Scribd
Upload
14 views5 pages

1 8 April 2 0 1 3 Linear ALG Multivariable Calc: T T X y T X T y X y T X T X X

Uploaded by

rodrchacaliaza
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
14 views5 pages

1 8 April 2 0 1 3 Linear ALG Multivariable Calc: T T X y T X T y X y T X T X X

Uploaded by

rodrchacaliaza
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

18 A P R I L 2 0 13

L I N E A R A L G & M U LT I VA R I A B L E C A L C §6
6.1 Review Linear Tr ansfor mation

Definition 1 (Linear Transformation). A linear transformation is a map


T ∶ Rn → Rm such that

• T (x + y) = T (x) + T ( y) for all x, y ∈ Rn (additivity)

• T (cx) = cT (x) for all x ∈ Rn and c ∈ R (homogeneity) •

How to describe a linear transformation? One difficulty is that the


domain Rn is infinite whenever n > 0 so we cannot write all of the
outputs T (x) for every x in Rn . Instead, we describe T in terms of a
basis for each of the domain and codomain, and this will be sufficient
to recover all possible outputs T (x) for all x in Rn . For now (until §21
on change of basis) we shall use the standard bases for Rn and Rm . Even
after learning about change of basis, the bases will be presumed to be the
standard bases when no bases are indicated.

Proposition 1. A linear transformation T ∶ Rn → Rm is equivalent to left


multiplication by the m × n matrix:

⎡⎢ ∣ ∣ ∣ ⎤⎥
A = ⎢⎢T (e 1 ) T (e 2 ) ⋯ T (e n )⎥⎥⎥
⎢ •
⎢⎣ ∣ ∣ ∣ ⎥⎦

Note 1. You can save yourself a lot of trouble down the road by recog-
nizing early the importance of the above proposition, and what exactly
it says. When asked for the matrix of the linear transformation T , the
desired matrix is A as above. •

Note 2. The proposition says that if x = x1 e 1 + ⋯ + x n e n , then:

∣ ∣ ∣ x1
T (x) = [ T(e 1 ) T(e 2 ) ⋯ T(e n ) ][ x⋮2 ] •
∣ ∣ ∣ xn

Example 1. What is the matrix of the linear transformation T defined


by T ([ xy ]) = [ −x
y ]? •
2

Solution. If unsure where to start, one thing that might help is to under-
stand the dimensions of the desired matrix. Since T maps R2 to R2 , the
size of the matrix of T is 2 × 2.
The first column of the matrix should be:

T (e 1 ) = T ([ 10 ]) = [ −1
0 ]

and the second column should be:

T (e 2 ) = T ([ 01 ]) = [ 01 ]

The matrix of T is thus:


−1 0
[ ] ∎
0 1

Example 2. What is the matrix of the linear transformation T defined


ax+b y
by T ([ xy ]) = [ cx+d y ] for constants a, b, c, d ∈ R? •

Solution. The first column of the matrix should be:


a(1)+b(0)
T (e 1 ) = T ([ 10 ]) = [ c(1)+d(0) ] = [ ac ]

and the second column should be:


a(0)+b(1)
T (e 2 ) = T ([ 01 ]) = [ c(0)+d(1) ] = [ db ]

The matrix of T is thus:


a b
[ ] ∎
c d

6.2 Examples of Linear Tr ansfor mations

Example 3 (Identity). The identity map T ∶ Rn → Rn (notice domain


and codomain are the same) given by T (x) = x for all x ∈ Rn is a linear
transformation. The matrix for T is the identity matrix:
⎡⎢1 0 ⋯ 0⎤⎥
⎢⎢ ⎥
⎢⎢0 1 ⋯ 0⎥⎥⎥ •
⎢⎢ ⋮ ⋮ ⋱ ⋮ ⎥⎥
⎢⎣0 0 ⋯ 1⎥⎦
3

Example 4 (Scaling Transformation). A scaling transformation is a lin-


ear transformation T ∶ Rn → Rn (notice domain and codomain are the
same) given by T (x) = αx for some scalar α ∈ R. (The identity map is
the special case α = 1.) The behavior of the above scaling transformation
depends on the size of ∣α∣:

