Homework 6

Solutions
 
a+b
 a 
1. Consider the plane P in R4 containing all vectors of the form 
 3b , as a and b range


b
 
0
0
over real numbers. Since this plane contains the origin (
0 comes from setting a = 0 and


0
b = 0), and we know that any plane containing the origin is a subspace, P is a subspace
of R4 . Find a basis for P.

Since a and b parameterize all of P,

we  just need to choose two independent
  values for
1 1
1
; taking a = 0, b = 1 we get v2 = 0. These
 
a,b. Taking a = 1, b = 0 we get v1 =  0 3
0 1
two vectors form a basis for P. To check, just take a general linear combination av1 + bv2
and observe that all vectors in P are of this form.
NOTICE: that if we focus only on the 2nd and 4th entries
 
a+b  
 a  a
 
 3b  b
b

we get a 2-dimensional vector space parameterized by the ‘free’ variables a, b. (Everything

else is determined by a and b.) The basis above for P comes from taking the standard basis
for this 2d space, then filling out the remaining entries (a + b and 3b) as appropriate:
   
  0   1
1  1  = v1 ,
  0  0  = v2 .
 
0 0 1 3
0 1
 
2
2. Let L ⊆ R2 be the line containing all vectors perpendicular to t =
3
a) Use the definition of a subspace to show that L is a subspace of R2 .
(Hint: you must show, for arbitrary vectors v and w and a scalar c, that if v is in L then
cv is in L; and that if v and w are in L then v + w are in L.)

If v · t = 0 then (c v) · t = c(v · t) = c(0) = 0 so c v ∈ L.

If v · t = 0 and w · t = 0 then (v + w) · t = v · t + w · t = 0 + 0 = 0 so v + w ∈ L.

1
 b) Find a basis for L.
 
3
Choose any nonzero vector perpendicular to t . v = will do. Since all other vectors
−2
perpendicular to t are multiples of v, we are done.
(To justify this process more carefully, do the following. We are looking for a vector v
in L. Suppose the first entry of v is nonzero. Then  by  scaling, we can set it to 1. The
1
second entry is determined by v · t = 0 ⇒ v = . We ask: are there any other
−2/3
vectors
 w in L not proportional to v? Suppose the first entry of w is a. Then w · t = 0 ⇒
a
w= = av. So, no, there no other vectors in L that are not contained in hvi. We
−2a/3
have L = hvi.)

3. Let M be the vector space of 2×2 matrices, and

 letW contain all matrices whose diagonal
a b
entries sum to zero, i.e. matrices of the form with a + d = 0. The matrices in W
c d
are called “traceless” matrices.
a) Find a basis for M.

Repeat the general idea behind to standard basis for R4 to get a basis with four elements
       
1 0 0 1 0 0 0 0
m1 = , m2 = , m3 = , m4 = .
0 0 0 0 1 0 0 1
 
a b
Notice that = a m1 + b m2 + c m3 + d m4 . They are independent because the only
c d
solution to a m1 + b m2 + c m3 + d m4 = 0 is a = b = c = d = 0.

b) Use the definition of subspace to show that W is a subspace of M.

 
a b
Let v = . If v ∈ W (meaning a + d = 0) then α a + α d = 0 for any α, so
 c d
αa αb
αv = ∈ W.
αc αd    0 0
a b a b
Similarly, let v = and w = 0 0 . If v ∈ W and w ∈ W (meaning a + d = 0
c d c d
a + a0 b + b 0
 
0 0 0 0
and a + d = 0) then (a + a ) + (d + d ) = 0, which implies v + w = ∈ W.
c + c0 d + d 0
c) What do you expect to be the dimension of W? (This should be the dimension of the
M, minus the number of conditions or relations that are obeyed in W.)

There is a single constraint. The expected dimension is 4-1=3.

2
 d) Find a basis for W.

Using the m’s above: notice that m2 and m3 are in W. Also n = m1 − m4 is in W. Since
all the m’s are independent, the three vectors m2 , m3 , n must also be independent, and its
not hard to see that they span W. So they form a basis. Explicitly:
     
0 1 0 0 1 0
m2 = , m3 = , n= .
0 0 1 0 0 −1
 
1
4. Let P ⊆ R3 be the set of vectors perpendicular to t = 1. Let Pxy by the xy plane, i.e.
  1
x
the set of vectors of the form y  for any real numbers x, y. Both P and Pxy are subspaces.
0
a) Find a basis for P.
   
