Lecture Note 11 (Sec 3.

1) Space of Vectors

Vector Space
o A vector space is a set “V” of elements called vectors, with the operations of addition and
scalar multiplication defined on it so that it satisfies the following 10 axioms.
A. Closure Axioms:
1. Closure under addition: 2. Closure under scalar multiplication:
If 𝒖, 𝒗 ∈ 𝑉, then, 𝒖 + 𝒗 ∈ 𝑉. If 𝒖 ∈ 𝑉, and if “c” is a scalar, then, 𝑐𝒖 ∈ 𝑉.

B. Addition Axioms:
3. Commutativity under addition: 4. Associativity under addition:
If 𝒖, 𝒗 ∈ 𝑉, then, 𝒖 + 𝒗 = 𝒗 + 𝒖. If 𝒖, 𝒗, 𝒘 ∈ 𝑉, then, (𝒖 + 𝒗) + 𝒘 = 𝒖 + (𝒗 + 𝒘).
5. Existence of additive identity: 6. Existence of additive inverse:
There exists a unique zero vector “0” in V, There exists a unique vector “-u” in V for each
such that for any vector u in “V”, 0+u=u. element u in “V”, such that u+(-u)=0.

C. Scalar multiplication Axioms:

7. Distributivity for vector addition: 8. Distributivity for scalar addition:
If c is a scalar, & 𝒖, 𝒗 ∈ 𝑉 then, c(u+v)=cu+cv. If c & d are scalars & 𝒖 ∈ 𝑉, then (c+d)u=cu+du.
9. Associativity for the product of scalars: 10. Existence of unity:
If c & d are scalars & 𝒖 ∈ 𝑉, then (cd)u= c(du). For any element u in “V”, “1u=u” holds true.

Example 1: The set ℝ𝑛 denoted by 𝑉 is a vector space under the following usual addition and scalar
multiplication of n-tuples, where u and v belong to ℝ𝑛 , and c is a scalar.

𝒖 + 𝒗 = (𝑢1 , … , 𝑢𝑛 ) + (𝑣1 … . , 𝑣𝑛 ) = (𝑢1 + 𝑣1 , … . . , 𝑢𝑛 + 𝑣𝑛 ) 𝑐𝒖 = 𝑐(𝑢1 , … , 𝑢𝑛 ) = (𝑐𝑢1 , … , 𝑐𝑢𝑛 )

o Let “+”, and “∙” denote the operation of addition and scalar multiplication (by real numbers).

1. For 𝒖, 𝒗 ∈ ℝ𝑛 . Then, the “u+v” belongs to ℝ𝑛 as well. 2. Let u∈ ℝ𝑛 . Let c be a scalar. Then, the
𝒖 + 𝒗 = (𝑢1 , … , 𝑢𝑛 ) + (𝑣1 … . , 𝑣𝑛 ) = (𝑢1 + 𝑣1 , … . . , 𝑢𝑛 + 𝑣𝑛 ) “cu”, belongs to ℝ𝑛 as well.
𝑐𝒖 = 𝑐(𝑢1 , … , 𝑢𝑛 ) = (𝑐𝑢1 , … , 𝑐𝑢𝑛 )

3. Let 𝒖, 𝒗 ∈ ℝ𝑛 . Clearly, “u+v=v+u” as we can see below:

𝒖 + 𝒗 = (𝑢1 , . . , 𝑢𝑛 ) + (𝑣1 . . , 𝑣𝑛 ) = (𝑢1 + 𝑣1 , . . , 𝑢𝑛 + 𝑣𝑛 ) = (𝑣1 + 𝑢1 , . . , 𝑣𝑛 + 𝑢𝑛 ) = (𝑣1 . . , 𝑣𝑛 ) + (𝑣1 . . , 𝑣𝑛 ) = 𝒗 + 𝒖
4. Let 𝒖, 𝒗, 𝒘 ∈ ℝ𝑛 . Clearly, (u+v)+w=u+(v+w).
(𝒖 + 𝒗) + 𝒘 = [(𝑢1 + 𝑣1 ), . . , ( 𝑢𝑛 + 𝑣𝑛 )] + (𝑤1 . . , 𝑤𝑛 ) = [(𝑢1 + 𝑣1 ) + 𝑤1 , . . , ( 𝑢𝑛 + 𝑣𝑛 ) + 𝑤𝑛 ]
= [(𝑢1 + (𝑣1 + 𝑤1 ), . . , 𝑢𝑛 + ( 𝑣𝑛 + 𝑤𝑛 )] = 𝒖 + (𝒗 + 𝒘)

5. 6.
𝟎 + 𝒖 = (0, . . , 0) + (𝑢1 . . , 𝑢𝑛 ) = (𝑢1 , … . . , 𝑢𝑛 ) = 𝒖 𝒖 + (−𝒖) = (𝑢1 , . . , 𝑢2 ) + (−𝑢1 . . , −𝑢𝑛 ) = (0, . . , 0) = 𝟎
We saw, there exists a “0” in “ℝ𝑛 ”, such that We saw, there exists a “0” in “ℝ𝑛 ”, such that for
for any “u∈ ℝ”, “0+u=u” holds true. any “u∈ ℝ𝑛 ”, “u+(-u)=0” holds true.

C. Scalar multiplication Axioms:

7. Let 𝒖, 𝒗 ∈ ℝ𝑛 . Let c be a scalar. 8. Let 𝒖 ∈ ℝ𝑛 . Let c and d be a scalar.
We can easily show “c(u+v)=cu+cv” holds true. We can easily show “(c+d)u=cu+du” holds true.
9. Let c & d be scalars and let 𝒖 ∈ ℝ𝑛 . 10. Let 𝒖 ∈ ℝ𝑛 .
We can easily show “(cd)u= c(du)” holds true. We can easily show that “1u=u” holds true.

Page 1 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Zero vector Space

A nonempty set “V” with just a single object “0” with the following two operations
𝟎 + 𝟎 = 0 𝑎𝑛𝑑 𝑐𝟎 = 0 𝑓𝑜𝑟 𝑒𝑣𝑒𝑟𝑦 𝑠𝑐𝑎𝑙𝑎𝑟 𝑐
is a vector space as it satisfies all the 10 axioms of a vector space. It is called zero-vector space.

Example 2: Show the set V of all polynomial functions of “degree 2 or less” in the real variable “x”
with real coefficients is a vector space.

• Let f and g be two elements of the set V of all polynomial functions of “degree 2 or less” in
the real variable “x” with real coefficients, where,
𝑓(𝑥) = 𝑎2 𝑥 2 + 𝑎1 𝑥 + 𝑎0 𝑔(𝑥) = 𝑏2 𝑥 2 + 𝑏1 𝑥 + 𝑏0
where,
𝑎2 , 𝑎1 , 𝑎0 , 𝑏2 , 𝑏1 , & 𝑏0 are real numbers, meaning they can be negative, or positive, or even zero.
working the following two operations in a pointwise manner
(𝑓 + 𝑔)(𝑥) = 𝑓(𝑥) + 𝑔(𝑥) (𝑐𝑓)(𝑥) = 𝑐𝑓(𝑥)

• Then, we can show that all the above mentioned 10 axioms of a vector space will be satisfied !

Closure Axioms
(𝑓 + 𝑔)(𝑥) = 𝑓(𝑥) + 𝑔(𝑥) = 𝑎2 𝑥 + 𝑎1 𝑥 + 𝑎0 + 𝑏2 𝑥 2 + 𝑏1 𝑥 + 𝑏0 = (𝑎2 + 𝑏2 )𝑥 2 + (𝑎1 + 𝑏1 )𝑥 + (𝑎0 + 𝑏0 )
2

This proves that “f+g” is also a polynomial function of degree “2” or less.
Note: If “𝑎2 = −𝑏2 ”, then, “f+g” will be a 1st degree polynomial. When can “f+g” be 0th degree ?
(𝑐𝑓)(𝑥) = 𝑐𝑓(𝑥) = 𝑐𝑎2 𝑥 2 + 𝑐𝑎1 𝑥 + 𝑐𝑎0 = (𝑐𝑎2 )𝑥 2 + (𝑐𝑎1 )𝑥 + (𝑐𝑎0 )
This proves that “cf” is also a polynomial function of degree “2” or less.

• We have already proved the two closure axioms, you can prove the remaining 8 axioms.

Example 3: Show that the set of polynomial functions of degree “2” is not a vector space.

