Definitions and Basic Properties

Throughout this chapter, the scalar field $ {\mathbb{F}}$ is either always the set $ {\mathbb{R}}$ or always the set $ {\mathbb{C}}.$

DEFINITION 4.1.1 (Linear Transformation)   Let $ V$ and $ W$ be vector spaces over $ {\mathbb{F}}.$ A map $ T: V {\longrightarrow}W$ is called a linear transformation if

$\displaystyle T(\alpha {\mathbf u}+ \beta {\mathbf v}) = \alpha T({\mathbf u}) ...
...ha, \beta \in {\mathbb
{F}}, \; \; {\mbox{and }} {\mathbf u},{\mathbf v}\in V.$

We now give a few examples of linear transformations.

EXAMPLE 4.1.2  
  1. Define $ T: {\mathbb{R}}{\longrightarrow}{\mathbb{R}}^2$ by $ \; T(x) = (x, 3x)$ for all $ x \in {\mathbb{R}}.$ Then $ T$ is a linear transformation as

    $\displaystyle T(x + y) = (x+y, 3(x+y)) =
(x,3x) + (y,3y) = T(x) + T(y).$

  2. Verify that the maps given below from $ {\mathbb{R}}^n$ to $ {\mathbb{R}}$ are linear transformations. Let $ {\mathbf x}= (x_1, x_2, \ldots, x_n).$
    1. Define $ \; T( {\mathbf x}) = \sum\limits_{i=1}^n x_i.$
    2. For any $ i, \; \; 1 \leq i \leq n,$ define $ \; T_i( {\mathbf x}) = x_i.$
    3. For a fixed vector $ {\mathbf a}= (a_1, a_2, \ldots, a_n) \in {\mathbb{R}}^n,$ define $ \; T( {\mathbf x}) = \sum\limits_{i=1}^n a_i x_i.$ Note that examples $ (a)$ and $ (b)$ can be obtained by assigning particular values for the vector $ {\mathbf a}.$
  3. Define $ T: {\mathbb{R}}^2 {\longrightarrow}{\mathbb{R}}^3$ by $ \;
T((x,y)) = (x+y, 2 x - y, x + 3 y).$
    Then $ T$ is a linear transformation with $ T((1,0)) = (1,2,1)$ and $ T((0,1)) =
(1,-1,3).$
  4. Let $ A$ be an $ m \times n$ real matrix. Define a map $ T_A: {\mathbb{R}}^n {\longrightarrow}{\mathbb{R}}^m$ by

    $\displaystyle T_A({\mathbf x}) = A
{\mathbf x}\;\; {\mbox{ for every }} \;\; {\mathbf x}^{t} = (x_1, x_2, \ldots, x_n)
\in {\mathbb{R}}^n.$

    Then $ T_A$ is a linear transformation. That is, every $ m \times n$ real matrix defines a linear transformation from $ {\mathbb{R}}^n$ to $ {\mathbb{R}}^m.$
  5. Recall that $ {\cal P}_n({\mathbb{R}})$ is the set of all polynomials of degree less than or equal to $ n$ with real coefficients. Define $ \; T: {\mathbb{R}}^{n+1} {\longrightarrow}
{\cal P}_n ({\mathbb{R}})$ by

    $\displaystyle T( (a_1, a_2, \ldots, a_{n+1}) ) = a_1 + a_2 x + \cdots + a_{n+1} x^n$

    for $ \; (a_1, a_2, \ldots, a_{n+1}) \in {\mathbb{R}}^{n+1}.$ Then $ T$ is a linear transformation.

PROPOSITION 4.1.3   Let $ T: V {\longrightarrow}W$ be a linear transformation. Suppose that $ {\mathbf 0}_V $ is the zero vector in $ V$ and $ {\mathbf 0}_W$ is the zero vector of $ W.$ Then $ T({\mathbf 0}_V)
= {\mathbf 0}_W.$

Proof. Since $ {\mathbf 0}_V = {\mathbf 0}_V + {\mathbf 0}_V,$ we have

$\displaystyle T({\mathbf 0}_V) = T({\mathbf 0}_V + {\mathbf 0}_V)=
T({\mathbf 0}_V) + T({\mathbf 0}_V).$

So, $ T({\mathbf 0}_V) = {\mathbf 0}_W$ as $ T({\mathbf 0}_V) \in W.$ height6pt width 6pt depth 0pt

From now on, we write $ {\mathbf 0}$ for both the zero vector of the domain space and the zero vector of the range space.

