Math 320 Lia Vas

The Laplace Transform

The Laplace transform is an integral operator (meaning that it is dened via an integral and that it maps one function to the other). It can be useful when solving dierential equations because it transforms a dierential equation into an ordinary equation. Laplace transform

Dierential Equation in y Get solution y .

Ordinary Equation in Y

Inverse Laplace transform

Find solution Y.

The Laplace transform of a piecewise continuous and exponentially bounded function f (t), dened for non-negative t-values, is dened to be

L[f (t)] =
0

est f (t)dt

Note that the integral on the right side is not a function of t any more but a function of variable s. We shall denote the resulting function L[f (t)] by F (s). Example. Find the Laplace transform of the following functions. a) f (t) = 1 d) f (t) = t2

b) f (t) = eat e) f (t) = sin at

c) f (t) = t f) f (t) = teat

1 st e |0 = 1 for s > 0. Solutions. a) L[1] = 0 est dt = s s atst 1 1 1 at atst dt = as e b) L[e ] = 0 e |0 = as = sa for s > a. t st st c) L[t] = 0 te dt To evaluate this integral, use the integration by parts. We have e s 1 st 1 e | = for s > 0 . 0 s2 s2 d) Using the integration by parts twice, obtain that L[t2 ] = s23 . Note that using inductive argun! ment, it can be shown that L[tn ] = sn +1 . st a e) L(sin at) = 0 e sin(at)dt. Using the integration by parts twice, we obtain s2 + for s > 0. a2 1 at atst f) L[te ] = 0 te dt similarly as in part c) we obtain (sa)2 for s > a.

The Laplace transform is linear (since the integral which denes it is linear) thus L[af (t) + bg (t)] = aL[f (t)] + bL[g (t)] for any constants a and b.

In order to transform a dierential equation using the Laplace transform, we also need to know the Laplace transform of the derivatives of a function. Let Y denotes L[y ] for some function y (t). Then, L[y ] = sL[y ] y (0) = sY y (0), L[y ] = sL[y ] y (0) = s(sL[y ] y (0)) y (0) = s2 Y sy (0) y (0), L[y ] = sL[y ] y (0) = s3 Y s2 y (0) sy (0) y (0). Continuing on this way we obtain that L[y (n) ] = sn Y sn1 y (0) . . . sy (n2) (0) y (n1) (0). The following table summarizes the Laplace transform of some frequently used functions.

Function f (t) Laplace transform F (s) 1 1 s n! n t sn+1 1 at e sa n! tn eat (sa)n+1 a sin at s2 +a2 s cos at s2 +a2 b eat sin bt (sa)2 +b2 sa eat cos bt (sa)2 +b2 y sL[y ] y (0) (n) n n1 y s L[y ] s y (0) . . . sy (n2) (0) y (n1) (0) uc (t) ecs 1 s uc (t)f (t c) ecs F (s) ect f (t) F (s c) (t c) ecs t F (s)G(s) 0 f (t )g ( )d
The Laplace transform of a dierential equation will turn out to be an ordinary equation. However, in order to obtain a solution from the original equation from the solution of the transformed equation, we need to use the inverse Laplace transform L1 . Note that L1 [F (s)] = f (t) if L[f (t)] = F (s) Thus, the inverse Laplace transform of a function in the right column of the table above is the corresponding function in the left column. Example. 1. Find the Laplace transform of the following. a) t2 e6t 7t3 + 8, 2 b) e2t cos 3t + 2 sin 3t 5.

2. Find the inverse Laplace transform of the following functions. a)

5 , s2 +4

b)

8 , (s2)4

c)

10 , s2 +3s4

d)

s+4 , s2 +2s+5

e)

5s2 +3s2 s3 +2s2

f)

3s2 4s+5 . (s1)(s2 +1)

3. Solve the following initial value problems. a) y 2y + 2y = et , y (0) = 0, y (0) = 1 b) y + y = sin 2t, y (0) = 2, y (0) = 1.

