ECE

EE 8443
3512 – PatternContinuous
– Signals: Recognition
and Discrete

LECTURE 22: INVERSE LAPLACE TRANSFORM

• Objectives:
Rational Transforms
Partial Fractions Expansion
Distinct Roots
Complex Poles
Repeated Roots
Transforms With Exponentials

• Resources:
MIT 6.003: Lecture 18
Wiki: Inverse Laplace Transform
Wiki: Partial Fractions
DM: Partial Fractions
DK: Integration By Partial Fractions
OSU: Application to Laplace Transforms

URL: Audio:
Rational Transforms
• Consider the problem of finding the inverse Laplace transform for:
B( s) b0  b1 s  b2 s 2 ...  bM s M
X ( s)  
A( s) a0  a1 s  a 2 s 2 ...  a N s N
where {ai} and {bi} are real numbers, and M and N are positive integers.
• One straightforward approach is to factor B(s) into a sum of simpler terms
whose Laplace transforms can be easily computed (or located in a table of
transform pairs). The method of Partial Fractions is one approach:
b0  b1 s  b2 s 2 ...  bM s M c1 c2 cN
X ( s)     ... 
a N ( s  p1 )( s  p 2 )...( s  p N ) s  p1 s  p 2 s  pN
• The details of the expansion depends on the properties of A(s). For example,
the presence of repeated roots slightly complicates things.
• Fortunately, MATLAB can be used to find the roots of the polynomial:
A(s)  s 3  4s 2  6s  4
A = [1 4 6 4]
p = roots(A)
p=
-2
-1.0000 + 1.0000i A( s)  ( s  2)( s  1  j )( s  1  j )
-1.0000 – 1.0000i
EE 3512: Lecture 22, Slide 1
Distinct Roots: Method of Residues
• Suppose there are distinct or non-repeated roots: pi  pj when i  j.
c1 c2 cN
X (s)    ... 
s  p1 s  p 2 s  pN
ci  [( s  pi ) X ( s)] s  pi , i  1, 2,..., N
• The constants, ci, are called residues, and this method of computation is
called the residue method.
• The constant ci is real if the corresponding pole is real. If two poles appear as
complex conjugate pairs, the ci must also appear in complex conjugate pairs.
• Once the partial fractions expansion is completed, the inverse Laplace
transform can be easily found as:
x(t )  c1e p1t  c 2 e p2t  ...  c N e p N t t  0
• MATLAB can be used to find poles as residues:
num = [bM bM-1 … b1 b0]
den = [aN aN-1 … a1 a0]
[r,p] = residue(num, den)

• Example:
s2 s2 c1 c2 c3 c1 c2 c3
X ( s)        
s 3  4s 2  3s s( s  1)( s  3) s  0 s  (1) s  (3) s s  1 s  3
EE 3512: Lecture 22, Slide 2
Distinct Roots (Cont.)
• Example:
s2 c1 c2 c3
X ( s)    
s 3  4s 2  3s s s  1 s  3
s2 2
c1  [ sX ( s )] s 0  
( s  1)( s  3) s 0 3
s2 1
c 2  [( s  1) X ( s )] s  1  
s ( s  3) s  1
2
s2 1
c3  [( s  3) X ( s )] s  3  
s ( s  1) s  3
6
(2 / 3) (1 / 2) (1 / 6) 2 1 1
X ( s)     x(t )   e t  e 3t , t  0
s s 1 s3 3 2 6
• This can be checked with MATLAB:
num = [1 2]
den = [1 4 3 0]
[r,p] = residue(num, den)

r= p=
-0.1667 -3
-0.5000 -1
0.6667 0

EE 3512: Lecture 22, Slide 3

Distinct Complex Poles
• Recall if all coefficients of the denominator are real, then the polynomial must
have a combination of real and/or complex conjugate poles.
• We can factor this polynomial into the following form:
c1 c1 cN p1    j
X ( s)    ...  where
s  p1 s  p1 s  pN p1    j
• The inverse transform is given by:
x(t )  c1e p1t  c1e p1t  ...  c N e pN t  2 c1 et cos(t  c1 )  ...  c N e pN t
• How can this be visualized in the s-plane? (Hint: frequency and bandwidth)
• Example:
s 2  2s  1 c1 c1 c3
X (s)  3   
s  3s 2  4s  2 s  (1  j ) s  (1  j ) s  (1)
s 2  2s  1 3
c1  [( s  1  j ) X ( s)] s  1 j     j2
( s  1  j )( s  1) s  1 j
2
s 2  2s  1
c3  [( s  1) X ( s)] s  1  4
( s  1  j )( s  1  j ) s  1

c1  5 / 2 c1  126.87   Re{ p1 }  1

x(t )  5e t cos(t  126.87)  4e t , t  0
EE 3512: Lecture 22, Slide 4
Repeated Poles
• If the denominator has repeated poles, the expansion is of the form:
c c2 cr cr 1 cN
X ( s)  1   ...   ...   ... 
s  p1 (s  p1 ) 2 (s  p1 ) r s  pr 1 s  pN
• The residues for the nonrepeated poles are calculated the same as before:
ci  [( s  pi ) X ( s)] s  pi , i  r  1, r  2, ..., N
• The residues for the repeated poles are calculated by:
c r  [( s  p1 ) r X ( s)] s  p1
1 di
c r i  [ i [( s  p1 ) r X ( s)]] s  p1 , i  1, 2, ..., r  1
i! ds
• If the poles are all real, the inverse can be compactly written using:
t N 1 1
e at 
( N  1)! ( s  a) N
• Repeated complex poles and complex conjugate poles can also be dealt with
by combining the complex conjugate poles into one quadratic term, and
solving for the residues by equating terms of a polynomial:

c1 c cs  d cs  d
X ( s)   1  ...   ...   ...
s  p1 s  p1 (s  p1 )( s  p1 ) s  p1
2 2

