EE-603 Linear System Theory

Lecture 06

Dr. Sheraz Khan

University of Engineering & Technology Mardan

November 11, 2022

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 1 / 15
 Remarks about Computation of e A (t)
As discussed in Section 3.6 (chen), three methods of computing functions
of a matrix. They can all be used to compute e A (t) :

Using Theorem 3.5

First, compute the eigenvalues of A; next, find a polynomial h(λ) of
degree n − 1 that equals e λt on the spectrum of A; then e At = h(A)

Using Jordan form of A

−1
Let A = QÂQ ; then e At = QeÂt Q−1 , where A is in Jordan form and
e At can readily be obtained by using (3.48).

Using the infinite power series in (3.51)

Using the infinite power series in (3.51): Although the series will not, in
general, yield a closed-form solution, it is suitable for computer
computation, as discussed following (3.51).
Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 2 / 15
 In addition, (3.58) can be used to compute e At

e At = L−1 (sI − A)−1 (4.8)

The inverse of (sI − A) is a function of A; therefore, again, we have many
methods to compute it:
1 Taking the inverse of (sI − A).
2 Using Theorem 3.5.
 −1
3 Using (sI − A)−1 = Q sI − Â Q−1
4 Using the infinite power series in (3.57).
5 Using the Leverrier algorithm discussed in Problem 3.26.

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 3 / 15
 Computation of e A (t)

Computing Structured e At via Taylor expansion

convenient when A is a diagonal or Jordan matrix

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 4 / 15
 Computation of e A (t)

Computing Structured e At via Taylor expansion

convenient when A is a diagonal or Jordan matrix

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 5 / 15
 Computation of e A (t)

Computing Structured e At via Taylor expansion

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 6 / 15
 Computation of e A (t)

Computing Structured e At via Taylor expansion

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 7 / 15
 Example

Mass moving on a straight line

Consider a mass moving on a straight line with zero friction and no
external force. Let x1 and x2 be be the position and the velocity of the
mass, respectively. The state-space description of the system is

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 8 / 15
 Example

Mechanical system with strong damping

Consider a spring-mass-damper system with m = 1, k = 2, b = 3. Let x1
and x1 be the position and velocity of the mass, respectively. We have

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 9 / 15
 Example

Mechanical system with strong damping

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 10 / 15
 Computation of e A (t)

Computing Low-Order e At Column Solutions

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 11 / 15
 Computation of e A (t)

Computing Low-Order e At Column Solutions

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 12 / 15
 Columns of The State-Transition Matrix

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 13 / 15
 Computation of e A (t)

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 14 / 15
 Online Computation using Octave

https://octave-online.net/

Dr. Sheraz Khan (UET Mardan) EE-603 LST November 11, 2022 15 / 15