MTH 102: Linear Algebra

Department of Mathematics and Statistics Indian Institute of Technology - Kanpur

Problem Set 4

Problems marked (T) are for discussions in Tutorial sessions.

1. Determine whether the following sets of vectors are linearly independent or not

(a) {(1, 0, 0), (1, 1, 0), (1, 1, 1)} of R3

 
1 1 1
Solution: Yes. Look at the null space, N (A) of A =  0 1 1 . One can show that
0 0 1
N (A) = {0} by computing the RREF.
(b) {(1, 0, 0, 0), (1, 1, 0, 0), (1, 2, 0, 0), (1, 1, 1, 1)} of R4
   
1 1 1 1 −1
 0 1 2 1   2 
Solution: No. The null space N (A) of A =   0
 contains 
 −1
.
0 0 1  
0 0 0 1 0
(c) {(1, 0, 2, 1), (1, 3, 2, 1), (4, 1, 2, 2)} in R4 .
Solution: Yes. Similar to (a).

2. Find a maximal linearly independent subset of

               
 1 −1 0 1 0 2 2 −1 
 2   0  2 −1 0 1 3 −2

 

S = −1,  1 , 2, −1, 0, 1, 0, −1 .
               
   1  2 −3 2 1 1  0 
 0

 
 

1 2 1 1 1 0 1 1

Find another. And another. Do they have the same cardinality?

Solution: Compute the RREF. Verify that RREF([S]) = [I5 , u1 , u2 , u3 ]. Since all the entries
of ui , for i = 1, 2, 3, are non-zero and hence any five vectors form a maximal independent set.
 11 5 1 
1 0 0 0 0 − 21 7 − 21
 0 1 0 0 0 − 11 − 6 6 
7 7 7 
3

RREF([S]) =   0 0 1 0 0 2 1 −1 

20 3 2
7 − 21
 0 0 0 1 0 
21
17 4 3
0 0 0 0 1 14 7 7

3. Give 2 bases for the trace 0 real symmetric matrices of size 3 × 3. Extend these bases to bases
of the real matrices of size 3 × 3.

Solution: {e12 + e21 , e13 + e31 , e32 + e23 , e11 − e22 , e11 − e33 },
{e12 +e21 +e13 +e31 , e13 +e31 −(e32 +e23 ), e12 +e21 +2(e13 +e31 )−3(e32 +e23 ), e11 −e22 , e11 −e33 }.
Just add the vectors {e11 , e12 , e13 , e23 }

4. Consider W = {v ∈ R6 : v1 + v2 + v3 = 0, v2 + v3 + v4 = 0, v5 + v6 = 0}. Supply a basis for W

and extend it to a basis of R6 .
 2

   
1 0 0 0
5. Let M be the vector space of all 2 × 2 matrices and let A = ,B = .
0 0 0 −1

(a) Give a basis of M .

       
1 0 0 1 0 0 0 0
Solution: One basis would be , , , .
0 0 0 0 1 0 0 1
(b) Describe a subspace of M which contains A and does not contain B.
Solution: The subspace consisting of all multiples of A is a subspace which contains A
but not B.
(c) Prove that if a subspace of M contains A and B, it must contain the identity matrix.
Solution: True : If a subspace contains A and B then it also contains A − B = I.

6. [T] Let {w1 , w2 , . . . , wn } be a basis of the finite dimensional vector space V. Let v 6= 0 be any
vector in V. Show that there exists wi such that if we replace wi by v then we still have a basis.
n n
P 1 1 P
Solution: Since v 6= 0, let v = αi wi with α1 6= 0. So, w1 = α1 v − α1 αi wi . Thus
i=1 i=2
n
P
{v, w2 , . . . , wn } also spans V. Now, β1 v+β2 w2 +· · ·+βn wn = 0 ⇒ β1 α1 w1 + (β1 αi +βi )wi = 0
i=2
As {w1 , . . . , wn } is a basis, ⇒ β1 α1 = 0 ⇒ β1 = 0 (α1 6= 0) ⇒ βi = 0, i ≥ 2.

7. Show that {u, v} is linearly independent if and only if {u + v, u − v} is linearly independent.

8. (T) Show that {u1 , . . . , uk } ⊂ Rn is linearly independent if and only if {Au1 , . . . , Auk } is linearly
independent for any invertible matrix A ∈ Mn (R), i.e., suppose we have an n×n invertible matrix
A and consider the map f : Rn → Rn defined by f (x) = Ax. Then, ‘{u1 , . . . , uk } is linearly
independent if and only if the set consisting of their images is also linearly independent’.
P
Solution: SupposeP {u1 , . P . . , un } is linearly dependent. Then ∃αi ’s, not all 0 s.t. αi ui = 0.
So 0 = A0 = A αi ui = αi (Aui ). Hence, {Au1 , . . . , Aun } is linearly dependent.
P
Now, supposeP that {Au1 , . P . . , Aun } is linearly dependent. P Then ∃αi ’s not all 0 s.t. αi (Aui ) =
αi (Aui ) = A αi ui . Hence, A−1 0 = A−1 A αi ui =
P
0. So 0 = αi ui . Thus, {u1 , . . . , un }
is linearly dependent.
 k k

P P
9. Show that {u1 , . . . , uk } ⊂ V is linearly independent if and only if ai1 ui , . . . , aik ui is
i=1 i=1
linearly independent for any invertible matrix A ∈ Mk (R). This means: In LS(u1 , . . . , uk ) the
k
P
set {u1 , . . . , uk } is a basis if and only if the vectors wj = aij ui (which are nothing but some
i=1
linear combinations of ui ’s given by the matrix A) is a basis.

w1 a11 · · · ak1 u1 u1
      
k
P ..  .. .. .. T ..
Solution: Put wr = air ui . Then  . = . . . =A
 . .
i=1
wr a1k ··· akk uk uk
 3

w1
 
..
Let {w1 , . . . , wk } be linearly dependent. Then ∃αi ’s, not all 0 s.t. [α1 · · · αk ] . = 0. So
wk

a11 · · · ak1 u1 u1
    

0 = [α1 · · · αk ] ... .. ..