EE 3512: Lecture 22, Slide 5

Repeated Poles (Cont.)
• Example:
5s  1 c1 c1 c3
X ( s)    
s 3  3s  2 s  1 ( s  1) 2 s  2
d
 
c1   ( s  1) 2 X ( s )    d  5s  1 
  
 ds  s  1  ds  s  2   s  1
 1  1    9 
 (5s  1)( )    
(5)   2 
 1
 s  2  s  2   s  1  ( s  2)  s 1
2

  5s  1 
c2  ( s  1) 2 X ( s) s 1  
  2
 s 2  s 1
c3  ( s  2) 2
X ( s )
s 2 1
Compute the inverse of each term individually and sum them:
x(t )  e t  2tet  e 2t , t  0
• This can be easily checked using MATLAB.

EE 3512: Lecture 22, Slide 6

Numerator Has A Larger Order (M  N)
• What if the numerator has a degree larger than the denominator?
• The polynomials can be decomposed using long division:
B( s) b0  b1 s  b2 s 2 ...  bM s M R( s )
X ( s)    Q ( s ) 
A( s) a0  a1 s  a 2 s 2 ...  a N s N A( s)
• The inverse transform of Q(s) is computing using the transform pair:
dN
N
 (t )  s N , N  1, 2, ...
dt
• Example:
s 3  2s  4 20s  12
X (s)   s4 2
s 2  4s  2 s  4s  2
d
x(t )   (t )  4 (t )  ...
dt
The long division can be performed using the MATLAB command deconv:
num = [1 0 2 -4]
den = [1 4 -2]
[Q,R] = deconv(num, den)

EE 3512: Lecture 22, Slide 7

Pole Locations Determine the Behavior of a Signal
• We have seen that the structure of the poles in the denominator polynomial
play a significant role in the determining the structure of the signal:
distinct real pole  ce pt
repeated pole  c1e pt  c2 te pt
complex conjugate poles  cet cos(t   )
repeated complex conjugate poles  c1et cos(t  1 )  c2 tet cos(t   2 )

• As a result, the signal can be determined directly from the poles. Modifying
the numerator does not change the signal significantly, it just changes the
values of the constants associated with the above terms.
• We can observe that x(t) converges to zero as t   if and only if the poles of
the signal all have real parts strictly less than zero:
Repi   0 for i  1, 2, ..., N
Hence, the poles play a critical role in the stability of a signal. The field of
Control Systems deals with stability issues in signals and systems.
• If X(s) has a single pole at s = 0, the limiting value of x(t) is equal to the residue
corresponding to that pole:
lim x(t )  [ sX ( s )] s 0
t 

EE 3512: Lecture 22, Slide 8

Transforms Containing Exponentials
• One last case we want to consider:
B0 ( s ) B1 ( s )  h1s Bq ( s )  hq s
X ( s)   e  ...  e
A0 ( s ) A1 ( s ) Aq ( s )
• The function is not rational in s. Such functions are referred to as
transcendental functions of s.
• Our approach is to use the methods previously described to compute the
inverse of the rational transforms, and then to apply the time shift property:
q
x(t )  x0(t )   xi (t  hi )u(t  hi ), t  0
i 1

• Such functions arise when the Laplace transform is applied to a piecewise-

continuous function, such as a pulse:
1 1 cs
u (t )  u (t  c)   e
s s
• Example:
s 1 1  s s  2 3 s / 2
X (s)  2  e  2 e
s 1 s 1 s 1
s 1 s2
Using : ( cos t  sin t)u(t)  2 ( cos t  2 sin t)  2
s 1 s 1
gives : x(t)  cos t  sin t  e (t 1) u (t  1)  [cos(t  1.5)  2 sin( t  1.5)]u (t  1.5), t  0
EE 3512: Lecture 22, Slide 9
Summary
• Introduced a method for funding the inverse Laplace transform using partial
fractions expansion:
1) Factor the denominator.
2) Assess the complexity of the poles (e.g., distinct vs. repeated, real vs.
complex conjugate pairs).
3) Compute the coefficients of the expansion using the method of residues.
4) Write the inverse Laplace transform by inspection.
• Discussed the influence poles have on the resulting signal. We will study this
more carefully later in the course.
• Note that the MATLAB Symbolic Toolbox can be used to find the inverse
transforms:
syms X s x
X = (s+2)/(s^3+4*s^2+3*s);
x = ilaplace(X);
x=
-1/6*exp(-3*t)-1/2*exp(-t)+2/3
ezplot(x,[0,10])

EE 3512: Lecture 22, Slide 10