數心 線性代數讀書會 線性變換（一）

Lukaswuuuuuuu

Fall, 2024

What will we learn from this session?

• Linear Map

• Fundamental Subspaces: Null Space and Range

• Rank-Nullity Theorem

• Invertible Matrix Theorem

• Row and Column Rank

Contents
1 Linear Maps and Fundamental Subspaces 2
1.1 Linear Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 The Algebra of Linear Map . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Null Space and Range . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Related Theorems 6
2.1 Rank-Nullity Theorem and Its Application . . . . . . . . . . . . . . . 6
2.2 Inverse of a Linear Map . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Matrix Representation of Linear Maps . . . . . . . . . . . . . . . . . 8
3 Selected Exercises 9

Notation:

• V: vector space

• F=R or C

• Theorem: All of the statement that requires a proof or explanation

I (the author) selected a few exercises from Linear Algebra by Stephen H. Fried-
berg, Arnold J. Insel, Lawrence E. Spence, Linear Algebra by Hoffman and Kunze
and Linear Algebra and Its Applications by David C. Lay, Steven R. Lay, Judi J.
McDonald. Readers are encouraged to finish at least half of them. If there is any
question about the exercises or this handout, please contact me by

• Discord: lukaswuuuuuuu

1
1 Linear Maps and Fundamental Subspaces
1.1 Linear Map

Definition 1.1.1 (Linear).

If a function T : V → W , or named transformation, is linear, then it satisfies

• Preserve addition: T (x1 + x2 ) = T (x1 ) + T (x2 ).

• Preserve scalar multiplication: T (cx1 ) = cT (x1 ).

where c ∈ F and x1 , x2 ∈ V .

One may identify if a map T is linear by checking if T (cx1 + x2 ) = cT (x1 ) + T (x2 ),

which actually requires a proof, but is left for readers as exercise.

Definition 1.1.2 (Linear Map).

If a map T : V → W is called a linear map, then it is linear.

The set of linear map from V to W is denoted as L(V, W ).

Example I : V → V , v 7→ v is linear and is called identity.

Linear map can be between different dimensions, we will give examples at our last part.

Corollary 1.1.1.
∀T ∈ L(V, W ), we have

1. T (0) = 0.
∑n ∑n
2. ∀vi ∈ V and ci ∈ F , T ( k=1 ck vk ) = k=1 ck T (vk ).

Proof. 1. T (0) = T (0 + 0) = T (0) + T (0), then T (0) = 0.

2. This proof will require the definition of linear map,

∑
n ∑
n
T( ck vk ) = T (ck vk )
k=1 k=1
∑
n
= ck T (vk )
k=1

Remark 1.1.1. The proof of the corollary is not as hard as the readers would think.
However, as the first proof we wrote, the point the author want to show is to always
use the definition to deduce any theorem.

2
 Theorem 1.1.1.
Given V, W are vector spaces over F . v1 , . . . , vn is a basis of V and w1 , . . . , wn ∈ W .
Then ∃! linear map T : v → W such that T (vj ) = wj , j = 1, . . . , n.

Proof.
(Existence) ∀v ∈ V, ∃! c1 , . . . , cn ∈ F such that v = c1 v1 + · · · + cn vn .

Define T : V → W, v = c1 v1 + · · · + cn vn 7→ T (v) = c1 w1 + · · · + cn wn . Then

T (vj ) = wj , j = 1, . . . , n

We check if T is a linear map:

1. Suppose u = d1 v1 + · · · + dn vn ,
T (u + v) = (d1 + c1 )T (v1 ) + · · · + (dn + cn )T (vn )
= (d1 + c1 )w1 + · · · + (dn + cn )wn
= T (u) + T (v)
2. Suppose a ∈ F , T (au) = T (ad1 v1 + · · · + adn vn ) = ad1 w1 + · · · + adn wn = aT (u).

(Uniqueness) If T ′ : V → W is a linear map such that T ′ (vj ) = wj , j = 1, . . . , n. For

some ci ∈ F
T ′ (c1 v1 + · · · + cn vn ) = c1 T ′ (v1 ) + · · · + cm T ′ (vn ) = c1 w1 + · · · + cn wn
which implies
T ′ (u) = d1 w1 + · · · + dn wn = T (u)
Hence, we obtain the unique map.

1.2 The Algebra of Linear Map

Theorem 1.2.1.
1. T, S ∈ L(V, W ), (T + S) : V → W, v 7→ T v + Sv is a linear map.

2. a ∈ F , T ∈ L(V, W ), aT : V → W, v 7→ aT (v), is a linear map.

3. associativity: (T1 T2 )T3 = T1 (T2 T3 )

4. multiplicative inverse: T I = IT = T

5. distributive property: (S1 + S2 )T = S1 T + S2 T

6. not always commutative

Example (of 6.) D, T ∈ L(P (R)), D : p(x) 7→ p′ (x), T : p(x) 7→ x2 p(x).