• ∣α∣ > 1 stretch lengths

• ∣α∣ = 1 same lengths

• ∣α∣ < 1 contract lengths

and on the sign of α:

• α > 0 direction preserved—no reflection

• α = 0 (every vector goes to zero)

• α < 0 direction reversed—reflect through the origin •

Example 5 (Transformations with Diagonal Matrices). A diagonal ma-


trix is a square matrix of the form
⎡⎢d 0 ⋯ 0 ⎤⎥
⎢⎢ 1 ⎥
⎢⎢ 0 d2 ⋯ 0 ⎥⎥⎥
⎢⎢ ⋮ ⋮ ⋱ ⋮⎥
⎢⎣ 0 0 ⋯ d n ⎥⎥⎦

whose only nonzero entries are along the diagonal (among the entries
d1 , . . . , d n ). (The entries d1 , . . . , d n may be zero.) The corresponding
linear transformation has the effect of scaling and reflecting (when neg-
ative), but different axes need not share the same behavior as they did
for scaling transformations. The ith axis becomes scaled/rotated by a
factor of d i , but (except in cases where the diagonal entries are the same)
an arbitrary vector might do something more complicated than just
scale/rotate. •

Example 6 (13.7 in Levandosky). Assume T ∶ R2 → R2 is a linear transfor-


mation satisfying

T ([ 21 ]) = [ 21 ] and T ([ −1 1
2 ]) = [ −2 ]

(a) For which matrix A is T equivalent to multiplication by A?

(b) Describe T geometrically. •


4

Solution.
(a) The difficulty is that we do not know the value of T on e 1 and e 2 ,
although we do know the value of T on two linearly independent
vectors. We will see later (see §21) a systematic way to handle this
difficulty. For now, we seek expressions
[ 10 ] = a[ 21 ] + b[ −1
2 ]
[ 01 ] = c[ 21 ] + d[ −1
2 ]

because then we could use linearity of T to write


T ([ 10 ]) = aT ([ 21 ]) + bT ([ −1
2 ])
T ([ 01 ]) = cT ([ 21 ]) + d T ([ −1
2 ])

which by substitution of the given information becomes:


1 ] = [ 2a+b ]
T ([ 10 ]) = a[ 21 ] + b[ −2 a−2b
1 ] = [ 2c+d ]
T ([ 01 ]) = c[ 21 ] + d[ −2 c−2d

Then the desired matrix for T could be obtained as the juxtapo-


sition of the above two column vectors. Thus our goal is to find
the numbers a, b, c, d. This is just two separate systems in two
variables and two equations. (The first system is in a, b and the
second system is in c, d.) The solution to the systems is:
a = 2⁄5 , b = −1⁄5 , c = 1⁄5 , d = 2⁄5
We already did most of the work as motivation for the solution,
but instead of substituting in to find the answer, let’s do the work
again (pretending we hadn’t) just for clarity. We have determined
that:
[ 10 ] = 2⁄5 [ 21 ] − 1⁄5 [ −1
2 ]
[ 01 ] = 1⁄5 [ 21 ] + 2⁄5 [ −1
2 ]

so applying the linear transformation T gives


T ([ 10 ]) = 2⁄5 T ([ 21 ]) − 1⁄5 T ([ −1
2 ])
T ([ 01 ]) = 1⁄5 T ([ 21 ]) + 2⁄5 T ([ −1
2 ])

or
1 ] = [ 3⁄5 ]
T ([ 10 ]) = 2⁄5 [ 21 ] − 1⁄5 [ −2 4⁄5
1 ] = [ 4⁄5 ]
T ([ 01 ]) = 1⁄5 [ 21 ] + 2⁄5 [ −2 −3⁄5

so the matrix for T is:


[ 3⁄4⁄55 4⁄5
−3⁄5 ]
5

(b) It is difficult to describe T from its matrix with respect to the


standard bases, but the original description in the problem state-
ment is very helpful. The transformation T fixes the vector [ 21 ]
and negates (reflects through the origin) the vector [ −12 ]. The
transformation T is reflection over the line through the origin
with direction [ 21 ]. ∎

You might also like