Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
0% found this document useful (0 votes)
33 views

Linear System Theory: Dr. Vali Uddin

The document discusses linear system theory and linear algebra concepts. It introduces linear spaces and vector spaces, defining them as sets with vector addition and scalar multiplication operations that satisfy certain properties like closure, commutativity, associativity, and the existence of additive and multiplicative inverses. Examples of linear spaces include sets of vectors over fields. A subspace is defined as a subset of a linear space that itself forms a linear space under the given operations.

Uploaded by

Muhammad Hassan
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
33 views

Linear System Theory: Dr. Vali Uddin

The document discusses linear system theory and linear algebra concepts. It introduces linear spaces and vector spaces, defining them as sets with vector addition and scalar multiplication operations that satisfy certain properties like closure, commutativity, associativity, and the existence of additive and multiplicative inverses. Examples of linear spaces include sets of vectors over fields. A subspace is defined as a subset of a linear space that itself forms a linear space under the given operations.

Uploaded by

Muhammad Hassan
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 41

Linear System Theory

Dr. Vali Uddin


Hamdard University

Vali.uddin@hamdard.edu.pk

Lecture 3 1
Linear Algebra
 The above are review materials on systems
 To further study systems beyond by using Laplace and z-
transforms, we need background on linear algebra
 We shall review/study the following:
– Scalars  Vectors  Matrices
– Vector space  Representation of Vectors
– Functions on vectors  Representation of functions
– Linear algebraic equations
– Eigenvectors, Generalized eigenvectors, and Jordan-form
– Functions of a square matrix
– Norms and inner products

Lecture 3 2
Linear Spaces and Linear Operators

 Consider a set of scalars. What properties to have?


 Field F: A set of elements (scalars) and two operations
"+" (addition) & "" (multiplication) s.t. for all , , and 
 F:
– Closedness:  +   F, and     F
– Commutative:  +  =  + , and    =   
– Associative: (+)+ = +(+), and () =()
– Distributive:   ( + ) =    +   
– Unit elements:  0 and 1 s.t.  + 0 =  and   1 = 
– Inverse elements: For any ,   s.t.  +  = 0
(additive inverse). For any   0,   s.t.    = 1
(multiplicative inverse)
Lecture 3 3
Examples
 Examine the following, and see if they are fields or not:
(N, +, ), (I, +, ), (R, +, ), (C, +, )
– (N, +, ): No additive inverse, no multiplicative inverse
– (I, +, ): No multiplicative inverse
– (R, +, ): Yes
– (C, +, ): Yes
 (The set of 2 by 2 real matrices, +, )
– No multiplicative inverse
 (R(s), +, ) (the set of rational functions with real
coefficients)
– Yes

Lecture 3 4
Linear Spaces over a Field
Q. Vector? What properties to have for a set of vectors?
 A vector space (or a linear space) over a field, denoted as
(X, F), consists of
– X ~ A set of elements (vectors);  F ~ A field
– "+" ~ Vector Addition;  "" ~ Scalar multiplication
s.t. for all x1, x2, x3  X and ,   F:
– Closedness: x1 + x2  X, and   x1  X
– Commutative: x1 + x2 = x2 + x1
– Associative: (x1 + x2) + x3 = x1 + (x2 + x3), (x) = ()x
– Distributive: (x1 + x2) = x1 + x2, (+)x1 = x1 + x1
– Unit elements:  0  X & 1  F s.t. x1+0 =x1 and 1x1 = x1
– Inverse: For all x1,  x2 s.t. x1 + x2 = 0 (additive inverse)

Lecture 3 5
Examples
 Examine the following, and see if they are fields or
not: (R2, R), (Rn, R), (Cn, C), (Cn, R), (Rn, C) with the
standard matrix addition and scalar-matrix
multiplication
 x11   x 21   x11  x 21   x11 
       
 x12   x 22   x12  x 22   x12 
x1  ,x  , x  x2  ,   x1 
 :  2  :  1  :   : 
       
 x1n   x 2n   x1n  x 2 n   x1n 

– (Rn, R), (Cn, C), and (Cn, R): Yes


– (Rn, C): No. No closedness under scalar multiplication
• The idea of a linear space goes beyond Euclidean spaces
Lecture 3 6
 What is a piecewise continuous function?
Which of the following are piecewise
continuous?

