跳到主要内容

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 2014年8月実施 問題8

Author

zephyr

Description

Let Z={(z1z2):complex number,z12+z220}Z^* = \left\{ \begin{pmatrix} z_1 \\ z_2 \end{pmatrix} : \text{complex number}, |z_1|^2 + |z_2|^2 \neq 0 \right\} be the set of non-zero complex two-dimensional vectors. Let M=(abbd)M = \begin{pmatrix} a & b \\ b & d \end{pmatrix} be a 2 by 2 real symmetric matrix, and I=(1001)I = \begin{pmatrix} 1 & 0 \\ 0 & 1 \end{pmatrix} be the unit matrix.

(1) Find all the eigenvalues λ1,λ2\lambda_1, \lambda_2 of MM.

(2) Under the assumption of λ1λ2\lambda_1 \neq \lambda_2, answer i) and ii).

  • i) Let U=(v1,v2)U = (v_1, v_2) be the matrix whose first and second columns consist of the eigenvectors v1v_1 and v2v_2 for the eigenvalues λ1\lambda_1 and λ2\lambda_2, respectively. Show that UU is invertible and satisfies M=U(λ100λ2)U1M = U \begin{pmatrix} \lambda_1 & 0 \\ 0 & \lambda_2 \end{pmatrix} U^{-1}.
  • ii) Prove that the set {U1xxZ}\{ U^{-1} x | x \in Z^* \} and ZZ^* are equal.

(3) For each of the statements A), B), and C), answer the conditions on matrix elements a,b,da, b, d for the statement to hold.

  • A) Every yZy \in Z^* can be expressed as y=Mxy = Mx with some xZx \in Z^*.
  • B) No yZy \in Z^* can be expressed as y=Mxy = Mx with some xZx \in Z^*.
  • C) At least one yZy \in Z^* can be expressed as y=(Mλ1I)xy = (M - \lambda_1 I)x with some xZx \in Z^*.

Z={(z1z2):复数,z12+z220}Z^* = \left\{ \begin{pmatrix} z_1 \\ z_2 \end{pmatrix} : \text{复数}, |z_1|^2 + |z_2|^2 \neq 0 \right\} 为非零复二维向量的集合。设 M=(abbd)M = \begin{pmatrix} a & b \\ b & d \end{pmatrix} 为一个 2×2 的实对称矩阵,I=(1001)I = \begin{pmatrix} 1 & 0 \\ 0 & 1 \end{pmatrix} 为单位矩阵。

(1) 找出 MM 的所有特征值 λ1,λ2\lambda_1, \lambda_2

(2) 在假设 λ1λ2\lambda_1 \neq \lambda_2 的条件下,回答 i) 和 ii)。

  • i) 设 U=(v1,v2)U = (v_1, v_2) 为一个矩阵,其第一列和第二列分别由特征值 λ1\lambda_1λ2\lambda_2 的特征向量 v1v_1v2v_2 组成。证明 UU 是可逆的,并且满足 M=U(λ100λ2)U1M = U \begin{pmatrix} \lambda_1 & 0 \\ 0 & \lambda_2 \end{pmatrix} U^{-1}
  • ii) 证明集合 {U1xxZ}\{ U^{-1} x | x \in Z^* \}ZZ^* 是相等的。

(3) 对于每个陈述 A), B), 和 C),回答矩阵元素 a,b,da, b, d 的条件使该陈述成立。

  • A) 每个 yZy \in Z^* 都可以表示为 y=Mxy = Mx,其中 xZx \in Z^*
  • B) 没有 yZy \in Z^* 可以表示为 y=Mxy = Mx,其中 xZx \in Z^*
  • C) 至少有一个 yZy \in Z^* 可以表示为 y=(Mλ1I)xy = (M - \lambda_1 I)x,其中 xZx \in Z^*

题目描述

Z={(z1z2):z1,z2C, z12+z220},M=(abbd),I=(1001),Z^*=\left\{\binom{z_1}{z_2}:z_1,z_2\in\mathbb C,\ |z_1|^2+|z_2|^2\ne0\right\}, \quad M=\begin{pmatrix}a&b\\b&d\end{pmatrix}, \quad I=\begin{pmatrix}1&0\\0&1\end{pmatrix},

其中 MM2×22\times2 实对称矩阵。回答下列问题:

  1. MM 的全部特征值 λ1,λ2\lambda_1,\lambda_2
  2. 假设 λ1λ2\lambda_1\ne\lambda_2
    1. 取对应特征向量 v1,v2v_1,v_2 并令 U=(v1,v2)U=(v_1,v_2),证明 UU 可逆且
      M=U(λ100λ2)U1.M=U\begin{pmatrix}\lambda_1&0\\0&\lambda_2\end{pmatrix}U^{-1}.
    2. 证明 {U1xxZ}=Z\{U^{-1}x\mid x\in Z^*\}=Z^*
  3. 分别求矩阵元素 a,b,da,b,d 应满足的条件,使下列陈述成立:
    • 每个 yZy\in Z^* 均可写为 y=Mxy=Mx,其中 xZx\in Z^*
    • 不存在可写为 y=Mxy=MxxZx\in Z^*)的 yZy\in Z^*
    • 至少存在一个 yZy\in Z^* 可写为 y=(Mλ1I)xy=(M-\lambda_1I)x,其中 xZx\in Z^*

