Lecture 08 recap

1) Geometric vectors and Euclidean n  space.

2) Identifying a vector with a matrix.
n
3) Subsets of . Implicit and explicit representations.
4) Linear combination
5) Linear span (set of all linear combinations).
n
6) Determining whether a set spans .
Lecture 09
Linear combinations and
linear spans (cont’d)
Subspaces
 Discussion
Suppose S  { u1 , u2 ,..., uk }  n
. We want to determine
if span(S )  n
.
u1  (a11 , a12 ,..., a1n ) u2  (a21 , a22 ,..., a2 n ) ... uk  (ak1 , ak 2 ,..., akn )

For any v  (v1 , v2 ,..., vn )  n

, consider the equation:
c1u1  c2 u2  ...  ck uk  v

c1(a11 , a12 ,..., a1n )  c2 (a21 , a22 ,..., a2 n )  ...  ck (ak1 , ak 2 ,..., akn )
 (v1 , v2 ,..., vn )
 Discussion
Suppose S  { u1 , u2 ,..., uk }  n
. We want to determine
if span(S )  n
.
c1u1  c2 u2  ...  ck uk  v
c1(a11 , a12 ,..., a1n )  c2 (a21 , a22 ,..., a2 n )  ...  ck (ak1 , ak 2 ,..., akn )
 (v1 , v2 ,..., vn )

a11c1  a21c2  ...  ak1ck  v

a c  a c  ...  ak 2ck  v2
 12 1 22 2

: : : :
a1n c1  a2 n c2  ...  akn ck  vn
 Discussion
Suppose S  { u1 , u2 ,..., uk }  n
. We want to determine
if span(S )  n
.
c1u1  c2 u2  ...  ck uk  v (*)
c1(a11 , a12 ,..., a1n )  c2 (a21 , a22 ,..., a2 n )  ...  ck (ak1 , ak 2 ,..., akn )
 (v1 , v2 ,..., vn )
If a row-echelon form of A
k columns
does not have a zero row,
 a11 a21 ... ak1 v1 
n  a a ... a v  (*) is always consistent
 12 22 A k2 2

rows  : : : :  regardless of v
a a   span( S )  n
 1n 2 n ... akn vn 
 Discussion
Suppose S  { u1 , u2 ,..., uk }  n
. We want to determine
if span(S )  n
.
c1u1  c2 u2  ...  ck uk  v (*)
c1(a11 , a12 ,..., a1n )  c2 (a21 , a22 ,..., a2 n )  ...  ck (ak1 , ak 2 ,..., akn )
 (v1 , v2 ,..., vn )
If a row-echelon form of A
k columns
has at least one zero row,
 a11 a21 ... ak1 v1 
n  a a ... a v  (*) is not always consistent
 12 22 A k2 2

rows  : : : :   span(S )  n
 a0 0a ... ...0 a0 v 
 1n 2 n kn n
 Example
From earlier example:
Show that span{(1, 0,1),(1,1, 0),(0,1,1)}  3
.
1 1 0 1 1 0
0 1 1 0 1 1 No zero row
   
1 0 1 0 0 2
   
Show that span{(1,1,1),(1, 2, 0),(2,1,3),(2, 3,1)}  3.
1 1 2 2  1 1 2 2
1 2 1 3   0 1 1 1  Has zero row
   
1 0 3 1  0 0 0 0
   
 Theorem
Let S  { u1 , u2 ,..., uk } be a set of vectors in n
.
If k  n , then S cannot span n
.
If k  n, then a row-echelon form of A has at
least one zero row  S cannot span n .
Suppose S  { u1 , u2 ,..., uk }  n
. We want to determine
if span(S )  n
.
c1u1  c2 u2  ...  ck uk  v (*)
k columns
 a11 a21 ... ak1 v1  If a row-echelon form of A
n  a a ... a v2 
 12 22
A
k2
 has at least one zero row,
rows  : : : : 
 a0 0a ... ...0 a0 v 
 1n 2 n kn n (*) is not always consistent
 Example
2
1) One vector cannot span .
3
2) One or two vectors cannot span .
 Theorem
Let S  { u1 , u2 ,..., uk }  n
.

1) 0  span(S )

2) For any v1 , v2 ,..., vr  span(S ) and c1 , c2 ,..., cr  ,

c1v1  c2v2  ...  cr vr  span(S ).

 Theorem
Let S1  {u1 , u2 ,..., uk } and S 2  {v1 , v2 ,..., vm } be subsets of n
.
n n

Then span(S1)  span(S2 ) each ui is a linear combination of

v1 , v2 ,..., vm .
n
 Example
Let u1  (1, 0,1), u2  (1,1, 2), u3  (1, 2,1), v1(1, 2, 3), v2  (2, 1,1).
Show that span{ u1 , u2 , u3 }  span{v1 , v2 }.

Idea: The previous theorem gives us a

necessary and sufficient condition for
span{ u1 , u2 , u3 }  span{v1 , v2 }.

