You are on page 1of 5

Math 206, Spring 2016

Assignment 5 Solutions

Due: February 26, 2016

Part A.
(1) Let

w1 =

1
1


,

w2 =

1
2


,

and

Prove that there cannot be T L(R2 , R3 ) that satisfies


1
3
T (w1 ) = 2 , T (w2 ) = 2 , and
3
5

w3 =

1
0


.

4
T (w3 ) = 1 .
7

[Hint: Find two distinct ways to express 0 R2 as a linear combination of w1 , w2 and w3 . If you
assume that T is linear, what do these two representations for 0 tell you about the value of T (0)?]
Solution. Suppose, for the sake of contradiction, that such a linear transformation T exists. Observe
that 0 = 0w1 + 0w2 + 0w3 , but also that
  

 
1
1
1
0=2

3
= 2w1 w2 3w3 .
1
2
0
Now the linearity of T tells us that, on the one hand,

3
4
0
1
T (0) = T (0w1 + 0w2 + 0w3 ) = 0 2 + 0 2 + 0 1 = 0 .
5
7
0
3
On the other hand, linearity of T gives

1
3
4
13
3 .
T (0) = T (2w1 w2 3w3 ) = 2 2 1 2 3 1 =
3
5
7
20
Hence our assumption leads to the conclusion that T (0) takes two separate values; but this contradicts
the fact that T is a function. So our initial assumption must be false, and we conclude that there is no
T L(R2 , R3 ) that satisfies the given properties.

(2) Suppose that T L(Rc , Rr ) is given. Prove that im(T ) satisfies the following properties:
(a) The vector 0 Rr is an element of im(T ).
Solution. There are a few ways one could approach this. Heres one: since T L(Rc , Rr ), there
exists some A Rrc so that for all x Rc we have T (x) = Ax. Hence
T (0) = A0 = 0Col1 (A) + 0Col2 (A) + + 0Colc (A) = 0.
Hence 0 Rr .
Heres a different proof. We know that for any n Z+ and any x Rn , we have 0x = 0. Hence
by linearity we have
T (0) = T (0x) = 0T (x) = 0
(where the initial 0 we plugged into T is an element of Rc , and the latter 0 is an element of Rr ).

http://palmer.wellesley.edu/~aschultz/w16/math206

Page 1 of 5

Math 206, Spring 2016

Assignment 5 Solutions

Due: February 26, 2016

(b) If v, w im(T ), then v + w im(T ).


Solution. Let v, w im(T ) be given. This means that there exist vectors x1 , x2 Rc so that
T (x1 ) = v and T (x2 ) = w. But then observe that
T (x1 + x2 ) = T (x1 ) + T (x2 )

(linearity of T )

= v + w.
Hence v + w im(T ), since it is the output of the vector x1 + x2 Rc .

(c) If v im(T ) and c R, then cv im(T ).


Solution. Let v im(T ) and c R be given. By the definition of image, this means that there
exists some x Rc so that T (x) = v. Note that we therefore have
T (cx) = cT (x)

(linearity of T )

= cv.
Hence cv im(T ), since it is the output associated to the vector cx Rc .

(3) For each of the following functions, determine (with proof) if the function is linear. If so, give the matrix
to which it corresponds.



x
3x + y
3
2

y
=
(a) The function T : R R defined by T
x + y 2z
z
Solution. Well check that T is linear.
First, let

x1
x = x2
x3

and

w1
w = w2
w3

be given. We then have

x1 + w 1
T (v + w) = T x2 + w2
x3 + w 3


3(x1 + w1 ) + (x2 + w2 )
=
(x1 + w1 ) + (x2 + w2 ) 2(x3 + w3 )

 

3x1 + x2
3w1 + w2
=
+
x1 + x2 2x3
w1 + w2 2w3

x1
w1
= T x2 + T w2
x3
w3

(definition of vector addition)

(definition of T )
(definition of vector addition)

(definition of T )

= T (x) + T (w).
This verifies the first property of linearity.
http://palmer.wellesley.edu/~aschultz/w16/math206

Page 2 of 5

Math 206, Spring 2016

Assignment 5 Solutions

Due: February 26, 2016

For the second, let

x1
x = x2
x3

and

cR

be given. Then we have

cx1
T (cx) = T cx2
cx3


3(cx1 ) + (cx2 )
=
(cx1 ) + (cx2 ) 2(cx3 )


3x1 + x2
=c
x1 + x2 2x3

(definition of scaling)

(definition of T )
(definition of scaling)

= cT (x).
This verifies the second property of linearity.
Now that we know T is linear, we can compute the matrix to which it corresponds. By a theorem
from class, we know this matrix is



3 1 0
T (e1 ) T (ec ) =
.
1 1 2

(b) The function T : R2 R3 defined by T

x
y

x
= 1
y

Solution. This is not linear. Note that



0
0
T
= 1 .
0
0
Of course we know that for 0 R2 we have 0 + 0 = 0, but observe that

0
T (0 + 0) = T (0) = 1
0
whereas


0
0
0
T (0) + T (0) = 1 + 1 = 2 .
0
0
0
Since T (0 + 0) 6= T (0) + T (0), we see that T fails to be linear.
Part B.

