1.8 Introduction to Linear Transformations

TipSummary of the Chapter
  • A transformation \(T\) (not necessarily a linear transformation) that maps a vector \(\mathbf{x}\in\mathbb{R}^n\) to a vector \(T(\mathbf{x})\in\mathbb{R}^m\) is denoted as \(T: \mathbb{R}^n\to\mathbb{R}^m\). The \(\mathbb{R}^n\) is called the domain, and the \(\mathbb{R}^m\) is called the codomain.
    Note that although the range of \(T\), \(\text{Range}(T)=\{T(\mathbf{x})\mid\mathbf{x}\in\mathbb{R}^n\}\), may not cover the entire codomain, the dimensionality is the same: \(\text{Range}(T)\subseteq \mathbb{R}^m\).
  • A matrix transformation is denoted as \(T(\mathbf{x})=A\mathbf{x}\) or \(\mathbf{x}\mapsto A\mathbf{x}\). Suppose \(A\) is an \(m\times n\) matrix, then the domain is \(\mathbb{R}^n\), and the codomain is \(\mathbb{R}^m\).
  • A linear transformation, \(T\), satisfies the following conditions for all \(\mathbf{u}\), \(\mathbf{v}\), and \(c\):
    1. \(T(\mathbf{u}+\mathbf{v})=T(\mathbf{u})+T(\mathbf{v})\);
    2. \(T(c\mathbf{u})=c T(\mathbf{u})\).
  • The above two conditions are equivalent to \(T(c\mathbf{u}+d\mathbf{v})=c T(\mathbf{u})+d T(\mathbf{v})\).
  • If \(T: \mathbb{R}^n \to \mathbb{R}^m\) is a linear transformation, \(T(\mathbf{0})=\mathbf{0}\), but the zero vector on the left hand side is in \(\mathbb{R}^n\), while the one on the right hand side is in \(\mathbb{R}^m\). This characteristic can be proved by substituting \(\mathbf{u}\) and \(\mathbf{v}\) with zero vectors and using either of the two conditions.
TipA Simple Interaction
Enter a \(3 \times 2\) matrix:

Exercise 1 (Example 1.c) Let \(A=\begin{bmatrix}1 & -3 \\ 3 & 5 \\ -1 & 7\end{bmatrix}, \mathbf{b}=\begin{bmatrix}3 \\ 2 \\ -5\end{bmatrix}\), and define a transformation \(T: \mathbb{R}^2 \to \mathbb{R}^3\) by \(T(\mathbf{x})=A\mathbf{x}\). Is there more than one \(\mathbf{x}\) whose image under \(T\) is \(\mathbf{b}\)?

NoteSolution (Solution to Exercise 1)

Solution (Solution to Exercise 1). \[\begin{bmatrix}1 & -3 & 3 \\ 3 & 5 & 2 \\ -1 & 7 & -5\end{bmatrix} \sim \begin{bmatrix}1 & -3 & 3 \\ 0 & 1 & -.5 \\ 0 & 0 & 0\end{bmatrix}\]

Since there are no free variables (the last column is augmented), there is exactly one \(x\) such that \(T(\mathbf{x})=\mathbf{b}\).

TipRemark

Remark. Having a row of zeros does not imply that the equation has multiple solutions. It all depends on free variables—namely, the coefficient columns.

Exercise 2 (T19) Let \(\mathbf{e}_1=\begin{bmatrix}1\\ 0\end{bmatrix}\), \(\mathbf{e}_2=\begin{bmatrix}0\\ 1\end{bmatrix}\), \(\mathbf{y}_1=\begin{bmatrix}2\\ 5\end{bmatrix}\), and \(\mathbf{y}_2=\begin{bmatrix}-1\\ 6\end{bmatrix}\), and let \(T: \mathbb{R}^2 \to \mathbb{R}^2\) be a linear transformation that maps \(\mathbf{e}_1\) into \(\mathbf{y}_1\) and maps \(\mathbf{e}_2\) into \(\mathbf{y}_2\). Find the images of \(\begin{bmatrix}5\\ -3\end{bmatrix}\) and \(\begin{bmatrix}x_1\\ x_2\end{bmatrix}\).

