東京大学 新領域創成科学研究科 メディカル情報生命専攻 2024年1月実施 問題8
Author
zephyr , 祭音Myyura
Description
Suppose that the eigenvalues and the corresponding eigenvectors of an n × n n \times n n × n square matrix A \mathbf{A} A are λ 1 , … , λ n \lambda_1, \dots, \lambda_n λ 1 , … , λ n and α 1 , … , α n \mathbf{\alpha}_1, \dots, \mathbf{\alpha}_n α 1 , … , α n respectively.
Suppose that I n \mathbf{I}_n I n is the n × n n \times n n × n identity matrix, and the inverse matrix of an invertible matrix C \mathbf{C} C is C − 1 \mathbf{C}^{-1} C − 1 .
Answer the following questions.
Show all the eigenvalues and the corresponding eigenvectors of A 2 \mathbf{A}^2 A 2 .
If λ 1 , … , λ n \lambda_1, \dots, \lambda_n λ 1 , … , λ n are mutually different, show that P − 1 A P \mathbf{P}^{-1} \mathbf{A} \mathbf{P} P − 1 AP is a diagonal matrix, using P = ( α 1 , … , α n ) \mathbf{P} = (\mathbf{\alpha}_1, \dots, \mathbf{\alpha}_n) P = ( α 1 , … , α n ) that is a matrix of concatenated eigenvectors.
Show all the eigenvalues and the corresponding eigenvectors of B \mathbf{B} B .
B = ( 3 0 0 − 2 3 2 0 0 1 ) \mathbf{B} = \begin{pmatrix}
3 & 0 & 0 \\
-2 & 3 & 2 \\
0 & 0 & 1
\end{pmatrix} B = 3 − 2 0 0 3 0 0 2 1
Suppose that μ \mu μ is the maximum eigenvalue of B \mathbf{B} B , and γ = ( 1 0 0 ) \mathbf{\gamma} = \begin{pmatrix} 1 \\ 0 \\ 0 \end{pmatrix} γ = 1 0 0 .
Calculate δ = ( B − μ I 3 ) γ \mathbf{\delta} = (\mathbf{B} - \mu \mathbf{I}_3)\mathbf{\gamma} δ = ( B − μ I 3 ) γ .
Suppose that β \beta β is the eigenvector of B \mathbf{B} B corresponding to the minimum eigenvalue. Calculate Q − 1 B Q \mathbf{Q}^{-1}\mathbf{B}\mathbf{Q} Q − 1 BQ using Q = ( δ , γ , β ) \mathbf{Q} = (\mathbf{\delta}, \mathbf{\gamma}, \beta) Q = ( δ , γ , β ) that is a matrix concatenating δ , γ , β \mathbf{\delta}, \mathbf{\gamma}, \beta δ , γ , β .
Suppose that m m m is an arbitrary positive integer. Calculate B m \mathbf{B}^m B m .
假设 n × n n \times n n × n 方阵 A \mathbf{A} A 的特征值及相应的特征向量分别为 λ 1 , … , λ n \lambda_1, \dots, \lambda_n λ 1 , … , λ n 和 α 1 , … , α n \mathbf{\alpha}_1, \dots, \mathbf{\alpha}_n α 1 , … , α n 。
假设 I n \mathbf{I}_n I n 是 n × n n \times n n × n 的单位矩阵,并且可逆矩阵 C \mathbf{C} C 的逆矩阵为 C − 1 \mathbf{C}^{-1} C − 1 。
回答以下问题。
展示 A 2 \mathbf{A}^2 A 2 的所有特征值及相应的特征向量。
如果 λ 1 , … , λ n \lambda_1, \dots, \lambda_n λ 1 , … , λ n 是互不相同的,证明 P − 1 A P \mathbf{P}^{-1} \mathbf{A} \mathbf{P} P − 1 AP 是一个对角矩阵,其中 P = ( α 1 , … , α n ) \mathbf{P} = (\mathbf{\alpha}_1, \dots, \mathbf{\alpha}_n) P = ( α 1 , … , α n ) 是由特征向量构成的矩阵。
展示 B \mathbf{B} B 的所有特征值及相应的特征向量。
B = ( 3 0 0 − 2 3 2 0 0 1 ) \mathbf{B} = \begin{pmatrix}
3 & 0 & 0 \\
-2 & 3 & 2 \\
0 & 0 & 1
\end{pmatrix} B = 3 − 2 0 0 3 0 0 2 1
假设 μ \mu μ 是 B \mathbf{B} B 的最大特征值,并且 γ = ( 1 0 0 ) \mathbf{\gamma} = \begin{pmatrix} 1 \\ 0 \\ 0 \end{pmatrix} γ = 1 0 0 。
计算 δ = ( B − μ I 3 ) γ \mathbf{\delta} = (\mathbf{B} - \mu \mathbf{I}_3)\mathbf{\gamma} δ = ( B − μ I 3 ) γ 。
假设 β \beta β 是 B \mathbf{B} B 对应于最小特征值的特征向量。计算 Q − 1 B Q \mathbf{Q}^{-1}\mathbf{B}\mathbf{Q} Q − 1 BQ ,其中 Q = ( δ , γ , β ) \mathbf{Q} = (\mathbf{\delta}, \mathbf{\gamma}, \beta) Q = ( δ , γ , β ) 是由 δ , γ , β \mathbf{\delta}, \mathbf{\gamma}, \beta δ , γ , β 构成的矩阵。
假设 m m m 是任意正整数。计算 B m \mathbf{B}^m B m 。
题目描述
设 n × n n\times n n × n 方阵 A \mathbf A A 的特征值及对应特征向量分别为
λ 1 , … , λ n , α 1 , … , α n . \lambda_1,\ldots,\lambda_n,\qquad
\boldsymbol\alpha_1,\ldots,\boldsymbol\alpha_n. λ 1 , … , λ n , α 1 , … , α n .