2 s+2 6 5 Solution. 1. a) (s 42 +8 b) (s+2) 2 +9 + s2 +9 s 6)3 s4 s 5 1 8 8 1 3! 4 3 2t 2. a) L1 [ s25 ]= 2 L [ s22 ]= 5 sin 2t b) L1 [ (s ]= 6 L [ (s2) 3+1 ] = 3 t e +4 +4 2 2)4 c) Since s2 + 3s 4 factors as (s + 4)(s 1), in order to nd the Laplace transform, we need to A B 2 2 nd the partial fractions s+4 + s . We obtain A = 2, B = 2. So, L1 [ s + s ] = 2e4t + 2et . 1 +4 1 d) Since s2 +2s +5 cannot be factored in a product of two linear real terms, we have to complete to squares and to use the formulas for eat cos bt and eat sin bt. s2 +2s +5 = (s +2s +1)+4 = (s +1)2 +22 . s+1+3 s+1 3 2 3 t s+4 t = (s+1) 2 +22 = (s+1)2 +22 + 2 (s+1)2 +22 . Hence the inverse Laplace is e cos 2t + 2 e sin 2t. s2 +2s+5 +3s2 B C 2 1 3 1 e) Using the partial fractions again, we have that 5s =A +s is 2 + s+2 = s s2 + s+2 . L s3 +2s2 s 2 t + 3e2t . 3s2 4s+5 A +C 2 3 f) Using the partial fractions, we have that (s = s + Bs = s + ss2 L1 is 2et + 1)(s2 +1) 1 s2 +1 1 +1 cos t 3 sin t. 3. a) Find the Laplace transform of the entire equation. Denoting L[y ] = Y , we get s2 Y 1 +2 1 . Solving for Y we get Y = (s+1)(s . The partial fraction decomposition is 2sY + 2Y = s+1 s2 2s+2) 1 1 s+8 1 1 s1 7 1 Y = 5(s+1) + 5 (s1)2 +1 = 5(s+1) 5 (s1)2 +1 + 5 (s1)2 +1 . From here y = 1 (et et cos t + 7et sin t). 5 b) Find the Laplace transform of the entire equation. Denoting L[y ] = Y , we get s2 Y 2s1+Y = +5/3 2/3 2 s3 +s2 +8s+6 . Solving for Y we get Y = 2 . The partial fraction decomposition is Y = 2s + . s2 +4 (s2 +4)(s2 +1) s2 +1 s2 +4 5 1 From here y = 2 cos t + 3 sin t 3 sin 2t.
2

Step, Boxcar and Impulse Functions

In the last two examples we could have used other methods to solve the dierential equations (undetermined coecients or variation of parameters) instead of the Laplace transform. The real importance of the Laplace transform is in its use for nding solutions of the equations for which the other methods fail to be applicable. In particular, Laplace transform can be used to nd solutions of dierential equations involving discontinuous functions. Most frequently used are step, boxcar and impulse functions which frequently appear in physics. Step Function. The unit step or Heaviside function is the function The Laplace transform of uc is

uc (t) =

1, t c, 0, t < c.

L[uc (t)] =
c

st

1 st ecs e |c = dt = s s

Using the unit step function, the Laplace transform of a discontinuous function dened by g (t) = f (t c), t c, = uc (t)f (t c) 0, t < c. 3

can be found to be L[g (t)] = L[uc (t)f (t c)] =

c

est f (t c)dt = est f (t)dt = ecs L[f (t)]

=
0

es(t+c) f (t)dt = esc

0

Note that from this follows that L1 [ecs F (s)] = uc (t)f (t c) Another property of the Laplace transform often used is that

L[ect f (t)] =
0

est+ct f (t)dt =
0

e(sc)t f (t)dt = F (s c)

1, a t < b, can be represented as a combination 0, t < a and t b. of two unit step functions ua (t) ub (t). Indeed if t < a < b, both step functions are 0, so f (t) = 0. If a t < b, ua (t) = 1 and ub (t) = 0, so ua (t) ub (t) = 1 0 = 1. If t > b, both step functions are 1 so ua (t) ub (t) = 1 1 = 0. The Laplace transform of a boxcar function f (t) = ua (t) ub (t) can be found as L[ua (t) ub (t)] = bs eas es . s The boxcar function f (t) = The general boxcar function f (t) = A, a t < b, 0, t < a and t b.

Boxcar Function.

can be represented as the dierence ua (t) ub (t) multiplied by A. So, f (t) = A(ua (t) ub (t)) = Aua (t) Aub (t). The general square wave function is a periodic extension of the boxcar function. For example, the function dened by 1, 2k t < 2k + 1, f (t) = 0, 2k + 1 t < 2k + 2. for k a non-negative integer is a general square wave function. It can be represented as f (t) = k 1 u1 (t) + u2 (t) u3 (t) + . . . = 1 + k=1 (1) uk (t).