t0

lim y( t ) ~ Right hand lim it lim y( t ) ~ Left hand lim it


t t 0 ,t  t 0 t t 0 ,t  t 0

– y(t) is piecewise continuous iff y(t) is continuous


except at a finite number of points. At those points,
y(t) has well defined left and right-hand limits
– Yes. No. No.
Lecture 3 7
 Consider (Y, R), where Y is the set of all real-valued
piecewise continuous functions defined over (-, ) with
the following operations:

y1

y2

y1+y2 y1
y1

• Is (Y, R) a vector space?


• Yes. It satisfies all the requirements
Lecture 3 8
• Now define (Y, R) with Y as the set of solutions to
y  3y  2y  2

– Is this a vector space? Why or why not?


– Suppose that y1  Y and y2  Y. How about y1 + y2
and y1?
y1  3y 1  2 y1  2
y 2  3y 2  2 y 2  2

y1  y2   3y 1  y 2   2y1  y 2   4  2


y1  3y 1  2y1  2  2 except when   1

– (Y, R) is not a vector space ~ No closedness

Lecture 3 9
• How about let Y be the set of solutions to
y  3y  2y  0

– In this case, (Y, R) is a vector space


• How about (Rn[s], R), where Rn[s] is the set of
polynomials of degree less than n
n 1 n 1
i
x1   c1is , x 2   c2isi
i 0 i 0
n 1 n 1
x1  x 2   c1i  c2i s , x1   c1isi
i
i 0 i 0

– It is a vector space, meeting all the


requirements
Lecture 3 10
 How about (Rn[s], R(s)), where where Rn[s] is the
set of polynomials of degree less than n and R(s) is
the set of rational functions with real coefficients?
– No. No closedness
 How about (Rn(s), R), where Rn(s) is the set of n-
vectors of rational functions with real coefficients?
– Yes
 How about (Rn(s), R(s))?
– Yes

Lecture 3 11
Subspace
 (X, F) is a linear space, and Y  X. (Y, F) is a
subspace of (X, F) iff (Y, F) itself is a linear space
– (Y, F) is a subspace if 1y1 + 2y2  Y for all y1,
y2  Y and 1, 2  F
– Other conditions are either automatically satisfied
or are covered by this one
– Consider (R2, R). Is the set of elements satisfying
x1 - 2x2 + 1 = 0 a subspace?

Y: x1-2x2+1=0

– Not a subspace ~ 0 is not included. No closedness


Lecture 3 12
 Then how about the set of elements
satisfying x1 - 2x2 = 0?
Y: x1-2x2=0

– Yes. In fact, any straight line passing through 0


• What would be a subspace for (R3, R)?
– Any plane or straight line passing through 0
• Recall that (Rn[s], R) is a vector space, where
Rn[s] is the set of polynomials of degree < n
– (R4[s], R) is a subspace of (R5[s], R)

Lecture 3 13
 Let (X, F) be a vector space
– Given any set of vectors {xi}i=1 to n, xi X.
– Form the set of linear combinations
n 
Y    i x i , i  F 
i1 

