \frac{D(x, y)}{D(u, v)} \right| \, du \, dv$$
In polar coordinates:
$$\begin{cases} x = \rho \cos \varphi \\ y = \rho \sin \varphi \end{cases} (\rho \geq 0, 0 \leq \varphi < 2\pi) \quad d\sigma = dx \, dy = \rho \, d\rho \, d\varphi$$
Triple integral by cylindrical coordinates:
$$M(\Omega) = \iiint_{\Omega} \rho(x, y, z) \, dV = \int_{a}^{b} dx \int_{y_1(x)}^{y_2(x)} dy \int_{z_1(x, y)}^{z_2(x, y, z)} \rho(x, y, z) \, dz$$
Triple integral variable substitution:
$$\iiint_{\Omega} f(x, y, z) \, dV = \iiint_{\Omega} f(x(r, s, t), y(r, s, t), z(r, s, t)) \left| \frac{D(x, y, z)}{D(r, s, t)} \right| \, dr \, ds \, dt$$
In cylindrical coordinates:
$$\begin{cases} x = \rho \cos \varphi \\ y = \rho \sin \varphi \\ z = z \end{cases} \quad \frac{D(x, y, z)}{D(\rho, \varphi, z)} = \rho$$
In spherical coordinates:
$$\begin{cases} x = r \sin \varphi \cos \theta \\ y = r \sin \varphi \sin \theta \\ z = r \cos \varphi \end{cases} \quad \frac{D(x, y, z)}{D(r, \varphi, \theta)} = -r^2 \sin \varphi \quad \begin{cases} r \geq 0 \\ 0 \leq \varphi < \pi \\ 0 \leq \theta < 2\pi \end{cases}$$
Example: Volume of a unit sphere in Rn:
$$V_n = \int_{\Omega_n} dV = \int_{-1}^{1} dx_n \int_{-1}^{1} \ldots \int_{-1}^{1} dx_1 \ldots dx_{n-1} \quad \Omega_{n-1} = \{ (x_1, \ldots, x_{n-1}) \mid x_1^2 + \ldots + x_{n-1}^2 \leq 1 - x_n^2 \}$$
$$\therefore V_n = \int_{\Omega_n} (\sqrt{1 - x_n^2})^{n-1} \, dx_n \int_{-1}^{1} \ldots \int_{-1}^{1} du_1 \ldots du_{n-1}$$
$$\therefore \frac{D(x_1, \ldots, x_{n-1})}{D(u_1, \ldots, u_{n-1})} = (\sqrt{1 - x_n^2})^{n-1} \quad \therefore V_n = \int_{\Omega_n} (\sqrt{1 - x_n^2})^{n-1} \, dx_n \int_{-1}^{1} \ldots \int_{-1}^{1} du_1 \ldots du_{n-1}$$
Page 12
$$V_n = V_{n-1} \int_{-1}^{1} (\sqrt{1-t^2})^n dt = V_{n-1} \cdot B\left(\frac{1}{2}, \frac{n+1}{2}\right)$$
$$\therefore V_n = \frac{\Gamma\left(\frac{n+1}{2}\right)}{\Gamma\left(\frac{n+2}{2}\right)} \sqrt{\pi} V_{n-1}$$
$$\therefore V_{2m} = \frac{\pi^m}{m!}, \quad V_{2m+1} = \frac{2^m \pi^{m-1}}{(2m-1)!}$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_2 & y_2 & 1
\end{vmatrix} = A$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_3 & y_3 & 1
\end{vmatrix} = B$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_4 & y_4 & 1
\end{vmatrix} = C$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_5 & y_5 & 1
\end{vmatrix} = D$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_6 & y_6 & 1
\end{vmatrix} = E$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_7 & y_7 & 1
\end{vmatrix} = F$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_8 & y_8 & 1
\end{vmatrix} = G$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_9 & y_9 & 1
\end{vmatrix} = H$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{10} & y_{10} & 1
\end{vmatrix} = I$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{11} & y_{11} & 1
\end{vmatrix} = J$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{12} & y_{12} & 1
\end{vmatrix} = K$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{13} & y_{13} & 1