Each of u1 , u2 , u3 is a linear combination of v1 , v2 .

 Example
Let u1  (1, 0,1), u2  (1,1, 2), u3  (1, 2,1), v1(1, 2, 3), v2  (2, 1,1).
Show that span{ u1 , u2 , u3 }  span{v1 , v2 }.

Each of u1 , u2 , u3 is a linear combination of v1 , v2 .

 a(1, 2, 3)  b(2, 1,1)

 Example
 1 2 1 1 1 Gauss-Jordan  1 0 15 53 53 
 2 1 0 1 2   0 1 2 1 4 
   5 5 5 
 3 1 1 2 1  Elimination 0 0 0 0 0 
   

(1, 0,1)  (1, 2, 3)  (2, 1,1)

Since each of u1 , u2 , u3
(1,1, 2)  (1, 2, 3)  (2, 1,1) is a linear combination of v1 , v2 ,
span{ u1 , u2 , u3 }  span{v1 , v2 }.
(1, 2,1)  (1, 2, 3)  (2, 1,1)
 Example
Let u1  (1, 0,1), u2  (1,1, 2), u3  (1, 2,1), v1(1, 2, 3), v2  (2, 1,1).
Show that span{ u1 , u2 , u3 }  span{v1 , v2 }.

Can we show Shown:

span{ u1 , u2 , u3 }  span{v1 , v2 }? span{ u1 , u2 , u3 }  span{v1 , v2 }.
Each of v1 , v2 is a linear combination of u1 , u2 , u3.
 a(1, 0,1)  b(1,1, 2)  c(1, 2,1)
 Example
Let u1  (1, 0,1), u2  (1,1, 2), u3  (1, 2,1), v1(1, 2, 3), v2  (2, 1,1).

 1 1 1 1 2  Gauss-Jordan  1 0 3 1 3 
 0 1 2 2 1  0 1 2 2 1
   
 1 2 1 3 1  Elimination  
   0 0 0 0 0 

v1  a(1, 0,1)  b(1,1, 2)  c(1, 2,1)

 Example
Let u1  (1, 0,1), u2  (1,1, 2), u3  (1, 2,1), v1(1, 2, 3), v2  (2, 1,1).

 1 1 1 1 2  Gauss-Jordan  1 0 3 1 3 
 0 1 2 2 1  0 1 2 2 1
   
 1 2 1 3 1  Elimination  
   0 0 0 0 0 
Since each of v1 , v2 is a linear combination of u1 , u2 , u3 ,
span{ u1 , u2 , u3 }  span{v1 , v2 }.

Together with span{ u1 , u2 , u3 }  span{v1 , v2 }, we have shown

span{ u1 , u2 , u3 }  span{v1 , v2 }.
 Example
Let u1  (1,1, 0, 2), u2  (1, 0, 0,1), u3  (0,1, 0,1),
v1(1,1,1,1), v2  (1,1, 1,1), v3  (1,1,1, 1).
Show that span{u1 , u2 , u3 }  span{v1 , v2 , v3 } but
span{u1 , u2 , u3 }  span{v1 , v2 , v3 }.

To show span{u1 , u2 , u3 }  span{v1 , v2 , v3 }, How did this

1 1 matrix come
1 1 1 0
1 1 1 1 0 1 about?
 
1 1 1 0 0 0
1 1 1 
 1 2 1
 Example
Let u1  (1,1, 0, 2), u2  (1, 0, 0,1), u3  (0,1, 0,1),
v1(1,1,1,1), v2  (1,1, 1,1), v3  (1,1,1, 1).
Show that span{u1 , u2 , u3 }  span{v1 , v2 , v3 } but
span{u1 , u2 , u3 }  span{v1 , v2 , v3 }.

To show span{u1 , u2 , u3 }  span{v1 , v2 , v3 },

1 1 1 1 1 0  1 1 1 1 1 0
1 1 1 1 0 1 0 2 2 0 1 1
   
1 1 1 0 0 0 0 0 2 1 1 0
1 1 1  0 0 0 
 1 2 1  0 0 0
 Example
Let u1  (1,1, 0, 2), u2  (1, 0, 0,1), u3  (0,1, 0,1),
v1(1,1,1,1), v2  (1,1, 1,1), v3  (1,1,1, 1).
Show that span{u1 , u2 , u3 }  span{v1 , v2 , v3 } but
span{u1 , u2 , u3 }  span{v1 , v2 , v3 }.
To show span{u1 , u2 , u3 }  span{v1 , v2 , v3 },

v1  (1,1,1,1)  a(1,1, 0, 2)  b(1, 0, 0,1)  c(0,1, 0,1)

has no solution.
So v1 is not a linear combination of u1 , u2 , u3.
 Example
Let u1  (1, 0, 0,1), u2  (0,1, 1, 2), u3  (2,1, 1, 4),
v1(1,1,1,1), v2  (1,1, 1,1), v3  (1,1,1, 1).