– Then (Y, F) is a linear space, and is a subspace of


(X, F).
– It is the space spanned by {xi}i=1 to n

Lecture 3 14
Introduction to System
Theory and Linear Algebra

Lecture 3 15
Linear Spaces and Linear
Operators
 Field F: A set of elements (scalars) and two operations
"+" (addition) & "" (multiplication) s.t. for all , , and 
 F:
– Closedness:  +   F and     F
– Commutative:  +  =  +  and    =   
– Associative: (+)+ = +(+) and () = ()
– Distributive:   ( + ) =    +   
– Unit elements:  0 and 1 s.t.  + 0 =  and   1 = 
– Inverse elements: For any ,   s.t.  +  = 0
(additive inverse). For any   0,   s.t.    = 1
(multiplicative inverse)
Lecture 3 16
Linear Spaces over a Field
 A vector space (or a linear space) over a field, denoted as
(X, F), consists of
– X ~ A set of elements (vectors);  F ~ A field
– "+" ~ Vector Addition;  "" ~ Scalar multiplication
s.t. for all x1, x2, x3  X and ,   F:
– Closedness: x1 + x2  X and   x1  X
– Commutative: x1 + x2 = x2 + x1
– Associative: (x1 + x2) + x3 = x1 + (x2 + x3), (x) =
()x
– Distributive: (x1 + x2) = x1 + x2, (+)x1 = x1 +
x1 0
Lecture 3 17
- Unit elements:  0  X & 1  F s.t. x1+ 0 = x1 and
1x1 = x1
- Inverse: For all x1,  x2 s.t. x1 + x2 = 0 (additive
inverse)
 (X, F) is a linear space, and Y  X. (Y, F) is a
subspace of (X, F) iff (Y, F) itself is a linear space
– (Y, F) is a subspace if 1y1 + 2y2  Y for all y1, y2
 Y and 1, 2  F
– Given any set of vectors {xi}i=1 to m, xi X, form the
set of linear combinations
m 
Y    i x i , i  F 
i 1 
– Then (Y, F) is a linear space, and is a subspace of (X, F)
– It is the space spanned by {xi}i=1 to m
Lecture 3 18
Linear Independence
 A set of vectors {x1, x2, .., xm} in (X, F) is linearly
dependent iff  {1, 2, .., m} in F, not all zero, s.t.
1x1 + 2x2 + .. + nxm = 0 (*)
– If (*) holds and assume for example that 1  0, then
x1 = -[2x2 + .. + nxm]/1
i.e., x1 is a linear combination of {i}i=2 to m
 If the only set of {i}i=1 to m s.t. the above holds is
1 = 2 = .. = m = 0
then {xi}i=1 to m is said to be linearly independent
– None of xi can be expressed as a linear combination
of the rest
Lecture 3 19
– A linearly dependent set ~ Some redundancy in the
set
Example. Consider the following vectors:
x3 x4
x1
x2
• For the following sets, are they linearly dependent
or independent?
– {x1} ~ Linearly independent
– {x1, x2} ~ Linearly dependent
– {x1, x3} ~ Linearly independent
– {x1, x3, x4} ~ Linearly dependent
Lecture 3 20
 Are the following vectors LD or LI?
 2 1  1 
x1  3, x 2  2, x 3  4 How to find it out?
     
4 5 7
– {x1, x2, .., xn} are LD iff  {1, 2, .., n}, not all
zero, s.t. 1x1 + 2x2 + .. + nxn = 0
 1   1 
   
x1 x 2 ... x n   2   0, or A  2   0, with A  x1 x 2 ... x n 
 :   : 
    Need |A| = 0 to be LD
 n  n
2 1 1
A3 2 4   10  0  These vectors are LI
4 5 7
Lecture 3 21
Example:  1   s  10 
 s5   s  5s  6 
x1   , x2   
1  1
   
 s  10   s6 

– Are they LI in (R2(s), R), where R(s) is the set of


rational functions with real coefficients?
 1 s  6  2 s  10   1  2 s  61  102 
 s  5s  6   s  5s  6 
1x1  2 x 2    
1 s  6  2 s  10  1  2 s  61  102 
   
 s  6s  10    s  6s  10  

– For the above to be zero, 1 = 2 = 0


– Consequently x1 and x2 are linearly independent

Lecture 3 22
 1   s  10 
Example (Continued)  
x1   s  5 , x 2   s  5s  6 
 
1 1
   
 s  10   s6 

• Are the same vectors LI in (R2(s), R(s))?


