211 Homework7 Solutions

MATH 211 HOMEWORK ASSIGNMENT 7 SOLUTIONS
FALL 2006, WILLIAMS COLLEGE
Abstract. These are the instructor’s solutions.
1. Nonsingularity
Show that the composition of two nonsingular linear transformations is also
nonsingular.
1.1. Solution. A linear transformation is nonsingular when its nullity is zero, that
is, when its kernel is the trivial subspace {0}. (Note that this is equivalent to
injectivity!) Let T : U → V and S :→ W be the linear transformations in question.
Suppose that u ∈ U lies in the kernel of S ◦ T . We must show that u = 0. Since
u ∈ ker(S ◦ T ) we see that, as elements of W ,
0 = (S ◦ T )(u) = S(T (u)).
Thus, T (u) lies in the kernel of S. Since S is nonsingular, we deduce that T (u) =
0 ∈ V . This means that u lies in the kernel of T . Since T is nonsingular, we see
that u = 0.
Therefore, the null space of S ◦ T consists of only the vector u = 0, and hence
S ◦ T is nonsingular.
2. Rank-Nullity
Find bases of the kernel and image and verify the Rank-Nullity theorem for the
linear transformation TA : R5 → R4 associated to the matrix
 
2 1 1 2 −2
1 0 3 5 1
A=  6 2 1 0 −9 .

−3 3 −1 7 8
2.1. Solution. We put A into reduced row echelon form:
 
1 0 0 −1 −2
0 1 0 2 1
A∼ 0 0

1 2 1
0 0 0 0 0
So the kernel is
   

 1 2 


−2 −1 
   
ker(TA ) = null(A) = span  −2  , −1 ,


  
 1   0  

 

 
0 1
1
2 FALL 2006, WILLIAMS COLLEGE
and the image is

     

 2 1 1 
      
1 0
, , 3  .
im(TA ) = col(A) = span      

 6 2 1 
 
−3 3 −1
We see by construction that these sets are bases for the relevant spaces, so
rank(TA ) + nullity(TA ) = 3 + 2 = 5 = dim(R5 )
which verifies the result of the rank-nullity theorem in this case.
3. Hyperplanes
Is it possible to find a family of 4 hyperplanes in R4 so that any subset of three
hyperplanes must intersect in exactly a line, but that the common intersection of
all four hyperplanes is empty? If so, give a concrete example. If not, explain why
it is not possible in terms of some of the tools we have developed.
3.1. Solution. This is possible, but it can be tricky to see how at first. Suppose
that the hyperplanes are defined by normal vectors n1 , n2 , n3 , n4 ∈ R4 . (We’ll view
these as column vectors.) That is, the ith hyperplane is the set of solutions to an
equation hni , xi = bi .
Take for example the first three hyperplanes. Their
 tcommon
 intersection is the
n1
solution set to a matrix equation Ax = b where A = nt2 . To have exactly a line’s
nt3
worth of solutions, we must have nullity(A) = 1. So by the rank-nullity theorem,
rank(A) = 4 − nullity(A) = 3. Since the matrix A has three rows, we see that these
rows are linearly independent.
This is true for any choice of three hyperplanes. But to get no solution for the
intersection of four planes, we have to have
 t  t 
n1 n1 b1
nt2  nt2 b2 
rank    
nt3  < rank nt3 b3  .
nt4 nt4 b4
Now the matrix on the right has to have rank at least 3 (since it contains our
A above as a submatrix). And if it were 4, then it would be impossible to have
the inequality, since each matrix has 4 rows. So, we must arrange things so that
the four vectors are linearly dependent, but any subset of three of them is linearly
independent.
I’ll use n1 = e1 , n2 = e2 , n3 = e3 , n4 = e1 + e2 + e3 . All that is left is to choose
the bi ’s. This is not so tricky now that we understand what we’re up against. I will
use
H1 = {x ∈ R4 | x1 = 1}
H2 = {x ∈ R4 | x2 = 1}
H3 = {x ∈ R4 | x3 = 1}
H4 = {x ∈ R4 | x1 + x2 + x3 = 6}.
It is not difficult to verify that this set of planes has the required properties.
HOMEWORK 7 3
4. An interesting new space

Let hom(R3 , R4 ) be the collection of all linear mappings from R3 into R4 .
• What should addition and scalar multiplication be in this space?
• Show that hom(R3 , R4 ) is a vector space under these operations.
• Find a basis for hom(R3 , R4 ) and use it to compute the dimension of this
space.
• Describe two different, more easily understood spaces which are isomorphic
to hom(R3 , R4 ). Write down explicit isomorphisms between hom(R3 , R4 )
and the two spaces.
4.1. Solution. Let S and T be elements of hom(R3 , R4 ), and α a real number (that
is, a scalar). We define the sum S + T to be the function
(S + T )(x) = S(x) + T (x),
where we take the sum on the right in R4 , and we define the scalar multiple αT to
be the function
(αT )(x) = αT (x),
where the scalar multiplication on the right is happening in R4 .

