0% found this document useful (0 votes)
138 views6 pages

Convolution PDF

The document discusses convolution and the Laplace transform. Convolution is defined as the integral of one function multiplied by the other function shifted in time. The Laplace transform of a convolution is equal to the Laplace transforms of the individual functions multiplied together. Convolution can be used to solve differential equations by finding the impulse response function and convolving it with the forcing term.

Uploaded by

Hary Kriz
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
138 views6 pages

Convolution PDF

The document discusses convolution and the Laplace transform. Convolution is defined as the integral of one function multiplied by the other function shifted in time. The Laplace transform of a convolution is equal to the Laplace transforms of the individual functions multiplied together. Convolution can be used to solve differential equations by finding the impulse response function and convolving it with the forcing term.

Uploaded by

Hary Kriz
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

170 CHAPTER 5.

LAPLACE TRANSFORMS

10. Solve y 00 + 2y 0 + 4y = f (t), y(0) = 0, y 0 (0) = 0, where f (t) is given


in the previous problem.
11. Graph the function f (t) = t − (2t − 2)u(t − 1) + (2t − 4)u(t − 2) −
(2t − 6)u(t − 3) + ....
12. Solve y 00 + 2y 0 + 4y = f (t), y(0) = 0, y 0 (0) = 0, where f (t) is given
in the previous problem.
13. Consider f (t) = e2t made into a periodic function fe(t) by taking
fT (t) where T = 1.
(a) Plot fe(t) for 0 < t < 4.
(b) Find L[fe(t)]
(c) y 00 + 2y 0 + 3y = fe(t), y(0) = 0, y 0 (0) = 0,
14. Use the differentiation theorem to verify that L[t u(t−a)] = e−as s12
15. Use appropriate theorems to compute L[t sin tet u(t − a)]

5.5 Convolution and the Laplace Trans-


form
We introduce a new operation ∗ between two functions called the
convolution.

Convolution of Two Functions


Let f (t) and g(t) be two functions. Define a new function:
Z t
f ∗ g(t) = f (t − w)g(w) dw
0

Note that f ∗ g is itself a function of t. Moreover note that if we


substitute v = t − w then dv = −dw and the integral becomes
Z w=0 Z w=t
f (v)g(t − v)(−1) dw = f (v)g(t − v) dw
w=t w=0

which is g ∗ f .

Example 5.29 Find t ∗ 1


5.5. CONVOLUTION AND THE LAPLACE TRANSFORM 171

Solution: Since f ∗ g = g ∗ f, we can compute 1 ∗ t easier, so let


f (t) = 1 and g(t) = t. Then
Z Z ¯t
t t
w2 ¯¯ t2
f ∗g = f (t − w)g(w) dw = w dw = =
0 0 2 ¯0 2
¤

We see that convolution is not the same as regular multiplication.

Example 5.30 Find t ∗ et

Solution: We set f (t) = t and g(t) = et .


Then
Z t Z t
f ∗g = f (t − w)g(w) dw = (t − w)ew dw
0 0

Z t
=t tew − wew dw
0

= (tew − wew + ew ))|t0 = et − t − 1


¤

The following result shows why convolutions are important:

Laplace Transform of the Convolution


Let f (t) and g(t) be two functions with Laplace transforms F (s) and G(s),
respectively. Then:

L[f ∗ g] = F (s)G(s)
and
L−1 [F (s)G(s)] = f ∗ g

Example 5.31 Find L[t2 ∗ et ]


172 CHAPTER 5. LAPLACE TRANSFORMS

Solution: Since L[f ∗ g] = F (s)G(s), we have

2 1
L[f ∗ g] =
s3 s − 1
¤

We can use convolution as an alternative to partial fractions as


shown next.

Example 5.32 Solve y 00 + y = e2t , y(0) = 0, y 0 (0) = 0.

