2.4. GEOMETRICALLY DEFINED LINEAR TRANSFORMATIONS 57

Theorem 2.4.1 Let T be a linear transformation from Fn to Fm. Then the matrix A sat-isfying 2.23 is given by  | |

T (e1) · · · T (en)

| |

where Tei is the ith column of A.

Proof: It remains to verify uniqueness. However, if A is a matrix which works, A =(a1 · · · an

), then Tei ≡ Aei = ai and so the matrix is of the form claimed above. ■

Example 2.4.2 Determine the matrix for the transformation mapping R2 to R2 whichconsists of rotating every vector counter clockwise through an angle of θ.

Let e1 ≡

(1

0

)and e2 ≡

(0

1

). These identify the geometric vectors which point

along the positive x axis and positive y axis as shown.

e1

e2

From Theorem 2.4.1, you only need to find Te1 and Te2, the first being the first columnof the desired matrix A and the second being the second column. From drawing a pictureand doing a little geometry, you see that

Te1 =

(cos θ

sin θ

), Te2 =

(− sin θ

cos θ

).

Therefore, from Theorem 2.4.1,

A =

(cos θ − sin θ

sin θ cos θ

)Example 2.4.3 Find the matrix of the linear transformation which is obtained by firstrotating all vectors through an angle of ϕ and then through an angle θ. Thus you want thelinear transformation which rotates all angles through an angle of θ + ϕ.

Let Tθ+ϕ denote the linear transformation which rotates every vector through an angleof θ + ϕ. Then to get Tθ+ϕ, you could first do Tϕ and then do Tθ where Tϕ is the lineartransformation which rotates through an angle of ϕ and Tθ is the linear transformationwhich rotates through an angle of θ. Denoting the corresponding matrices by Aθ+ϕ, Aϕ,and Aθ, you must have for every x

Aθ+ϕx = Tθ+ϕx = TθTϕx = AθAϕx.

Consequently, you must have

Aθ+ϕ =

(cos (θ + ϕ) − sin (θ + ϕ)

sin (θ + ϕ) cos (θ + ϕ)

)= AθAϕ

=

(cos θ − sin θ

sin θ cos θ

)(cosϕ − sinϕ

sinϕ cosϕ

).

2.4. GEOMETRICALLY DEFINED LINEAR TRANSFORMATIONS 57Theorem 2.4.1 Let T be a linear transformation from F" to F™. Then the matrix A sat-isfying 2.28 is given bythwhere Te; is the i’" column of A.Proof: It remains to verify uniqueness. However, if A is a matrix which works, A =( al oct: an ) , then Te; = Ae; = a; and so the matrix is of the form claimed above.Example 2.4.2 Determine the matrix for the transformation mapping R? to R? whichconsists of rotating every vector counter clockwise through an angle of 0.1 0Let e, = ( 0 ) and eg = ( 1 ) . These identify the geometric vectors which pointalong the positive x axis and positive y axis as shown.e€2e1From Theorem 2.4.1, you only need to find Te; and Teg, the first being the first columnof the desired matrix A and the second being the second column. From drawing a pictureand doing a little geometry, you see thatTe; = cos 0 Tex = —sin@sin 0 cos 0Therefore, from Theorem 2.4.1,A= cos 6 —sindsin@ cos@Example 2.4.3 Find the matrix of the linear transformation which is obtained by firstrotating all vectors through an angle of ¢ and then through an angle 0. Thus you want thelinear transformation which rotates all angles through an angle of 0+ ¢.Let Tg+¢ denote the linear transformation which rotates every vector through an angleof 0+ ¢. Then to get 744, you could first do Ty and then do Tg where Ty is the lineartransformation which rotates through an angle of ¢ and Tg is the linear transformationwhich rotates through an angle of 6. Denoting the corresponding matrices by Agig, Ag,and Ag, you must have for every xAg+oX = T946X = ToT gx = Ag Agx.Consequently, you must have_ cos(9+¢) —sin(@+¢) \ _Aor = (orn? cos (0 + ¢) ) = oa,_ cos? —sindé cosd —sing7 sin@ cos sing cos¢