• Note: To show that a set with the addition and scalar multiplication is not a vector space, it is
enough to show just a single explicit counterexample to one of 10 axioms for a vector space.
• Below, we show that the above-mentioned set does not satisfy closure axiom of vector space.
o Consider the following two polynomial functions of degree “2”:

𝑓(𝑥) = 2𝑥 2 + 3𝑥 + 5 𝑔(𝑥) = −2𝑥 2 − 𝑥 + 1

Then,
𝑓(𝑥) + 𝑔(𝑥) = (2𝑥 2 + 3𝑥 + 5) + ( −2𝑥 2 − 𝑥 + 1) = 2𝑥 + 6
The sum of two 2nd degree functions did not produce a 2nd degree function !
It means, the axiom #1 i.e. the closure property under addition is not satisfied !

o If we take “c=0”, then, “c f(x)” produces “0”, a polynomial of 0 th degree. It does not produce a
polynomial of degree “2” as we have sought. It means, the closure property under scalar
multiplication is not satisfied! No need to check other axioms. The set is not a vector space.

Note: More examples on showing if a set is (or is not) a vector space are in Extra Example section.

Page 2 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Subspace
o A subset W of a vector space V is a subspace of V if the following conditions are satisfied:
I. The vector “0” of “V” is also an element of W. [This also means, W is a nonempty set.]
II. If u and v are two vectors in W, then, u+v is also an element in W.
III. If c is a scalar and u is a vector in W, then cu is also an element in W.
Note: Above requirements can also be combined to construct the following definition of subspace:
o A subset W of a vector space V is a subspace of V if the following condition is satisfied:
“If u and v are two vectors in W, and if c and d are two scalars, then cu+dv is also in W”.
A subspace containing u and w must contain all linear combinations cu+dv.

Trivial subspaces: We are talking the following two subspaces.

a) Let V be a vector space. Then, “V” is its own subspace.
b) Let V be a vector space. The set W={0}, with solo element “0”, is a subspace of V because
0+0=0 c0=0 (for any scalar c)
shows W is closed under addition, and shows W is closed under scalar multiplication.
Such subspace “W” is called the zero space of the vector space V.

Example 4: Set W of 2 × 2 upper triangular matrices is a subspace of 𝑀22 , the set of 2 × 2 matrices.

Let
𝑊={ [
𝑎 𝑏
]| 𝑎, 𝑏, 𝑑 ∈ ℝ }
0 𝑑
o We know that the set 𝑀22 of 2 × 2 matrices is a vector space. We will show W is its subspace.
o Clearly, W is a non-empty subset of 𝑀22 . Let c be a scalar. Also, let 𝒖, 𝒗 ∈ 𝑊 so that
𝑒 𝑓 𝑝 𝑞
𝒖=[ ] 𝒗=[ ]
0 ℎ 0 𝑠
Then,
𝑒 𝑓 𝑝 𝑞 𝑒+𝑝 𝑓+𝑞 𝑒 𝑓 𝑐𝑒 𝑐𝑓
𝒖+𝒗 =[ ]+[ ]=[ ] 𝑐𝒖 = 𝑐 [ ]=[ ]
0 ℎ 0 𝑠 0 ℎ+𝑠 0 ℎ 0 𝑐ℎ
We see, “u+v” is a 2 × 2 upper triangular matrix. We see, “𝑐𝒖” is a 2 × 2 upper triangular matrix.
Thus, “u+v” is also an element of W. Thus, “𝑐𝒖” is also an element of W. It shows
It shows that W is closed under addition. that W is closed under scalar multiplication.

• Since the subset W of 𝑀22 is closed under addition & scalar multiplication it is a subspace of 𝑀22 .
• This subspace W of 𝑀22 is therefore itself a vector space of matrices, embedded in 𝑀22 .
• Note that a set D of 2 × 2 diagonal matrices is the subspace of the above vector space W.

Example 5: Show the subset W of vector space 𝑉 = ℝ3 = {(𝑥, 𝑦, 𝑧)|𝑥, 𝑦, 𝑧 ∈ ℝ } is not a subspace of V.
𝑊 = {(𝑥, 𝑦, 𝑧)|𝑥, 𝑦, 𝑧 ∈ ℝ, 𝑦 = 𝑥 + 𝑧 + 1, 𝑡ℎ𝑎𝑡 𝑖𝑠, 𝑦 − 𝑥 − 𝑧 = 1}
Note: A vector x is in W only if its “(2nd component y)– (1st component x) – (3rd component z)”=1

Answer: Let k be a scalar. Also, let 𝒖, 𝒗 ∈ 𝑊 so that

𝒖 = {(𝑎, 𝑏, 𝑐) such that 𝑏 − 𝑎 − 𝑐 = 1 𝒗 = {(𝑑, 𝑒, 𝑓) such that 𝑒 − 𝑑 − 𝑓 = 1

Then,
𝒖 + 𝒗 = (𝑎, 𝑏, 𝑐) + (𝑑, 𝑒, 𝑓) = (𝑎 + 𝑑, 𝑏 + 𝑒, 𝑐 + 𝑓) 𝑘𝒖 = 𝑘(𝑎, 𝑏, 𝑐) = (𝑘𝑎, 𝑘𝑏, 𝑘𝑐)
Then, Then,
( 𝑏 + 𝑒) − (𝑎 + 𝑑) − (𝑐 + 𝑓) = (𝑏 − 𝑎 − 𝑐) + (𝑒 − 𝑑 − 𝑓) = 2 (𝑘𝑏) − ( 𝑘𝑎) − (𝑘𝑐) = 𝑘(𝑏 − 𝑎 − 𝑐) = 𝑘(1) = 𝑘
It proves, “u+v” is not an element of the set W. It proves, “𝑘u” is not an element of W.

• Subset W of 𝑉 is not closed under addition & scalar multiplication so it is not a subspace of 𝑉.

Page 3 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Spanning set and Span

o A set {𝒗𝟏 , 𝒗𝟐 , … . , 𝒗𝒏 } of vectors is called the spanning set of vectors for a vector space V if every
vector “v” in that vector space “V” can be expressed as a linear combination of these vectors as
𝒗 = 𝑐1 𝒗𝟏 + 𝑐2 𝒗𝟐 + ⋯ + 𝑐𝑛 𝒗𝒏
o The set of all linear combinations of 𝒗𝟏 , 𝒗𝟐 , … . , 𝒗𝒏 is called the span of 𝒗𝟏 , 𝒗𝟐 , … . , 𝒗𝒏 .
o The span of 𝒗𝟏 , 𝒗𝟐 , … . , 𝒗𝒏 is denoted as: span{ 𝒗𝟏 , 𝒗𝟐 , … . , 𝒗𝒏 }.

o We know that W={(𝑐1 , 𝑐2 , 𝑐1 + 𝑐2 )} is the subset of a vector space ℝ3 .

o We can easily show that W={(𝑐1 , 𝑐2 , 𝑐1 + 𝑐2 )} is the subspace of a vector space ℝ3 .
What are the vectors that span W ?

o Consider an arbitrary vector v in W such that v=(𝑐1 , 𝑐2 , 𝑐1 + 𝑐2 ). Then,

𝒗 = (𝑐1 , 𝑐2 , 𝑐1 + 𝑐2 ) = (𝑐1 + 0, 0 + 𝑐2 , 𝑐1 + 𝑐2 ) = (𝑐1 , 0, 𝑐2 ) + (0, 𝑐2 , 𝑐2 ) = 𝑐1 (1,0,1) + 𝑐2 (0,1,1) = 𝑐1 𝒗𝟏 + 𝑐2 𝒗𝟐
o Since any arbitrary vector v in W is expressed as the linear combination of the vectors (1,0,1)
and (0,1,1), these two vectors are said to span W. This W is a plane through the origin.

𝑐1 𝒗𝟏 + 𝑐2 𝒗𝟐
span{ 𝒗𝟏 , 𝒗𝟐 }
that is
𝑐1 𝒗𝟏 𝑐2 𝒗𝟐 span{(1,0,1), (0,1,1)}

𝒗𝟏 = (1,0,1) 𝒗𝟐 = (0,1,1)

Example 6: Show that {(1,0), (0,1) } is the spanning set for ℝ2 .

