Math 19b: Linear Algebra with Probability Oliver Knill, Spring 2011 The pivot columns of A span the

The pivot columns of A span the image of A.

Proof. You can see this by deleting the other columns. The new matrix B still allows to
Lecture 13: Image and Kernel solve Bx = b if Ax = b could be solved.

5 Find the image of

1 1 1 1 1
The image of a matrix 2

2 2 2 2



3
A= 3 3 3 3 .

4 4 4 4 4


If T : Rm Rn is a linear transformation, then {T (~x) | ~x Rm } is called the 5 5 5 5 5
image of T . If T (~x) = A~x, where A is a matrix then the image of T is also called
the image of A. We write im(A) or im(T ).

1 The map T (x, y, z) = (x, y, 0) maps the three dimensional space into itself. It is linear The kernel of a matrix

x 1 0 0 x
because we can find a matrix A for which T (~x) = A y = 0 1 0 y . The image

z 0 0 0 z If T : Rm Rn is a linear transformation, then the set {x | T (x) = 0 } is called

of T is the xy-plane. the kernel of T . These are all vectors which are annihilated by the transformation.
If T (~x) = A~x, then the kernel of T is also called the kernel of A. The kernel of A
2 If T (x, y) = (cos()x sin()y, sin()x + cos()y) is a rotation in the plane, then the image are all solutions to the linear system Ax = 0. We write ker(A) or ker(T ).
of T is the whole plane.

3 The averaging map T (x, y, z) = (x + y + z)/3 from R3 to R has as image the entire real axes 6 The kernel of T (x, y, z) = (x, y, 0) is the z-axes. Every vector (0, 0, z) is mapped to 0.
R.
7 The kernel of a rotation in the plane consists only of the zero point.
The span of vectors ~v1 , . . . , ~vk in Rn is the set of all linear combinations c1~v1 +
. . . ck~vk .
8 The kernel of the averaging map consists of all vector (x, y, z) for which x + y + z = 0.
The kernel is a plane. In the language of random variables, the kernel of T consists of the
centered random variables.
4 The span of the standard basis vectors e1 , e2 is the xy-plane.
Also the kernel of a matrix A is a linear space.
n
A subset V of R is called a linear space if it is closed under addition scalar
multiplication and contains 0. How do we compute the kernel? Just solve the linear system of equations A~x = ~0. Form
rref(A). For every column without leading 1 we can introduce a free variable si . If ~x is
the solution to A~xi = 0, where all sj are zero except si = 1, then ~x = j sj ~xj is a general
P

vector in the kernel.

1 3 0

The image of a linear transformation ~x 7 A~x
kernel 2 6 5
is the span of the column vectors of A. The

image
9 Find the kernel of A =

3 9 1
. Gauss-Jordan elimination gives: B = rref(A) =

image is a linear space. domain

codomain 2 6 0
1 3 0

0 0 1
. There are two pivot columns and one redundant column. The equation B~ x=0

How do we compute the image? If we are given a matrix for the transformation, then the 0

0 0
image is the span of the column vectors. But we do not need all of them in general. 0 0 0
is equivalent to the system x + 3y = 0, z = 0. After fixing z = 0, can chose y = t freely and
3
A column vector of A is called a pivot column if it contains a leading one after 1 .
obtain from the first equation x = 3t. Therefore, the kernel consists of vectors t
row reduction. The other columns are called redundant columns. 0
 Step II) Transmission.
Homework due March 2, 2011 Now add an error by switching one entry in each vector:
   

 . 


 . 

1 . .
   
Find the image and kernel of the chess matrix: 








 .   . 
1 0 1 0 1 0 1 0
   
u = Mx + e = 

. 
 v = My + f = 

. 


0 1 0 1 0 1 0 1
 .

  .


1 0 1 0 1 0 1 0
   





 . 


 . 

0 1 0 1 0 1 0 1 
.
 
.

A= .
   
1 0 1 0 1 0 1 0
Step III) Detect the error e and f. Form


0 1 0 1 0 1 0 1

1 0 1 0 1 0 1 0
   
. .
   
0 1 0 1 0 1 0 1
   
. .
   
   
     
 1 0 0 1 0 1 1  
 .  .  1 0 0 1 0 1 1  
 .  .
2 Find the image and kernel of the following Pascal triangle matrix:

Hu =  0 1 0 1 1 0 1   .
 
 = . , Hv =  0
 
1 0 1 1 0 1   .

= . .

