📚 Matrix Diagonalization: Conditions and Steps Explained | 矩阵对角化：条件与步骤详解
Matrix diagonalization is a powerful technique in linear algebra that simplifies many computations, especially when raising matrices to high powers or solving systems of differential equations. In this article, we will explore what diagonalization means, the conditions required for it to succeed, and the step-by-step procedure to perform it.
矩阵对角化是线性代数中一项强大的技术，能够极大地简化许多计算，尤其是矩阵的高次幂或微分方程组的求解。本文将探讨对角化的含义、所需的条件，以及逐步执行对角化的具体方法。
1. What is Diagonalization? | 什么是对角化？
A square matrix A of size n×n is said to be diagonalizable if there exists an invertible matrix P such that P⁻¹AP = D, where D is a diagonal matrix. In this case, A is similar to a diagonal matrix, and P is called the matrix of eigenvectors.
一个 n×n 的方阵 A 如果存在可逆矩阵 P，使得 P⁻¹AP = D（其中 D 为对角矩阵），则称 A 可对角化。此时称 A 与对角矩阵相似，P 称为特征向量矩阵。
The diagonal entries of D are the eigenvalues of A, and the columns of P are the corresponding eigenvectors. Once A is written as A = PDP⁻¹, many operations become trivial.
对角矩阵 D 的对角元素就是 A 的特征值，而 P 的列向量是对应的特征向量。一旦将 A 写成 A = PDP⁻¹，许多运算就变得十分简便。
2. Eigenvalues and Eigenvectors | 特征值与特征向量
An eigenvalue λ of a matrix A satisfies the equation Av = λv for some non-zero vector v, called an eigenvector. Equivalently, (A – λI)v = 0, so λ must satisfy the characteristic equation det(A – λI) = 0.
矩阵 A 的特征值 λ 满足方程 Av = λv，其中 v 是非零向量，称为特征向量。等价地，(A – λI)v = 0，因此 λ 必须满足特征方程 det(A – λI) = 0。
For an n×n matrix, the characteristic polynomial is of degree n, giving n eigenvalues counted with multiplicity. Each eigenvalue yields at least one eigenvector, but sometimes not enough to form a basis.
对于 n×n 矩阵，特征多项式是 n 次的，因此按重数计有 n 个特征值。每个特征值至少给出一个特征向量，但有时数量不足以构成整个空间的一组基。
3. The Diagonalization Condition | 可对角化的条件
The fundamental condition for diagonalization is that an n×n matrix A must have n linearly independent eigenvectors. If such a basis of eigenvectors exists, then the matrix formed by these vectors as columns is invertible, and diagonalization is possible.
对角化的基本条件是：一个 n×n 矩阵 A 必须具有 n 个线性无关的特征向量。如果存在这样的特征向量基，那么以这些向量为列构成的矩阵可逆，从而可以实现对角化。
Equivalently, A is diagonalizable if and only if the sum of the geometric multiplicities of all distinct eigenvalues equals n. Each eigenvalue’s geometric multiplicity is the dimension of its eigenspace, i.e., the number of free variables in (A – λI)v = 0.
等价地，A 可对角化当且仅当所有不同特征值的几何重数之和等于 n。每个特征值的几何重数就是其特征子空间的维数，也就是 (A – λI)v = 0 中自由变量的个数。
4. Algebraic and Geometric Multiplicity | 代数重数与几何重数
For each eigenvalue λ, the algebraic multiplicity is the number of times λ appears as a root of the characteristic polynomial. The geometric multiplicity is the dimension of the eigenspace corresponding to λ.
对于每个特征值 λ，代数重数是指 λ 作为特征多项式根的出现次数；几何重数则是 λ 对应特征子空间的维数。
In general, the geometric multiplicity is always less than or equal to the algebraic multiplicity. The matrix A is diagonalizable if and only if the geometric multiplicity equals the algebraic multiplicity for every distinct eigenvalue.
一般地，几何重数总是小于或等于代数重数。矩阵 A 可对角化当且仅当对每个不同的特征值，几何重数都等于代数重数。
For example, if a 3×3 matrix has eigenvalues 2, 2, 5, then for diagonalization we need two linearly independent eigenvectors for λ=2 and one for λ=5.
例如，如果一个 3×3 矩阵的特征值为 2, 2, 5，那么要对角化，就需要 λ=2 对应两个线性无关的特征向量，λ=5 对应一个特征向量。
5. Step-by-Step Procedure | 逐步步骤
The diagonalization process can be summarized in a clear algorithm:
对角化过程可以总结为以下清晰的算法：
- Step 1: Compute the characteristic polynomial det(A – λI) and solve for eigenvalues λ.