DEFINITION 4.1.4 (Zero Transformation)   Let $ V$ be a vector space and let $ T: V {\longrightarrow}W$ be the map defined by

$\displaystyle T({\mathbf v}) = {\mathbf 0}\; {\mbox{ for every }} \; {\mathbf v}\in V.$

Then $ T$ is a linear transformation. Such a linear transformation is called the zero transformation and is denoted by $ {\mathbf 0}.$

DEFINITION 4.1.5 (Identity Transformation)   Let $ V$ be a vector space and let $ T: V {\longrightarrow}V$ be the map defined by

$\displaystyle T({\mathbf v}) = {\mathbf v}\; {\mbox{ for every }} \; {\mathbf v}\in V.$

Then $ T$ is a linear transformation. Such a linear transformation is called the Identity transformation and is denoted by $ I.$

We now prove a result that relates a linear transformation $ T$ with its value on a basis of the domain space.

THEOREM 4.1.6   Let $ T: V {\longrightarrow}W$ be a linear transformation and $ {\cal B}= ( {\mathbf u}_1, {\mathbf u}_2, \ldots, {\mathbf u}_n )$ be an ordered basis of $ V.$ Then the linear transformation $ T$ is a linear combination of the vectors $ T({\mathbf u}_1), T({\mathbf u}_2), \ldots, T({\mathbf u}_n).$

In other words, $ T$ is determined by $ T({\mathbf u}_1), T({\mathbf u}_2), \ldots, T({\mathbf u}_n).$

Proof. Since $ {\cal B}$ is a basis of $ V,$ for any $ {\mathbf x}\in V,$ there exist scalars $ \alpha_1, {\alpha}_2, \ldots, \alpha_n$ such that $ {\mathbf x}= \alpha_1 {\mathbf u}_1 + \alpha_2 {\mathbf u}_2 + \cdots + \alpha_n {\mathbf u}_n.$ So, by the definition of a linear transformation

$\displaystyle T({\mathbf x}) = T(\alpha_1 {\mathbf u}_1 + \cdots + \alpha_n {\mathbf u}_n) = \alpha_1 T({\mathbf u}_1) + \cdots
+ \alpha_n T({\mathbf u}_n).$

Observe that, given $ {\mathbf x}\in V,$ we know the scalars $ {\alpha}_1, {\alpha}_2, \ldots, {\alpha}_n.$ Therefore, to know $ T({\mathbf x}),$ we just need to know the vectors $ T({\mathbf u}_1), T({\mathbf u}_2), \ldots, T({\mathbf u}_n)$ in $ W.$

That is, for every $ {\mathbf x}\in V,$ $ T({\mathbf x})$ is determined by the coordinates $ (\alpha_1, \alpha_2, \ldots, \alpha_n)$ of $ {\mathbf x}$ with respect to the ordered basis $ {\cal B}$ and the vectors $ T({\mathbf u}_1), T({\mathbf u}_2), \ldots, T({\mathbf u}_n) \in W.$ height6pt width 6pt depth 0pt

EXERCISE 4.1.7  
  1. Which of the following are linear transformations $ T : V {\longrightarrow}W?$ Justify your answers.
    1. Let $ V = {\mathbb{R}}^2$ and $ W = {\mathbb{R}}^3$ with $ T\bigl( \; (x,y) \; \bigr) = (x + y + 1, 2 x - y, x + 3 y)$
    2. Let $ V = W = {\mathbb{R}}^2$ with $ T \bigl( \;(x,y) \; \bigr) = (x-y, x^2 - y^2)$
    3. Let $ V = W = {\mathbb{R}}^2$ with $ T \bigl( \;(x,y) \; \bigr) = (x-y, \vert x\vert)$
    4. Let $ V = {\mathbb{R}}^2$ and $ W = {\longrightarrow}{\mathbb{R}}^4$ with $ T\bigl( \;(x,y) \; \bigr)= (x + y, x-y, 2x + y, 3x - 4 y)$
    5. Let $ V = W = {\mathbb{R}}^4$ with $ T\bigl( \;(x,y,z,w)\; \bigr)= (z,x,w,y)$
  2. Recall that $ M_2({\mathbb{R}})$ is the space of all $ 2 \times 2$ matrices with real entries. Then, which of the following are linear transformations $ T : M_2({\mathbb{R}}) {\longrightarrow}M_2({\mathbb{R}})?$
    $ (a) \;\; T(A) = A^t \hspace{1in} (b) \;\; T(A) = I + A \hspace{1in} (c) \;\; T(A) = A^2$
    $ (d) \;\; T(A) = B A B^{-1},$ where $ B$ is some fixed $ 2 \times 2$ matrix.
  3. Let $ T: {\mathbb{R}} \longrightarrow {\mathbb{R}}$ be a map. Then $ T$ is a linear transformation if and only if there exists a unique $ c \in {\mathbb{R}}$ such that $ \; T({\mathbf x}) = c {\mathbf x}$ for every $ {\mathbf x}\in {\mathbb{R}}.$
  4. Let $ A$ be an $ n \times n$ real matrix. Consider the linear transformation