General boxcar function 4

Square wave function

Example. Find the solution of dierential equation y +y = 1, 5 t < 20, 0, t < 5 and t 20.

with the initial conditions y (0) = 0 and y (0) = 0. Graph the solution on interval [0, 30]. Solution. The function on the right side is a boxcar function given by u5 (t) u20 (t). Taking 20s 5s the Laplace transform we obtain s2 Y + Y = e s e s . From here Y = (e5s e20s ) s(s21+1) . In order to nd the inverse Laplace transform, it is sucient to nd the inverse Laplace transform of 1 and to use the property L1 [ecs F (s)] = uc (t)f (t c). So let F (s) = s(s21+1) and let us nd s(s2 +1) +C its partial fractions decomposition F (s) = A + Bs = 1 s2s . Then take the Laplace inverse s s2 +1 s +1 L1 [F (s)] = 1 cos t. Denote this last expression by f (t). Thus, the solution is y = L1 [Y ] = L1 [(e5s F (s) e20s F (s)] = u5 (t)f (t 5) u20 (t)f (t 20) = = u5 (t)(1 cos(t 5)) u20 (t)(1 cos(t 20)). In order to graph the solution using Matlab or a graphing calculator, we can represent it using curly bracket notation. The resulting function will consist of three branches. 1. Since both u5 (t) and u20 (t) are zero for t < 5, we obtain the rst branch of the function to be zero for t < 5. 2. On the interval 5 t < 20, the function u5 (t) is 1 and u20 (t) is 0. Thus, the second branch is 1(1 cos(t 5)) 0(1 cos(t 20)) = 1 cos(t 5) for 5 t < 20. 3. On the interval t 20, both functions u5 (t) and u20 (t) are 1. Thus, the third branch is 1(1 cos(t 5)) 1(1 cos(t 20)) = cos(t 5) + cos(t 20) on t 20. Hence 0, t<5 1 cos(t 5), 5 t < 20, y= cos(t 5) + cos(t 20), t 20. To graph y using Matlab, you can use syms x ezplot(0*x, [0, 5]) hold on ezplot(1-cos(x-5), [5, 20]) ezplot(-cos(x-5)+cos(x-20), [20, 40]) hold o axis([0 40 -3 3]) To graph y using TI83+, you can enter 0(X<5)+(1-cos(X-5))(5 X<20)+(-cos(X-5)+cos(X-20))(X 20) as a function and graph it. The inequality signs <, and other can be found in 2nd Math menu.

Impulse function. In many applications, it is necessary to represent phenomena of an impulsive nature using mathematical models. For example, voltages of large magnitude that act over a very short period of time. Consider a function h dened by h (t) = Note that
1 , 2h

0,
h

h < t < h, t h and t h.

h (t)dt =

Since this integral does not depend on h, Now let us dene (t) = lim h (t).
h0

h (t)dt h

1 1 1 h dt = t|h = 2h = 1 2h 2h 2h = 1 even if we let h 0.

The impulse function (t) is characterized by the following properties 1. 2. (t) = 0 for all values of t = 0

(t)dt = 1.

Since no ordinary function satises both of these properties, is not a function in the usual sense of the word. It is an example of generalized functions. (t) is called unit impulse function or Dirac delta function. Alternate denition of Dirac delta function can be found on wikipedia. The Laplace transform of (t) can be found to be 1 (to see this, nd the Laplace transform of h (t) and let h 0). More generally, L[ (t c)] = ecs Example. Find the solution of dierential equation y + y = (t ) with the initial conditions y (0) = 0 and y (0) = 0. Sketch the graph of the solution. and so Solution. Taking the Laplace transform, we obtain s2 Y + Y = es . Thus Y = es s21 +1 y = u (t) sin(t ). To graph the solution, represent it as 0, t< sin(t ), t .

y=

Use ezplot in combination with hold on and hold o to graph it in Matlab. On TI83, you can use 0(X< )+sin(X- )(X ).

Integral Equations. The Convolution

The convolution is an operation on two functions f (t) and g (t) dened as follows
t t