\end{vmatrix} = L$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{14} & y_{14} & 1
\end{vmatrix} = M$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{15} & y_{15} & 1
\end{vmatrix} = N$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{16} & y_{16} & 1
\end{vmatrix} = O$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{17} & y_{17} & 1
\end{vmatrix} = P$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{18} & y_{18} & 1
\end{vmatrix} = Q$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{19} & y_{19} & 1
\end{vmatrix} = R$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{20} & y_{20} & 1
\end{vmatrix} = S$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{21} & y_{21} & 1
\end{vmatrix} = T$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{22} & y_{22} & 1
\end{vmatrix} = U$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{23} & y_{23} & 1
\end{vmatrix} = V$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{24} & y_{24} & 1
\end{vmatrix} = W$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{25} & y_{25} & 1
\end{vmatrix} = X$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{26} & y_{26} & 1
\end{vmatrix} = Y$$
$$\begin{vmatrix}
x & y & 1 \\
x_1 & y_1 & 1 \\
x_{27} & y_{27} & 1
\end{vmatrix} = Z$$
$$\therefore A \cdot B \cdot C \cdot D \cdot E \cdot F \cdot G \cdot H \cdot I \cdot J \cdot K \cdot L \cdot M \cdot N \cdot O \cdot P \cdot Q \cdot R \cdot S \cdot T \cdot U \cdot V \
Page 13
Chapter 4: First Kind of Line Integrals and First Kind of Surface Integrals
First Kind of Line Integrals:
$$\int_{AB} f(x, y, z) dl = \int_{BA} f(x, y, z) dl$$
$$\int_{L} f(x, y, z) dl = \int_{L_1} f(x, y, z) dl + \int_{L_2} f(x, y, z) dl$$
$$\int_{L} f(x, y, z) dl = \int_{a}^{b} f(x(t), y(t), z(t)) \sqrt{x'(t)^2 + y'(t)^2 + z'(t)^2} dt$$
First Kind of Surface Integrals: If the surface S's equation is parametric form
$$\begin{cases} x = x(u, v) \\ y = y(u, v) \\ z = z(u, v) \end{cases}$$
Then the surface's normal vector is
$$\begin{pmatrix} A = \begin{vmatrix} y_v & z_v \\ x_v & z_v \\ x_v & y_v \end{vmatrix} \\ B = \begin{vmatrix} z_u & x_u \\ z_v & x_v \\ z_v & y_v \end{vmatrix} \\ C = \begin{vmatrix} x_u & y_u \\ x_v & y_v \end{vmatrix}$$
Then $|\cos(\vec{n}, \vec{z})| = \frac{C}{\sqrt{A^2 + B^2 + C^2}}$, $ds = \frac{\partial x \partial y}{|\cos(\vec{n}, \vec{z})|} = \frac{1}{\cos(\vec{n}, \vec{z})} \cdot \frac{\partial(x, y)}{\partial(u, v)} du dv = \sqrt{A^2 + B^2 + C^2} du dv$
And $A^2 + B^2 + C^2 = EG - F^2$, where $E = 1 + (\frac{\partial z}{\partial x})^2$, $G = 1 + (\frac{\partial z}{\partial y})^2$, $F = \frac{\partial z}{\partial x} \frac{\partial z}{\partial y}$.
$\therefore ds = \sqrt{EG - F^2} du dv$.
$\therefore \iint_{S} f(x, y, z) ds = \iint_{D_{xy}} f(x(u, v), y(u, v), z(u, v)) \sqrt{EG - F^2} du dv$.
Example:
$\iint_{S} f(x, y, z) ds = \iint_{D_{xy}} f(x, y, z(x, y)) \sqrt{1 + (\frac{\partial z}{\partial x})^2 + (\frac{\partial z}{\partial y})^2} dx dy$.
Example:
The surface area element $ds = \sqrt{EG - F^2} d\varphi dv = r^2 \sin\varphi d\varphi dv$.