Note: Some books express the vectors (1,0), (0,1) as (1,0)𝑇 , (1,1)𝑇 as well.
We show, it is possible to find constants 𝑐1 & 𝑐1 so that any arbitrary vector v=(𝑎, 𝑏) in ℝ2 can be
expressed as the linear combination
𝑐1 𝒗𝟏 + 𝑐2 𝒗𝟐 = 𝒗
So that the following holds true.
1 1 𝑎
𝑐1 [ ] + 𝑐2 [ ] = [ ] 𝐸𝑞. 1
0 1 𝑏
Solving we get
𝑐 𝑐 𝑎 ⟹ 𝑐 + 𝑐2 𝑎 ⟹ 𝑐1 + 𝑐2 = 𝑎 ⟹ 𝑐1 = 𝑎 − 𝑏
[ 1] + [ 2] = [ ] [ 1 ]=[ ]
0 𝑐 𝑏 𝑐2 𝑏 𝑐2 = 𝑏 𝑐2 = 𝑏
𝑐1 (1,0) + 𝑐2 (1,1) = (𝑎 − 𝑏)(1,0) + (𝑏)(1,1) = (𝑎 − 𝑏, 0) + (𝑏, 𝑏) = (𝑎 − 𝑏 + 𝑏, 𝑏) = (𝑎, 𝑏)
When 𝑐1 value of (𝑎 − 𝑏), & 𝑐2 value of 𝑏 are put in Eq.1, it becomes satisfied as shown below:
1 1 𝑎
(𝑎 − 𝑏) [ ] + (𝑏) [ ] = [ ] 𝐸𝑞. 2
0 1 𝑏
Thus, we just showed that {(1,0), (0,1) } is the spanning set for ℝ2 .
Alternatively, using the given two vectors as the columns of a matrix, solve the following system:
1 1 𝑐1 𝑎
[ ][ ] = [ ]
0 1 𝑐2 𝑏
Consider the augmented matrix and reduce it by elementary row operations.
1 1 𝑎 𝑅 : 𝑅 − 𝑅2 1 0 𝑎 − 𝑏
[ | ]≈ 1 1 [ | ]
0 1 𝑏 0 1 𝑏
which has solution 𝑐1 = (𝑎 − 𝑏), & 𝑐2 = 𝑏 for every (𝑎, 𝑏)in ℝ2 . It is consistent for any (𝑎, 𝑏).
So, for the constants 𝑐1 = (𝑎 − 𝑏), 𝑐2 = 𝑏, we can express (𝑎, 𝑏) as: (𝑎, 𝑏) = (𝑎 − 𝑏)(1,0) + (𝑏) (1,1)
o Since the “2 by 2” matrix has its reduced row echelon form as 𝐼2 , the system is consistent for all
right side vector (a,b). There is no restriction imposed on a and b.

Page 4 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Example 7: Show that {(1,1,0)𝑇 , (0, 1, −2)𝑇 , (1,3,1)𝑇 } is the spanning set for ℝ3 .
Note: Our book expresses (1,1,0)𝑇 , (0, 1, −2)𝑇 , (1,3,1)𝑇 as (1,1,0), (0,1, −2), (1,3,1). Both notations are okay.

We show, it is possible to find constants 𝑐1 , 𝑐2 , & 𝑐3 so that any arbitrary vector v=(𝑎, 𝑏, 𝑐)𝑇 in ℝ3 can
be expressed as the linear combination
𝑐1 𝒗𝟏 + 𝑐2 𝒗𝟐 + 𝑐3 𝒗𝟑 = 𝒗
So that the following holds true.
1 0 1 𝑎
𝑐1 [1] + 𝑐2 [ 1 ] + 𝑐3 [3] = [𝑏 ] 𝐸𝑞. 1
0 −2 1 𝑐
Express it as augmented matrix & apply the row elimination to reduced it to upper triangular form
1 0 1 𝑎 1 0 1 𝑎 1 0 1 𝑎 𝑐1 + 𝑐3 = 𝑎
𝑅 : 𝑅 − 𝑅1
[1 1 3| 𝑏 ] ≈ 2 2 [0 1 2| (𝑏 − 𝑎)] ≈ [0 1 2| (𝑏 − 𝑎) ] ⟹ 𝑐2 + 2𝑐3 = (𝑏 − 𝑎)
𝑅3 : 𝑅3 + 2𝑅2
0 −2 1 𝑐 0 −2 1 𝑐 0 0 5 𝑐 + 2(𝑏 − 𝑎) 5𝑐3 = −2𝑎 + 2𝑏 + 𝑐
We got three pivots for the in upper triangular form of 3 by 3 matrix. That is a good news !
Solving the back substitution, we get the following solutions for every (a, b, c) in ℝ3 :
1 1 1
𝑐1 = (7𝑎 − 2𝑏 − 𝑐) 𝑐2 = (−𝑎 + 𝑏 − 2𝑐) 𝑐3 = (−2𝑎 + 2𝑏 + 𝑐)
5 5 5
When the above calculated 𝑐1 , 𝑐2 , & 𝑐3 are put in Eq.1, the Eq.1 becomes satisfied as shown below:
1 1 1 0 1 1 𝑎
(7𝑎 − 2𝑏 − 𝑐) [1] + (−𝑎 + 𝑏 − 2𝑐) [ 1 ] + (−2𝑎 + 2𝑏 + 𝑐) [3] = [𝑏]
5 5 5 𝑐
0 −2 1
With the help of above found constants 𝑐1 , 𝑐2 , & 𝑐3 , any vector 𝒗 = (𝑎, 𝑏, 𝑐)𝑇 in ℝ3 can be expressed as
the linear combination of 𝑣1 = (1,1,0)𝑇 , 𝑣2 = (0,1, −2)𝑇 , & 𝑣3 = (1,3,1)𝑇 .Thus, given vectors span ℝ3 .

Example 8: Show that {(1, −2,3)𝑇 , (−2,4, −6)𝑇 , (0,6,4)𝑇 } is not the spanning set for ℝ3 .
Apply the row elimination to reduced it to upper triangular form
1 −2 0 𝑎 𝑎 𝑎 𝑎
𝑅 : 𝑅 + 2𝑅1 1 −2 0 𝑅2 : 𝑅2 /6 1 −2 0 1 −2 0
[−2 4 6| 𝑏 ] ≈ 2 2 [ 0 6| (𝑏 + 2𝑎)] ≈ [0 0 1| (𝑏 + 2𝑎)/6] ≈ [0 0| (13𝑎 + 2𝑏 − 3𝑐)/12]
𝑅3 : 𝑅3 − 3𝑅1 0 𝑅3 : 𝑅3 /4 𝑅2 : 𝑅2 − 𝑅3
0
3 −6 4 𝑐 0 0 4 (𝑐 − 3𝑎) 0 0 1 (𝑐 − 3𝑎)/4 0 0 1 (𝑐 − 3𝑎)/4
The system has no solution if
(13𝑎 + 2𝑏 − 3𝑐) ≠ 0
We just encountered a restriction !
We are unable to express an arbitrary vector v=(𝑎, 𝑏, 𝑐)𝑇 in ℝ3 as a combination of 𝒗𝟏 , 𝒗𝟐 & 𝒗𝟑 when
(13𝑎 + 2𝑏 − 3𝑐) is nonzero. So, given vectors 𝒗𝟏 = (1, −2,3)𝑇 , 𝒗𝟐 = (−2,4, −6)𝑇 ), 𝒗𝟑 = (0,6,4)𝑇 do not span ℝ3 .
Alternatively, not that vectors (1,-2,3) and (-2,4,-6) are linearly dependent as (-2,4,-6) =2(1,-2,3).
We need at 3 linearly independent vectors to span ℝ3 . So, these 3 vectors do not span ℝ3 . However,
1 −2 0 𝑎
𝑐1 [−2] + 𝑐2 [ 4 ] + 𝑐3 [6] = [𝑏 ]
3 −6 4 𝑐
gives
1 1 0 𝑎 1 0 𝑎
𝑐1 [−2] + (−2)𝑐2 [−2] + 𝑐3 [6] = [𝑏 ] ⟹ (𝑐1 − 2𝑐2 ) [−2] + 𝑐3 [6] = [𝑏 ]
3 −3 4 𝑐 3 4 𝑐
Thus, we have expressed (a, b, c) as linear combinations of 2 linearly independent vectors (1,-2,3) &
(0,6,4) because these are not collinear. These 2 vectors span a 2-dimensional subspace of ℝ3 .

o We need 3 linearly independent vectors to span ℝ3 . If there are 3 vectors but one vectors is a multiple of
the other, then, these 3 vectors of course cannot span ℝ3 . Can 2 vectors span ℝ3 ? Of course not !
o Can the 4 vectors (3,1,0), (-5,-2,1), (1,0,0), and (3,1,1) span ℝ𝟒 ?
No. We can express (3,1,1) as linear combination of other 3 vectors as: (3,1,1)=(3)(3,1,0)+(1)(-5,-2,1)+(-1)(1,0,0).
Also, we can show that the three vectors (3,1,0), (-5,-2,1) and (1,0,0) are linearly independent. So, they span ℝ3 .
Thus, (3,1,1) lies in the span of (3,1,0), (-5,-2,1) and (1,0,0). Therefore, these four vectors rather span just ℝ3 .
o Please find more “spanning examples” in Extra Example section.