1 1
Example: v1 = −1 and v2 =
   0 .
0 −1
b) Find a basis for Pxy .   
1 0
Standard basis: w1 = 0 and w2 = 1.
  
0 0
c) Find a basis for the intersection of P and Pxy : the set of vectors that are both in P
and Pxy .

We expect the intersection to be one-dimensional. v1 is in it, and in fact spans it, so v1 is

a basis. (Notice that v1 = w1 −w2 , making explicit the fact that v1 is in Pxy in addition to P.)

5. Strang Section 3.4, problem 41, reworded here:

Consider the six 3 × 3 permutation matrices:

       
1 0 0 0 1 0 1 0 0 0 0 1
I = 0 1 0 , P1 = 1 0 0 , P2 = 0 0 1 , P3 = 0 1 0 ,
0 0 1 0 0 1 0 1 0 1 0 0
   
0 1 0 0 0 1
P4 = 0 0 1 , P5 = 1 0 0 .
1 0 0 0 1 0
a) Show that P1 , P2 , P3 , P4 , P5 are independent (as vectors in the vector space of all 3 × 3
matrices). Do this by writing down the matrix c1 P1 + c2 P2 + c3 P3 + c4 P4 + c5 P5 and showing
that if this matrix is zero then necessarily c1 = c2 = ... = c5 = 0.

3
  
c2 c1 + c4 c3 + c5
We compute: c1 P1 + c2 P2 + c3 P3 + c4 P4 + c5 P5 = c1 + c5 c3 c2 + c4 . Ev-
c3 + c4 c2 + c5 c1
ery entry of this matrix must vanish (meaning: equals zero). That immediately implies
c1 = c3 = c2 = 0. Substituting back in then gives c4 = c5 = 0.

The matrices P1 , P2 , P3 , P4 , P5 form a basis for a vector space W, which contains 3 × 3

matrices such that the entries of each row and each column all sum to the same number.
b) Notice that I is in W. Find a linear combination of the P ’s that equals I.
   
c2 c1 + c4 c3 + c5 1 0 0
Use the same computation as above, and now solve c1 + c5
 c3 c2 + c4 = 0 1
  0.
c3 + c4 c2 + c5 c1 0 0 1
We get c1 = c2 = c3 = 1 and c4 = c5 = −1; in other words: P1 + P2 + P3 − P4 − P5 = I.

 
1 2 3
6. Let A = .
4 8 10
a) Find the reduced row-echelon form Rof A. 
1  0
Hint: in this case, it will look like R = .
0 0 1
Indicating the row-operation matrices above the arrows:

 1 0    1 3/2    1 0  
1 2 3 ( −4 1 ) 1 2 3
 
0 1 1 2 0 0 −1/2 1 2 0
−→ −→ −→ =R
4 8 10 0 0 −2 0 0 −2 0 0 1
 
x1
b) Use back-substitution to find all solutions x = x2  to A x = 0. You should find here

x3
that x2 is a free variable (because the second column of R does not contain a pivot). The re-
maining variables x1 and x3 , whose columns do contain pivots, are determined in terms of x2 .

There are equations: x1 + 2x2 = 0 and x3 = 0. The second equation (in an obvious way)
determines x3 = 0. We are missing an equation involving only x2 and x3 (but not x1 ), so x2
is a free variable. The first equation determines
 x1 in terms of x2 : x1 = −2x2 .
−2x2
Therefore: Every solution looks like x =  x2 , with x2 free.
0
c) Find a basis for the nullspace N (A).