Page 14
Chapter 5: Second Kind Line Integrals and Second Kind Surface Integrals
Second Kind Line Integral:
$$\int_{L(A)}^{(B)} \vec{F} \cdot d\vec{r} = -\int_{L(B)}^{(A)} \vec{F} \cdot d\vec{r}$$
Vector Form:
$$\int_{L(A)}^{(B)} \vec{F} \cdot d\vec{r} = \int X dx + Y dy + Z dz = \int_{a}^{b} X \cdot x'(t) dt + \int_{a}^{b} Y \cdot y'(t) dt + \int_{a}^{b} Z \cdot z'(t) dt$$
Converted to First Kind Line Integral:
$$\int_{L(A)}^{(B)} \vec{F} \cdot d\vec{r} = \int_{AB} (X \cos\alpha + Y \cos\beta + Z \cos\gamma) ds$$
Second Kind Surface Integral:
$$\iint_{S^+} \vec{F} \cdot d\vec{S} = \iint_{S^+} (X \cos\alpha + Y \cos\beta + Z \cos\gamma) ds$$
Properties:
$$\iint_{S^+} \vec{F} \cdot d\vec{S} = -\iint_{S^+} \vec{F} \cdot d\vec{S}$$
Parametric Form:
$$\vec{r}(u, v) = (x(u, v), y(u, v), z(u, v)) \quad (u, v) \in D_{uv}$$
When $\vec{r}_u \times \vec{r}_v \neq 0$, unit tangent vector $\vec{r}_s = \pm \frac{\vec{r}_u \times \vec{r}_v}{\|\vec{r}_u \times \vec{r}_v\|} = \pm \frac{(A, B, C)}{\sqrt{A^2 + B^2 + C^2}} = \pm (\cos\alpha, \cos\beta, \cos\gamma)$
And $ds = \sqrt{A^2 + B^2 + C^2} \, du dv$
Therefore:
$$dy \wedge dz = \cos\alpha \, ds = \pm A \, du dv$$
$$dz \wedge dx = \cos\beta \, ds = \pm B \, du dv$$
$$dx \wedge dy = \cos\gamma \, ds = \pm C \, du dv$$
$$\therefore \iint_{S^+} \vec{F} \cdot d\vec{S} = \pm \iint_{D_{uv}} (XA + YB + ZC) \, du dv$$
Page 15
Green's Theorem: $$\int_S \nabla \times \vec{V} \cdot d\vec{S} = \oint_{\partial S} \vec{V} \cdot d\vec{\ell}$$ where $$\nabla \times \vec{V} = \begin{vmatrix} \hat{i} & \hat{j} & \hat{k} \\ \frac{\partial}{\partial x} & \frac{\partial}{\partial y} & \frac{\partial}{\partial z} \\ V_x & V_y & V_z \end{vmatrix}$$
If $$\nabla \times \vec{V} = 0$$, then the second kind of line integral $$\int_{A_1}^{A_2} \vec{V} \cdot d\vec{\ell}$$ is path-independent. And $$\int_{A_1}^{A_2} \vec{V} \cdot d\vec{\ell} = u(X) \Big|_{A_1}^{A_2} = u(A_2) - u(A_1)$$: primitive function.
Gauss's Theorem: $$\int_{\Omega} \nabla \cdot \vec{F} \cdot dV = \oint_{\partial \Omega} \vec{F} \cdot d\vec{S}$$ where $$\nabla \cdot \vec{F} = \frac{\partial F_x}{\partial x} + \frac{\partial F_y}{\partial y} + \frac{\partial F_z}{\partial z}$$
Theorem: In a simply connected region, a vector field $$\vec{V} = (X, Y, Z)$$ is a gradient field if and only if there exists a function $$u(x, y, z)$$ such that $$du = Xdx + Ydy + Zdz$$ ($$R.P. X = \frac{\partial u}{\partial x}, Y = \frac{\partial u}{\partial y}, Z = \frac{\partial u}{\partial z}, du = \nabla u \cdot d\vec{r}$$). $$u$$ is the potential function of $$\vec{V}$$.
$$\nabla = \hat{e_x} \frac{\partial}{\partial x} + \hat{e_y} \frac{\partial}{\partial y} + \hat{e_z} \frac{\partial}{\partial z}$$
$$\nabla u \cdot \vec{f} = \frac{\partial}{\partial x} f_x + \frac{\partial}{\partial y} f_y + \frac{\partial}{\partial z} f_z$$
$$\text{rot} \vec{f} = \begin{vmatrix} \hat{e_x} & \hat{e_y} & \hat{e_z} \\ \frac{\partial}{\partial x} & \frac{\partial}{\partial y} & \frac{\partial}{\partial z} \\ f_x & f_y & f_z \end{vmatrix}$$
$$\text{grad} \varphi = \frac{\partial \varphi}{\partial x} \hat{e_x} + \frac{\partial \varphi}{\partial y} \hat{e_y} + \frac{\partial \varphi}{\partial z} \hat{e_z}$$
Theorem: $$\nabla \times \nabla \varphi = 0 \Rightarrow$$ a solenoidal field can be expressed as the gradient of a scalar potential. (Continuous and differentiable). $$\nabla \cdot \nabla \times \vec{f} = 0 \Rightarrow$$ a solenoidal field can be expressed as the curl of another vector field.