Page 5 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

The column space of a matrix A

o Below, we introduce a “column space” for a “3 × 2” matrix A. Then, we can define it generally.
o Consider a “3 × 2” matrix A with “3” rows and “2” columns:
3 4
𝐴 = [1 0]
2 1

o Column space:
o Each column of the “3 × 2” matrix “A” has “3” components so each of its columns belongs to ℝ3 .
o For this “3 × 2” matrix A, the set of all possible linear combinations (or the ‘span’) of the
columns of a matrix A is called the column space of A. Thus, it is the subspace of “ℝ3 ” spanned
by the column vectors of “A”. It is denoted by C(A). Thus, for above matrix,
3 4
{𝒄
𝑪(𝐴) = 𝟏 𝑥1 + 𝒄𝟐 𝑥2 ∶ 𝑥1 , 𝑥2 ∈ ℝ} where, 𝒄 𝟏 = [1 ] and 𝒄 𝟐 = [0]
2 1
That means,

3 4 3 4 𝑥
1
𝑪(𝐴) = {[1] 𝑥1 + [0] 𝑥2 ∶ 𝑥1 , 𝑥2 ∈ ℝ} ⟹ 𝑪(𝐴) = {[1 0] [𝑥 ] ∶ 𝑥1 , 𝑥2 ∈ ℝ } That is, 𝑪(𝐴) = {𝐴𝒙 }
2
2 1 2 1

• We just saw, a “3 × 2” matrix with “3” rows, has a column space of vectors with 3 components.
• To solve Ax=b means to express the right-side vector “b” as a combination “Ax” of columns of A.
Therefore,
3 4 𝑥 𝑏1 𝑏1
1
the system “Ax=b” i.e. [1 0] [𝑥 ] = [𝑏2 ]is solvable if and only if [𝑏2 ] belongs to column space of A.
2
2 1 𝑏3 𝑏3
Notes:
Each column of the “3 × 2” matrix “A” has “3” components so each of its columns belongs to ℝ3 .
It means, the column space of this “3 × 2” matrix “A” is a subspace of ℝ3 (not ℝ2 .) We can show it !
It means, the column space of a “𝑚 × 𝑛” matrix is a subspace of ℝ𝑚 (It is not the subspace of ℝ𝑛 .)
In section 3.4, we will discuss the dimension and basis of the column space.

Example 9: Describe the column spaces for the following matrices A and B:

1 4
𝐴=[ ]
3 12

Note that here,

Column 2 Vector = 𝒄𝟐 = [
4 1
] = 4 [ ] = 4𝒄𝟏 = 4 times Column 1 vector
12 3

Therefore, the column space consists of only those vectors the form (k, 3k) that lie along a line
“y=3x” in which each of the column vectors (1,3) & (4,12) lies.

Equation Ax=b is solvable only if the right-side vector b is on the above-mentioned line. We will
discuss this fact in the next page.

Page 6 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Condition when the system Ax=b is solvable

• Now, consider the system Ax=b.
• We already know, the left side “Ax” belongs to the column space of the matrix “A”.
• So, to solve “Ax=b”, we must show that the right-side b also belongs to the column space of A.
In other words,
the vector b is in the column space of the matrix A if and only if the system Ax = b is solvable.

• Note: In the Example 9, the column space of the given matrix “A” was only a line. So, if we have
equation “Ax=b” with that matrix “A”, then, “Ax=b” is solvable only when b is on that line.

Example 10: Check if the following vector b belongs to the column space of the following matrix A.

5 0 1 2
𝒃 = [13] 𝐴 = [1 2 5]
4 1 0 2

• We know, “vector b belongs to the column space of a matrix A if and only if Ax = b is solvable”.
• So, we would like to see if Ax = b is solvable, i.e., test if there exists a unique solution to Ax = b.
• For that, we will find the reduced row echelon form of the augmented matrix [𝐴|𝒃] of Ax = b as

0 1 2 𝟓 ≈ 1 2 5 𝟏𝟑 ≈ 1 1 3 𝟖 ≈ 1 0 1 𝟑 ≈ 1 0 0 𝟐
[𝐴|𝒃] = [1 2 5| 𝟏𝟑 ] 𝑅1 ⟺ 𝑅2 [0 1 2| 𝟓 ] 𝑅1 − 𝑅2 [0 1 2 | 𝟓 ] 𝑅1 − 2𝑅2 [0 1 2| 𝟓] 𝑅1 − 𝑅3 [0 1 0| 𝟑 ]
1 0 2 𝟒 1 0 2 𝟒 𝑅3 − 𝑅1 0 −2 −3 −𝟗 𝑅3 + 2𝑅2 0 0 1 𝟏 𝑅2 − 2𝑅3 0 0 1 𝟏

• Since [𝐴|𝒃] is reduced to reduced row echelon form, the unique solution is: 𝑥1 = 2, 𝑥2 = 3, and 𝑥3 = 1.
• Thus, the system Ax = b is solvable, so it proves that the vector b belongs to column space of A.
• Then, we can express b as the linear combination of the column vectors of A as shown below:
0 1 2 𝟓
(2) [1] + (3) [2] + (1) [5] = [𝟏𝟑] = 𝒃
1 0 2 𝟒

Example 11: Find the condition on the components 𝑏1 , 𝑏2 , & 𝑏3 of vector b so that Ax = b is solvable.
𝑏1 1 2
𝒃 = [𝑏2 ] 𝐴=[ 3 7]
𝑏3 −1 −2
• We know, “Ax = b is solvable if and only if the vector b belongs to column space of matrix A”.

• Suppose b belongs to column space of A. Then, Ax=b is solvable. Our Job: Find components of b.
• Let’s see what type of “b” will make the system “Ax=b” solvable?
• For that, we will find the reduced row echelon form of the augmented matrix [𝐴|𝒃] of Ax = b as
1 2 𝑏1 ≈ 1 2 𝑏1 ≈ 1 0 𝑏1 − 2𝑏2 + 6𝑏1 ≈ 1 0 7𝑏1 − 2𝑏2
[3 𝑅
7 | 𝑏2 ] 2 = 𝑅3 − 3𝑅 𝑅
1 [0 1| 𝑏2 − 3𝑏1 ] 1 = 𝑅 1 − 2𝑅2 [0 1| 𝑏2 − 3𝑏1 ] [0 1| 𝑏2 − 3𝑏1 ]
−1 −2 𝑏3 𝑅3 = 𝑅3 + 𝑅1 0 0 𝑏3 + 𝑏1 0 0 𝑏3 + 𝑏1 0 0 𝑏3 + 𝑏1
We get
𝑥1 = 7𝑏1 − 2𝑏2 𝑥2 = 𝑏2 − 3𝑏1 (0)𝑥1 + (0)𝑥2 = 𝑏3 + 𝑏1 ⟹ 𝑏3 + 𝑏1 = 0
𝑬𝒒. 𝟏 𝑬𝒒. 𝟐 𝑬𝒒. 𝟑
• From Eq. 3, we see, the solution is possible only when 𝑏3 + 𝑏1 = 0. This is the condition on b.
Note: You will see one more example like this in the last page of this lecture note.

Note: You will see lots of extra examples starting page 6. Read them. I will do them if time permits.

Page 7 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Extra Examples (EE)

Extra Example Part I: Extra Examples (EE1 to EE2) on “proving vector spaces”

EE1: Show that the set of all 𝟐 × 𝟐 real diagonal matrices of the form
𝑝 0
[ ]
0 𝑞
with the standard matrix operations of addition and scalar multiplication is a vector space.
Let “V” be a set of set of all 2 × 2 real diagonal matrices of the form
𝑝 0
[ ]
0 𝑞
Let “c” and “d” be real scalars. And, let u, v, and w be three matrices in the set V.
𝑒 0 𝑔 0 𝑠 0
𝒖=[ ] 𝒗=[ ] 𝒘=[ ]
0 𝑓 0 ℎ 0 𝑡

A. Closure Axioms:
1. 2.
𝑒 0 𝑔 0 𝑒+𝑔 0+0 𝑒 0 𝑐𝑒 0
𝒖+𝒗= [ ]+[ ]=[ ] 𝑐𝒖 + 𝒗 = 𝑐 [ ]=[ ]
0 𝑓 0 ℎ 0+0 𝑓+ℎ 0 𝑓 0 𝑐𝑓
Of course, Of course,
“u+v” is also a 2 × 2 real diagonal matrix “cu” is also a 2 × 2 real diagonal matrix.