NoteSolution (Solution to Exercise 2)

Solution (Solution to Exercise 2). Though it’s possible to solve the transformation matrix because all are in \(\mathbb{R}^2\), the forms of \(\mathbf{e}_1\) and \(\mathbf{e}_2\) make us easy to represent any \(\mathbb{R}^2\) vectors using \(\mathbf{e}_1\) and \(\mathbf{e}_2\).

\[\begin{bmatrix}x_1\\ x_2\end{bmatrix}=x_1\mathbf{e}_1+x_2\mathbf{e}_2\]

Thus,

\[\begin{align*} T(\begin{bmatrix}x_1\\ x_2\end{bmatrix}) &= T(x_1\mathbf{e}_1+x_2\mathbf{e}_2) \\ &= x_1 T(\mathbf{e}_1) + x_2 T(\mathbf{e}_2) \\ &= x_1\mathbf{y}_1 + x_2\mathbf{y}_2 \end{align*}\]

A more general form will be discussed in the next chapter.

Exercise 3 (T25) Given \(\mathbf{v}\neq\mathbf{0}\) and \(\mathbf{p}\) in \(\mathbb{R}^n\), the line through \(\mathbf{p}\) in the direction of \(\mathbf{v}\) has the parametric equation \(\mathbf{x} = \mathbf{p} + t\mathbf{v}\). Show that a linear transformation \(T: \mathbb{R}^n \to \mathbb{R}^n\) maps this line onto another line or onto a single point (a degenerate line).

NoteSolution (Solution to Exercise 3)

Solution (Solution to Exercise 3). Two points to notice at first:

  1. This is mentioned not because it could cause misunderstanding, but because it clarifies the meaning of vectors. The textbook is rigorously phrasing the problem as “the line through \(\mathbf{p}\) in the direction of \(\mathbf{v}\)”;
  2. The equation is a parametric equation, so all vectors and lines in this problem are in \(\mathbb{R}^n\). Do not confuse this with \(y=a+bx\), though the line equation can also be written as \(\begin{bmatrix}x\\ y\end{bmatrix}=\begin{bmatrix}0\\ a\end{bmatrix} + x\begin{bmatrix}1\\ b\end{bmatrix}\).

A paraphrase of the problem would be “does applying a linear transformation to this line always yield another line, or possibly a single point?”

\[T(\mathbf{x})=T(\mathbf{p}+t\mathbf{v})\] According to the definition of linear transformations, \[T(\mathbf{p}+t\mathbf{v})=T(\mathbf{p})+t T(\mathbf{v}).\]

Since \(T:\mathbb{R}^n\to\mathbb{R}^n\) maps a vector in \(\mathbb{R}^n\) to another vector in \(\mathbb{R}^n\), \(T(\mathbf{p})\) and \(T(\mathbf{v})\) are both vectors in \(\mathbb{R}^n\). Thus, the new line can be described as a line through \(T(\mathbf{p})\) in the direction of \(T(\mathbf{v})\).

When \(T(\mathbf{v})=\mathbf{0}\), the line will be mapped onto the single point at \(T(\mathbf{p})\).

TipRemark

Remark. We now know that applying a linear transformation on a line generates a new line or a point (a degenerate line). Quite linear, right?

TipIs every linear transformation a matrix transformation?

We know that every matrix transformation is a linear transformation since \(A(\mathbf{u}+\mathbf{v})=A\mathbf{u}+A\mathbf{v}\) and \(A(c\mathbf{u})=c(A\mathbf{u})\).

If we are working in finite-dimensional vector spacese with chosen bases (e.g. \(\mathbb{R}^n\to\mathbb{R}^m\)), then yes — every linear transformation is a matrix transformation. The matrix depends on the choice of basis. (We will also learn this in the next chapter.)
If we are working in infinite-dimensional vector spaces (like function spaces), then not every linear transformation corresponds to a finite matrix. For example, the derivative operator \(D: C^\infty (\mathbb{R})\to C^\infty(\mathbb{R}),D(f)=f'\), is linear but cannot be represented by a finite matrix.
— ChatGPT