Vector Spaces and Subspaces

Subspaces and Vector Spaces.

What are these?

There are two operations we can perform with vectors.

1. We can add them together.

2. We can multiply them by scalars.

To have "spaces" or collections of vectors, we should be able to continue to do

these operations - we should be able to form linear combinations of our vectors
in a space. Also, there should be some reasonable rules satisfied by the
elements in our space.
Examples of vector spaces:

ℝ 2 - the standard xy- plane is the set of all 2-dimensional real vectors.
 x
It is a vector space because it contains all possible vectors  y  including the
→
zero vector, 0 0 .
0


Suppose we have ℝ 2 but without the zero vector. Do we still have a space?

NO! We have to be able to multiply any vector in ℝ 2 by any scalar, including 0.

0
If   is missing we cannot multiply by the scalar 0.
0


Also we need to be able to add vector  x to its negative x . If the origin
y y
is missing we can not perform this
addition.
So every vector space must at least contain the zero
vector.

Other easy examples:

ℝ 3 which is all 3-dimensional vectors   xy .
 z

 x1 
 
 x2
ℝ n which is all n-dimensional vectors  
 ⁝ 
. x 
n


In all these spaces ℝ 2 , ℝ 3 , … , ℝ n we can add vectors and we stay
in the space. We can also multiply by scalars and we stay in the
space. That is, we can form linear combinations of vectors and still
stay in the space.

These vector additions and scalar multiplications obey the standard

rules, there are 8 of them, that addition and multiplication must satisfy.

What about ℝ 0 , and ℝ 1 ... are these vector spaces?

What are some examples which are not vector spaces?

Suppose we take just the first quadrant in ℝ 2 we then have all vectors   x

y
such that x  0 and y  0. Does Q1 form a vector
space?

We can add vectors and we stay in Q1.

What about multiplication by scalars?

 x 
Here there is a problem. Suppose we want to form (1)  x . We get
y   y  
but this vector does not lie in Q1. So we can not perform
scalar
multiplication. Q1 is not a vector space - it is not closed
under scalar multiplication.
 x
Suppose we take Q1 and Q 2 in ℝ 2 we then have all vectors
  y 
such that y  0. Do we have a vector
space?

We can add vectors and we stay in Q1 Q

2. What about multiplication by scalars?

Here there is again a problem. Suppose we want to form (1)  x .
y

We get x  but this vector does not lie in Q1 Q 2. So we can
 y 
not
perform scalar multiplication.

Q1 Q 2 is not a vector space - it is not closed under scalar

multiplication.
So to have a vector space we require that vector addition and
scalar multiplication both be closed operations. Then we can
form all linear combinations of the vectors in the space.

ℝ n gives us the most important vector spaces.

We are also interested in vector spaces inside ℝ n and these are

called subspaces of ℝ n .
Consider ℝ 2 again. We want to construct a vector space inside ℝ 2 .
 
It will be a subspace of ℝ 2 . Suppose the vector vu  is a vector in our
subspace.

u
 
v

If we are to construct a vector space we need all vectors

k  u , k any scalar, to be in the space and they clearly are.

 
v 
So our subspace is closed under scalar multiplication.

u
 
v


u 
  
v

Also, the sum of any vector on the line and any other vector on the
line is also on the line. So vector addition is closed - our
subspace, which is a vector space, is the line containing our original
vector and passing through the origin.

u u
   
v  v

Not every line in ℝ 2 is a subspace of ℝ 2 . Consider the line below.

Not a subspace since the line does not include the zero vector.
Here are all the subspaces of
ℝ2 :
1. ℝ 2 itself is a subspace - the largest
one.
0
2. Any line passing through   .
0


0 
3. The vector   alone.
0


Here are all the subspaces of ℝ 3 :
1. ℝ 3
itself.
0 
 
2. Any plane passing through  0 .
0

0 
 
3. Any line passing through  0 .
0

0 
 
4. The vector 0 alone.
0
You might wonder about the combination of a line and a

3 0
plane in ℝ . Suppose P is any plane passing through  0 and
0

0
 
L is any line passing through 0 . Form the union of the two,
0
written P  L.
a) Is P  L a subspace of ℝ3 ?

Now form the intersection of the two, written P 

L.
b) Is P  L a subspace of ℝ3 ?
a) Is P  L a subspace of ℝ3 ?

No. Pick a vector in P and a vector on L and add

them. In general, the sum will not be in P  L.

b) Is P  L a subspace of ℝ3 ?

Yes. It contains only the zero vector which is a

subspace.
In general, suppose V and W are subspaces.

Is V W a subspace?
→ →
Yes. Suppose x and y are in V W . (Note V W  ).

Then x→  y→V W since V and W are both subspaces.

→ →
Also, for scalar k, k x V W and 0 V W for the same
reason.
 A vector space (V,F, +, .)
• F a field
• V a set (of objects called vectors)
• Addition of vectors (commutative,
associative)

• Scalar multiplications
 Examples

– Other laws are easy to show

– This is just written differently

• The space of functions: A a set, F a field

– If A is finite, this is just F|A|. Otherwise this is infinite dimensional.

• The space of polynomial functions

• The following are different.