B. Addition Axioms:
3.
𝑒 0 𝑔 0 𝑒+𝑔 0+0 𝑔+𝑒 0+0 𝑔 0 𝑒 0
𝒖+𝒗= [ ]+[ ]=[ ]=[ ]=[ ]+[ ] =𝒗+𝒖
0 𝑓 0 ℎ 0+0 𝑓+ℎ 0+0 ℎ+𝑓 0 ℎ 0 𝑓
4.
(𝑒 + 𝑔) + 𝑠 (0 + 0) + 0 (𝑒 + (𝑔 + 𝑠) 0 + (0 + 0)
(𝒖 + 𝒗) + 𝒘 = [ ]=[ ] = 𝒖 + (𝒗 + 𝒘)
(0 + 0) + 0 (𝑓 + ℎ) + 𝑡 0 + (0 + 0) 𝑓 + (ℎ + 𝑡)
5. Take the following element of V. 6. For already defined matrix u, define “-u” as
0 0 −𝑒 0
𝟎=[ ] −𝒖 = [ ]
0 0 0 −𝑓
Then, we can clearly see, Then, we can clearly see,
0 0 𝑒 0 0+𝑒 0+0 𝑒 0 𝑒 0 −𝑒 0 𝑒−𝑒 0−0 0 0
𝟎+𝒖=[ ]+[ ]=[ ]=[ ]=𝒖 𝒖 + (−𝒖) = [ ]+[ ]=[ ]=[ ]=𝟎
0 0 0 𝑓 0+0 𝑓+0 0 𝑓 0 𝑓 0 −𝑓 0−0 0−0 0 0

C. Scalar multiplication Axioms:

𝑒+𝑔 0+0 𝑐(𝑒 + 𝑔) 𝑐(0 + 0) 𝑐𝑒 + 𝑐𝑔 𝑐0 + 𝑐0 𝑐𝑒 𝑐0 𝑐𝑔 𝑐0
𝑐(𝒖 + 𝒗) = 𝑐 [ ]=[ ]=[ ]=[ ]+[ ] = 𝑐𝒖 + 𝑐𝒗
7 0 + 0 𝑓 + ℎ 𝑐(0 + 0) 𝑐(𝑓 + ℎ) 𝑐0 + 𝑐0 𝑐𝑓 + 𝑐ℎ 𝑐0 𝑐𝑓 𝑐0 𝑐ℎ

8 𝑒 0 (𝑐 + 𝑑)𝑒 (𝑐 + 𝑑)0 𝑐𝑒 + 𝑑𝑒 𝑐0 + 𝑑0 𝑐𝑒 𝑐0 𝑑𝑒 𝑑0
(𝑐 + 𝑑)𝒖 = (𝑐 + 𝑑) [ ]=[ ]=[ ]=[ ]+[ ] = 𝑐𝒖 + 𝑑𝒖
0 𝑓 (𝑐 + 𝑑)0 (𝑐 + 𝑑)𝑓 𝑐0 + 𝑑0 𝑐𝑓 + 𝑑𝑓 𝑐0 𝑐𝑓 𝑑0 𝑑𝑓
9 𝑒 0 (𝑐𝑑)𝑒 (𝑐𝑑)0 𝑐(𝑑𝑒) 𝑐(𝑑0) 𝑑𝑒 𝑑0 𝑒 0
(𝑐𝑑)𝒖 = (𝑐𝑑) [ ]=[ ]=[ ] = 𝑐[ ] = 𝑐 {𝑑 [ ]} = 𝑐(𝑑𝒖)
0 𝑓 (𝑐𝑑)0 (𝑐𝑑)𝑓 𝑐(𝑑0) 𝑐(𝑑𝑓) 𝑑0 𝑑𝑓 0 𝑓
10 Surely, for any element u in of “ℝ”, the condition “1u=u” holds true as we can see below.
𝑒 0 1𝑒 0 𝑒 0
1𝒖 = 1 [ ]=[ ]=[ ]𝒖
0 𝑓 0 1𝑓 0 𝑓
We just showed that set V of 2 × 2 diagonal real matrix satisfies all 10 properties of a vector space.

Page 8 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

EE2: Let V be a set of all real-valued functions that are defined at each “x” in the interval (−∞, ∞).
If f=f(x) and g=g(x) are two functions in “V” and if “c” is a scalar, then define the operations of
addition and scalar multiplication by

𝐴𝑑𝑑𝑖𝑡𝑖𝑜𝑛: (𝒇 + 𝒈)(𝑥) = 𝑓(𝑥) + 𝑔(𝑥)

𝑆𝑐𝑎𝑙𝑎𝑟 𝑀𝑢𝑙𝑡𝑖𝑝𝑙𝑖𝑐𝑎𝑡𝑖𝑜𝑛: (𝑐𝒇)(𝑥) = 𝑐𝑓(𝑥)

Then, show that the above set V is a vector space under the above two operations.

A. Closure Axioms:
1. Let 𝒇 and 𝒈 be the elements in V. By the 2. Let 𝒇 be an element in V, and “c” be a scalar.
addition, “f+g” is defined as By the scalar multiplication, “cf” is defined as
𝒇 + 𝒈 = 𝑓(𝑥) + 𝑔(𝑥) 𝑐𝒇 = 𝑐𝑓(𝑥)
Since the sum of two real valued functions Since the product of a real valued function with a
produces a real valued function, “𝑓(𝑥) + 𝑔(𝑥)” scalar produces a real valued function, “𝑐𝑓(𝑥)”
is also a real valued function. Is also a real valued function.
Thus, “f+g” is also an element in V. Thus, “cf” is also an element in V.

B. Addition Axioms:
3.
(f+g)(x) = f(x)+g(x) = g(x)+f(x) = (g+f)(x) Thus, we get
Commutativity of real numbers (f+g)=(g+f)

4.
((f+g)+h)(x)=(f+g)(x)+h(x)=(f(x)+g(x))+h(x)=f(x)+((g(x)+h(x))=f(x)+(g+h)(x)=(f+(g+h)(x)
Associativity of real numbers
Thus, we get, (f+g)+h=f+(g+h)
5. There exists an entry “0” of V such that 6. For already defined matrix f, define “-f” as
𝟎(𝑥) = 0 (−𝒇)(𝒙) = −[𝑓(𝑥)]
Then, we can clearly see, Then, we can clearly see,
(𝟎 + 𝒇)(𝑥) = 0(𝑥) + 𝑓(𝑥) = 𝑓(𝑥) [(𝒇 + (−𝒇))](𝑥) = 𝑓(𝑥) + (−𝑓)(𝑥) = 𝑓(𝑥) = 𝑓(𝑥) = 0 = 𝟎𝑥
Thus, we get, 0+f=f Thus, we get, f+(-f)=0

C. Scalar multiplication Axioms:

7. Show 𝑐(𝒇 + 𝒈) = 𝑐𝒇 + 𝑐𝒈.

8. Show (𝑐 + 𝑑)𝒇 = 𝑐𝒇 + 𝑑𝒇.

9. Show (𝑐𝑑)𝒇 = 𝑐(𝑑𝒇)

10. Sure, for any f in “𝑉”, we will be able to get “1f=f”.

Page 9 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Extra Example Part II: Extra Examples (EE3 to EE8) on “proving not a vector space”

EE3: Show why the following set is not a vector space.

• Consider a set V where 𝑉 = ℝ2 . Define the addition and scalar multiplication in V as follows:

For 𝒖 = (𝑢1 , 𝑢2 ) and 𝒗 = (𝑣1 , 𝑣2 ) , For 𝒖 = (𝑢1 , 𝑢2 ) and any real number 𝒌 ,
𝒖 + 𝒗 = (𝑢1 , 𝑢2 ) + (𝑣1 , 𝑣2 ) = (𝑢1 + 𝑣1 , 𝑢2 + 𝑣2 ) 𝑘𝒖 = 𝑘(𝑢1 , 𝑢2 ) = (𝑘𝑢1 , 0)

• The above set does not satisfy the 10th axiom (i.e. the existence of unity) because
1𝒖 = 1(𝑢1 , 𝑢2 ) = (1𝑢1 , 0) = (𝑢1 , 0) ≠ 𝒖

• The condition “1u=u” does not hold true for any u in “V”. So, above set V s not a vector space.