Page 16
Chapter 6: Space Curves and Space Surfaces
In R^3, a vector function is defined as:
$$\vec{r}(t) = (x(t), y(t), z(t)), t \in [a, b]$$
The derivative at t0:
$$d\vec{r}(t_0) = \left(\frac{dx}{dt}, \frac{dy}{dt}, \frac{dz}{dt}\right)^T_{t_0} dt$$
The derivative at t0:
$$\vec{r}'(t_0) = (x'(t_0), y'(t_0), z'(t_0)) : R^1 \rightarrow R^3$$
Properties:
$$\frac{d}{dt}(\vec{r}_1 \times \vec{r}_2) = \frac{d\vec{r}_1}{dt} \times \vec{r}_2 + \vec{r}_1 \times \frac{d\vec{r}_2}{dt}$$
$$\frac{d}{dt}(\vec{r}_1 \cdot \vec{r}_2 \cdot \vec{r}_3) = (\frac{d\vec{r}_1}{dt}, \vec{r}_2, \vec{r}_3) + (\vec{r}_1, \frac{d\vec{r}_2}{dt}, \vec{r}_3) + (\vec{r}_1, \vec{r}_2, \frac{d\vec{r}_3}{dt})$$
where $$(\vec{r}_1, \vec{r}_2, \vec{r}_3) = \vec{r}_1 \cdot (\vec{r}_2 \times \vec{r}_3)$$
If $\vec{F}(t)$ is n times continuously differentiable, then each component of $\vec{F}(t)$ is n times continuously differentiable.
Vector function integration:
$$\int \vec{F}(t) dt = \vec{F}(t) + \vec{C}$$
Properties:
$$\int \vec{a} \cdot \vec{F}(t) dt = \vec{a} \cdot \int \vec{r}(t) dt$$
$$\int \vec{a} \times \vec{F}(t) dt = \vec{a} \times \int \vec{F}(t) dt$$
Arc length:
$$s = \int_a^b ||\vec{r}(t)|| dt = \int_a^b \sqrt{x^2(t) + y^2(t) + z^2(t)} dt$$
Since $ds > 0$, the inverse function $t = t(s)$ exists, allowing us to parameterize the curve by arc length:
$$\vec{F} = \vec{F}(s) = (x(s), y(s), z(s))$$
Convention:
$$\vec{F} = \frac{d\vec{F}}{dt}, \vec{F}' = \frac{d\vec{F}}{ds}$$
Page 17
The tangent vector of the curve: $\vec{T} = \vec{r}'(s_0)$, $\|\vec{T}\| = 1$
The tangent line equation: $\vec{p} = \vec{r}(s_0) + \lambda\vec{T} = \vec{r}(s_0) + \lambda\vec{r}'(s_0)$. General parameter: $\vec{T} = \frac{\vec{r}(t)}{\|\vec{r}(t)\|}$
The tangent plane equation: $(\vec{p} - \vec{r}(s_0)) \cdot \vec{T} = (\vec{p} - \vec{r}(s_0)) \cdot \vec{r}'(s_0) = 0$.
The unit normal vector: $\vec{B} = \frac{\vec{r}'(s_0) \times \vec{r}''(s_0)}{\|\vec{r}'(s_0) \times \vec{r}''(s_0)\|}$
The normal line equation: $\vec{p} = \vec{r}(s_0) + \lambda\vec{B}$ General parameter: $\vec{B} = \frac{\vec{r}'(t) \times \vec{r}''(t)}{\|\vec{r}'(t) \times \vec{r}''(t)\|}$
The tangent plane equation: $(\vec{p} - \vec{r}(s_0)) \cdot \vec{B} = 0$.
The unit binormal vector: $\vec{N} = \frac{\vec{r}'''(s_0)}{\|\vec{r}'''(s_0)\|}$
The binormal line equation: $\vec{p} = \vec{r}(s_0) + \lambda\vec{N}$ General parameter: $\vec{N} = \frac{(\vec{r} \times \vec{r}') \times \vec{r}'}{\|\vec{r} \times \vec{r}'\| \|\vec{r}'\|}$
The tangent plane equation: $(\vec{p} - \vec{r}(s_0)) \cdot \vec{N} = 0$.