- Step 2: For each eigenvalue, solve the homogeneous system (A – λI)v = 0 to find a basis for the eigenspace.
- Step 3: Combine all eigenvectors obtained. If the total number of linearly independent eigenvectors is n, proceed; otherwise, A is not diagonalizable.
- Step 4: Form the matrix P whose columns are the n independent eigenvectors.
- Step 5: Form the diagonal matrix D with corresponding eigenvalues on the diagonal in the same order as the columns of P.
- Step 6: Verify by checking that A = PDP⁻¹, or equivalently AP = PD.
步骤 1：计算特征多项式 det(A – λI) 并解出特征值 λ。
步骤 2：对每个特征值，解齐次方程组 (A – λI)v = 0，求特征子空间的一组基。
步骤 3：合并所有得到特征向量。如果线性无关特征向量的总数为 n，则继续；否则 A 不可对角化。
步骤 4：以这 n 个无关特征向量为列构造矩阵 P。
步骤 5：按照 P 的列顺序，将对应特征值放在对角线上，构造对角矩阵 D。
步骤 6：通过验证 A = PDP⁻¹ 或 AP = PD 来检查结果。
6. Worked Example: 2×2 Matrix | 2×2矩阵示例
Consider the matrix A = | 4 1 |. Let us diagonalize it.
考虑矩阵 A = [4 1; 2 3]。下面我们对它进行对角化。
A = [ 4 1 ; 2 3 ]
First, find the characteristic polynomial:
首先求特征多项式：
det(A – λI) = (4-λ)(3-λ) – 2 = λ² – 7λ + 10 = 0
Solve: λ = 5 and λ = 2. For λ = 5, solve (A – 5I)v = 0:
解得 λ = 5 和 λ = 2。对于 λ = 5，解 (A – 5I)v = 0：
[-1 1 ; 2 -2] → eigenvector v₁ = [1; 1]
For λ = 2, solve (A – 2I)v = 0:
对于 λ = 2，解 (A – 2I)v = 0：
[2 1 ; 2 1] → eigenvector v₂ = [1; -2]
Thus P = [ 1 1 ; 1 -2 ] and D = [ 5 0 ; 0 2 ]. One can check that AP = PD.
因此 P = [1 1; 1 -2]，D = [5 0; 0 2]。可以验证 AP = PD。
7. Worked Example: 3×3 Matrix | 3×3矩阵示例
Now consider a 3×3 matrix with repeated eigenvalues to see how multiplicity matters.
现在考虑一个具有重复特征值的 3×3 矩阵，以观察重数的影响。
Let B = | 3 1 0 ; 0 3 0 ; 0 0 5 |. Its eigenvalues are λ = 5 and λ = 3 (with algebraic multiplicity 2).
设 B = [3 1 0; 0 3 0; 0 0 5]。其特征值为 λ = 5 和 λ = 3（代数重数为 2）。
For λ = 3, solve (B – 3I)v = 0:
对于 λ = 3，解 (B – 3I)v = 0：
[0 1 0 ; 0 0 0 ; 0 0 2] → equation v₂ = 0 and 2v₃ = 0
This leaves v₁ free, so the eigenspace has dimension 1. We find only one eigenvector for λ = 3, but we need two. Therefore B is not diagonalizable.
这里只有 v₁ 自由，因此特征子空间维数为 1。对于 λ = 3 我们只找到一个特征向量，但需要两个，所以 B 不可对角化。
In contrast, if we modify B to C = | 3 0 0 ; 0 3 0 ; 0 0 5 |, each repeated eigenvalue gives two independent eigenvectors, so C is diagonalizable.
相比之下，如果将 B 改为 C = [3 0 0; 0 3 0; 0 0 5]，每个重复特征值都能给出两个无关特征向量，因此 C 可对角化。
8. Matrix Powers via Diagonalization | 用对角化计算矩阵幂
One of the most important applications is computing A^k. If A = PDP⁻¹, then A² = PD²P⁻¹, and in general A^k = PD^kP⁻¹. Since D is diagonal, D^k is obtained by raising each diagonal entry to the k-th power.
最重要的应用之一就是计算 A^k。如果 A = PDP⁻¹，那么 A² = PD²P⁻¹，一般地 A^k = PD^kP⁻¹。因为 D 是对角矩阵，D^k 只需将对角线每个元素取 k 次幂即可。
This turns a complicated matrix multiplication into three simpler steps: multiply P, D^k, and P⁻¹. For large k, this is dramatically more efficient than repeated multiplication.