考点

  • 特征值与特征向量:由实对称二阶矩阵的特征方程求两个特征值,并利用不同特征值对应向量的线性无关性。
  • 矩阵对角化:以两组特征向量组成可逆矩阵 UU,证明相似对角分解及可逆线性变换对非零向量集的保持。
  • 核与像:把三个存在性陈述分别转化为 MMMλ1IM-\lambda_1I 的满射性、零映射性与像空间非平凡性,再落实为 a,b,da,b,d 的条件。

Kai

(1)

To find the eigenvalues of the matrix M=(abbd)M = \begin{pmatrix} a & b \\ b & d \end{pmatrix}, we solve the characteristic equation:

det(MλI)=0\det(M - \lambda I) = 0

The characteristic polynomial of MM is:

det(aλbbdλ)=(aλ)(dλ)b2=0\det \begin{pmatrix} a - \lambda & b \\ b & d - \lambda \end{pmatrix} = (a - \lambda)(d - \lambda) - b^2 = 0

This simplifies to:

λ2(a+d)λ+(adb2)=0\lambda^2 - (a + d)\lambda + (ad - b^2) = 0

The eigenvalues λ1\lambda_1 and λ2\lambda_2 are the roots of this quadratic equation:

λ1,2=(a+d)±(a+d)24(adb2)2\lambda_{1,2} = \frac{(a + d) \pm \sqrt{(a + d)^2 - 4(ad - b^2)}}{2}

(2)

i) Showing UU is Invertible

Let v1v_1 and v2v_2 be the eigenvectors ==corresponding== to λ1\lambda_1 and λ2\lambda_2, respectively. Define the matrix U=(v1,v2)U = (v_1, v_2). Since λ1λ2\lambda_1 \neq \lambda_2, the eigenvectors v1v_1 and v2v_2 are ==linearly independent==, and thus UU is invertible.

To show that M=U(λ100λ2)U1M = U \begin{pmatrix} \lambda_1 & 0 \\ 0 & \lambda_2 \end{pmatrix} U^{-1}, consider the action of MM on the eigenvectors:

Mv1=λ1v1andMv2=λ2v2Mv_1 = \lambda_1 v_1 \quad \text{and} \quad Mv_2 = \lambda_2 v_2

Therefore,

M(v1,v2)=(Mv1,Mv2)=(λ1v1,λ2v2)=(v1,v2)(λ100λ2)M(v_1, v_2) = (Mv_1, Mv_2) = (\lambda_1 v_1, \lambda_2 v_2) = (v_1, v_2) \begin{pmatrix} \lambda_1 & 0 \\ 0 & \lambda_2 \end{pmatrix}

Thus, we have:

M=U(λ100λ2)U1M = U \begin{pmatrix} \lambda_1 & 0 \\ 0 & \lambda_2 \end{pmatrix} U^{-1}

ii) Proving Set Equality

To prove that the set {U1xxZ}\left\{ U^{-1} x | x \in Z^* \right\} and ZZ^* are equal, consider any xZx \in Z^*. Then U1xZU^{-1}x \in Z^* if and only if z12+z220|z_1|^2 + |z_2|^2 \neq 0. Since UU is invertible and ZZ^* consists of all non-zero complex vectors, applying U1U^{-1} to any vector in ZZ^* yields another non-zero complex vector, ensuring the sets are equal.

(3)

A) For every yZy \in Z^* to be expressible as y=Mxy = Mx for some xZx \in Z^*, MM must be invertible. This requires λ10\lambda_1 \neq 0 and λ20\lambda_2 \neq 0, ensuring a0a \neq 0, d0d \neq 0, and adb20ad - b^2 \neq 0.

B) No yZy \in Z^* can be expressed as y=Mxy = Mx for some xZx \in Z^* if MM is singular and its image does not cover ZZ^*. This happens when MM has a zero eigenvalue, i.e., adb2=0ad - b^2 = 0 and one of the eigenvalues is zero.

C) At least one yZy \in Z^* can be expressed as y=(Mλ1I)xy = (M - \lambda_1 I)x for some xZx \in Z^* if a=da=d and b=0b=0 do not both hold true. This requires Mλ1IM - \lambda_1 I to be invertible or have a non-trivial image, which is true if λ1\lambda_1 is not an eigenvalue of MM, ensuring λ2λ1\lambda_2 \neq \lambda_1.

Knowledge

特征值和特征向量 矩阵分解

解题技巧和信息

  1. 特征值问题中,特征多项式是重要的工具,通过求解特征多项式可以得到特征值。
  2. 当矩阵的特征值不同时,其特征向量是线性无关的,这使得特征向量矩阵是可逆的。
  3. 在处理复杂矩阵时,注意到特征向量的规范性及其在不同基底下的表示。

重点词汇

eigenvalue 特征值

eigenvector 特征向量

invertible 可逆的

characteristic polynomial 特征多项式

quadratic equation 二次方程

参考资料

  1. 《线性代数及其应用》 第 5 章 特征值和特征向量