    $\displaystyle T_A({\mathbf x}) = A {\mathbf x}\;\; {\mbox{ for every }} \; {\mathbf x}\in {\mathbb{R}}^n.$

    Then prove that $ T^2({\mathbf x}) := T(T({\mathbf x}))
= A^2 {\mathbf x}.$ In general, for $ k \in {\mathbb{N}},$ prove that $ T^k ({\mathbf x}) = A^k {\mathbf x}.$
  5. Use the ideas of matrices to give examples of linear transformations $ T,S : {\mathbb{R}}^3 {\longrightarrow}{\mathbb{R}}^3$ that satisfy:
    1. $ T \neq {\mathbf 0}, \;\; T^2 \neq {\mathbf 0}, \;\; T^3 = {\mathbf 0}.$
    2. $ T \neq {\mathbf 0}, \;\; S \neq {\mathbf 0}, \;\; S\circ T \neq {\mathbf 0}, \;\; T \circ S = {\mathbf 0};$ where $ T\circ S({\mathbf x}) = T\bigl(S({\mathbf x})\bigr).$
    3. $ S^2 = T^2, \;\; S \neq T.$
    4. $ T^2 = I, \;\; T \neq I.$
  6. Let $ T : {\mathbb{R}}^n \longrightarrow {\mathbb{R}}^n$ be a linear transformation such that $ T \neq {\mathbf 0}$ and $ T^2 = {\mathbf 0}.$ Let $ {\mathbf x}\in {\mathbb{R}}^n$ such that $ T ({\mathbf x}) \neq {\mathbf 0}.$ Then prove that the set $ \{{\mathbf x}, T({\mathbf x})\}$ is linearly independent. In general, if $ T^k \neq
{\mathbf 0}$ for $ 1 \leq k \leq p$ and $ T^{p+1} = {\mathbf 0},$ then for any vector $ {\mathbf x}\in {\mathbb{R}}^n$ with $ T^p ({\mathbf x}) \neq {\mathbf 0}$ prove that the set $ \{{\mathbf x},
T({\mathbf x}), \ldots, T^p ({\mathbf x}) \}$ is linearly independent.
  7. Let $ T : {\mathbb{R}}^n \longrightarrow {\mathbb{R}}^m$ be a linear transformation, and let $ {\mathbf x}_0 \in {\mathbb{R}}^n$ with $ T({\mathbf x}_0) =
y.$ Consider the sets

    $\displaystyle S = \{ {\mathbf x}\in {\mathbb{R}}^n : T({\mathbf x}) = {\mathbf ...
... }} \;\; N = \{ {\mathbf x}\in {\mathbb{R}}^n : T({\mathbf x}) = {\mathbf 0}\}.$

    Show that for every $ {\mathbf x}\in S$ there exists $ {\mathbf z}\in N$ such that $ {\mathbf x}=
{\mathbf x}_0 + z.$
  8. Define a map $ T: {\mathbb{C}} \longrightarrow
{\mathbb{C}}$ by $ T(z) = \overline{z},$ the complex conjugate of $ z.$ Is $ T$ linear on
    (a) $ \;$ $ {\mathbb{C}}$ over $ {\cal R}\;\; $ (b) $ \;$ $ {\mathbb{C}}$ over $ {\mathbb{C}}.$
  9. Find all functions $ f : {\mathbb{R}}^2 \longrightarrow {\mathbb{R}}^2$ that satisfy the conditions
    1. $ f(\;(x,x)\;) = (x,x)$ and
    2. $ f(\;(x,y)\;) = (y,x)$ for all $ (x,y) \in {\mathbb{R}}^2.$
    That is, $ f$ fixes the line $ y = x$ and sends the point $ (x_1, y_1)$ for $ x_1 \neq y_1$ to its mirror image along the line $ y = x.$

    Is this function a linear transformation? Justify your answer.

THEOREM 4.1.8   Let $ T: V {\longrightarrow}W$ be a linear transformation. For $ {\mathbf w}\in W,$ define the set

$\displaystyle T^{-1}({\mathbf w}) = \{{\mathbf v}\in V : T({\mathbf v}) = {\mathbf w}\}.$

Suppose that the map $ T$ is one-one and onto.
  1. Then for each $ {\mathbf w}\in W,$ the set $ T^{-1}({\mathbf w})$ is a set consisting of a single element.
  2. The map $ T^{-1} : W {\longrightarrow}V$ defined by

    $\displaystyle T^{-1}({\mathbf w}) = {\mathbf v}\; {\mbox{ whenever }} \; T({\mathbf v}) = {\mathbf w}.$

    is a linear transformation.

Proof. Since $ T$ is onto, for each $ {\mathbf w}\in W$ there exists a vector $ {\mathbf v}\in V$ such that $ T({\mathbf v}) = {\mathbf w}.$ So, the set $ T^{-1}({\mathbf w})$ is non-empty.