Remark 1.2.1. 1. and 2. implies that L(V, W ) is a vector space over F .

3
1.3 Null Space and Range

Definition 1.3.1 (Null Space and Nullity).

If T ∈ L(V, W ), we define

nullT = {T (v) = 0 | v ∈ V }

and
dim(nullT ) = nullityT

Example
Let p(x) denotes any polynomial and let D be the differential operator.

Then nullD = {p′ (x) = 0} = {p(x) = c | c ∈ R}.

Theorem 1.3.1 (Null Space is a Subspace).

If T ∈ L(V, W ), then nullT is a subspace of V .

Remark 1.3.1. Let us recall the conditions for A to be a subspace of V :

1. A ⊆ V

2. 0 ∈ A

3. there is closed under addition

4. there is closed under multiplication

Proof. Obviously, nullT ⊂ V .

1.(∃ 0 ∈ nullT ) T (0) = 0.

2.(closed under addition) Suppose T (u) = 0 and T (v) = 0,

T (u + v) = T (u) + T (v) = 0 + 0 = 0 ∈ nullT

3.(closed scalar multiplication) Suppose T (u) = 0, a ∈ F ,

T (au) = aT (u) = a · 0 = 0 ∈ nullT

Thus, nullT is a subspace of V .

4
 Theorem 1.3.2. T ∈ L(V, W ) is injective if and only if nullT = {0}.

Remark 1.3.2. Let f : A → B, if f is called injective or one to one (1-1), then

∀ x1 , x2 ∈ A and x1 6= x2 , f (x1 ) 6= f (x2 ).

which is
∀ x1 , x2 ∈ A, f (x1 ) = f (x2 ) ⇒ x1 = x2

Proof.
(⇒) Let u ∈ nullT , T (u) = 0 = T (0) implies u = 0 since T is injective.

Then nullT ⊆ {0} and obviously {0} ⊆ nullT , hence nullT = {0}.

(⇐) Let u, v ∈ V and T (u) = T (v). We have

T (u) − T (v) = 0
⇒ T (u − v) = 0 = T (0)
⇒u−v =0

which implies that T is injective.

Definition 1.3.2 (Range and Rank).

If T ∈ L(V, W ), we define

rangeT = {T (v) ∈ W | v ∈ V }

and
dim(rangeT ) = rankT

Example
Let E = {2n | n ∈ N}. Then there is a linear map T ∈ L(N, E) and v 7→ 2v,
rangeT = E.

Theorem 1.3.3 (Range is a Subspace).

If T ∈ L(V, W ), then rangeT is a subspace of W .

Proof.
Obviously rangeT ⊂ W .

1.(∃ 0 ∈ rangeT ) T (0) = 0.

2.(closed under addition) Suppose v1 , v2 ∈ V , T (v1 ) + T (v2 ) = T (v1 + v2 ) ∈ W .

3.(closed scalar multiplication) Suppose a ∈ F , aT (v1 ) = T (av1 ) ∈ W .

Thus, rangeT is a subspace of W .

5
2 Related Theorems
2.1 Rank-Nullity Theorem and Its Application

Theorem 2.1.1 (Rank-Nullity Theorem).

Suppose V is finite-dimensional and let T ∈ L(V, W ), then,

dimV = rankT + nullityT

Proof.
Let {u1 , . . . , um } be a basis of nullT then let u1 , . . . , um , v1 , . . . , vn as a basis of V , which
implies dimV = m + n.

Let v = a1 u1 + · · · + am um + b1 v1 + · · · + bn vn ∈ V where ai , bi ∈ F . Applying T to

both sides, we have

T (v) = a1 T (u1 ) + · · · + am T (um ) + b1 T (v1 ) + · · · + bn T (vn )

= b1 T (v1 ) + · · · + bn T (vn )

which implies span{T (v1 ), . . . , T (vn )} = range(T ).

Suppose c1 T (v1 ) + · · · + cn T (vn ) = 0 = T (c1 v1 + · · · + cn vn ) = T (ṽ) thus ṽ ∈ null(T ).

We write c1 v1 + · · · + cn vn = d1 u1 + · · · + dm um and di ∈ F since ṽ ∈ nullT ⊂ V .

Since {u1 , . . . , um , v1 , . . . , vn } is a basis of V , ci , di are all 0.⇒ {T v1 , . . . , T vn } is a basis

of W .

Thus rankT = n and dimV = m + n = rankT + nullity T .

Application
Solve the system of equations


 A1,1 x1 + A1,2 x2 + · · · + A1,m xm = c1


A2,1 x1 + A2,2 x2 + · · · + A2,m xm = c2
 ..

 .

