You are on page 1of 11

Mid-Semester Exam: Math-I (Linear Algebra)

Indraprastha Institute of Information Technology, Delhi

20th September, 2019

Duration: 60 minutes Maximum Marks: 50

Question 1.
(a) (6 marks) Determine whether the following vectors form a linearly in-
dependent set:
     
1 −2 1
−1  1 −6
 3  , v2 = −6 , v3 =  3  .
v1 =      

1 −4 −9

(b) (4 marks) Which of the following vectors are in Span{v1 , v2 , v3 }?


       
1 0 0 0
0 1 0 0
0 ,
e1 =  
0 ,
e2 =  
1 ,
e3 =  
0 .
e4 =  

0 0 0 1
Question 2. All parts of this question carry equal marks.
(a) Find an LU -factorization of the following matrix:
 
7 −1 0
A = 14 0 1
7 −3 3

(b) Use the LU -factorization method to solve the linear system Ax = b


where A is the matrix given in part (a) and b is the vector given below.
Write down all the calculations involved.
 
0
b=  1 .
−1
Question 3 (10 marks). In each of the following parts, say whether the set
W is a subspace of the vector space V . Justify your answer in each case.
All parts of the question carry equal marks.
(a) V = Rn , W is the set of solutions of the nonhomogeneous linear system
Ax = b, where A is an m × n matrix and b ∈ Rm .
(b) V = C[0, 1], the space of all continuous functions defined on the interval
[0, 1], and W is the set of all continuous functions f defined on [0, 1]
such that f (c) ∈ Q for a fixed c ∈ [0, 1].

(Recall that the interval [0, 1] is the set of numbers 0 ≤ x ≤ 1 and that
Q is the set of rational numbers).
(c) V = R2 , W is the graph of the function f (x) = x.

(Recall the the graph of a function f : R → R is defined as the set of


points
{(x, y) : y = f (x)}.
The graph of f is a subset of R2 .)
(d) V = R2 , W is the graph of the function f (x) = x + x2 .
Question 4.
(a) (5 marks) Show that for any positive integer n,
n  n
λ nλn−1
 
λ 1
= .
0 λ 0 λn

(b) (5 marks) Given  


2 −1 3
A = 1 2 4 ,
3 1 1
show that A3 − 5A2 − 4A + 30I = 0. Hence deduce that A is
invertible and find A−1 .
Question 5 (10 marks). Let A and B be m × n matrices that are both in
reduced row echelon form (RREF), such that A 6= B. Suppose that the first
n − 1 columns of A and B are identical. Assume further that neither A nor
B have pivot positions in the last column.
Prove or disprove: There exists a vector x in Rn such that Ax = 0 but
Bx 6= 0.
Sample answer(s) for each question are provided below, along with a sug-
gested grading rubric. In case the student comes up with an alternative
solution, please use your discretion in marking.

Answer 1.

(a) • Step 1 (2 marks). The given set of vectors is a linearly dependent


set if and only if the equation Ax = 0 has a nontrivial solution,
where A = [v1 v2 v3 ].
• Step 2 (2 marks). In order to determine whether the system Ax = 0
has a solution, we reduce matrix A to RREF:
   
1 −2 1 1 −2 1
−1 R2 = R2 + R1,
1 −6 0 −1 −5 

 3 −6
 R3 = R3 − 3R1,  
3  0 0 0
R4 = R4 − R1
1 −4 −9 0 −2 −10
 
1 −2 1
0 −1 −5
R4 = R4 − 2R2 
0

0 0
0 0 0
 
1 0 11
0 1 5
R1 = R1 − 2R2; R2 = −R2 
0 0

0
0 0 0
• Step 3 (2 marks). As there is only two pivot columns in A, there
is one free variable. Therefore nontrivial solutions of Ax = 0 exist.
Hence the given set of vectors is linearly dependent.
(b) For each j = 1, . . . , 4, ej is in Span{v1 , v2 , v3 } if the system Ax =
ej has a solution. We solve each of these systems.

Step 1 (1 mark). We solve the system Ax = e1 . Write the aug-


mented matrix and perform the same sequence of row operations
used in part (a) to see if the last column contains a pivot.
 
1 −2 1 1
−1 1 −6 0
 
 3 −6 3 0
1 −4 −9 0
I will show the row operations only on the last column as we already
know their effect on the first three columns
       
1 1 1 −1
0 1 1 −1
       
0 −3 −3 −3
0 −1 −3 −3
As the last column of the augmented matrix is a pivot column, the
system has no solution. Therefore e1 is not in the span of the given
vectors.
• Step 2 (1 mark). We do the same for e2 .
       
