T1A/T1B, hmcheungae Math 2121 Tutorial 5 October 11, 2018

Mathematics is the cheapest science. Unlike physics or chemistry, it does not require
any expensive equipment. All one needs for mathematics is a pencil and paper.———
—-George Polya

Inverse of a matrix:
• Let A be an n × n matrix and define T : Rn → Rn by T (x) = Ax. The matrix A is invertible if the
function T is invertible. Its inverse is the unique matrix A−1 such that T −1 (x) = A−1 x.
• To show an n × n matrice A is invertible, suffices to find an n × n matrix B such that
AB = BA = In .
When such the matrix B exists, it is called the inverse of A, and B is denoted by A−1 .
Subspace of Rn :
• Let W be a subset of Rn . W is called a subspace of Rn if it satisfies
1. 0 ∈ W .
2. If u, v ∈ W , then u + v ∈ W .
3. If u ∈ W and c ∈ R, then cu ∈ W .
Col(A), Nul(A)
Suppose A is an m × n matrix. Write
A = [a1 a2 · · · an ],
where ai denotes the i-th column of A.
• The null space of A is Nul A = {x ∈ Rn | Ax = 0}.
• The columns space of A is Col A = Span{a1 , a2 , · · · , an } = {Ax ∈ Rn | x ∈ Rn }.

Nul(A) Col(A)
1 Nul(A) is a subspace of Rn . Col(A) is a subspace of Rm .
2 Nul(A) is implicitly defined, Col(A) is explicitly defined,
i.e., you are only given a condition Ax i.e., you are told how to bulid vectors in Col(A).
that vectors in Nul(A) must satisfy.
3 It takes time to find vectors in Nul(A): It is easy to find vectors in Col(A):
Row operations are required. the columns of A are displayed
and others are formed from them.
4 There is no obvious relation There is an obvious relation
between Nul(A) and the entries of A. between Col(A) and the entries of A,
since each of column of A is in Col(A).
5 A typical vector v in Nul(A) has the property A typical vector v in Col(A) has the property
that Av = 0. that Ax = v is consistent.
6 Given a specific vector v, it is easy to tell It takes time to tell whether a given v is in Col(A):
if v is in Nul(A). Compute Av. Row operations are required.
7 Nul(A)={0} iff Col(A)= Rm iff
Ax = 0 has only the trivial solution Ax = b has a solution for every b ∈ Rm
iff x 7→ Ax is one-to-one. iff x 7→ Ax is onto.
8 A basis for Nul(A) is obtained by A basis for Col(A) is obtained by
collecting basic solutions from solving Ax = 0. collecting the pivot columns of A.
T1A/T1B, hmcheungae Math 2121 Tutorial 5 Page 2 of 5

Example 1
True or false:
  
 x 
1. S1 = y  ∈ R3 : 2x − 3y = 1 + z is a subspace of R3 .
z
 

2. Given A be a n × n matrix and let S2 = {v ∈ Rn : Av = λv}. Then S2 is a subspace of Rn .

S2 is called the λ-eigenspace of A.
  
 x1 
3. S3 = x2  ∈ R3 : x1 ≥ 0 is a subspace of R3
x3
 
  
x
4. S4 = ∈ R2 : y = x 2 is a subspace of R2 .
y
5. Let W ⊆ Rn be a subspace of Rn . Then the complement of W in Rn :
S5 = Rn \ W = {v ∈ Rn : v 6∈ W } is a subspace of Rn .
6. If v1 , · · · , vp are in Rn , then S6 =Span{v1 , · · · , vp } is a subspace of Rn .
7. S7 = {(a + 2b, 2a − 3b : a, b ∈ R} is a subspace of R2 .
8. Given A is n × n matrix. Then S8 = the set of solution of Ax = 0 is a subspace of Rn .
9. Given A is n × n matrix. Then S9 = the set of solution of Ax = b is a subspace of Rn for each b ∈ Rn .
Answer:
Proof. 1. False:
The most obvious reason is that S1 does not contain the origin (i.e., the zero vector).
2. True:
By the fact that null space is a subspace of domain and note that S2 =Nul(A − λIn ), we get
W = Nul(A − λIn ) ≤ Rn .

3. False: e1 ∈ S3 but −e1 6∈ S3 .

         
1 1 2 1 1
4. False: , ∈ S4 but = + 6∈ S4 since 22 6= 0.
−1 1 0 −1 1
5. False: 0 ∈ W implies 0 6∈ S5 .
6. True:
• 0 ∈ S6 since 0v1 + · · · + 0vp ∈ S6 .
• Given u, w ∈ S6 . Let u = a1 v1 + · · · + ap vp and w = b1 v1 + · · · + bp vp .
Then u + w = (a1 + b1 )v1 + · · · + (ap + bp )vp ∈ S6 .
• Let c ∈ R, u ∈ S6 . Let u = a1 v1 + · · · + ap vp . Then Let cu = (ca1 )v1 + · · · + (cap )vp ∈ S6
7. True:         
a + 2b 1 2 1 2
Since =a +b . Then S7 =span , ≤ R2 .
2a − 3b 2 −3 2 −3
8. True:
S8 = N ul(A) ≤ Rn .
9. False: For b 6= 0, 0 6∈ S9 .