We must also check that S + T and αS are elements of hom(R3 , R4 ), that is that
they are linear transformations. For any vectors v, w ∈ R3 and scalars a, b ∈ R we
have
(S + T )(av + bw) = S(av + bw) + T (av + bw) = aS(v) + bS(w) + aT (v) + bT (w)
= a(S(v) + T (v)) + b(S(w) + T (w)) = a(S + T )(v) + b(S + T )(w)
and
(αS)(av + bw) = α(S(av + bw)) = α(aS(v) + bS(w))
= a(αS(v)) + b(αS(w)) = a(αS)(v) + b(αS)(w),
hence both these maps are linear transformations.

The fact that this is a vector space follows just like in the problem from the
last assignment. You have a set of functions which have a target equal to a vector
space, so all of the vector space rules get inherited by the set of functions. It is a
bit tedious to type one more time, but you should write this part out for practice.
Recall that there are eight rules to check. Just for clarity in what we do below,
note that the zero element, Z, of hom(R3 , R4 ) is the function defined by
Z(x) = 0 for all x ∈ R3 .
To find a basis for our vector space, we first try to find a spanning set. Recall
that we have canonical bases {e1 , e2 , e3 } for R3 and {e01 , e02 , e03 , e04 } for R4 . I’m using
primes for the second set to avoid confusion with the first set, there is nothing new
going on.
We know that every linear transformation is determined by what it does to a
basis of the domain. So each linear transformation is determined by what it does
to the set {e1 , e2 , e3 }. That is, a linear transformation T : R3 → R4 is defined by
the three vectors T (e1 ), T (e2 ), T (e3 ). Alternately, we can think of it as spanned
4 FALL 2006, WILLIAMS COLLEGE
by the 12 constants which appear in the vectors when expanded like below:
T (e1 ) = a11 e01 + a21 e20 + a31 e03 + a41 e04
(1) T (e2 ) = a12 e01 + a22 e20 + a32 e03 + a42 e04
T (e3 ) = a13 e01 + a23 e20 + a33 e03 + a43 e04 .
Thus, if we let Tij be the linear mapping which carries ej to e0i and the other ek ’s
to zero, we see that
hom(R3 , R4 ) = span{Tij | i = 1, 2, 3, j = 1, 2, 3, 4}.
Now suppose that we have a linear combination of our spanning set which represents
the zero vector Z:
X X
(2) aij Tij = Z.
i=1,2,3,4 j=1,2,3
If we evaluate this equation at the vector e1 we see that

X X
0 =Z(e1 ) = aij Tij (e1 )
i=1,2,3,4 j=1,2,3
= a11 e01 + a21 e02 + a31 e03 + a41 e04 + 0

Since the set {e01 , e02 , e03 , e04 } is linearly independent, we deduce that a11 = a12 =
a13 = a14 = 0. Similarly, by evaluating equation (2) at e2 and e3 , we see that all
of the other coefficients are also zero. Thus, hom(R3 , R4 ) has the set of Tij ’s as a
basis, and therefore has dimension 12.
Since hom(R3 , R4 ) a 12 dimensional vector space over the real numbers, we know
that it is isomorphic to R12 using the coordinate representation map with respect
to the basis {T11 , T12 , . . . , T42 , T43 }. Explicitly, this is the map which sends a linear
¡ ¢t
transformation T to the vector a11 a12 . . . a42 a43 , where the coefficients
are chosen like in equations (1), above.
The other representation is also hinted at by the equations (1). The space
hom(R3 , R4 ) is isomorphic to the vector space
M43 (R) = {real 4 × 3 matrices}.
The isomorphism is given by the linear mapping
 
a11 a12 a13
a21 a22 a23 
M : T 7→ M (T ) = a31 a32
.
a33 
a41 a42 a43
That this defines a linear map which is injective and surjective is a simple direct
check using the theorem about describing a linear transformation by its action on
a basis. (Only this time M is the transformation, and Tij is the basis.)

211 Homework7 Solutions

Uploaded by

Copyright:

Available Formats

211 Homework7 Solutions

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

211 Homework7 Solutions

Uploaded by

Copyright:

Available Formats

MATH 211 HOMEWORK ASSIGNMENT 7 SOLUTIONS

FALL 2006, WILLIAMS COLLEGE

Abstract. These are the instructor’s solutions.

and the image is

4. An interesting new space

(S + T )(x) = S(x) + T (x),

where the scalar multiplication on the right is happening in R4 .

hence both these maps are linear transformations.

If we evaluate this equation at the vector e1 we see that

= a11 e01 + a21 e02 + a31 e03 + a41 e04 + 0

You might also like