Linear Algebra and Differential Equations

Sartaj Ul Hasan

Department of Mathematics
Indian Institute of Technology Jammu
Jammu, India - 181221

Email: sartaj.hasan@iitjammu.ac.in

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 0 / 15

 Lecture 09
(March 24, 2021)

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 1 / 15

Subspaces (Conti . . . )
Test for Subspaces
Proposition 8: A subset W of V is a subspace if and only if it
satisfies the following three properties:
1 The zero vector 0 is in W .
2 W is closed under addition. That is, for each u and v in W , the sum
u + v is in W .
3 W is closed under scalar multiplication. That is, for each u in W , and
each scalar c, the scalar product cu is in W .

Note: In some books, this is treated as the definition of a subspace.

0
It is also possible to replace the condition 1 above by the condition 1
that W be non-empty. However, in practice, this is not so easy to use.
Proposition 9: A non-empty subset W of V is a subspace if and only
if for each u and v in W , and each scalar c, the sum cu + v is in W .
Remark: It is left as an exercise to show that the two tests are
equivalent. In some books, the above is taken as the definition of a
subspace. We may use either of the tests whichever is convenient.
Sartaj Ul Hasan (IIT Jammu) SMD002U1M 2 / 15
Subspaces (Conti . . . )

Some More Examples of Subspaces

For any vector space V , the subset consisting of the zero vector alone
is a subspace of V , called the zero subspace.
V is of course a subspace of itself. Subspaces other than V and {0}
are known as proper subspaces.
The set X = {(x, y ) ∈ R2 : x + y = 0} is a subspace of R2 –verify:
exercise.
The set Symn of all symmetric (square) matrices of size n is a
subspace of the space Rn×n of square n × n matrices.

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 3 / 15

Subspaces (Conti . . . )
But . . .

Something to think about: Is R2 is a subspace of R3 ?

Answer

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 4 / 15

Subspaces (Conti . . . )
But . . .

Something to think about: Is R2 is a subspace of R3 ?

Answer
However, R2 is not a subspace of R3 ! This is because R2 is not even a
subset of R3 . The set
  
 x 
W = y  : x, y ∈ R
0
 

is a subspace of R3 , which behaves much like R2 , but is logically distinct

from R2 .
Sartaj Ul Hasan (IIT Jammu) SMD002U1M 4 / 15
Span of a Set

Span of a Set of Vectors

Definition 1: A linear combination of finitely many given vectors is
any sum of scalar multiples of the vectors.
Definition 2: Let S = {v1 , v2 , . . . , , vp } be a finite set of vectors in a
vector space V . Then the Span of S is the set of all vectors that can
be written as linear combinations of the vectors v1 , v2 , . . . , vp .
Symbolically, Span S = {c1 v1 + c2 v2 + · · · + cp vp : ci ∈ F}, where F
is the underlying field. In our case, usually F = R.

Subspace Spanned by a Set

Proposition 10: If S = {v1 , v2 , . . . , , vp } is a set of vectors in a vector
space V , then Span S =Span {v1 , v2 , . . . , , vp } is a subspace of V .

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 5 / 15

Span of a Set(Conti . . . )

Proof of Proposition 10: Let us use the first test for subspaces (Prop 8):
The zero vector 0 = 0v1 + 0v2 + · · · + 0vp is a linear combination of the
v ’s. If w1 = c1 v1 + c2 v2 + · · · + cp vp and w2 = d1 v1 + d2 v2 + · · · + dp vp
are two linear combinations, then so is
w1 + w2 = (c1 v1 + c2 v2 + · · · + cp vp ) + (d1 v1 + d2 v2 + · · · + dp vp )
= (c1 + d1 )v1 + (c2 + d2 )v2 + · · · + (cp + dp )vp
(Note: we have here used some of the axioms of vector space without
specifically mentioning them)
If c is any scalar, and w1 is a linear combination as above, then
cw1 = c(c1 v1 + c2 v2 + · · · + cp vp ) = cc1 v1 + cc2 v2 + · · · + ccp vp is again
a linear combination.

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 6 / 15

Span of a Set(Conti . . . )
Corollary10.1: Let V be a vector space.
(a) If U and W are two subspaces of V , then U ∩ W (i.e. the
intersection of U and W ) is also a subspace of V .
(b) If S = {v1 , v2 , . . . , vp } is a set of vectors in a vector space V , then
Span S = Span {v1 , v2 , . . . , vp } is the smallest subspace which
contains S, i.e. if W is a subspace such that S ⊆ W , then
Span S ⊆ W .
Remark 1: Proof is left as an exercise (must do! )
Remark 2: In terms of this, Span S is sometimes described as the
intersection of all subspaces of V containing S. (Also left as an exercise. )
Remark 3: Show by means of an example that union of two subspaces of
a vector space need not be a subspace (Hint: Take two line passing
through the origin in R2 ).
In fact, if If U and W are two subspaces of V , then U ∪ W is subspace of
V ⇐⇒ either U ⊆ W or W ⊆ U. (Left as an exercise!)
Sartaj Ul Hasan (IIT Jammu) SMD002U1M 7 / 15
Vector Formulation

A system of linear equations can also be expressed in a vector form:

X1 v1 + X2 v2 + · · · + Xn vn = b, where the Xi are scalar unknowns and
the vi are column vectors formed from the coefficients of the original
linear system. The vectors vi are the columns of the coefficient matrix
A and so we can write A = [v1 v2 . . . vn ].
This formulation can be interpreted as: if we can find scalars Xi
satisfying the equation, then the given vector b can be expressed in
terms of the given vectors vi . In terms of the concept of the span of a
set of vectors, we can say that the non-homogeneous system Ax = b
has a solution if and only if the vector b ∈ Span {v1 , v2 , . . . vn },
where the vi are the columns of A. This formulation is not useful for
solving the system, but becomes useful when we are discussing the
subspaces associated with a given matrix A.