0 0 −2 −2
1  1  1 −1
       
0  0  0  0
0 −2 −2 −2
One further step of swapping rows 3 and 4 show that the last
column is a pivot column. Therefore e2 is not in the span of the
given vectors.
• Step 3 (1 mark). We do the same for e3 . The same sequence of
row operations leave the vector unchanged. So the last column is a
pivot column. Therefore e3 is not in the span of the given vectors.
• Step 4 (1 mark). We do the same for e4 . The same sequence
of row operations leave the vector unchanged. One further step of
swapping rows 3 and 4 show that the last column is a pivot column.
Therefore e4 is not in the span of the given vectors.
Answer 2.
First Solution:
(a) • Step 1 (1 mark). The first column of L is
 
1
2
1
• Step 2 (1 mark). Row reduce A to clear out entries in the first
column that are below the pivot. R2=R2-2R1; R3=R3-R1.
   
7 −1 0 7 −1 0
14 0 1 0 2 1
7 −3 3 0 −2 3
• Step 3 (1 mark). The second column of L is
 
0
 1
−1
and the third column of L is e3 .
• Step 4 (1 mark). Continue row reduction to clear out entries in the
second column that are below the pivot. R3 = R3 + R2.
   
7 −1 0 7 −1 0
0 2 1 0 2 1
0 −2 3 0 0 4
• Step 5 (1 mark). The required factorisation is
    
7 −1 0 1 0 0 7 −1 0
A = 14 0 1 = LU = 2 1 0 0 2 1
7 −3 3 1 −1 1 0 0 4

(b) • Step 1 (1 mark) Solving Ax = b for x is equivalent to first solving


Ly = b for y and then solving U x = y for x.
• Step 2 (2 marks). Since b is the second column of L, we obtain
y = e2 by inspection. Alternatively, we solve the system
y1 = 0
2y1 + y2 = 1
y1 − y2 + y3 = 1
to give
y1 = 0
y2 = 1
y3 = −1 + y2 = 0,
obtaining y = e2 .
• Step 3 (2 marks). We solve the system U x = e2 for x using back
substition. Solving
7x1 − x2 = 0
2x2 + x3 = 1
4x3 = 0
gives x3 = 0, x2 = (1 − x3 )/2 = 1/2 and x1 = x2 /7 = 1/14.
Second solution:

(a) • Step 1 (3 marks). Row reduce A to row echelon form with no


interchange or scaling:
 
7 −1 0
A (R2 → R2 − 2R1 ; R3 → R3 − R1 ) 0 2 1
0 −2 3
 
7 −1 0
(R3 → R3 + R2 ) 0 2 1
0 0 4

So  
7 −1 0
U = 0 2 1 .
0 0 4
• Step 2 (2 marks). Obtain L by applying the inverses of the above
steps to the identity matrix I3 . So
 
1 0 0
L = 2 1 0 .
1 −1 1

Check:
    
1 0 0 7 −1 0 7 −1 0
LU = 2 1 0 0 2 1 = 14 0 1
1 −1 1 0 0 4 7 −3 3

(b) Same as first solution.

Answer 3. Grading Rubric: 1 mark for the correct answer and 1.5 marks
for the correct reason for eact part.

(a) Since the zero vector is not a solution of a non-homogeneous linear


system, the zero vector does not belong to W . This implies W will not
form a subspace of vector space V .

(b) Suppose f ∈ W , i.e. f is a nonzero function such that f (c) ∈ Q,


f (c) 6= 0. It is always possible to find such a function f , for example
the constant function f (x) = 1 for all x ∈ [0, 1].
If we take α to be an irrational number then αf (c) becomes irrational.
=⇒ αf ∈ / W . Hence W is not closed w.r.t to scalar multiplication and
therefore W will not form a subspace of vector space V .
(c)
W = {(x, x) : x ∈ R},
i.e. W represent the collection of points on a line passing through origin.
Let u = (x, x) and v = (y, y) be two arbitrary point of W where x and
y are real no. and let α be a scalar (a real no.)
Now u + v = (x, x) + (y, y) = (x + y, x + y) ∈ W
αu = α(x, x) = (αx, αx) ∈ W , as αx is again a real no
=⇒ W is closed w.r.t to addition and scalar multiplication
=⇒ W will form a subspace of vector space V .
(d) W = {(x, x + x2 ) : x ∈ R} i.e. W represents the collection of points
that lie on the curve y = x + x2 . Let x = 1 and x = −1.
=⇒ (1, 2) and (−1, 0) belong to W
But (1, 2) + (−1, 0) = (0, 2) ∈
/ W =⇒ W is not closed w.r.t addition.

