Está en la página 1de 17

# NATIONAL CHENG KUNG UNIVERSITY

## Instructor: Student: Student ID: Department: Class:

Prof. Szu Chi Tien Nguyen Van Thanh P96007019 Inst. of Manufacturing & Information Systems 1001- N154000 Linear System

## September 25, 2011

Problem 1........................................................................................................................ 2 Problem 2........................................................................................................................ 5 Problem 3........................................................................................................................ 6 Problem 4...................................................................................................................... 10 Problem 5...................................................................................................................... 13 Problem 6...................................................................................................................... 15

## Linear system class HW1

Page 1

Problem 1
1. Given a matrix A size of , with . Let matrix A be real. According to the

## Where, U and V are two orthogonal matrices. That means:

I is identity matrix (hence, U and V have full rank). From the property of the singular value decomposition, ( ) ( ) (1.1)

Where r is the number of nonzero singular values and they satisfy the condition:

( )

(1.3)

] (1.4) Page 2

[ ( ) ( )

] (1.5)

## From equations (1.1), (1.2), (1.3), (1.4) and (1.5), we have ( ) ( ) ( ) ( ) ( ) ( )

2. If the above matrix A has full rank that means: 3. Check the result with the following matrices.

How to find matrices U, S and V: in Matlab, it is easy to do, but we want to know the way to get the result Step 1: calculate ( ) [ ( ] ). And then, we get . ).

## Where, Xi is the eigenvector of ATA corresponding to i(

matrix VT (V) by orthogonalizing X. Hows about matrix S? We have Step 2: similarly, calculate ( [ ) ] ( ).

Where, Yi is the eigenvector of AAT corresponding to i(AAT). And then, we get matrix UT (U) by orthogonalizing Y.And matrix S doesnt change because s dont change, but note that get the result. After that, we get the form: , size(S) maybe changes, we need to adjust to

## Linear system class HW1

Page 3

(a)

We can expressed,

+* ( ) [ ]

+[ ( )

(b)

## Linear system class HW1

Page 4

Problem 2
If A and C are nonsingular, find the inverse of the matrix, * Solution: Assume, * From the property of the matrix inverse, we have, * Hence, * + * + +* + * + (I is identity matrix) + +

## Linear system class HW1

Page 5

Problem 3
Let, [ ]

1. Find the characteristic polynomial equation of the matrix A. The characteristic polynomial equation of the matrix A is, ( ) ( ) ([ ( )| | ]) ( )[( ) ]

2. Find the eigenvalues f A using the characteristic polynomial. Now we solve the characteristic polynomial equation of the matrix A. ( )[( ) ] ( { ) {

3. Determine the eigenvectors of A from the definition using the Matlab function eig(A). ( - With [ : ][ ] { ( ) ) [

Page 6

- With [

: ][ ] { ( ) {

## [ ] - With [ Matlab command: >> [V, D] = eig (A); [ ] [ [ We have, [ [ [ ] ] [ ] ( ) ] ] = -1: ][ ] { { [ ]

[ ]

4. Using only the eigenvalues of A, find the eigenvalues of the matrix B, where Firstly, we have, ( Linear system class HW1 ) ( ) Page 7

## { But, note that:

( )

( ) (a) ( ) - With

( )

( )

## (it is right for at least B=I or C = I)

( ) :

( )

( )

( ) - With : ( ) - With :

( ( ( )

) ) ( )

( ( ( )

) )

(b) From section 1, we have the characteristic polynomial equation of the matrix A is ( ) According to the Cayley Hamilton theorem, hence, ( ) For every (c) From property of similarity transformation, we choose, [ ] .

Where, V1,V2, and V3 are three eigenvectors of matrix A. Where, Linear system class HW1 Page 8

] [ ]

( ) ( )

( )

## Linear system class HW1

Page 9

Problem 4
Let ( ) ( ), find Y(s) the Laplace transform of y(t). Show that y(t) is the

solution of a homogenous linear differential equation with appropriate initial conditions. Solution ( ) [ ( )] Where, ( ) ( )) ( )) ( * ( ( ) ( ( ) ) ( )( ) + ( ( )) ) ( ) ( ) ( ) ( ( ))

( ) )

Page 10

( ) ( (

) )

( )

( )

( ( ) (

) )

## From, ( ) ( )( Note that: ( (

( )

(4.1)

( )) ( ))

( ) ( ) ( ) ) ( ( ) )

( ) ( )

( ) ( )

( ) ( )

( )

( )

( )

( )

( ( )) Let (

## And then, we compare coefficients: ( ) ( ) ( )

( )

( ) ( )

( )
( )

( )
( )

( ) ( )
( )

( )

( )

( )

So, y(t) is the solution of a homogenous linear differential equation: Linear system class HW1 Page 11

( )

( )

( )

( )

( )

( )

( )

( )

( )

Page 12

Problem 5
* + * +

## 1. Find the spectral norm of A (i.e., ||A||). * ( ) We have, | | ( ) ( )( ) +* + * +

2. Find the Euclidian norm of the vector (i.e., ) and its normalized vector. * +* + * + * + Hence, ys normalized direction vector is, * + | | + [ ]* +

3. Describe geometrically the operations performed by the transformation A onto the vector x. Before transform * + | | ( ) After transform: * + ( ) Linear system class HW1 | | [ ]* +

( ) ( ) Page 13

* +

## , thus after transformed by matrix A, vector x

doesnt change the direction, but changes the Euclidean norm. 4. Is ||y|| greater or less than | | | | | | | ? || |

5. Using the Rayleigh - Ritz inequality, find a set of lower and upper bounds on for any arbitrary x. Obviously, , where 0 and are the

non useful lower and upper bounds on ||y||. Applying the Rayleigh - Ritz inequality to | ( ( ( ( )| )| )| )| | So for any arbitrary x, | For the given * + | | | | | | | | | | | | | | | | | | | | ( ( ( | ( )| )| )| ) | | |

Page 14

Problem 6
Let [ ]

## 1. Is the matrix A symmetric? [ ]

2. Can the eigenvalues of A be complex? Explain. Verify your answers using the Matlab function eig(A). (6.1)

## (6.2) From equations (6.1) and (6.2): ( ) ( )

We can obtain the eigenvalues of a matrix A by using the Matlab command eig(A): {

So, they are all real as they should be. 3. Find the quadratic expression [ ]

( ( )

][

][ ]

4. Can you find a vector x such that: (a) J = 0? If yes, find x. If not, explain why? Linear system class HW1 Page 15

From the Rayleigh Ritz inequality: ( ) The eigenvalues of matrix A are: { ( ) Hence, yes, we can find a vector x such that J = 0. Since ( ) that ( eigenvalue ) . ( ( ) ) [ ] is along the eigenvector ( )

## (6.3) , the vector x such

corresponding to the

(b) J < 0? Explain. No. There is no vector x that can produce J < 0, because the matrix A is positive semi-definite ( ( ) (c) J > 0? Explain. Yes. We can find a vector x such that J > 0, because the matrix A is positive semidefinite ( ( ) ( [ ] ). Clearly see that, for any vector x is not in the null space of A ) such that J > 0. ) with all its eigenvalues non-negative.

For example,

[ ] yields J = 2 > 0.

Page 16