. . = [β1 ··· βk ] ...,
a1k ··· akk uk uk

where [α1 · · · αk ]AT = [β1 ··· βk ] 6= 0. Thus {u1 , . . . , uk } is linearly dependent.

Converse: Similar.

10. (T) If {v1 , v2 , . . . , vd } is a basis for a vector space V, then show that any set of n vectors in V
with n > d, say {w1 , w2 , . . . , wn }, is linearly dependent.

Solution: Since {v1 , v2 , . . . , vd } is a basis for V and wj ∈ V for j = 1, . . . , n, there exist

constants aij , 1 ≤ i ≤ d, 1 ≤ j ≤ n such that
d
X
wj = aij vi .
i=1

n
P
Consider a linear combination of wj ’s that equals zero, that is, cj wj = 0. Then,
j=1
 
n n d d n
!
X X X X X
cj wj = 0 ⇐⇒ cj aij vi = 0 ⇐⇒  aij cj  vi = 0.
j=1 j=1 i=1 i=1 j=1

As vi ’s are linearly independent, we have Ac = 0 where the matrix A is a d by n matrix and c

is a column vector of size n with [A]ij = aij . As A is a rectangular matrix with more columns
n
P
than rows, its null space is non-trivial. We therefore have non-zero cj ’s with cj wj = 0. Thus,
j=1
vectors {w1 , w2 , . . . , wn } is linearly dependent.

11. Suppose V is a vector space of dimension d. Let S = {w1 , w2 , . . . , wn } be a set of vectors from
V. Then show that S does not span V if n < d.

Solution: Let B be a basis of V . Since dim(V) = d, the definition implies that B is a linearly
independent set of d vectors that spans V.
Now, suppose on the contrary that S does span V. Then B is a larger set of vectors which is
linearly independent. This contradicts the result in the previous problem.

12. (T) Let T = {1, x2 − x + 5, 4x3 − x2 + 5x, 3x4 + 2}. Is LS(T ) = R[x; 4]?

Solution: The vector space R[x; 4] has dimension 5. Since T contains only 4 vectors, T does
not span R[x; 4].

13. Let W be a proper subspace of a finite dimensional vector space V.

 4

(a) Show that there is a subspace U of V such that W ∩ U = {0} and U + W = V.

Solution: Extend the basis of W to a basis of V and define U to be the span of new basis
elements.
(b) Show that there is no subspace U such that U ∩ W = {0} and dim U + dim W > dim V.
Solution: Follows from dim(U + W) = dim(U) + dim(W) − dim(U ∩ W) (just ask the
students to assume this result) and the fact that U + W is a subspace of V.
14. (T) Describe all possible ways in which two planes (passing through origin) in R3 could intersect.

Solution: Let U and V be planes. Then, dim(U + V) = dim(U) + dim(V) − dim(U ∩ V) implies
that dim(U + V) = 4 − dim(U ∩ V). Clearly, 2 ≤ dim(U + V) ≤ 3. If dim(U + V) = 2, then
dim(U ∩ V) = 2 which implies U + V = U = V = U ∩ V, i.e., U = V. If dim(U + V) = 3, then
dim(U ∩ V) = 1 which implies that U and V intersect on a line.
15. Construct a matrix with the required property or explain why this is impossible:
   
1 0    
1 2
(a) Column space contains 1, 0, row space contains , .
2 5
0 1
 
1 0
Solution: 1 0.
0 1
     
 1   3  3
(b) Column space has basis 1 , null-space has basis 1 . What if 1 belongs to
3 1 1
   
the null space (but not necessarily forms a basis)?
Solution: Not possible; dimension of the column  space andthe dimension of the null-space
1 1 −4
must add to 3. For the second part, take A = 1 1 −4  .
3 3 −12
(c) The dimension of null-space is one more than the dimension of left null-space.
   
  1 1 2 1 0 2
Solution: 1 1 or or
1 1 2 0 1 5
   
1 3
(d) Left null-space contains , row space contains .
3 1
 
-9 -3
Solution: .
3 1
16. Suppose A is a 3 by 4 matrix and B is a 4 by 5 matrix with AB = 0. Show that
rank(A) + rank(B) ≤ 4.

Solution: As AB = 0, col(B) ⊆ N (A). Therefore, the dim(col(B)) ≤ dim(N (A)). This implies
rank(B) ≤ 4 − rank(A).
 5

17. (T) Let A ∈ Mm,n (R) and B ∈ Mn,p (R) with rank(A) = rank(B) = n. Show that rank(AB) = n.

  Since rank(A) = n, m ≥ n. Thus, there exists an invertible matrix P such that

Solution:
I
P A = n . As P is invertible
0
    
In B
rank(AB) = rank(P AB) = rank B = rank = rank(B).
0 0