1
(a) Let v = 2 , and for a given w R3 define w = w projv w.
2

http://palmer.wellesley.edu/~aschultz/w16/math206

Page 3 of 5

Math 206, Spring 2016

Assignment 5 Solutions

Due: February 26, 2016

(i) Prove that w is orthogonal to v.


Solution. We will show that w v = 0. To help in this calculation, recall that projv w =
wv
vv v. Hence we get

wv 
v v
w v = w
vv
wv
=wv
vv
(linearity of dot product)
vv
=wvwv

(cancellation of v v)

= 0.

3

(ii) Prove that the function T : R R given by T (w) = w

is a linear transformation.

Solution. We will check the relevant axioms. First, suppose that w1 and w2 are given in
R3 . Then we have
T (w1 + w2 ) = (w1 + w2 ) projv (w1 + w2 )

(definition of T )

= (w1 + w2 ) projv w1 projv w2


= (w1 projv w1 ) + (w2 projv w2

(linearity of proj from class)


commutativity of vector addition)

= T (w1 ) + T (w2 ).
This verifies the first condition of linearity.
For the second, let w R3 and c R be given. Then we have
T (cw) = (cw) projv (cw)(definition of T )
= cw cprojv w

(linearity of proj from class)

= c(w projv w)

(scaling distributes across addition)

= cT (w)

(definition of T ).

This verifies the second condition of linearity.

(iii) Compute the matrix that corresponds to T .

T (e1 )

Solution. By a theorem from class, we know this matrix is


T (ec ) = e1 e1 v v e2 e2 v v e3 e3 v v =
vv

vv

vv

8/9 2/9
2/9 5/9
2/9 4/9

2/9
4/9 .
5/9


(iv) Is T injective?
Solution. Well compute the reduced row echelon form of the matrix associated to T :

1 0 1/2
0 1
1 .
0 0
0
(You can reach this form by performing the following sequence of operations: 89 1 , 29 1 +
2 , 92 1 + 3 , 22 , 12 2 + 3 , 41 2 + 1 .) Note that this matrix has rank 2. By a theorem
http://palmer.wellesley.edu/~aschultz/w16/math206

Page 4 of 5

Math 206, Spring 2016

Assignment 5 Solutions

Due: February 26, 2016

from class, since the rank is less than the number of columns, we have that T is not injective.

(b) Suppose that v span {w1 , w2 , , wk }. Prove that
span {w1 , w2 , , wk } = span {w1 , w2 , , wk , v} .
Solution. Well prove this result using a dual containment argument. First, we argue that
{w1 , w2 , , wk } span {w1 , w2 , , wk , v}. So let z {w1 , w2 , , wk } be given; our goal is
to show that z span {w1 , w2 , , wk , v}. By definition of span, our hypothesis tells us that z is
a linear combination of the vectors w1 , , wk . This means that there exist d1 , , dk R so that
z = d 1 w1 + + d k wk .
But observe that then we also have
z = d1 w1 + + dk wk + 0 = d1 w1 + + dk wk + 0v.
Hence we have exhibited z as a linear combination of the vectors w1 , , wk , v. Therefore z
span {w1 , , wk , v}. This verifies the first containment.
For the containment span {w1 , w2 , , wk , v} {w1 , w2 , , wk }, suppose that z span {w1 , , wk , v}.
This means that z is a linear combination of w1 , , wk , v, and hence (by definition of linear combination) there exist d1 , , dk , dk+1 R so that
z = d1 w1 + + dk wk + dk+1 v.
Furthermore, were told that v span {w1 , , wk }. Hence v is a linear combination of w1 , , wk ,
and so there are real numbers c1 , , ck satisfying
v = c1 w1 + + ck wk .
We plug this expression for v into our expression for z above to find
z = d1 w1 + + dk wk + dk+1 v
= d1 w1 + + dk wk + dk+1 (c1 w1 + + ck wk )
= (d1 + dk+1 c1 )w1 + + (dk + dk+1 ck )wk .
Now since each of the di and ci are real numbers, each coefficient di + dk+1 ci above is a real
number. So we have exhibited z as a linear combination of w1 , , wk . Therefore we have z
span {w1 , , wk }. This verifies the second containment.


http://palmer.wellesley.edu/~aschultz/w16/math206

Page 5 of 5

You might also like