LECTURE 6

Laplace Transform Techniques

1. Review of Laplace Transform
Theorem 6.1. Let f be a function on the half line 0; 1 with the following properties

1. 2. 3.

f; f 0 ; f 00 ; : : : ; f n,1 are continuous f n is piecewise continuous There exists constants K , a and M such that jf xj  Keat for t  M: Z1 e,st f tdt 0 Z1 0 L f s = e,st f 0 tdt

Then the Laplace transforms

L f s =
. . .

Z1 h i L f n s = e,st f n tdt

all exist for s a, and moreover
0

h i L f n s = sn L f s , sn,1f 0 ,


, sf n,2 0 , f n,1 0 : Theorem 6.2. If F s = L f s and Gs = L g s both exist for s a  0, then Zt Zt , 1 L F sGs = f t , g d = f  gt , d :
Theorem 6.3. If F s = L f s then
0 0

Z +i1 1 est F sds f t = 2i ,i1

Here the constant is chosen so that all the singularities of the integrand occur to the left of Res = . For t 0, the contour may be closed by an in nite semicircle in the left half plane. Example 6.4. The following example shows how the Laplace transform can be used to construct a solution of the boundary value problem @ , a2 @ 2 = 0 6.1 @t @x2 6.2 x; 0 = 0 6.3 0; t = f t in the region x 0, t 0.
22

1. REVIEW OF LAPLACE TRANSFORM

23

Multiplying 6.1 by e,st and integrating along t from 0 to 1 we get @   @2  2 L @t s , a L @x2 = 0 or  @2  2 sL , x; 0 , a L @x2 = 0 ; which becomes  @2  s L @x2 , a2 L = 0 6.4 once we employ the initial condition x; 0 = 0. If x; t is su ciently well behaved, and 6.4 becomes

 2  2 @ L @ @x2 = @x2 L

@ 2  x; s , s x; s = 0 : @x2 a2 Regarding this a second order linear di erential equation with constant coe cients, we have as a general solution ps ps 6.5 x; s = Ase, a x + B se a x :
In order that our solution be a bounded function of t for all x  0, we will demand that its Laplace transform go to zero as s ! 1. This will require the coe cient B s = 0. To determine As, we note that the boundary condition 0; t = f t has as its Laplace transform 6.6 0; s = L f s : On the other hand, when we evaluate 6.5 at x = 0, we have 6.7 0; s = As : Comparing this with 6.4 with 6.5 we see that the coe cient As must be exactly L f s. Thus, x; s = L f se, a x : 6.8 To nd the solution x; t we now note that  x  p 2 , 4x a2 t e 6.9 e, as x = L 2pat 3 =2 and apply the Convolution Theorem to get    x 2 , 4x a2 t 6.10 x; t = L,1 L f L 2pat e 3 =2 Z t x2 2 d p f t ,  ,3=2e, 4a = 2ax 6.11 0
ps

1. REVIEW OF LAPLACE TRANSFORM

24

@ , a2 @ 2 = 0 6.12 @t @x2 6.13 x; 0 = f x for ,1 x +1, t 0. As before, the Laplace transform of the heat equation with respect to the time variable t is @ 2 x; s = 0 sx; s , x; 0 , a2 @x 2 or 2 sx; s , f x , a2 @ @x2 = 0 or @ 2  , s  = , f x : 6.14 @x2 a2 a2 The general solution to a ordinary di erential equation of the form y00 , pxy0 + qxy = gx is  Z x y 
g
  Z x y 
g
 2 6.15 d
y  x  + B + W 1y ; y 
 d
y1 x yx = A , W 1 y1 ; y2 
 1 2 where y1 x and y2 x are any two linearly independent solutions of the corresponding homogeneous problem y00 + pxy0 + qxy = 0 : See Boyce and DiPrima, Section 3.6.2. In the case at hand, we can take
6.16 6.17 6.18 and
s a x y1 x = e, ps y2 x = e a x x gx = ,f a2