这将复杂的矩阵乘法转化为三步简单运算：乘以 P、D^k 和 P⁻¹。对于很大的 k，其效率远高于反复做矩阵乘法。
For example, if A is the 2×2 matrix above, then for any positive integer k:
例如，如果 A 是上面的 2×2 矩阵，则对任意正整数 k：
A^k = [ 1 1 ; 1 -2 ] [ 5^k 0 ; 0 2^k ] [ 1 1 ; 1 -2 ]⁻¹
9. Symmetric Matrices and Orthogonal Diagonalization | 对称矩阵与正交对角化
A real symmetric matrix (Aᵀ = A) is always diagonalizable. Moreover, its eigenvectors corresponding to distinct eigenvalues are orthogonal. In fact, every symmetric matrix can be diagonalized by an orthogonal matrix P, meaning P⁻¹ = Pᵀ.
实对称矩阵（满足 Aᵀ = A）总是可对角化的。此外，其不同特征值对应的特征向量相互正交。事实上，每个对称矩阵都可以被正交矩阵 P 对角化，即 P⁻¹ = Pᵀ。
This special form is called orthogonal diagonalization. It is widely used in data science (PCA), physics (principal axes), and engineering because it preserves lengths and angles.
这种特殊形式称为正交对角化。它在数据科学（PCA）、物理学（主轴）和工程学中广泛使用，因为它保持长度和角度不变。
To achieve orthogonal diagonalization, normalize each eigenvector to unit length and place them as columns of P. Then P is orthogonal and D is diagonal.
实现正交对角化时，将每个特征向量标准化为单位向量，并把它们作为列构成 P。此时 P 是正交矩阵，D 是对角矩阵。
10. When Diagonalization Fails | 不能对角化的情况
A matrix fails to be diagonalizable when it does not have enough linearly independent eigenvectors. This usually happens when some eigenvalue has geometric multiplicity strictly less than its algebraic multiplicity.
当矩阵没有足够多的线性无关特征向量时，它就不能对角化。这通常发生在某个特征值的几何重数严格小于其代数重数时。
Classic examples include matrices with a Jordan block, such as [ λ 1 ; 0 λ ]. For this matrix, λ is the only eigenvalue and there is only one eigenvector, so it cannot be diagonalized.
典型例子包括具有 Jordan 块的矩阵，例如 [λ 1; 0 λ]。对于这个矩阵，λ 是唯一特征值且只有一个特征向量，因此它不能被对角化。
However, every matrix can be put into Jordan normal form, which is a block diagonal matrix that is “almost” diagonal. Studying this form extends the idea of simplification to non-diagonalizable cases.
然而，每个矩阵都可以化为 Jordan 标准形，这是一种“几乎”对角的块对角矩阵。研究这种形式将化简思想推广到了不可对角化的情况。
11. Applications | 应用
Diagonalization appears in many areas. In solving linear differential equations, if a system can be written as x’ = Ax, then diagonalizing A decouples the system into independent scalar equations.
对角化在多个领域中出现。在求解线性微分方程组时，如果系统可以写成 x’ = Ax，那么对角化 A 就能将方程组解耦为独立的标量方程。
In Markov chains, the long-term behavior of a transition matrix can be studied through its eigenvalues. Diagonalization allows us to compute high powers of the transition matrix quickly to find steady-state distributions.
在马尔可夫链中，转移矩阵的长期行为可以通过其特征值来研究。对角化允许我们快速计算转移矩阵的高次幂，从而找到稳态分布。
In machine learning, algorithms such as Principal Component Analysis (PCA) rely on the diagonalization of a covariance matrix. The eigenvectors define the principal directions, and their corresponding eigenvalues give the variance along each direction.
在机器学习中，主成分分析（PCA）等算法依赖于协方差矩阵的对角化。特征向量定义了主要方向，而对应的特征值给出了沿每个方向的方差。
Finally, diagonalization is one of the key ideas that leads to the spectral theorem, which describes how operators on finite-dimensional spaces can be represented by diagonal matrices under appropriate conditions. Mastering it deepens your understanding of linear algebra and equips you for advanced topics.
最后，对角化是通向谱定理的关键思想之一，谱定理描述了有限维空间上的算子如何在适当条件下用对角矩阵表示。掌握对角化能加深你对线性代数的理解，为高阶主题做好准备。
Published by TutorHao | Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导