Linear Algebra I (2020 Fall)

Lecture 7.

Studying Contents : Chapter 1. Vector Spaces

- Section 1.6 Bases and Dimension

Section 1.6 Bases and Dimension

Recall : (1) Let  be a vector space and let  be a subspace of  .

Let  be a subset of  such that   span   . Then

No proper subset of  generates  ⇔  is linearly independent.

(2) Definition : Let  be a vector space over a field  .

We say that a subset B of  is a basis for  if

  B is linearly independent and    B generates  (i.e.,   span(B)).

(3) For    , the set B       …         …   ⋯      …     is a basis

called the standard basis. Note that we can write an element      …    ∈   as

    ⋯   .

Furthermore, the expression     ⋯    of  is unique.

Theorem 1.8. : Let  be a vector space over a field  and let B    …   be a subset of  .

Then    B    …  is a basis for 

if and only if

   each ∈  can be uniquely expressed as a linear combination of vectors of B    …  .

(That is,      ⋯     for unique scalars   …  ∈  .)

Theorem 1.9. : Let  be a vector space over a field  . If  is generated by a finite set  , then

some subset of  is a basis for  .

In particular,  has a finite basis.

Theorem 1.10 (Replacement Theorem) : Let  be a vector space

which is generated by a set  containing exactly  elements.

Let  be a subset of  which is linearly independent. Put    . Then

(1)     ≤   .

(2) There exists a subset  of  such that       and span  ∪     .

Corollary 1. : Let  be a vector space which has a finite basis, say B    …  .

Then every basis for  contains  vectors.

Proof : Let ℬ     …   and ℬ     …   be two bases for  .

  span ℬ   and ℬ  is linearly independent. ⇒  ≥  by Theorem 1.10.

  span ℬ   and ℬ  is linearly independent. ⇒  ≥  by Theorem 1.10.

⇒  ■
Definition : A vector space is called finite dimensional

if it has a basis consisting of a finite number of vectors.

The unique number of vectors in each basis for  is called the dimension of 

and is denoted by dim    .

A vector space that is not finite-dimensional is called infinite-dimensional.

Example 8. : In    , let

     …         …   ⋯      …    .

Then B   …  is a basis for  .

So, the vector space     has dimension .
Corollary 2. : Let  be a vector space of dimension . Let  ⊂  be a finite subset.

(1) If   span   , then  ≥ . Also

   if and only if  is a basis for  .

(2) If  is linearly independent, then  ≤ . Also

   if and only if  is a basis for  .

(3) If  is linearly independent, then it can be extended to a basis for  .

Proof : Let B    …  be a basis for  .

(1) Since  generates  and B is linearly independent, it holds by “Replacement Theorem” that

 ≥ |B|  .

 ⇐     by Corollary 1.

 ⇒  If    and  is not a basis for  , then it is linearly dependent since it spans  .

Then  can be spanned by    vectors.

Then Theorem 1.10 shows that    ≥ , a contradiction.

In consequence,  must be linearly independent and hence it is a basis for  .

Corollary 2. : Let  be a vector space of dimension . Let  ⊂  be a finite subset.

(2) If  is linearly independent, then  ≤ . Also

   if and only if  is a basis for  .

Proof : Since  is linearly independent and B generates  , it holds by “Replacement Theorem”

that

|B|   ≥  .

 ⇐  If  is a basis for  , then    by Corollary 1.

 ⇒  Suppose that    and  is not a basis for  .

Since  is linearly independent, it follows that

 fails to generate  and hence span   is a proper subset of  .

In particular, we can choose a vector  ∈   span   .
Then,  ∪ is linearly independent since  is linearly independent.

But this is impossible since  ∪        ℬ .

Corollary 2. : Let  be a vector space of dimension . Let  ⊂  be a finite subset.

(3) If  is linearly independent, then it can be extended to a basis for  .

Proof : For the basis B   …   for  , we have   span ℬ  .

Now, put   . Since  is a linearly independent, Theorem 1.10 says that

 ≥  and there exists a subset  of ℬ such that       and   span  ∪   .

⇒  ∪  ≤       and  ∪  spans  .

⇒  ∪  ≥ 

⇒  ∪   
That is,

  spa   ∪  and  ∪   .
By (1),  ∪  is a basis for  . ■

Corollary 2. : Let  be a vector space of dimension . Let  ⊂  be a finite subset.

(1) If   span   , then  ≥ . Also

   if and only if  is a basis for  .
◎ The Dimension of Subspaces

Theorem 1.11 : Let  be a vector space of dimension . Let  ⊂  be a subspace. Then

(1)  is finite dimensional.

(2) dim  ≤ dim  and equality is attained if and only if    .

Proof of (1) : (1) Let  ⊂  be a linearly independent subset.

Since  ⊂  and dim   , it holds that  ≤ .

Now, let  ⊂  be a maximal linearly independent subset. Then

 ≤ .

Also we claim that span    .

If not, we can choose ∈   span   . Then ∪   is linearly independent.

This contradicts to the assumption that  is maximal.

This shows that  is a basis for  .