 0 0 1 1 1 1 0   . .  0 0 1 1 1 1 0   . .
     
0 0 0 0 1 0 0 0 0 



0

0 0 1 0 1 0 0 0



 . 


 . 

. .
   
0
A= 0 1 0 2 0 1 0 0 .    

0 1 0 3 0 3 0 1 0

. .
1 0 4 0 6 0 4 0 1 Now look in which column Hu or Hv is. Assume this column

.

.


is k. Place 0s everywhere in the vectors e except at the kth . .
3

We work on the error correcting code as in the book (problem 53-54 in 3.1). Your task entry where you put 1. For example if Hu is the second e =

. , f = . .

is to do the encoding and decoding using the initials from your name and write in one column, then put a 1 at the second place. We obtain vectors

.

.

sentence using the terminology of image and kernel, what the essence of this error

e and f :
.

.

correcting code is. . .
Step I) Encoding. To do so, we encode the letters of the alphabet by pairs of three Step IV) Decode the message.
vectors containing zeros and ones:
x1 . .
A = (0, 0, 0, 1), (0, 0, 0, 1) B = (0, 0, 0, 1), (0, 0, 1, 0) C = (0, 0, 0, 1), (0, 0, 1, 1) x2

.

.

x4 . .

D = (0, 0, 0, 1), (0, 1, 0, 1) E = (0, 0, 0, 1), (0, 1, 1, 0) F = (0, 0, 0, 1), (0, 1, 1, 1)

x . .
3 x . .
G = (0, 0, 0, 1), (1, 0, 0, 1) H = (0, 0, 0, 1), (1, 0, 1, 0) I = (0, 0, 0, 1), (1, 0, 1, 1) Use P

x =
5
to determine P e = P . = , P f = P . =

. In

4 x6 . .
J = (0, 0, 0, 1), (1, 1, 0, 1) K = (0, 0, 0, 1), (1, 1, 1, 0) L = (0, 0, 0, 1), (1, 1, 1, 1) x5 . .

M = (0, 0, 1, 0), (0, 0, 0, 1) N = (0, 0, 1, 0), (0, 0, 1, 0) O = (0, 0, 1, 0), (0, 0, 1, 1)

x
x7
. .
. .
6
P = (0, 0, 1, 0), (0, 1, 0, 1) Q = (0, 0, 1, 0), (0, 1, 1, 0) R = (0, 0, 1, 0), (0, 1, 1, 1) x7 . .
S = (0, 0, 1, 0), (1, 0, 0, 1) T = (0, 0, 1, 0), (1, 0, 1, 0) U = (0, 0, 1, 0), (1, 0, 1, 1) an error-free transmission (P u, P v) would give the right result back. Now
V = (0, 0, 1, 0), (1, 1, 0, 1) W = (0, 0, 1, 0), (1, 1, 1, 0) X = (0, 0, 1, 0), (1, 1, 1, 1)
. .

Y = (0, 0, 1, 1), (1, 0, 0, 1) Z = (0, 0, 1, 1), (1, 0, 1, 0) ? = (0, 0, 1, 1), (1, 0, 1, 1) . .
! = (0, 0, 1, 1), (1, 0, 0, 1) . = (0, 0, 1, 1), (1, 0, 1, 0) , = (0, 0, 1, 1), (1, 0, 1, 1) Pu = ,

Pv =

. .
. . . .

. .
Choose a letter to get a pair of vectors (x, y). x = , y = . Use 1 + 1 = 0 in
satisfy P u = x + P e, P v = y + P f . We recover the original message (x, y) and so the
. .
letter from
. . . .

the matrix multiplications to build . .
x = Pu Pe = ,
y = Pv Pf =
. .
       

 1 0 1 1 


 . 


 1 0 1 1 


 . 


1 1 0 1
 
.
 
1 1 0 1
 
.
 . .
.   .  
         
     
 1 1 1 0  .   1 1 1 0  .  Assignment: perform the encoding with your initials and check that you got the
  .  
    .  
 
Mx = 

1 0 0 0 
= . 
My = 

1 0 0 0 
= . 
right letter back. Tell then in one sentence what the essence of this error correction is.
.   .  
   
0 1 0 0 . 0 1 0 0 .
   



 .  





 .  

 In particular, how does the image of the encoding matrix M fit with the kernel of

 0 0 1 0 
  . 


 0 0 1 0 
  . 
 the healing matrix H?
0 0 0 1 . 0 0 0 1 .