 Subspaces
• V a vector space of a field F. A subspace W
of V is a subset W s.t. restricted operations of vector
addition, scalar multiplication make W into a vector
space.
– +:WxW -> W, :FxW -> W.
– W nonempty subset of V is a vector subspace iff
for each pair of vectors a,b in W, and c in F, ca+b is in W.
(iff for all a,b in W, c, d in F, ca+db is in W.)
• Example:
•
is a vector subspace with field F.
• Solution spaces: Given an mxn matrix A

– Example x+y+z=0 in R3. x+y+z=1 (no)

• The intersection of a collection of vector
subspaces is a vector subspace
• is not.
 Span(S)

• Theorem 3. W= Span(S) is a vector subspace and is

the set of all linear combinations of vectors in S.
• Proof:
• Sum of subsets S1, S2, …,Sk of V

• If Si are all subspaces of V, then the

above is a subspace.
• Example: y=x+z subspace:

• Row space of A: the span of row

vectors of A.
• Column space of A: the space of
column vectors of A.
 Linear independence
• A subset S of V is linearly dependent if

• A set which is not linearly dependent is called

linearly independent:
The negation of the above statement
 Basis
• A basis of V is a linearly independent set of
vectors in V which spans V.
• Example: Fn the standard basis

• V is finite dimensional if there is a finite basis.

Dimension of V is the number of elements of
a basis. (Independent of the choice of basis.)
• A proper subspace W of V has dim W < dim
V. (to be proved)
• Example: P invertible nxn matrix. P1,…,Pn
columns form a basis of Fnx1.
– Independence: x1P1+…+xnPn=0, PX=0.
Thus X=0.
– Span Fnx1: Y in Fnx1. Let X = P-1Y. Then Y = PX. Y=
x1P1+…+xnPn.
• Solution space of AX=0. Change to RX=0.

– Basis Ej uj=1, other uk=0 and solve above

 – Thus the dimension is n-r:

• Infinite dimensional example:

• V:={f| f(x) = c0+c1x+c2x2 + …+ cnxn}.
– Given any finite collection g1,…,gn there is
a maximum degree k. Then any polynomial
of degree larger than k can not be written
as a linear combination.
• Theorem 4: V is spanned by
Then any independent set of vectors in V is
finite and number is  m.
– Proof: To prove, we show every set S with more than m
vectors is linearly dependent. Let
be elements of S with n > m.

– A is mxn matrix. Theorem 6, Ch 1, we can solve for x1,x2,

…,xn not all zero for

– Thus
• Corollary. V is a finite d.v.s. Any two bases
have the same number of elements.
– Proof: B,B’ basis. Then |B’||B| and |B||B’|.
• This defines dimension.
– dim Fn=n. dim Fmxn=mn.
• Lemma. S a linearly independent subset of V.
Suppose that b is a vector not in the span of
S. Then S{b} is independent.
– Proof:
Then k=0. Otherwise b is in the span.
Thus,
and ciare all zero.
• Theorem 5. If W is a subspace of V, every
linearly independent subset of W is finite and
is a part of a basis of W.
• W a subspace of V. dim W  dim V.
• A set of linearly independent vectors can be
extended to a basis.
• A nxn-matrix. Rows (respectively columns) of
A are independent iff A is invertible.
(->) Rows of A are independent. Dim Rows A = n. Dim Rows
r.r.e R of A =n. R is I -> A is inv.
(<-) A=B.R. for r.r.e form R. B is inv. AB-1 is inv. R is inv. R=I.
Rows of R are independent. Dim Span R = n. Dim Span A = n.
Rows of A are independent.
• Theorem 6.
dim (W1+W2) = dim W1+dimW2-dimW1W2.
• Proof:
– W1W2 has basis a1,…,ak. W1 has basis a1,..,ak,b1,…,bm. W2
has basis a1,..,ak,c1,…,cn.
– W1+W2 is spanned by a1,..,ak,b1,…,bm ,c1,…,cn.
– There are also independent.
• Suppose

• Then

• By independence zk=0. xi=0,yj=0 also.

 Coordinates
• Given a vector in a vector space, how
does one name it? Think of charting
earth.
• If we are given Fn, this is easy? What
about others?
• We use ordered basis:
One can write any vector uniquely
• Thus,we name

Coordinate (nx1)-matrix (n-tuple) of a vector.

For standard basis in Fn, coordinate and vector are the same.
• This sets up a one-to-one correspondence
between V and Fn.
– Given a vector, there is unique n-tuple of
coordinates.
– Given an n-tuple of coordinates, there is a unique
vector with that coordinates.
– These are verified by the properties of the notion
of bases. (See page 50)
 Coordinate change?
• If we choose different basis, what
happens to the coordinates?
• Given two bases
– Write
• X=0 iff X’=0 Theorem 7,Ch1, P is invertible

• Thus, X = PX’, X’=P-1X.

• Example {(1,0),(0,1)}, {(1,i), (i,1)}

– (1,i) = (1,0)+i(0,1)
(i,1) = i(1,0)+(0,1)
– (a,b)=a(1,0)+b(1,0): (a,b)B =(a,b)
– (a,b)B’ = P-1(a,b) = ((a-ib)/2,(-ia+b)/2).
– We check that (a-ib)/2x(1,i)+
(-ia+b)/2x(i,1)=(a,b).
End
presentation