A x + A x + · · · + A x = c
n,1 1 n,2 2 1,m m n

We define a linear map T : F m → F n by

∑
m ∑
m
T (x1 , . . . , xm ) = ( A1,i xi , . . . , An,i xi )
i=1 i=1

1. If (c1 , . . . , cn ) ∈ rangeT , then a solution to the system exists.

2. If nullityT = 0, then any solution is unique.

6
 Figure 1: Visualized linear map between vector spaces

Remark 2.1.1. By figure 1, we see the relation between the two subspaces mentioned
previously. It will probably help readers better understand rank-nullity theorem. Since
0 ∈ rankT seems to be distilled from nullT to 0, the theorem show a idea of making
up for the loss of rangeT .

2.2 Inverse of a Linear Map

Definition 2.2.1 (Inverse).

For a map T , if there exists a map T1 such that T T1 = T1 T = I, then we say that
T1 is the inverse of T and denoted as T −1 .

T −1 exists if and only if T is a bijection.

Remark 2.2.1. If a map T is call a bijection, then it is injective and surjective.

We say T ∈ L(V, W ) is surjective, or onto, if

∀w ∈ W, ∃v ∈ V such that T (v) = w.

Corollary 2.2.1 (T −1 is Linear).

If T ∈ L(V, W ) and T −1 exists, then T −1 is linear.

Proof. We want to show that

T −1 (cw1 + w2 ) = cT −1 (w1 ) + T −1 (w2 )

Let vi = T −1 (wi ), since T is linear, we have

T (cv1 + v2 ) = cT (v1 ) + T (v2 )

= cw1 + w2

which implies cv1 + v2 = T −1 (cw1 + w2 ) = cT −1 (w1 ) + T −1 (w2 )

7
 Definition 2.2.2 (Non-singular).
If a linear map T ∈ L(V, W ) is called non-singular, then nullT = {0}.

Theorem 2.2.1 (Invertible Linear Map Theorem). Let T ∈ L(V, W ). V and W are
finite dimensional and dimV = dimW , The following statements are equivalent

1. T is invertible

2. T is non-singular

3. rangeT = W

Proof.
(2.⇔3.) Let dimV = dimW = n, by rank-nullity theorem, we have

dimV = n = nullityT + rankT = rankT = dimV

thus T is non-singular if and only if rangeT = W .

Since nullityT = 0 and T (V ) = W , then T is invertible.

Remark 2.2.2. The theorem of equivalent statements of invertibility is a gigantic

one. One who has interested in it is encouraged to read P.134 of Linear Algebra and
Its Applications by David C. Lay, Steven R. Lay, Judi J. McDonald.

2.3 Matrix Representation of Linear Maps

Theorem 2.3.1 (Column and Row Rank).

If A ∈ Mm×n (R), then row rankA = column rankA.

Proof. Let T be linear map from n-dimensional column space Xn to m-dimensional

column space Xn . T (Xn ) is defined as AXn .

nullT is the solution space of AXn = 0

rangeT is the set of all m-dimensional column matrices Y such that AX = Y has a
solution. If A1 , . . . , An are the columns of A, then

AX = a1 A1 + · · · + xn An

so that the rangeT is the column space of A. Therefore,

rankT = column rankA

By rank-nullity theorem, nullityT + column rankA = n.

Similarly, by nullityT = n − row rankA. Hence, row rankA = column rankA.

8
3 Selected Exercises
Remark 3.0.1.
The answer of these exercises will be given the week after the handout is posted.

Problem 1.
Show that T is linear if and only if T (cx1 + x2 ) = cT (x1 ) + T (x2 ) where c ∈ F .

Problem 2.
Represent the map T : R2 → R2 where T is called the ’reflection’ about the x-axis.

Problem 3.
Represent the map T : R2 → R2 where T is called the ’projection’ on the x-axis.

Problem 4.
Let A, B ∈ Mn (R), show that if T (A) = AB − BA, then T is linear.

Problem 5. Let V = C(R), the vector space of continuous real-valued functions on R.

Let a, b ∈ R, a < b. Define T : V →R by

∫ b
T (f ) = f (t)dt
a

Show that T is linear.

Problem 6.
Let us recall the inner product. Let T : Rn × Rn → R, T (v, w) = hv, wi. Is inner
product linear?

Problem 7.
If the rank of a 7 × 6 matrix A is 4, what is the dimension of the solution space Ax = 0?

Problem 8. Let T : P2 (R) → P3 (R) be the linear map defined by

∫ x
′
T (f (x)) = 2f (x) + 3f (t)dt
0

Compute a list of vectors that span rangeT .

Problem 9. Let T : F2 → F2 be the linear transformation defined by

T (a1 , a2 ) = (a1 + a2 , a1 )

Is T invertible? If it is, find its inverse, by any method reader prefers.

Problem 10.
Let V and W be finite-dimensional vector spaces and T : V → W be linear.

Prove that if dim(V ) > dim(W ), then T cannot be injective.