We work exactly the same way we did in Problem 1, finding

 
−2
N (A) = hv1 i , with v1 =  1  .
0

4
  
1 2 3
7. Repeat the steps in Problem 6 for A = .
4 5 6

 1 0    1 2/3    1 0  
1 2 3 ( −4 1 ) 1 2
 
3 0 1 1 0 −1 0 −1/3 1 0 −1
−→ −→ −→ =R
4 5 6 0 −3 −6 0 −3 −6 0 1 2
 
x3
x3 is free (since the third column has no pivot), x2 = −2x3 , and x1 = x3 , so x = −2x3 .
  x3
1
We have N (A) = hv1 i with v1 = −2.
1
 
1 2 3
8. Repeat the steps in Problem 6 for A = .
2 4 6

 1 0 
1 2 3 ( −2 1 ) 1 2 3
 
−→ =R
2 4 6 0 0 0
The second and third columns are free, so x2 , 
x3 are free variables;
 the first equation (the
−2x2 − 3x3
only equation) sets x1 = −2x2 − 3x3 . Thus x =  x2 .
  x3  
−2 −3
We have N (A) = hv1 , v2 i with v1 =  1  and v2 =  0 .
0 1
 
2 −4
 6 −11
9. Repeat for A = 
1
.
2 
−1 1

 
1 0 0 0   1/2 0 0 0  
 −3 1 0 0 1 0 0 0!
      1 4 0 0!
 
2 −4  −1/2 0 1 0 2 −4 0 1 0 0 2 −4 0 1 0 0 2 0  0 1 0 0 1 0
 6 −11 0 1  0 −4 1 0 0 1  0 0 1 0 0 0 0 1 0
1/2 0 0 1 0 1 0 1 0 0 0 1 1 0−→ 0 0 1
0 1

1
 −→   −→   −→   
2  0 4  0 0  0 0 0 0
−1 1 0 −1 0 0 0 0 0 0
There are no free variables (all columns have a pivot). The third and four equations are
automatically
 satisfied (0=0). The second equation sets x2 = 0 and the first sets x1 = 0.
0
Thus x = .
0
The nullspace is the zero vector space, containing only the origin N (A) = O = h0i.
The basis is empty: N (A) has dimension zero, hence no basis elements.

5
  
2 −4
 6 −12
10. Repeat for A = 
 1 −2 .


−1 2

   1 0 0 0    1/2 0 0 0   
2 −4  −3 1 0 0  2 −4  0 1 0 0  1 −2
 6 −12 −1/2 0 1 0
1 0 0 1
0 0  00 00 10 01 0 0 

 1 −2 
 −→ 
0 0 
 −→  
0 0 
−1 2 0 0 0 0
 
2x2
Now x2 is free and x1 = 2x2 . Thus x = .
  x2
2
N (A) = hv1 i with v1 = .
1
       
1 −2 2 −1
−1 2 −2 1
11. Let v1 = 
 2 , v2 = −2, v3 =  6 , v4 = −6. Write down a matrix A as
      

1 1 5 −7
above whose rows are the v ’s. Find the rref R of A. What are the nonzero rows w1T , w2T , ...
T

of R? What’s the resulting basis for the span hv1 , v2 , v3 , v4 i ⊆ R4 ?

1 0 0 0!
 
   
v1T 1 −1 2 1 2 1 0 0 1 −1 2 1
v2T  −2 2 −2 1  −2 0 1 0 0 0
1 0 0 1 2 3
v3T  =  2 −2 6
Set A =     −→  
5 0 0 2 3
v4T −1 1 −6 −7 0 0 −4 −6

1 0 0 0!
  1 0 −1 0!
  1 0 0 0  
0 1 0 0 1 −1 2 1 0 1 0 0 1 −1 0 −2  0 1/2 0 0  1 −1 0 −2
0 −1 1 0 0 0 2 3 0 0 1 0 0 0 2 3  00 00 10 01 0 0 1 3/2
0 2 0 1 0 0 0 1
−→ 
0 0 0 0
 −→ 
0 0 0 0 
 −→ 0 0 0 0  = R
 

0 0 0 0 0 0 0 0 0 0 0 0
   
1 0
 −1 
 0 .
 
The nonzero rows of R contain w1T and w2T , where w1 =  0 and w 2 =  1 
−2 3/2
By the argument on the Homework, hv1 , v2 , v3 , v4 i = hw1 , w2 i; and since w1 and w2 are
independent (as any rows of the rref must be), they form the desired basis.
This also shows that the span of the v’s is two-dimensional.