Chapter 5 (Eigenvalues and Eigenvectors): Eigenvectors and linear transformations

本文为《Linear algebra and its applications》的读书笔记

The Matrix of a Linear Transformation

• Let $V$ be an $n$-dimensional vector space, let $W$ be an $m$-dimensional vector space, and let $T$ be any linear transformation from $V$ to $W$ . To associate a matrix with $T$ , choose (ordered) bases $\mathcal{B}$ and $\mathcal{C}$ for $V$ and $W$ , respectively. Given any $\mathbf{x}$ in $V$ , the coordinate vector $[\mathbf{x}{]}_{\mathcal{B}}$ is in ${\mathbb{R}}^n$ and the coordinate vector of its image, $[T(\mathbf{x}){]}_{\mathcal{C}}$, is in ${\mathbb{R}}^m$, as shown in Figure 1.
  
• Let { b 1 , . . . , b n } \{\boldsymbol b_1,..., \boldsymbol b_n\} {b1​,...,bn​} be the basis $\mathcal{B}$ for $V$. If $\mathbf{x}=r_1{\mathbf{b}}_1+...+r_n{\mathbf{b}}_n$, then
  and
  Since the coordinate mapping is linear, equation (1) leads toThe vector equation (2) can be written as a matrix equation, namely,where
  The matrix $M$ is a matrix representation of $T$, called the matrix for $\mathbf{T}$ relative to the bases $\mathcal{B}$ and $\mathcal{C}$. If $\mathcal{B}$ and $\mathcal{C}$ are bases for the same space $V$ and if $T$ is the identity transformation, then matrix $M$ is just a change-of-coordinates matrix.

Linear Transformations from $V$ into $V$

• In the common case where $W$ is the same as $V$ and the basis $\mathcal{C}$ is the same as $\mathcal{B}$, the matrix $M$ in (4) is called the matrix for $T$ relative to $\mathcal{B}$, or simply the $\mathcal{B}$-matrix for $T$, and is denoted by $[T{]}_{\mathcal{B}}$.
  

---

EXERCISES 7

Assume the mapping $T:{\mathbb{P}}^2\to {\mathbb{P}}^2$ defined by
is linear. Find the matrix representation of $T$ relative to the basis B = { 1 , t , t 2 } \mathcal B = \{1, t, t^2\} B={1,t,t2}.

SOLUTION

• A faster way is to realize that the information given provides the general form of $T(p)$ as shown in the figure below:
  The matrix that implements the multiplication along the bottom of the figure is easily filled
  in by inspection:
  

Since $[T{]}_{\mathcal{B}}$ can be represented by a matrix, $[T{]}_{\mathcal{B}}$ is linear. Thus $T$ is linear. ($[\mathbf{x}{]}_{\mathcal{B}}$ and $[\mathbf{x}{]}_{\mathcal{C}}^{-1}$ are all linear)

Linear Transformations on ${\mathbb{R}}^n$

• In an applied problem involving ${\mathbb{R}}^n$, a linear transformation $T$ usually appears first as a matrix transformation, $\mathbf{x}\mapsto A\mathbf{x}$. If $A$ is diagonalizable, then there is a basis $\mathcal{B}$ for ${\mathbb{R}}^n$ consisting of eigenvectors of $A$. Theorem 8 below shows that, in this case, the $\mathcal{B}$-matrix for $T$ is diagonal. Diagonalizing $A$ amounts to finding a diagonal matrix representation of $\mathbf{x}\mapsto A\mathbf{x}$.

• The mappings $\mathbf{x}\mapsto A\mathbf{x}$ and $\mathbf{u}\mapsto D\mathbf{u}$ describe the same linear transformation, relative to different bases.

Similarity of Matrix Representations

矩阵表示的相似性

• The proof of Theorem 8 did not use the information that $D$ was diagonal. Hence, if $A$ is similar to a matrix $C$, with $A=PC{P}^{-1}$, then $C$ is the $\mathcal{B}$-matrix for the transformation $\mathbf{x}\mapsto A\mathbf{x}$ when the basis $\mathcal{B}$ is formed from the columns of $P$.
  
• Conversely, if $T:{\mathbb{R}}^n\to {\mathbb{R}}^n$ is defined by $T(\mathbf{x})=A\mathbf{x}$, and if $\mathcal{B}$ is any basis for ${\mathbb{R}}^n$, then the $\mathcal{B}$-matrix for $T$ is similar to $A$. In fact, the calculations in the proof of Theorem 8 show that if $P$ is the matrix whose columns come from the vectors in $\mathcal{B}$, then $[T{]}_{\mathcal{B}}=P^{-1}AP$.Thus, the set of all matrices similar to a matrix $A$ coincides with the set of all matrix representations of the transformation $\mathbf{x}\mapsto A\mathbf{x}$.

---

Tip

• The $trace$ of a square matrix $A$ is the sum of the diagonal entries in $A$ and is denoted by $trA$. It can be verified that $tr(FG)=tr(GF)$ for any two $n\times n$ matrices $F$ and $G$. It can be shown that if $A$ and $B$ are similar, then $trA=trB$. If $B$ is diagonal, then the diagonal entries of $B$ must be the eigenvalues of $A$. So t r A = t r B = { s u m   o f   t h e   e i g e n v a l u e s   o f   A } tr A = tr B = \{sum\ of\ the\ eigenvalues\ of\ A\} trA=trB={sum of the eigenvalues of A}.
• In general, it can be shown that for any square matrix $A$, the trace of $A$ is the sum of the eigenvalues of $A$, counted according to multiplicities. You can use this fact to provide a quick check on your eigenvalue calculations. The sum of the eigenvalues must match the sum of the diagonal entries in $A$ (even if $A$ is not diagonalizable).