Suppose there exist vectors $ {\mathbf v}_1, {\mathbf v}_2 \in V$ such that $ T({\mathbf v}_1) =
T({\mathbf v}_2).$ But by assumption, $ T$ is one-one and therefore $ {\mathbf v}_1 = {\mathbf v}_2.$ This completes the proof of Part $ 1.$

We now show that $ T^{-1}$ as defined above is a linear transformation. Let $ {\mathbf w}_1, {\mathbf w}_2 \in W.$ Then by Part $ 1,$ there exist unique vectors $ {\mathbf v}_1, {\mathbf v}_2 \in V$ such that $ T^{-1}({\mathbf w}_1) = {\mathbf v}_1$ and $ T^{-1}({\mathbf w}_2) = {\mathbf v}_2.$ Or equivalently, $ T({\mathbf v}_1) = {\mathbf w}_1$ and $ T({\mathbf v}_2) = {\mathbf w}_2.$ So, for any $ {\alpha}_1, {\alpha}_2 \in {\mathbb{F}},$ we have $ T({\alpha}_1{\mathbf v}_1 + {\alpha}_2{\mathbf v}_2) = {\alpha}_1 {\mathbf w}_1 + {\alpha}_2 {\mathbf w}_2.$

Thus for any $ {\alpha}_1, {\alpha}_2 \in {\mathbb{F}},$

$\displaystyle T^{-1}({\alpha}_1{\mathbf w}_1 + {\alpha}_2{\mathbf w}_2) = {\alp...
...bf v}_2 = {\alpha}_1 T^{-1}({\mathbf w}_1)
+ {\alpha}_2 T^{-1}({\mathbf w}_2).$

Hence $ T^{-1}:W {\longrightarrow}V,$ defined as above, is a linear transformation. height6pt width 6pt depth 0pt

DEFINITION 4.1.9 (Inverse Linear Transformation)   Let $ T: V {\longrightarrow}W$ be a linear transformation. If the map $ T$ is one-one and onto, then the map $ T^{-1} : W {\longrightarrow}V$ defined by

$\displaystyle T^{-1}({\mathbf w}) = {\mathbf v}\;\; {\mbox{ whenever }} T({\mathbf v}) = {\mathbf w}$

is called the inverse of the linear transformation $ T.$

EXAMPLE 4.1.10  
  1. Define $ T: {\mathbb{R}}^2 {\longrightarrow}{\mathbb{R}}^2$ by $ T((x,y)) = (x+y, x-y).$ Then $ T^{-1}: {\mathbb{R}}^2 {\longrightarrow}{\mathbb{R}}^2$ is defined by

    $\displaystyle T^{-1}((x,y)) =
(\displaystyle\frac{x+y}{2}, \displaystyle\frac{x-y}{2}).$

    Note that
    $\displaystyle T\circ T^{-1} ((x,y))$ $\displaystyle =$ $\displaystyle T(T^{-1}((x,y))) = T( (\frac{x+y}{2},
\frac{x-y}{2}) )$  
      $\displaystyle =$ $\displaystyle (\frac{x+y}{2}+
\frac{x-y}{2}, \frac{x+y}{2}-
\frac{x-y}{2})$  
      $\displaystyle =$ $\displaystyle (x,y).$  

    Hence, $ T\circ T^{-1} = I,$ the identity transformation. Verify that $ T^{-1} \circ T = I.$ Thus, the map $ T^{-1}$ is indeed the inverse of the linear transformation $ T.$
  2. Recall the vector space $ {\cal P}_n({\mathbb{R}})$ and the linear transformation $ \; T: {\mathbb{R}}^{n+1} {\longrightarrow}
{\cal P}_n ({\mathbb{R}})$ defined by

    $\displaystyle T( (a_1, a_2, \ldots, a_{n+1}) ) = a_1 + a_2 x + \cdots + a_{n+1} x^n$

    for $ (a_1, a_2, \ldots, a_{n+1}) \in {\mathbb{R}}^{n+1}.$ Then $ \; T^{-1}: {\cal P}_n({\mathbb{R}}) {\longrightarrow}{\mathbb{R}}^{n+1}$ is defined as

    $\displaystyle T^{-1}( a_1 + a_2 x + \cdots + a_{n+1} x^n) = (a_1, a_2, \ldots, a_{n+1})$

    for $ a_1+ a_2 x + \cdots + a_{n+1} x^n \in {\cal P}_n({\mathbb{R}}).$ Verify that $ T\circ T^{-1} = T^{-1} \circ T = I.$ Hence, conclude that the map $ T^{-1}$ is indeed the inverse of the linear transformation $ T.$

A K Lal 2007-09-12