EE4: Show why the following set is not a vector space.

• Consider a set V with 𝑉 = ℝ3 . Define the addition and scalar multiplication in V as follows:

For 𝒖 = (𝑢1 , 𝑢2 , 𝑢3 ) and 𝒗 = (𝑣1 , 𝑣2 , 𝑣3 ), For 𝒖 = (𝑢1 , 𝑢2 , 𝑢3 ) and any real number 𝒌
𝒖 + 𝒗 = (𝑢1 , 𝑢2 , 𝑢3 ) + (𝑣1 , 𝑣2 , 𝑣3 ) = (𝑢1 + 𝑣1 , 𝑢2 + 𝑣2 , 𝑢3 + 𝑣3 ) 𝑘𝒖 = 𝑘(𝑢1 , 𝑢2 , 𝑢3 ) = (𝑘 3 𝑢1 , 𝑘 3 𝑢2 , 𝑘 3 𝑢3 )

• Above set does not satisfy the 8th axiom (i.e. the distributivity for scalar addition) because
if c & d are scalars, and u is an element of “V”, then,

(𝑐 + 𝑑)𝒖 = (𝑐 + 𝑑)(𝑢1 , 𝑢2 , 𝑢3 ) = ((𝑐 + 𝑑)3 𝑢1 , (𝑐 + 𝑑)3 𝑢2 , (𝑐 + 𝑑)3 𝑢3 )

But,

𝑐𝒖 + 𝑑𝒖 = (𝑐 3 𝑢1 , 𝑐 3 𝑢2 , 𝑐 3 𝑢3 ) + (𝑑 3 𝑢1 , 𝑑 3 𝑢2 , 𝑑 3 𝑢3 ) = ((𝑐 3 + 𝑑 3 ) 𝑢1 , (𝑐 3 + 𝑑 3 ) 𝑢2 , (𝑐 3 + 𝑑 3 ) 𝑢3 )

• Thus, in general, “(c+d)u” does not equal “cu+du”. Therefore, V is not a vector space.

EE5: Show why the following set is not a vector space.

• Consider a set V of 2 by 2 matrix with real positive entries. That is, let

𝑎 𝑏
𝑉 = {[ ] : 𝑎 > 0, 𝑏 > 0, 𝑐 > 0, 𝑎𝑛𝑑 𝑑 > 0}
𝑐 𝑑

• Above set does not satisfy the 2nd axiom (i.e. closure under scalar multiplication) because for “c”
as “-1’, and for a 2 by 2 matrix “A” of the above form (with 𝑝 > 0, 𝑞 > 0, 𝑟 > 0, 𝑠 > 0 ), we get

𝑝 𝑞 −𝑝 −𝑞
(−1)𝐴 = (−1) [ ]=[ ]
𝑟 𝑠 −𝑟 −𝑠

• Thus, we obtained a 2×2 matrix whose all four entries are negative.
• For “c=-1” & “u=A”, the “cu” turned out not to be an element of V. Thus, V is not a vector space.

Page 10 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

EE6: Show why the following set is not a vector space.

• Consider a set V with 𝑉 = {𝑥|𝑥𝜖ℝ, 𝑥 > 0}. Define addition and scalar multiplication in V as:

For 𝒙 ∈ 𝑉 & 𝒚 ∈ 𝑉, we have, 𝒙 + 𝒚 = 𝑥 ⋅ 𝑦 For 𝒙 ∈ 𝑉 & for a real number 𝑘, we have, 𝑘 𝒙 = (𝑥)𝑘

• The set V is not a vector space because the axiom #7 i.e. “k(𝒙+y)=k𝒙 +k𝒚” is not satisfied.

𝑘(𝒙 + 𝒚) = (𝑥 + 𝑦)𝑘 = (𝑥𝑦)𝑘 = (𝑥)𝑘 (𝑦)𝑘 k𝒙 +k𝒚 = (𝑥)𝑘 + (𝑦)𝑘 Thus, 𝑘(𝒙 + 𝒚 ≠ 𝑘𝒙 + 𝑘𝒚

Can “0” be a zero vector? Can “1” be a zero vector?

• No, because for any x in V, we get • Yes, because for any x in V, we get
𝟎 + 𝒙 = 0 ∙ 𝑥 = 0 ≠ 𝑥 = 𝒙 ⟹ 𝟎 + 𝒙 ≠ 𝒙. 𝟏 + 𝒙 = 1 ∙ 𝑥 = 𝑥 = 𝒙 ⟹ 𝟏 + 𝒙 = 𝒙.
• Moreover, note that 0 ∉ 𝑉 as well. • Moreover, note that 1 ∈ 𝑉 as well because 1>0.

EE7: Show why the following set is not a vector space.

𝑉 = {(𝑥, 𝑦, 𝑧) ∈ ℝ3 | 𝑥 + 𝑦 + 𝑧 = 1}

• We can show that the above set does not satisfy the axiom #5. That is, we can show, for any
vector u in “V”, there does not exist any zero vector “0” in V that would make “0+u=u” true.
o Let u= (p,q,r) be an element of vector V.
o Then, by the definition of vector V, we have p+q+r=1.
o Can (0,0,0)+u=u ? Yes. (0,0,0)+u=(0,0,0)+ (p,q,r)=(p,q,r)=u. So, we get (0,0,0)+u = u.
o However, for the axiom #5 to be satisfied, this (0,0,0) has to be a member of the set V.
o Note that, (0,0,0) is not a member of the set V because 0 + 0 + 0 = 0 ≠ 1.
o Thus, (0,0,0) is not a zero vector we are looking for.
o Is there any “zero vector” that belong to set V and that satisfy “zero vector+u =u” ? No.

• Based on the above discussed reasoning, we conclude that the given set V is not a vector space.

EE8: Show why the following set is not a vector space.

𝑎 𝑏
𝑉 = {( ) | 𝑎, 𝑏, 𝑐 ∈ ℝ}
𝑐 1
• We can show that the above set does not satisfy the axiom #5. That is, we can show, for any
vector u in “V”, there does not exist any zero vector “0” in V that would make “0+u=u” true.
• Let the vector “u” shown below be an element of the above vector V.
𝑎 𝑏
𝒖=( )
𝑐 1
• Will we be able to satisfy the axiom #5 ? That is, will we able to get a “zero vector” that can
make “zero vector”+ “vector u” = “vector u” ? The answer is “No” because, for
𝑎 𝑏 𝑎 𝑏
zero vector + 𝒖 = ( )=( )=𝒖
𝑐 1 𝑐 1
to happen, we must have our “zero vector” as
0 0
𝟎=( )
0 0
But, this “o” does not fit inside the above set V, that is, this “o” is not an element of set V.
Thus, this “o” is not a zero vector we are looking for.
• Is there any “zero vector” that belong to set V and that satisfy “zero vector+u =u” ? No.
• Based on the above discussed reasoning, we conclude that the given set V is not a vector space.

Page 11 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Extra Example Part III: Extra Examples (EE 9 to EE 11) on “proving subspaces”

EE9: Prove, the intersection U∩W of subspaces U and V of a vector space ℝ3 is also a subspace of ℝ3 .
𝑈 ∩ 𝑊 = {𝒙 | 𝒙 ∈ 𝑈 𝑎𝑛𝑑 𝒙 ∈ 𝑊
I. Since U and V are subspaces of ℝ3 , the zero vector 𝟎 = (0,0,0) lies in each of them. Therefore,
𝟎 = (0,0,0) ∈ 𝑈 ∩ 𝑊

II. Let
𝒗𝟏 = (𝑎, 𝑏, 𝑐) ∈ 𝑈 ∩ 𝑊, 𝒗𝟐 = (𝑝, 𝑞, 𝑟) ∈ 𝑈 ∩ 𝑊
Then,
𝒗𝟏 = (𝑎, 𝑏, 𝑐) ∈ 𝑈 𝒗𝟏 = (𝑎, 𝑏, 𝑐) ∈ 𝑈 𝒗𝟏 = (𝑎, 𝑏, 𝑐) ∈ 𝑊 𝒗𝟐 = (𝑝, 𝑞, 𝑟) ∈ 𝑊
Since is U is subspace, by the definition of subspace Since is W is subspace, by the definition of subspace
𝒗𝟏 + 𝒗𝟐 = (𝑎 + 𝑝, 𝑏 + 𝑞, 𝑐 + 𝑟) ∈ 𝑈 𝒗𝟏 + 𝒗𝟐 = (𝑎 + 𝑝, 𝑏 + 𝑞, 𝑐 + 𝑟) ∈ 𝑊
Thus, we get
𝒗𝟏 + 𝒗𝟐 = (𝑎 + 𝑝, 𝑏 + 𝑞, 𝑐 + 𝑟) ∈ 𝑈 ∩ 𝑊