– If we let 1 1
1  , 2  , then
s6 s  10
 1   1 
 s  5s  6   s  5s  6 
1x1  2 x 2      0
1 1
   
 s  6s  10   s  6s  10 

– They are linearly dependent in (R2(s), R(s))


– Linearly dependence or independence depends on the
vector space under consideration
Lecture 3 23
Dimension
 For (X, R), the maximum number of LI vectors is called
the dimension of the space
x3 x4
Examples. For the following, D = ?
x1
– (R , R)
2 ~ 2
x2
– (Rn, R) ~ n
• How about (X, R), where X  the set of real-valued
piecewise continuous functions over (-, )
– Let xi = ti, i = 0, 1, 2, ...
• If  iti  0, then i  0  i
• {ti} are LI  Dimension = 
• Let Y be the set spanned by {ti}i from 0 to n-1
– (Y, R) is a subspace of (X, R) with dimension = n
Lecture 3 24
 Consider (Y, R) with Y as the set of solutions
to
y  3y  2y  0
– We know that (Y, R) is a vector space. Dimension = ?
– A linear differential equation with constant coefficient. Use
Laplace transform to solve it

s2ŷ(s)  sy(0 )  y (0 ) 3sŷ(s)  y(0 )y  2ŷ(s)  0


s2  3s  2ŷ(s)  sy(0 )  y (0 )  3y(0 )
sy(0 )  y (0 )  3y(0 )
ŷ(s)  y( t )  c1e  t  c2e  2 t
s  1s  2 

Lecture 3 25
– 2nd order system  2 natural modes and 2 ICs  D = 2
– Generally, N'th order system with N ICs, D = N

– Basis, Representation, and Orthonormalization


 Relationship among a set of vectors: Linear
dependence and linear independence
 Dimension of a linear space

• The base of a linear space: Basis


 Representations of a vector in term of a basis

 Relationship among representations for different


bases

Lecture 3 26
 Generalization of the idea of length: Norms
 A sense of orientation: Inner Product

 The concept of perpendicularity:


Orthogonality
 Gram-Schmidt Process to obtain orthonormal
vectors
 Projection and Orthogonal Projection
Theorem
– Linear Operators and Representations

Lecture 3 27
Basis and Representations
 Basis: The base, foundation, or chief
supporting factor of anything
 A set of LI vectors {e1, e2, .., en} of (X, F)
is said to be a basis of X if every vector in
X can be expressed as a unique linear
combination of them

Lecture 3 28
• They span X
• For any x  X, then  {1, 2, .., n}
s.t.
n
x  1e1  2e2  ..  ne n   iei 1 
i 1  
x  e1 e 2 ... e n   2 
: 
x  e1 e2 ... e n     ~  
 n

– : Representation of x with respect to the basis

Lecture 3 29
Example: Consider (R3[s], R), where R3[s] is the set of
polynomials of real coefficient with degree less than 3
– D = ? An example of a basis?
– Dimension = 3
– A basis: {e1 = s2, e2 = s, e3 = 1}
– Suppose x = 5s2 + 4s + 7. What is its representation?
 5  5
 2

 
x  s s 1  4
 
    4
~ A representation
7 7
   
– A vector, such as x = 5s2 + 4s + 7, need not be in the form of
a vector
– But a vector in a finite dimensional linear space always has a
vector representation in terms of a basis
Lecture 3 30
Q. What qualifies to be a basis?
Theorem: In an n-dimensional vector space, any set of
n LI vectors qualifies as a basis
Proof:
– Let {e1, e2, .., en} be linearly independent
– For any x  X, {x, e1, e2, .., en} are linearly dependent
–  {0, 1, 2, .., n} such that

0x  1e1  2e2  ...  ne n  0


– 0  0. For otherwise, {e1, e2, .., en} are not linearly
independent. Thus
1 n
x   1e1  2e2  ...  n e n    iei
0 i 1
– Any x can be expressed as a linear combination of them

Lecture 3 31
• Is the combination unique here?
– Suppose that  another linear combination
~ n n
x   iei   i ei
i 1 i 1

 i  i  ei  0
n ~
i 1
~ ~
i  i  0 or i  i for all i ~ since {e1, e2, .., en} are LI

 What can be said now?