I n \mathbf I_n I n 表示单位矩阵,可逆矩阵 C \mathbf C C 的逆记为 C − 1 \mathbf C^{-1} C − 1 。回答:
列出 A 2 \mathbf A^2 A 2 的全部特征值及对应特征向量。
若 λ 1 , … , λ n \lambda_1,\ldots,\lambda_n λ 1 , … , λ n 两两不同,令
P = ( α 1 , … , α n ) , \mathbf P=(\boldsymbol\alpha_1,\ldots,\boldsymbol\alpha_n), P = ( α 1 , … , α n ) ,
证明 P − 1 A P \mathbf P^{-1}\mathbf A\mathbf P P − 1 AP 为对角矩阵。
求
B = ( 3 0 0 − 2 3 2 0 0 1 ) \mathbf B=
\begin{pmatrix}
3&0&0\\
-2&3&2\\
0&0&1
\end{pmatrix} B = 3 − 2 0 0 3 0 0 2 1
的全部特征值与相应特征向量。
设 μ \mu μ 为 B \mathbf B B 的最大特征值,
γ = ( 1 0 0 ) , \boldsymbol\gamma=\begin{pmatrix}1\\0\\0\end{pmatrix}, γ = 1 0 0 ,
计算
δ = ( B − μ I 3 ) γ . \boldsymbol\delta=(\mathbf B-\mu\mathbf I_3)\boldsymbol\gamma. δ = ( B − μ I 3 ) γ .
设 β \boldsymbol\beta β 是 B \mathbf B B 最小特征值对应的特征向量,并令
Q = ( δ , γ , β ) , \mathbf Q=(\boldsymbol\delta,\boldsymbol\gamma,\boldsymbol\beta), Q = ( δ , γ , β ) ,
计算 Q − 1 B Q \mathbf Q^{-1}\mathbf B\mathbf Q Q − 1 BQ 。
对任意正整数 m m m ,计算 B m \mathbf B^m B m 。
Kai
1. Eigenpairs of A 2 \mathbf A^2 A 2
For every i i i ,
A 2 α i = A ( λ i α i ) = λ i 2 α i . \mathbf A^2\boldsymbol\alpha_i
=\mathbf A(\lambda_i\boldsymbol\alpha_i)
=\lambda_i^2\boldsymbol\alpha_i. A 2 α i = A ( λ i α i ) = λ i 2 α i .
Thus the eigenvalues of A 2 \mathbf A^2 A 2 are λ i 2 \lambda_i^2 λ i 2 (with algebraic multiplicity), and each α i \boldsymbol\alpha_i α i is a corresponding eigenvector. If A \mathbf A A is diagonalizable, the full eigenspace is
E μ ( A 2 ) = ⨁ λ ∈ σ ( A ) λ 2 = μ E λ ( A ) . E_\mu(\mathbf A^2)=
\bigoplus_{\substack{\lambda\in\sigma(\mathbf A)\\\lambda^2=\mu}}
E_\lambda(\mathbf A). E μ ( A 2 ) = λ ∈ σ ( A ) λ 2 = μ ⨁ E λ ( A ) .
Without diagonalizability, the listed eigenvectors need not describe every eigenvector of A 2 \mathbf A^2 A 2 .