Solution: After taking Laplace transform of both sides we get:


1
(s2 + 1)Y (s) =
s−2
or

1 1
Y (s) =
s2 +1s−2
so setting F (s) = s21+1 and G(s) = s−2
1
we see that
Y (s) = F (s)G(s) so y(t) = f ∗ g where f (t) = sin t and g(t) = e2t .
The convolution is
Z t
e2(t−w) sin w dw
0

which is Z t
e2(t−w) sin w dw.
0

This is the same thing as


Z t
2t
e e−2w sin w dw.
0

At this point, we could integrate by parts to get the solution, but


we wish to introduce a slick trick to avoid integration by PARTS,
since the integrand looks like the definition of a Laplace transform,
5.5. CONVOLUTION AND THE LAPLACE TRANSFORM 173

where s = 2, and since 1 − u(w − t) is equal to zero for w > t and


is equal to one for w < t (we view t as fixed), we may rewrite
Z t
2t
e e−2w sin w dw
0
Z t Z ∞
2t −2w 2t
=e [1−u(w−t)]e sin w dw+e [1−u(w−t)]e−2w sin w dw
0 t
Z ∞
= e2t [1 − u(w − t)]e−2w sin w dw
0

= e2t L [(1 − u(w − t)) sin w] (2)


(note s = 2 in the Laplace transform definition).

= e2t (L [sin w](2) − L[u(w − t) sin w] (2))

= e2t (L [sin w](2) − L[u(w − t) sin(w − t + t)] (2))

µ ¶
2t 1
=e − L [u(w − t) sin(w − t) cos t + u(w − t) cos(w − t) sin t] (2)
22 + 1

µ ¶
2t 1
=e − L [u(w − t) sin(w − t) cos t + u(w − t) cos(w − t) sin t] (2)
22 + 1

µ ¶
2t 1
=e − cos tL[u(w − t) sin(w − t)](2) − sin tL[u(w − t) cos(w − t)](2)
5
µ ¶
2t 1 1 2
=e − cos t(e−2t 2 ) − sin t(e−2t 2 )
5 2 +1 2 +1

1 1 2
= e2t − cos t − sin t
5 5 5
174 CHAPTER 5. LAPLACE TRANSFORMS

Perhaps this was better done with PARTS, but we wanted to


illustrate the power of the Laplace transform ¤

The advantage of convolution is that we can solve any spring mass


system without actually having the forcing function, as illustrated
in the next example.

Example 5.33 Solve y 00 + y = g(t), y(0) = 0, y 0 (0) = 0 for any


g(t).

Solution: After taking Laplace transform of both sides we get:

(s2 + 1)Y (s) = G(s)

or

1
Y (s) = G(s)
s2 +1
so setting F (s) = s21+1 and we see that
Y (s) = F (s)G(s) so y(t) = f ∗ g where f (t) = sin t.
The convolution (and hence the solution) is
Z t
y(t) = sin(t − w)g(w) dw
0

Convolution and Second Order Linear with Constant Coefficients


Consider
ay 00 + by 0 + cy = g(t), y(0) = 0, y 0 (0) = 0.
1
The solution is y(t) = f ∗ g where F (s) = as2 +bs+c , which is called the
transfer function and we call f (t) impulse response function for this
second order DE.

By superposition, we obtain the following:


5.5. CONVOLUTION AND THE LAPLACE TRANSFORM 175

Convolution and Second Order Linear with Constant Coefficients


Consider
ay 00 + by 0 + cy = g(t), y(0) = c1 , y 0 (0) = c2 .
If we have the particular solution to the homogeneous yhomo part (t) that sat-
isfied the initial conditions y(0) = c1 and y 0 (0) = c2 then

y(t) = yhomo part (t) + f ∗ g(t)

will solve the nonhomogeneous IVP.

Exercises
In 1-5, find the convolution.
1. t2 ∗ t3
2. et ∗ e3t
3. cos t ∗ 1
4. cos t ∗ sin t
5. u(t − 1) ∗ 1
6. Use convolution to solve y 00 + y = 2, y(0) = 0, y 0 (0) = 0
7. Use convolution to solve y 00 − 4y = t, y(0) = 0, y 0 (0) = 0
8. For a fixed constant β, use convolution to solve y 00 + 4y =
sin βt, y(0) = 0, y 0 (0) = 0
9. Use an integral approximation to estimate y(1) for y 00 + 4y =
2
et , y(0) = 0, y 0 (0) = 0
10. Find the impulse response function for an overdamped spring
mass system my 00 + by 0 + ky = g(t)
11. Find the impulse response function for an underdamped spring
mass system my 00 + by 0 + ky = g(t)

You might also like