– The linear combination is unique
– {e1, e2, .., en} is a basis, and the proof is completed

Lecture 3 32
Example (Continued): The set of solutions to
y  3y  2y  0

– We showed that e1 = e-t and e2 = e-2t


form a basis
– Is it the unique one?
– If not, give another one

e1  e  t  e  2 t and e2  e  t  2e  2 t

– They form a basis since they are LI and span the two
dimensional space
– There are in fact an infinite number of bases
Lecture 3 33
Example (Continued): x = 5s2 + 4s + 7
– {e1 = s2, e2 = s, e3 = 1}. What would be another
basis?
e1  s2  s, e2  s  1, e3  1
– What is the representation of the above x? 5
 
x  5 s2  s  1s  1  8  1  e1 e2
 
e3    1
8
 
– When a basis is changed, the representation is
changed correspondingly
– What is the new representation? How do we formalize
this process? Why bother?
– Sometimes it is much easer to work with a particular
basis then working with others, e.g., controllable
canonical form or observable canonical form  Need
to know how to do it
Lecture 3 34
Change of Basis
 Any n LI vectors qualify as a basis
e1 e2 ... en    e1 e2 ... en   
– For a particular x, the representation is unique for each
basis
Q. Given , how to find?
– Express x in terms of e, e in terms ofe  x in terms
ofe
 11 
p  1i 
p
   
e1  e1 e2 ... en   p12 
ei  e1 e2 ... en   p1i 
 :   : 
  ~p   ~ pi
 p1n  1  p1i 
e1 e2 ... en   e1 e2 ... en  p1 p2 ... pn 
 P ~ nn
Lecture 3 35
e1 e2 ... en   e1 e2 ... en  P
x  e1 e2 ... e n    e1 e2 ... en  P

 e1 e2 ... en      P

ith column of P: Representation of ei


w.r.t. the set of new basis

– Conversely, ith column of Q: Representation ofei


  Q w.r.t. the set of existing basis
– What is the relationship between P and Q?
  P  PQ   PQ  I, or P  Q 1

Lecture 3 36
Example (Continued) (R3[s], R)
– {e1 = s2, e2 = s, e3 = 1}
– {e1  s 2  s, e2  s  1, e3  1}

• What is P?
1
2

2  
e1  s  s  s s  1 1   1 
 1  ~ p1
 
0
2  
e2  s  s  s s  1 1  1  
  1 ~ p2
 
 0 1 0 0
 2

 
e3  1  s  s s  1 1  0 
 
P    1 1 0
 1  ~ p3  1 1 1
   
Lecture 3 37
• With {e1 = s2, e2 = s, e3 = 1} and x = 5s2 + 4s + 7:

 5
 
   4 Now {e1  s2  s, e2  s  1, e3  1}
7
 

• What is?
1 0 0  5  5 
    
  P    1 1 0   4     1
 1 1 1  7  8  ~ As expected
    

Lecture 3 38
1  0
Example. Consider R) with e1   0 , e2   1 
(R2,
   
_  cos  
e2
e2 _ e1  cos e1  sin e2  e1 e 2   ~ q1
 e1  sin  
   sin  
e2   sin  e1  cos  e2  e1 e 2   ~ q2
e1  cos  
 cos   sin   1 1  cos  sin  
Q , P  Q  Q  
 sin  cos     sin  cos  
with | Q | cos  cos   sin  sin   cos(  )   P and   Q 
– For  =  = 45 and x = (1 1)T
 1 1  _e2
  x
  2 2 1   2  _ e1
 1 1 1  0  e2
 2 2 e1
Lecture 3 39
– Basis, Representation, and Orthonormalization
 Relationship among a set of vectors: Linear
dependence and linear independence
 Dimension of a linear space

 The base of a linear space: Basis

 Representations of a vector in term of a basis

 Relationship among representations for different


bases
• Generalization of the idea of length: Norms
 A sense of orientation: Inner Product

 The concept of perpendicularity: Orthogonality

 Gram-Schmidt Process to obtain orthonormal vectors

 Projection and Orthogonal Projection Theorem

– Linear Operators and Representations


Lecture 3 40
THE END

Lecture 3 41

You might also like