Show that span{v1 , v2 , v3 }  span{u1 , u2 , u3 }.

We try to write each vi as a linear combination of u1 , u2 , u3 .

1 0 2 1 1 1 1 0 2 1 1 1
0 1 1 1 1 1 0 1 1 1 1 1
   
 0 1 1 1 1 1 0 0 0 2 0 2
1 2 1 0 0 
 4 1 1  0 0 0 0
Which vi is NOT a linear combination of u1 , u2 , u3 ?
 Theorem (‘useless vector’)
n
Suppose u1 , u2 ,..., uk are vectors taken from .

If uk is a linear combination of u1 , u2 ,..., uk1 , then

span{u1 , u2 ,..., uk 1}  span{u1 , u2 ,..., uk 1 , uk }

set of ALL linear

combinations of  set of ALL linear
combinations of
 Example
Let u1  (1, 0, 0), u2  (0, 1, 0), u3  (2, 3, 0).
Clearly, u3  2u1  3u2 . So span{u1 , u2 , u3 }  span{ u1 , u2 }.

Can you describe span{ u1 , u2 , u3 } geometrically?

z

z 0
u2  (0, 1, 0)
y
u1  (1, 0, 0)
x
 Definition (Subspaces)
n
Let V be a subset of . n

If there exists a set of vectors

S  {u1 , u2 ,..., uk } in n
V span(S ) VV
{ }
such that span(S )  V ,
then V is said to be Remember: span(S )  set of all
a subspace of n
. linear combinations of u1 , u2 ,..., uk .
 Definition (Subspaces)
n
Let V be a subset of . n

If there exists a set of vectors

S  {u1 , u2 ,..., uk } in n
V span(S )  V
{ }
such that span(S )  V ,
then V is said to be More precisely, we say:
a subspace of n
. 1) V is the subspace spanned by S .
2) V is the subspace spanned by u1 , u2 ,..., uk .
3) S spans V .
 Remark
n
{0} = span{0} is the zero subspace of .
n
(Here, 0 is the zero vector of .)
n
It is also called the zero space of .

It is also the only subspace of n with a finite number

(in this case, one), and thus the least, of vectors.
 Remark
Consider the vectors in n :
e1  (1, 0,...., 0) e2  (0,1,...., 0) .... en  (0, 0,....,1)
n
 {(u1 , u2 ,..., un )|u1 , u2 ,..., un  }
 {u1e1  u2e2  ...  un en |u1 , u2 ,..., un  }
 the set of all linear combinations of e1 , e2 ,..., en
 span{e1 , e2 ,..., en }  a subspace of n

n
So is a subspace of itself and to some extent, it can be
n
thought of as the subspace of with the 'largest' number
of vectors.
 Remark
In more advance textbooks on linear algebra,
'subspace' is defined differently. We will discuss this
briefly later in today's lecture.
 Example (subspaces)
V1  {(a  2b , 3b)| a , b  }. V1 is a subset of 2
.
2
Is V1 a subspace of ?

Idea: Can you try to write V1 as a linear span?

Is V1  2
?
2
2
So V1 is a subspace of . V1
2
Is there a vector in that is not in V1 ?
 Example (subspaces)
V1  {(a  2b , 3b)| a , b  }.
Is V1  2
?
2
Idea: Is every vector in a linear combination
of (1, 0) and (2, 3)?
 Example (subspaces)
V2  {( x , y , z)| x  3 y  2 z  0}. V2 is a subset of 3
.
3
Describe V2 geometrically. Is V2 a subspace of ?

Idea: V2 is given in implicit form. Can you find

its explicit form?
 Example (subspaces)
V3  {( x , y , z)| x  3 y  2 z  1}. V3 is a subset of 3
.
3
Describe V3 geometrically. Is V3 a subspace of ?

(0, 0, 0) V3 since 0  3(0)  2(0)  1.

If V3 is a subspace, then it can be expressed as a
linear span, that is, V3  span(S ) for some finite set S .
By (first) theorem in today's lecture, (0, 0, 0)  span(S )  V3 .
3
Contradiction! So V3 is not a subspace of .
 Example (subspaces)
V4  {( x , y , z)| x  y  z}. V4 is a subset of 3
.

Is V4 a subspace of 3
? No!
If V4 is a subspace, then it can be expressed as a
linear span, that is, V4  span(T ) for some finite set T .
By (first) theorem in today's lecture,
For any v1 , v2 ,..., vr  span(T )  V4 and c1 , c2 ,..., cr  ,
c1v1  c2v2  ...  cr vr  span(T )  V4 .

(1,1, 2),(0, 2, 4) V4 but (1,1, 2)  2(0, 2, 4)  (1, 3, 6) V4 .
 End of Lecture 09
Lecture 10:
Subspaces (cont’d)
Linear independence
Bases (till Example 3.5.5.1)