In particular,  is a finite dimensional vector space.
Theorem 1.11 : Let  be a vector space of dimension . Let  ⊂  be a subspace. Then

(2) dim  ≤ dim  and equality is attained if and only if    .

Proof of (2) : From the above proof of (1), we know already that  is finite dimensional and
dim    ≤   dim  .

So, it remains to show that if dim   dim  then    .

To this aim, suppose that

dim   dim  .

Then  has a basis ℬ with ℬ   .

Since ℬ is a linearly independent subset of  , it can be extended to a basis ℬ′ of  .

But it holds that

ℬ′  dim     ℬ .

This can occur only when ℬ′  ℬ and hence it holds that

  span ℬ′   span ℬ   . ■
Corollary : Let  be a subspace of a finite dimensional vector space  . Then

any basis for  can be extended to a basis for  .

Proof : Suppose that dim    and let  be a basis for  .

Then Corollary 2.(3) says that

 can be extended to a basis for  .

Thus

any basis for  can be extended to a basis for  . ♣

Example 17 : Let         ∈                  .

(1)  is a subspace of  .

(2) If             is an element of  , then

         and     

and hence
                                                          .

This shows that  ⊂ span                   .

On the other hand, it holds that

the vectors                    are contained in 

since they satisfy the conditions           and     .

Therefore  is a subspace which is generated by                    .

(3)                     is linearly independent and hence a basis for  .

In particular, dim                        .

◎ The Lagrange Interpolation Formula

Problem : Let  ≥  be an integer. Also let       ⋯    and   …  be real numbers.

Then find a polynomial     ∈   ℝ  satisfying the condition

       for all  ≤  ≤ .

Example : If   , then let

  
   and        .
  

We can determine  from the condition       .

Namely,       and hence

               

          .
     
Problem : Let  ≥  be an integer. Also let       ⋯    and   …  be real numbers.

Then find a polynomial     ∈   ℝ  satisfying the condition

       for all  ≤  ≤ .

Ideal : Suppose that we have polynomials  … ∈   ℝ  such that

 if ≠  and
      
 if   
for all  ≤  ≤ . ------ (*)

Then the polynomial


         

 

is contained in   ℝ  and satisfies the desired condition that

       for all  ≤  ≤ .
Remark : Note that   ℝ  is of dimension      .

If there exist elements  … ∈   ℝ  satisfying

 if ≠  and
      
 if   
for all  ≤  ≤ . ------ (*),

then one can check that

they are linearly independent and hence  …  is a basis of  ℝ  .

Indeed, suppose that    ⋯     for some  … ∈ℝ.

For     , we get        and hence    .

Similarly, for     we get    .

Consequently, it is shown that     ⋯     and hence

 …  is linearly independent.

Construction of  … ∈   ℝ  satisfying

 if ≠  and
        if   
for all  ≤  ≤ . ------ (*)

    :      ⋯         ⇒       ×       ×⋯×       for some ∈ℝ.

Since       , we have


 ×        ⋯          and hence    .
    ⋯    

    ⋯    
That is,       .
    ⋯    
   
(Continued) Construction of  … ∈   ℝ  satisfying

 if ≠  and
      
 if   
for all  ≤  ≤ . ------ (*)

In general, let

    
≠

      ⋯                ⋯      
       
    
≠
 
       ⋯                  ⋯       

for each  ≤  ≤ . Then it satisfies the condition (*).

Indeed,
       ⋯                  ⋯       
        
     ⋯          ⋯     
       

and
     for  ≠  is equal to zero since       is a factor of     .
Summary : Let  ≥  be an integer. Also let       ⋯    and   …  be real numbers.

(1) For each  ≤  ≤ , the polynomials

    
≠

      ⋯                ⋯      
       
    
≠
 
     ⋯          ⋯     
       

are elements of   ℝ  satisfying

 if ≠  and
        if   
for all  ≤  ≤ .

(2)       …      is a basis for  ℝ  .

(3) (The Lagrange Interpolation Formula) For the element

        ⋯                ⋯      
     

       

 ×  of   ℝ  ,
       ⋯                  ⋯       

it holds that

       for all  ≤  ≤ .
Summary (Continued) :

(4) Let     ∈   ℝ  be an element such that        for all  ≤  ≤ .

Since       …      is a basis for  ℝ  , we can write

                 ⋯       .

Then, by the requirements        for all  ≤  ≤ , we have

                   .

Therefore           .

 

That is, if       ⋯    and   …  are real numbers then

there exists a unique polynomial     ∈   ℝ  such that        for all  ≤  ≤ .

(The Lagrange Interpolation Formula) It is written as

       ⋯                ⋯      
    

 ×  .
     ⋯          ⋯     
       
Example : Find a polynomial     ∈    ℝ  such that

     ,      ,       and      .

First, we want to construct  … ∈   ℝ  satisfying the following conditions :

                and      ;

                and      ;

                and      ;

                and      ;


Then       ×              such that       ×      and hence    .

Similarly,

  
                ,                 and                 .
  
Therefore, by Lagrange Interpolation Formula, it holds that

                        .