III. Let
𝒗 = (𝑎, 𝑏, 𝑐) ∈ 𝑈 ∩ 𝑊, 𝒂𝒏𝒅 𝒍𝒆𝒕 𝑘 𝒃𝒆 𝒂 𝒔𝒄𝒂𝒍𝒂𝒓
Then,
𝒗 = (𝑎, 𝑏, 𝑐) ∈ 𝑈 𝒗 = (𝑎, 𝑏, 𝑐) ∈ 𝑊
Since is U is subspace, by the definition of subspace Since is W is subspace, by the definition of subspace
𝑘𝒗 = (𝑘𝑎, 𝑘𝑏, 𝑘𝑐) ∈ 𝑈 𝑘𝒗 = (𝑘𝑎, 𝑘𝑏, 𝑘𝑐) ∈ 𝑊
Thus, we get
𝑘𝒗 = (𝑘𝑎, 𝑘𝑏, 𝑘𝑐) ∈ 𝑈 ∩ 𝑊
This is we saw that “ 𝑈 ∩ 𝑊” satisfies all three conditions needed for a subspace definition.

EE 10: Show the subset U of vector space 𝑉 = ℝ3 = {(𝑥, 𝑦, 𝑧)|𝑥, 𝑦, 𝑧 ∈ ℝ } is a subspace of V.

𝑈 = {(𝑥, 𝑦, 𝑧)|𝑥, 𝑦, 𝑧 ∈ ℝ, 𝑥 =𝑦+𝑧, 𝑡ℎ𝑎𝑡 𝑖𝑠, 𝑥 − 𝑦 − 𝑧 = 0}
Answer: Let k be a scalar. Also, let 𝒖, 𝒗 ∈ 𝑈 so that
𝒖 = {(𝑎, 𝑏, 𝑐) such that 𝑎 − 𝑏 − 𝑐 = 0 𝒗 = {(𝑑, 𝑒, 𝑓) such that 𝑑 − 𝑒 − 𝑓 = 0
Then,
𝒖 + 𝒗 = (𝑎, 𝑏, 𝑐) + (𝑑, 𝑒, 𝑓) = (𝑎 + 𝑑, 𝑏 + 𝑒, 𝑐 + 𝑓) 𝑘𝒖 = 𝑘(𝑎, 𝑏, 𝑐) = (𝑘𝑎, 𝑘𝑏, 𝑘𝑐)
Then, Then,
(𝑎 + 𝑑) − ( 𝑏 + 𝑒) − (𝑐 + 𝑓) = (𝑎 − 𝑏 − 𝑐) + (𝑑 − 𝑒 − 𝑓) = 0 (𝑘𝑎) − ( 𝑘𝑏) − (𝑘𝑐) = 𝑘(𝑎 − 𝑏 − 𝑐) = 𝑘(0) = 0
It proves, “u+v” is an element of the set U. It proves, “𝑘u” is an element of the set U.

• Since the subset U of 𝑉 is closed under addition & scalar multiplication it is a subspace of 𝑉.

EE 11: Show the set of all linear combinations of 𝒗 = (1,0,1) and 𝒘 = (0,1,2) is a subspace of ℝ3
Answer: Let 𝑆 = {𝑎𝒗 + 𝑏𝒘} be the set of all linear combinations of 𝒗 = (1,0,1) and 𝒘 = (0,1,2).

𝑆 = {𝑎𝒗 + 𝑏𝒘} = {𝑎(1,0,1) + 𝑏(0,1,2)} = {(𝑎, 𝑏, 𝑎 + 2𝑏)}

• Let k be a scalar. Also, let 𝒖𝟏 , 𝒖𝟐 ∈ 𝑆. Then,

𝒖𝟏 + 𝒖𝟐 = (𝑎1 , 𝑏1 , 𝑎1 + 2𝑏1 ) + (𝑎2 , 𝑏2 , 𝑎2 + 2𝑏2 ) = ((𝑎1 + 𝑎2 ), (𝑏1 + 𝑏2 ), (𝑎1 + 𝑎2 ) + 2(𝑏1 + 𝑏2 ))

It proves, “𝒖𝟏 +𝒖𝟐 ” is an element of the set S.
Next,
𝑘𝒖𝟏 = 𝑘(𝑎1 , 𝑏1 , 𝑎1 + 2𝑏1 ) = (𝑘𝑎1 , 𝑘𝑏1 , (𝑘𝑎1 ) + 2(𝑘𝑏1 ))
It proves, “𝑘𝒖𝟏 ” is an element of the set S.
• Since the subset S of ℝ is closed under addition & scalar multiplication it is a subspace of ℝ3 .
3

Page 12 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Extra Example Part IV: Extra Examples (EE 12 to EE 14) on “proving not subspaces”

EE 12: Show the subset W of vectors of the form 𝑊 = {(𝑥, 𝑥, 𝑥 + 3)|𝑥 ∈ ℝ} is not a subspace of ℝ3
by showing that this set does not contain the zero vector.
Answer: Suppose, this set W contains a zero vector. At the end, we expect a contradiction. Let’s
see.

• If 𝟎 = (0,0,0) is a vector in “W” then, there should be an “x” value for which a member (𝑥, 𝑥, 𝑥 + 3)
of W has to equal to (0,0,0). In such a situation, we get:
(𝑥, 𝑥, 𝑥 + 3) = (0,0,0) ⟹ 𝑥 = 0, and , 𝑥 = −3
We just saw that the system of equations has no solution. Therefore (0,0,0) is not an element of
W.
Therefore, W is not a subspace of ℝ3 .
• Subset W of 𝑉 is not closed under addition & scalar multiplication so it is not a subspace of 𝑉.

EE 13: Show the 𝑊 = {(𝑥, 𝑦, 0)|𝑥 ∈ ℝ, 𝑥 ≠ 𝑦 𝑢𝑛𝑙𝑒𝑠𝑠 𝑥 = 𝑦 = 0} is not a subspace of ℝ3

by showing that W is not closed under addition.
Answer:
• We know, 𝒖 = (2, −2,0) is a vector in W. We know, 𝒗 = (−1,3,0) is a vector in W.
But,
𝒖+𝒗= (2, −2,0) + (−1,3,0) = (1,1,0)
shows
𝒖 + 𝒗 is not a vector in W (because 1st and 2nd components of W should not equal unless both are
0.)
• Subset W of 𝑉 is not closed under addition so it is not a subspace of 𝑉.

EE 14: Show the 𝑊 = {(𝑥, 0,0)|𝑥 ∈ ℝ, 𝑥 𝑖𝑠 𝑛𝑜𝑛𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 } is not a subspace of ℝ3

by showing that W is not closed under scalar multiplication.
Answer:
• We know, 𝒖 = (2,0,0) is a vector in W. Let “k” be a negative number. That is, let “k=-3”. Then,
𝑘𝒖 = (−3)(2,0,0) = (−6,0,0)
shows
𝑘𝒖 is not a vector in W (because the 1st component of “𝑊” should not be a negative number).
• Subset W of 𝑉 is not closed under scalar multiplication.

Extra Example Part V: Extra Examples (EE 15 to EE 16) on “column spaces”

EE 15: Construct a matrix B whose column space consists of all vectors (x, y, z) where z=4x-y.
1 0
𝑪(𝐴) = {( 𝑥, 𝑦, 𝑧) ∶ 𝑥, 𝑦, 𝑧 ∈ ℝ, 𝑧 = 4𝑥 − 𝑦} = {( 𝑥, 𝑦, 4𝑥 − 𝑦) } = {𝑥( 1,0,4) + 𝑦(0,1 − 1)} ⟹ {𝑥 [0] + 𝑦 [ 3 ]}
4 −1
1 0
𝐵 = [0 3 ]
4 −1

EE 16: Construct a 3 by 3 matrix B whose column space consists of a line.

1 3 0
𝐵 = [3 9 0]
2 6 0
It contains the vectors (c, 3c,0) along a line.