Example 6.5. Let x; t satisfy the equation

0 , y0 y2 = 2 s W y1 ; y2 x = y1 y2 1 a

so we have 6.19 6.20

In order to avoid uncontrolled growth in x; s as x ! +1 we require Z1 p ps e, as

f 
d
B = 2a1 6.21 0 and similarly to control the behavior of x; s as x ! ,1, we require Z ,1 ps ps 6.22 e a
f 
d
A = , 2a1 0

 ps Z x ps 1
a x; s = A + 2aps e f 
d
e, a x 0  ps Z x ps ,
a + B , 2aps e f 
d
e a x
0

2. THE METHOD OF IMAGES

25

So 6.23 6.24 6.25 6.26 6.27 6.28

 1 Z ,1 ps ps Z x ps 0 1
x a a x; s = , 2aps e f 
d
+ 2aps e f 
d
e, a x 0 0 Z  1 Z 1 ps ps x ps 1 ,
,
a a + 2aps e f 
d
, 2aps e f 
d
e a x 0 0 ps  ps  Z ,1 ps Z 1 ps ps ps = , 2a1 e a
f 
d
e, a x + 2a1 e, a
f 
d
e a x x x
=

Z +1 1 ps ps 1 a 
,x f 
d
+ p p a x,
f 
d
e 2a s e ,1 2a s x Z +1 1 ps Zx 1 p , as j
,xj f 
d
+ p ps e, a j
,xj f 
d
6.29 e = 2 a s 2 a x Z,1 ps +1 1 p e, a j
,xj f 
d
= 6.30 ,1 2a s The third step comes form the observation that ps ps a 
,x a j
,xj 6.31 ep = e, p ; 8
2 ,1; x s s  x ,
 , j
, x j 6.32 ea = e a ; 8
2 x; +1 Now we employ the identity that   p xj2 1 e, j
4, a2 t p p ps e, as j
,xj 6.33 s = 2a1 L 2a1 t to write 6.34 L x; t s  x; s Z +1  1 1 j
,xj2  = L 2ap pt e, 4a2 t sf 
d
6.35 ,1 Z +1 1  xj2 a2 t f 
d
s p e, j
4, 6.36 = L ,1 2a t Taking the inverse Laplace transform of both sides, we get Z +1 1 j
,xj2 p e, 4a2 t f 
d
: 6.37 x; t = ,1 2a t
x

2. The Method of Images

Consider now the problem 6.38 6.39 6.40

@ , a2 @ 2 = 0 @t @x2 x; 0 = f x 0; t = 0

; 0 x 1 ; 0 t 1 :

Note that in the statement of this problem we are given no information and asked no questions about the behavior of solutions as x ! ,1. Yet in the preceding example, the assumption that x; s was well behaved as x ! 1, was crucial to the determination of the constants of integration As and B s; which in turn allowed us to construct an explicit solution. Thus, the technique used in the preceding example can not be applied directly to the case at hand.

2. THE METHOD OF IMAGES

26

But a simple trick will su ce. We simply extend the domain of f x to the whole real line by de ning f ,x  ,f x ; 8 x 2 R+ : The result of the preceding example then implies Z +1 1 j
,xj2 p e, 4a2 t f 
d
6.41 x; t = ,1 2a t will satisfy the di erential equation @ , a2 @ 2 = 0 @t @x2 and the boundary condition x; 0 = f x : We now verify that this solution also satis es the boundary condition 0; t = 0 ; 0 t +1 : Setting x = 0 in 6.41 yields Z +1 1
2 2 t f 
d
p e, 4a 6.42 : 0; t = ,1 2a t Noting that the exponential function is a even function of
and that the function f 
 is de ned to be an odd function of
, we conclude that the integrand in 6.42 is an odd function of
and so its integral between ,1 and +1 vanishes. Thus, 0; t = 0 :

Homework: 2.2.2, 2.2.3, 2.2.5