Answer 4.
(a) • Step 1 (2 marks). We use induction on n. Assume that
k  k
λ kλk−1
 
λ 1
=
0 λ 0 λk
holds true for every positive integer k < n.
• Step 2 (2 marks). The induction hypothesis gives us
 n  n−1  
λ 1 λ 1 λ 1
=
0 λ 0 λ 0 λ
 n−1
n − 1λn−2 λ 1
 
λ
=
0 λn−1 0 λ
 n n−1 + (n − 1)λn−1

λ λ
=
0 λn
 n
λ nλn−1

=
0 λn
• Step 3 (1 mark). Clearly the induction hypothesis holds true when
k = 1 so the base case is ok.
(b) • Step 1 (1 mark).  
12 −1 5
A2 = 16 7 15
10 0 14
• Step 2 (1 mark).  
38 −9 37
A3 = 84 13 91
62 4 44
• Step 3 (1 mark).
A3 − 5A2 − 4A + 30I
       
38 −9 37 12 −1 5 2 −1 3 30 0 0
= 84 13 91 − 5 16 7 15 − 4 1 2 4 +  0 30 0 
62 4 44 10 0 14 3 1 1 0 0 30
= 0.

• Step 4 (1 mark). Since A3 − 5A2 − 4A + 30I = 0 it follows that


−1 3
A − 5A2 − 4A = I.

30
Hence
   
−1 2  −1 2 
A A − 5A − 4I = A − 5A − 4I A = I
30 30
• Step 5 (1 mark). Hence
−1 2
A−1 =

A − 5A − 4I
30      
12 −1 5 2 −1 3 4 0 0
−1 
= 16 7 15 − 5 1 2 4 − 0 4 0
30
10 0 14 3 1 1 0 0 4
 
−2 4 −10
−1 
= 11 −7 −5 
30
−5 −5 5
 1 2 1

15 − 15 3
 11 7 1

=− 30 30 6
1 1
6 6 − 61
Answer 5. Three solutions are presented along with the grading rubric for
each.
First Solution.

• Step 1 (2 marks). We PROVE the given statement. No marks for


selecting disprove.
• Step 2 (3 marks). Let A0 be the matrix formed by taking the first n−1
columns of matrix A, in the same order as in the original matrices. Let
a be the last column of A. Consider the linear system

A0 y = a

where  
y1

 y2 

 . 
y= 

 . 

 . 
yn−1

• Step 3 (3 marks). The augmented matrix for the system A0 y = a is


A. As the augmented column is not a pivot column, this system has
a solution, say ξ.
• Step 4 (2 marks). Let x be the vector in Rn whose first n − 1 entries
are the same as the entries of the vector ξ and whose last entry is −1.
Then Ax = 0, but

Bx = A0 ξ − b = a − b 6= 0.

Second Solution:

• Step 1 (2 marks). We PROVE the given statement. No marks for


selecting disprove.
• Step 2 (2 marks). Let the nonzero entries in the last column of matrix
A be aj1 ,n , . . . , ajk ,n where

1 ≤ j1 < j2 < . . . < jk ≤ m.

• Step 3 (2 marks). For each i = 1, . . . , k let li denote the column of


matrix A which contains the pivot entry of row ji .
• Step 4 (2 marks). Since A is in reduced echelon form, the li -th column
of matrix A is the standard basis vector eji .

• Step 5 (2 marks). Let xli = −aji ,n for i = 1, . . . , k. Let xn = −1. Let


the remaining entries of the vector x be zero. Then Ax = 0. However
were Bx = 0, the last column of B would be the same as the last
column of A, which is not possible.

Third Solution:

• Step 1 (2 marks). We PROVE the given statement. No marks for


selecting disprove.

• Step 2 (2 marks). Let A = [aij ] and B = [bij ] with aij = bij for
1 ≤ j < n. Since the n-th column of A is not a pivot column, the
variable xn is a free variable. Hence the system Ax̄ = 0̄ has a nontrivial
solution.

• Step 3 (2 marks). Now: when we express the solution of Ax̄ = 0̄ in


standard vector form, the vector corresponding to the free variable
xn has 1 in its n-th position, corresponding to the dummy equation
xn = xn . Let this vector be
 
u1
 u2 
 
 . 
ū =  
 . 
un

with un = 1 so ū is a nontrivial solution of the equation Ax̄ = 0̄.

• Step 4 (4 marks). Since the n-th columns of A and B are different,

akn 6= bkn for some k ≤ n. (∗)

Consider the k-th entry in Aū. We must have

ak1 u1 + ak2 u2 + . . . + akn un = 0 (1)

since Aū = 0̄. Suppose B ū = 0̄ also. Then

bk1 u1 + bk2 u2 + . . . + bkn un = 0 (2)


Equating LHS of (1) and (2) we have

ak1 u1 + ak2 u2 + . . . + akn un = bk1 u1 + bk2 u2 + . . . + bkn un

or
akn = bkn (3)
since akj = bkj for j < n. But (3) contradicts (∗), Therefore B ū 6= 0̄,
and the result follows.

You might also like