Frenet's Frame (natural frame): $\{\vec{F}(s), \vec{T}(s), \vec{N}(s), \vec{B}(s)\}$
Curvature: $k(s) = \|\vec{r}'(s)\|$ General parameter: $k(t) = \frac{\|\vec{r}'(t) \times \vec{r}''(t)\|}{\|\vec{r}'(t)\|^3}$
Curvature radius: $R(s) = \frac{1}{k(s)}$. $\vec{T}'(s) = k(s)\vec{N}(s)$
Curvature center: $\vec{r}(s) + R(s)\vec{N}(s)$ $k(s) = \frac{d\varphi}{ds}$.
Torsion: $\tau(s) = -\vec{B}(s) \cdot \vec{N}(s)$. $|\tau(s)| = \left|\frac{d\varphi}{ds}\right|$
General parameter: $\tau = \frac{\left(\vec{F}, \vec{F}', \vec{F}''\right)}{\left(\vec{F} \times \vec{F}'\right)^2}$
$\vec{B}' = -\tau\vec{N}$
Frenet's Formula: $\begin{pmatrix} \vec{T}' \\ \vec{N}' \\ \vec{B}' \end{pmatrix} = \begin{pmatrix} 0 & k & 0 \\ -k & 0 & \tau \\ 0 & -\tau & 0 \end{pmatrix} \begin{pmatrix} \vec{T} \\ \vec{N} \\ \vec{B} \end{pmatrix}$
The arc length parameter s, curvature k(s), and torsion τ(s) are the basic quantities describing a curve.
Page 18
The parametric form of a surface is: $\vec{r} = \vec{r}(u, v) = (x(u, v), y(u, v), z(u, v))$
$u$ curve: Fix $v$ and vary $u$.
$v$ curve: Fix $u$ and vary $v$.
Parameter transformation: $\vec{r}_{u} \times \vec{r}_{v} = \frac{D(u, v)}{|\vec{r}_{u} \times \vec{r}_{v}|} (\vec{r}_{u} \times \vec{r}_{v})$. By the chain rule of composite functions.
$\vec{r}_{u}, \vec{r}_{v}$: Basis of the tangent space.
Normal vector: $\vec{n} = \frac{\vec{r}_{u} \times \vec{r}_{v}}{|\vec{r}_{u} \times \vec{r}_{v}|}$
Tangent plane: $\vec{n} \cdot (\vec{p} - \vec{r}) = 0$. Parametric equation: $\vec{p} = \vec{r} + \lambda \vec{r}_{u} + \mu \vec{r}_{v}$.
Normal equation: $\vec{p} = \vec{r} + \lambda \vec{n}$
The first fundamental form of the surface: $ds^2 = d\vec{r} \cdot d\vec{r} = (\vec{r}_{u} du + \vec{r}_{v} dv)^2 = \vec{r}_{u}^2 du^2 + 2 \vec{r}_{u} \cdot \vec{r}_{v} du dv + \vec{r}_{v}^2 dv^2$
$= (du, dv) \begin{bmatrix} E & F \\ F & G \end{bmatrix} \begin{bmatrix} du \\ dv \end{bmatrix}$.
The first fundamental form is independent of the parameter transformation.
The curvature of a curve on the surface: $k \vec{N} = k_n \vec{n} + k_g (\vec{n} \times \vec{r})$. $k_n$ is the normal curvature, $k_g$ is the geodesic curvature.
$k$: curvature
$\therefore k_n = k \vec{N} \cdot \vec{n} = \vec{r}' \cdot \vec{n} = \vec{r}_{u} \cdot \vec{n}$
$\begin{aligned}
\angle &= -\vec{r}_{u} \cdot \vec{r}_{u} \\
M &= -\vec{r}_{u} \cdot \vec{r}_{v} = -\vec{r}_{v} \cdot \vec{r}_{u} \\
N &= -\vec{r}_{v} \cdot \vec{r}_{v}
\end{aligned}$
$\begin{aligned}
k_n &= -\frac{d\vec{r} \cdot d\vec{r}_{u}}{ds^2} \\
&= \frac{(\vec{r}_{u} u + \vec{r}_{v} v)'}{\vec{r}_{uu} \cdot \vec{n}} \cdot \vec{n} \\
&= (\vec{r}_{uu} \cdot \vec{n}) u^2 + 2 \vec{r}_{uv} \cdot \vec{n} u v + (\vec{r}_{vv} \cdot \vec{n}) v^2 \\
&= \frac{du^2 + 2M du dv + N dv^2}{F} \\
&= \frac{du}{dv} \begin{bmatrix} \frac{2}{F} & M \\ M & N \end{bmatrix} \frac{du}{dv}
\end{aligned}$