Page 13 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

Extra Example Part VI: Extra Examples (EE 17 to EE 18) on “spanning sets”

EE 17: Show that {(2,2,2)𝑇 , (0, 0,3)𝑇 , (0,1,1)𝑇 } is the spanning set for ℝ3 .
Note: Our book expresses (2,2,2)𝑇 , (0, 0,3)𝑇 , (0,1,1)𝑇 as (2,2,2), (0,0,3), (0,0,1). Both notations are okay.
We show, it is possible to find constants 𝑐1 , 𝑐2 , & 𝑐3 so that any arbitrary vector v=(𝑎, 𝑏, 𝑐)𝑇 in ℝ3 can
be expressed as the linear combination
𝑐1 𝒗𝟏 + 𝑐2 𝒗𝟐 + 𝑐3 𝒗𝟑 = 𝒗
So that the following holds true.
2 0 0 𝑎
𝑐1 [2] + 𝑐2 [0] + 𝑐3 [1] = [𝑏 ] 𝐸𝑞. 1
2 3 1 𝑐
Express it as augmented matrix & apply the row elimination to reduced it to upper triangular form
2 0 0 𝑎 2 0 0 𝑎 𝑎 𝑎 2𝑐1 = 𝑎
𝑅 ⟷ 𝑅3 𝑅 : 𝑅 − 𝑅1 2 0 0 (𝑐 − 𝑎) 𝑅 : 𝑅 − 𝑅3
2 0 0
[2 0 1| 𝑏 ] ≈ 2 [2 3 1| 𝑐 ] ≈ 2 2 [0 3 1| ]≈ 2 2 [0 3 0| (𝑐 − 𝑏) ] ⟹ 3𝑐2 = (𝑐 − 𝑏)
𝑅3 : 𝑅3 − 𝑅1
2 3 1 𝑐 2 0 1 𝑏 0 0 1 (𝑏 − 𝑎) 0 0 1 (𝑏 − 𝑎) 𝑐3 = (𝑏 − 𝑎)
We got three pivots for the in upper triangular form of 3 by 3 matrix. That is a good news !
Solving the back substitution, we get the following solutions for every (a, b, c) in ℝ3 .
𝑎 1 𝑐3 = (𝑏 − 𝑎)
𝑐1 = 𝑐2 = (𝑐 − 𝑏)
2 3
When the above calculated 𝑐1 , 𝑐2 , & 𝑐3 are put in Eq.1, the Eq.1 becomes satisfied as shown below:
0 (𝑎 + 0 + 0 𝑎
𝑎 2 1 0 1
( ) [2] + (𝑐 − 𝑏) [0] + (𝑏 − 𝑎) [1] = [ 𝑎 + 0 + (𝑏 − 𝑎) ] = [𝑏]
2 3 6 𝑐
2 3 1 𝑎 + (𝑐 − 𝑏) + (𝑏 − 𝑎)
Hence, Using just found constants 𝑐1 , 𝑐2 , & 𝑐3 , any vector 𝒗 = (𝑎, 𝑏, 𝑐) in ℝ can be expressed as linear
𝑇 3

combination of 𝑣1 = (2,2,2)𝑇 , 𝑣2 = (0,0,3)𝑇 , & 𝑣3 = (0,1,1)𝑇 . Thus, the given sets are spanning sets for ℝ3 .

EE 18: Show that {(2, −1,3)𝑇 , (4,1,2)𝑇 , (8, −1,8)𝑇 } is not the spanning set for ℝ3 .
Note: Our book expresses (2, −1,3)𝑇 , (4,1,2)𝑇 , (8, −1,8)𝑇 as (2, −1,3), (4,1,2), (8, −1,8). Both notations are okay.
Consider
2 4 8 𝑎 The corresponding augmented matrix is: 2 4 8 𝑎
𝑐1 [−1] + 𝑐2 [1] + 𝑐3 [−1] = [𝑏 ] 𝐸𝑞. 1 [−1 1 −1| 𝑏 ]
3 2 8 𝑐 3 2 8 𝑐
Apply the row elimination to reduced it to upper triangular form
𝑎 𝑎
𝑎 2 2
2 4 8 𝑎 𝑅1 1 2 4
𝑅2 : 𝑅2 + 𝑅1 1 2 4 𝑎 4 1 2 4 𝑎
[−1 1 −1| 𝑏 ] ≈ 𝑅 1 : 2 𝑏 + 𝑅3 : 𝑅3 + ( )𝑅2 𝑏 +
2 [−1 1 −1| 𝑏] ≈ 𝑅 : 𝑅 − 3𝑅 0 3 3|
2
≈ 3 0 3 3|
2
3 3 1
3 2 8 𝑐 3 2 8 0 −4 −4 3𝑎 0 0 0 −5𝑎 + 8𝑏 + 6𝑐
𝑐
[ 𝑐− ] [ ]
2 6
The system has no solution if −5𝑎 + 8𝑏 + 6𝑐
≠0
6
We got a restriction! We will not be able to express those type of vectors v=(𝑎, 𝑏, 𝑐)𝑇 in ℝ3 as the
linear combination of 𝒗𝟏 , 𝒗𝟐 , and 𝒗𝟑 for which (−5𝑎 + 8𝑏 + 6𝑐) ≠ 0. Hence, the given vectors of 𝒗𝟏 , 𝒗𝟐 ,
and 𝒗𝟏 = (2, −1,3), 𝒗𝟐 = (4,1,2), 𝒗𝟑 = (8, 1, −8) do not span ℝ3 .
Alternatively, note that 2(2,-1,3)+1(41,2)=(8,-1,8) so this set of 3 vectors is linearly dependent. Also
it means the 3rd column vector (8, -1,8) belongs to the span of the other two column vectors (2,-
1,3) and (4,1,2). Hence, just including the vector (8,-1,8) does not make the span any larger.
Another alternative way to show this is that the determinant of the 3 by 3 matrix A of these 3
column vectors is zero. Therefore, these 3 vectors can not span ℝ3 . However, since we are going to
discuss the determinant until we reach chapter 5, I recommend you not so answer this type of
problem from the perspective of determinant. [Note: By the way, if the determinant were zero, then,
these 3 vectors would span ℝ3 . If such happens then the matrix A for the system Ax=b becomes invertible.].

Page 14 of 15
 Lecture Note 11 (Sec 3.1) Space of Vectors

The last two Extra Example

EE 19: Show that the lines through the origin (0,0) are the subspaces of ℝ2 .
• We will show this via a simple example with graph.
• We know of ℝ2 is a vector space.
• Consider a line “W”, passing through the origin (0,0), and represented by the equation y=2x.
• We will show this line “W” is a subspace of ℝ2 .

• Note that line “W” is a nonempty set because the zero vector 0=(0,0) belongs to it.
• Consider the two vectors u=(1,2) and v=(-2,-4) that belong to this line “W” represented by y=2x.
Then,
u+v=(1,2)+(-2,-4)=(1-2, 2-4)=(-1,-2)
shows that “W is closed under addition”
because the result vector (-1,-2) belongs to W as well. u
3u=3(1,2)=(3,6)
shows that “W is closed under scalar multiplication”
because the result vector (3,6) belongs to W as well.
v

o Notes:
I. We can show that a line “W” through the origin (0,0,0) is a subspace of ℝ3 .
II. We can show it geometrically that a plane through the origin (0,0,0) is a subspace of ℝ3 .

E 20: Find the condition on the components 𝑏1 , 𝑏2 , & 𝑏3 of vector b so that Ax = b is solvable.
𝑏1 1 3 2
𝒃 = [𝑏2 ] 𝐴 = [ 2 6 4]
𝑏3 −1 −3 −2
• We know, “Ax = b is solvable if and only if the vector b belongs to column space of matrix A”.

• Suppose b belongs to column space of A. Then, Ax=b is solvable. Our Job: Find components of b.
• Let’s see what type of “b” will make the system “Ax=b” solvable?
• For that, we will find the reduced row echelon form of the augmented matrix [𝐴|𝒃] of Ax = b as
1 3 2 𝑏1 ≈ 1 3 2 𝑏1
[3 9 12 | 𝑏2 ] 𝑅2 = 𝑅3 − 3𝑅1 [0 0 0| 𝑏2 − 3𝑏1 ]
−2 −6 −4 𝑏3 𝑅3 = 𝑅3 + 2𝑅1 0 0 0 𝑏3 + 2𝑏1
We get
𝑥1 + 3𝑥2 + 2𝑥3 = 𝑏1 0 = 𝑏2 − 3𝑏1 0 = 𝑏3 + 2𝑏1
𝑬𝒒. 𝟏 𝑬𝒒. 𝟐 𝑬𝒒. 𝟑
• From Eq. 2 and Eq. 3, we see, the solution is possible only when 𝑏2 = 3𝑏1 and 𝑏3 = −2𝑏1 .
• This is the condition on b.

Page 15 of 15