Cont.
T1A/T1B, hmcheungae Math 2121 Tutorial 5 Page 3 of 5

Example 2
True or False: Given T : Rn → Rm is a linear transformation.
1. If T is one-to-one, then m ≥ n.
2. If T is onto, then n ≥ m.
3. If T is invertible, then m = n.
4. If m < n, then T is one-to-one.
5. If m < n, then T is onto.
6. If n < m, then T is one-to-one.
7. If n < m, then T is onto.
8. If m = n, then T is invertible.
Proof. 1. True:
If T is one-to-one, then every column of A has pivot position, this implies the number of rows of A is
greater or equal to the number of columns of A, i.e., m ≥ n.
2. True:
If T is onto, then every row of A has pivot position, this implies the number of columns of A is greater
or equal to the number of rows of A, i.e., n ≥ m.
3. True:
If T is invertible, then T is one-to-one and T is onto, then by part (a), we get m ≥ n and by part (b),
we get n ≥ m. This implies m = n.
4. False:
Since T is one-to-one implies m ≥ n. Taking contrapositive, we get n > m implies T is not one-to-one.
5. False:
T (x) = 0 is not onto.
6. False:
T (x) = 0 is not one-to-one.
7. False:
Since T is onto implies n ≥ m. Taking contrapositive, we get n < m implies T is not onto.
8. False: T (x) = 0 is not invertible.

Example 3
True or False:
1. Given A is m × n matrix, B is n × m matrix. If m = n, AB = Im implies BA = In .
2. Given A is m × n matrix, B is n × m matrix. If m 6= n, AB = Im implies BA = In .
3. Given A is a n × n matrix. Then A − λIn is invertible iff AT − λIn is invertible.
4. Given two matrices A and B of the same size. There exists a matrix C whose nullspace is the intersection
of Nul(A) and Nul(B).

Cont.
T1A/T1B, hmcheungae Math 2121 Tutorial 5 Page 4 of 5

5. Given two matrices A and B of the same size. There exists a matrix C whose column space contains
both Col(A) and Col(B).
Proof. 1. True:
If A, B are n × n matrices and AB = In , then BA = In .
Now A, B are n × n matrices and AB = In , consider Bx = 0, then
x = In x = (AB)x = A(Bx) = A(0) = 0 =⇒ x = 0.

Also,
AB = In =⇒ B(AB)B −1 = BIn B −1 = BB −1 = In =⇒ BA = In .

Hence A = B −1 . The assertion is proved.

2. False:
Method 1:
For m 6= n, we disprove by finding counterexamples:
 
  1 0
Let A = 1 0 and let B = AT . Then AB = [1] but BA = .
0 0
 
  1 0 0
1 0 0
Or A = and let B = AT . Then AB = I2 but BA = 0 1 0.
0 1 0
0 0 0
Method 2: For m 6= n,
When m < n,
Claim: BA 6= In .
Note that m × n matrix A cannot have every column have a pivot position. Then is equivalent to
Ax = 0 has a nontrivial solution, say, x0 .
Thus Ax0 = 0. Then
(BA)x0 = B(Ax0 ) = B0 = 0.
Hence x0 is a nontrivial solution to (BA)x = 0.
This implies BA 6= In , since In x = 0 has only the trivial solution.
3. True:
Suppose A − λIn is invertible, let B be its inverse.
We get
(A − λIn )B = B(A − λIn ) = In .

Taking transpose on the both sides, we get

T T
((A − λIn )B) = (B(A − λIn )) = InT .
This implies
B T (A − λIn )T = (A − λIn )T B T = In .
and then
B T (AT − λIn ) = (AT − λIn )B T = In .
Thus AT − λIn is invertible and B T is its inverse.
Conversely, suppose AT − λIn is invertible, then replacing A by AT in the above argument.
We get (AT )T − λIn is invertible, i.e.,A − λIn is invertible.

Cont.
T1A/T1B, hmcheungae Math 2121 Tutorial 5 Page 5 of 5

 
A
4. True: Stack the two matrices on top of each other to form C = .
B
 
Av
Then Cv = = 0 if and only if Av = 0 and Bv = 0. Then v ∈ N ul(C) if and only if v ∈ N ul(A)
Bv
and v ∈ N ul(B). Therefore, N ul(C) = N ul(A) ∩ N ul(B).
5. True: Just put the two matrices side by side to form C = [A B]. The columns of C the include all
columns of A and B, and therefore Col(C) contains all linear combinations of these columns.

Example 4
Find a linear transformation T : R2 → R3 whose image is the plane x + 2y + 3z = 0.
Answer:
 us to find a A ∈ M3×2 (R)
The problem askes  such that T = LA
 x 
and Im(T ) = y  ∈ R3 : x + 2y + 3z = 0 .
z
 
In other word,
 we need to find A ∈ M3×2 (R) such that Col(A) = N ul(B),
where B = 1 2 3 ,

T =L L
R2 −−−−→
A
R3 −−→
B
R.

To find a basis for the Nul(B), let us compute set of solution of Bx = 0.

Let y = s, z = t, this implies x = −2s − 3t and then
       
x −2s − 3t −2 −3
y  =  s  = s 1  + t 0 .
z t 0 1
   
 −2 −3 
Then  1  ,  0  is clearly a basis of Nul(B).
0 1
 
 
−2 3
Consider T = LA where A =  1 0.
0 1
This works because
   
 −2 −3 
Im(T ) = T (x) ∈ R3 : x ∈ R 2
= Ax ∈ R3 : x ∈ R 2
 
= Col(A) = Span  1  ,  0  = N ul(B).
0 1
 

The End.