2. Diagonalization of A \mathbf A A
Distinct eigenvalues have linearly independent eigenvectors, so P \mathbf P P is invertible. Since
A P = ( λ 1 α 1 , … , λ n α n ) = P diag ( λ 1 , … , λ n ) , \mathbf A\mathbf P
=(\lambda_1\boldsymbol\alpha_1,\ldots,\lambda_n\boldsymbol\alpha_n)
=\mathbf P\operatorname{diag}(\lambda_1,\ldots,\lambda_n), AP = ( λ 1 α 1 , … , λ n α n ) = P diag ( λ 1 , … , λ n ) ,
we obtain
P − 1 A P = diag ( λ 1 , … , λ n ) . \mathbf P^{-1}\mathbf A\mathbf P
=\operatorname{diag}(\lambda_1,\ldots,\lambda_n). P − 1 AP = diag ( λ 1 , … , λ n ) .
3. Eigenpairs of B \mathbf B B
det ( B − λ I 3 ) = ( 3 − λ ) 2 ( 1 − λ ) . \det(\mathbf B-\lambda\mathbf I_3)=(3-\lambda)^2(1-\lambda). det ( B − λ I 3 ) = ( 3 − λ ) 2 ( 1 − λ ) .
Hence
λ = 3 : E 3 = span { ( 0 1 0 ) } , λ = 1 : E 1 = span { ( 0 − 1 1 ) } . \lambda=3:\quad E_3=\operatorname{span}\!\left\{\begin{pmatrix}0\\1\\0\end{pmatrix}\right\},
\qquad
\lambda=1:\quad E_1=\operatorname{span}\!\left\{\begin{pmatrix}0\\-1\\1\end{pmatrix}\right\}. λ = 3 : E 3 = span ⎩ ⎨ ⎧ 0 1 0 ⎭ ⎬ ⎫ , λ = 1 : E 1 = span ⎩ ⎨ ⎧ 0 − 1 1 ⎭ ⎬ ⎫ .
The eigenvalue 3 3 3 has algebraic multiplicity 2 2 2 and geometric multiplicity 1 1 1 .
4. Computing δ \boldsymbol\delta δ
Here μ = 3 \mu=3 μ = 3 , so
δ = ( B − 3 I 3 ) γ = ( 0 − 2 0 ) . \boldsymbol\delta=(\mathbf B-3\mathbf I_3)\boldsymbol\gamma
=\begin{pmatrix}0\\-2\\0\end{pmatrix}. δ = ( B − 3 I 3 ) γ = 0 − 2 0 .
5. Computing Q − 1 B Q \mathbf Q^{-1}\mathbf B\mathbf Q Q − 1 BQ
Choose β = ( 0 , − 1 , 1 ) T \boldsymbol\beta=(0,-1,1)^T β = ( 0 , − 1 , 1 ) T . Then
B δ = 3 δ , B γ = δ + 3 γ , B β = β . \mathbf B\boldsymbol\delta=3\boldsymbol\delta,
\qquad
\mathbf B\boldsymbol\gamma=\boldsymbol\delta+3\boldsymbol\gamma,
\qquad
\mathbf B\boldsymbol\beta=\boldsymbol\beta. B δ = 3 δ , B γ = δ + 3 γ , B β = β .
Therefore
Q − 1 B Q = ( 3 1 0 0 3 0 0 0 1 ) . \mathbf Q^{-1}\mathbf B\mathbf Q
=\begin{pmatrix}
3&1&0\\
0&3&0\\
0&0&1
\end{pmatrix}. Q − 1 BQ = 3 0 0 1 3 0 0 0 1 .
6. Computing B m \mathbf B^m B m
Raising the Jordan block above to the m m m -th power gives
B m = ( 3 m 0 0 − 2 m 3 m − 1 3 m 3 m − 1 0 0 1 ) ( m ≥ 1 ) . \boxed{
\mathbf B^m=
\begin{pmatrix}
3^m&0&0\\
-2m3^{m-1}&3^m&3^m-1\\
0&0&1
\end{pmatrix}}
\qquad(m\ge1). B m = 3 m − 2 m 3 m − 1 0 0 3 m 0 0 3 m − 1 1 ( m ≥ 1 ) .
Knowledge
重点词汇
singular value decomposition (SVD) 奇异值分解
pseudoinverse 广义逆
surjective 满射
injective 单射
orthogonal decomposition 正交分解
参考资料
"Linear Algebra and Its Applications" by Gilbert Strang, Chapter 7: The Singular Value Decomposition (SVD)
"Matrix Computations" by Gene H. Golub and Charles F. Van Loan, Chapter 2: Matrix Analysis