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 8 / 15

Span of a Set (Conti . . . )

Example:
   
1 1
Let S = {u, v}, where u = 3 and v = 1 .
2 4
Note that u ∈ Span S, v ∈ Span S and S ⊆ Span S.
   
2 1
u + v = 4 ∈ Span S, 2u + (−1)v = 5 ∈ Span S.
  
6 0
Clearly, while S is finite, Span S is infinite (assuming that the field F
is infinite).
By the way, 0 = 0u + 0v ∈ Span S.

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 9 / 15

Span of a Set (Conti . . . )
Example (Conti . . . ):
Note that constructing vector in Span S is easy.
What about the reverse question: given a vector w, does w ∈ Span
S?
If w ∈ Span S, then w = c1 v1 + c2 v2 + · · · + cp vp for some scalars ci .
So we have to solve a linear system!
   
 1 1 
As before, let S = 3 , 1
2 4
 
       
3 1 1 3
Put w = 2 → solve c1 3 + c2 1 = 2 
      
13 2 4 13
   
1 1 : 3 1 0 : −1/2
3 1 : 2  ∼ 0 1 : 7/2 .
2 4 : 13 0 0 : 0

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 10 / 15

Span of a Set (Conti . . . )

Example (Conti . . . ):
     
3 1 1
So YES → w =  2  = (−1/2) 3 + (7/2) 1.
13 2 4
 
−3
On the other hand (OTOH), consider w1 = −2.

7
   
1 1 : −3 1 0 : −3
Then, 3 1 : −2 ∼ 0 −2 : 7  → inconsistent!
  
2 4 : 7 0 0 : 20
So w1 ∈
/ Span S.

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 11 / 15

Linear Dependence

Definition 1: Let v1 , v2 , . . . , vp be a finite list of vectors in a vector

space V . Then the vectors are said to be (linearly) dependent if there
exist scalars c1 , c2 , . . . , cp , not all zero, such that
c1 v1 + c2 v2 + · · · + cp vp = 0.
Definition 2: If a list of vectors is not linearly dependent, it is said to
be (linearly) independent. In other words, if the list is linearly
independent, and c1 v1 + c2 v2 + · · · + cp vp = 0, then all the ci0 s must
be 0.
We usually simply say v1 , v2 , . . . , vp are dependent or independent.
We have given the definition in terms of list of vectors rather than set
of vectors, because list is a more general concept in this situation. A
finite set can easily be considered as a list and then the above
definition can be applied to it.

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 12 / 15

Linear Dependence (Conti . . . )
Consequences of the Definitions
Remark 1: Any list which contains the 0 vector has to be linearly
dependent. In fact, the single zero vector 0 is always linearly
dependent (LD).
Proof: Let v1 , . . . , vp be a list of vectors, and suppose vk = 0 for
some k, k = 1, 2, . . . , p. Put ck = 1 and ci = 0 for i 6= k. Then:
c1 v1 +· · ·+cp vp = 0v1 +· · ·+0vk−1 +1.vk +0Vk+1 +· · ·+0vp = 1.0 = 0.
Since not all the ci ’s are 0, in fact ck = 1, the list of vectors is LD as
required.

Moreover the single zero vector 0 is LD because 1.0=0.

Remark 2: A single non-zero vector is linearly independent.

[Hint: For a ∈ F and v ∈ V , av = 0 =⇒ a = 0 or v = 0.]

Remark 3: A list of two non-zero vectors is linearly dependent only if
one of the vectors is a scalar multiple of the other.

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 13 / 15

Linear Dependence (Conti . . . )
Consequences of the Definitions
Remark 4: A list of non-zero vectors is linearly dependent if and only
if at least one of the vectors is a linear combination of the others.
Proof: ( =⇒ ) Given: A list of vectors, say v1 , . . . , vp , is LD. We
have to show that at least one of the vector is expressible as a linear
combination of others. Since v1 , . . . , vp are LD. Hence,

c1 v1 + · · · + cp vp = 0, (1)

where not all ci = 0. Suppose ck 6= 0. Rewrite 1 as:

ck vk = −c1 v1 − c2 v2 − · · · − cp vp , (2)

where the RHS of 2 contains all vectors except vk . Since ck 6= 0,

multiplying by ck−1 on both sides, we get:

1.vk = −(ck−1 c1 )v1 − (ck−1 c2 )v2 − · · · − (ck−1 cp )vp , (3)

i.e. vk is a linear combination of the rest.

Sartaj Ul Hasan (IIT Jammu) SMD002U1M 14 / 15
Linear Dependence (Conti . . . )
Proof of Remark 4 (Cont’d)
Proof: ( ⇐= ) Given that at least one vector in a list is expressible as
a linear combination of the rest. We have to show that the list is
linearly dependent. Suppose
vk = c1 v1 + · · · + cp vp , (4)
where the RHS of 4 contains all the vectors other than vk . Rewrite 4
as: c1 v1 + · · · − vk + · · · + cp vp = 0, we get that the coeff of vk (ck )
satisfies ck = −1 6= 0, the list is LD.
Remark 5: Consequently, any list which contains a repeated vector
must be linearly dependent. A list which is linearly independent
corresponds to a set.
Remark 6: Any list which contains a linearly dependent list is linearly
dependent or Superset of a linearly dependent set is linearly
dependent.
Remark 7: Any subset of a linearly independent set is linearly
independent . [By convention, null-set φ is LI ]
Sartaj Ul Hasan (IIT Jammu) SMD002U1M 15 / 15