f (t) g (t) =
0

f (t )g ( )d =
0

f ( )g (t )d

The signicance of this operation for the Laplace transform is because L[f (t) g (t)] = L[f (t)] L[g (t)] This property implies that L1 [F (s) G(s)] = L1 [F (s)] L1 [G(s)]. Using convolution, the function equations of involving or yielding an integral could also be solved. Example. Solve the integral equation y (t) + 0 (t )y ( )d = 1. Solution. Note that the equation can be written as y + t y = 1. Let again Y = L[y ]. Taking the Laplace transform, we have that Y + s12 Y = 1 . Solving for Y, we have Y = s2s . Thus, y = cos t. s +1 The equation from the last example belongs to the class of integral equations known as Volterra integral equations. This class of equations was introduced in the early 1900s, by V. Volterra in his t study of population growth. The general form of these equations is y (t) + 0 f (t )y ( )d = g (t). Using convolution, one can also express solution of dierential equation involving any integrable function f (t) in terms of an integral involving f (t). Example. Find the solution of the equation y + y = f (t), with the initial conditions y (0) = 0 and y (0) = 0. Express your answer in terms of an integral involving function f (t). Solution. Let Y = L[y ] and F = L[f ]. The Laplace transform converts the equation to s2 Y + Y = = F s21 . Applying the inverse Laplace, you obtain F. From here Y = s2F +1 +1 1 1 ] = L1 [F ] L1 [ 2 ] = f sin t = y = L [Y ] = L [F 2 s +1 s +1
1 1 t t

f ( ) sin(t )d.
0

Solving Systems Using Laplace Transform

The Laplace transform can be used for solving systems of dierential equations. Taking Laplace transform of a system of dierential equations produces a system of ordinary equations. Solving this new system and taking the inverse Laplace transform of the solution produces the solutions of the original system. Example. Solve the following system. x = x 3y y = x + y x(0) = 1 y (0) = 0

Solutions. Let X denote L[x] and Y denote L[y ]. Start by taking the Laplace transform of both equations. Obtain the new system sX 1 = X 3Y sY = X + Y

Solving the second equation for X produces X = Y sY. Substitute that in the rst equation and 1 solve for Y. s(Y sY ) 1 = Y + sY 3Y sY s2 Y 1 = sY 4Y 1 = (s2 4)Y Y = s2 . 4 7

1 Thus X = Y sY = s2 + s2s . Finding the partial fractions decomposition for X and Y produces 4 4 1/4 1/4 1/4 3/4 1 2t 1 2t Y = s2 + s+2 and X = s2 + s+2 . Thus, x = L1 [X ] = 4 e +3 e2t and y = L1 [Y ] = e +1 e2t . 4 4 4

Practice Problems.
1. Find the Laplace transform of the following functions. a) t4 e2t + cos 5t 7 b)
t 0

3 et d

c)

t 0

sin(2 ) cos(2t 2 )d

2. Find the inverse Laplace transform of the following functions. a)

3 s2 +25

b)

5 (s+2)3

c)

s11 s2 +2s3

d)

s 2 s2 2s+5

e)

7s3 2s2 3s+6 s4 2s3

f)

7s2 41s+84 (s1)(s2 4s+13)

3. Use the Laplace transform to solve the following initial value problems. a) y + 3y + 2y = 0, y (0) = 1, y (0) = 0. b) y 4y + 4y = 0, y (0) = 1, y (0) = 1. c) y 6y + 5y = 2, y (0) = 0, y (0) = 1. t < 5, 0, t 5, 5 t < 10, with y (0) = 0 and y (0) = 0. 4. Solve the initial value problem y + 4y = 5, t 10. The function on the right side of the equation is known as ramp loading and can be represented as u5 (t)(t 5) u10 (t)(t 10). Assume that an undamped harmonic oscillator is described by the given dierential equation. Consider the graph of the motion and expand on its meaning.

5. Solve the initial value problem y + 4y = (t 4 ), y (0) = 1, y (0) = 0. 6. Assume that the initial value problem y + 3y + 4y = (t 3), y (0) = 0, y (0) = 0 models the motion y (in cm) of an oscillator as time t (in seconds) passes. Find the solution, write your answer as a piecewise function, sketch its graph and describe the motion of the oscillator. 7. Solve the initial value problem y + 4y = f (t), y (0) = 3, y (0) = 1. Express your answer in terms of an integral involving function f (t). 8. Solve the integral equation y (t) +
t (t 0

)y ( )d = t.
t 0

9. Solve the integro-dierential equation y (t) + 10. Solve the following system. x = x + y y = x y 8

y (t )e2 d = 1, y (0) = 1. x(0) = 1 y (0) = 2

Solutions.
24 s 7 1. a) (s+2) b) The function is the convolution of t3 and et . Thus the Laplace 5 + s2 +25 s 1 6 transform is L[t3 ]L[et ] = s64 s = s4 (s . c) The function is the convolution of sin 2t and 1 1) s s = (s22 . cos 2t. Thus the Laplace transform is L[sin 2t]L[cos 2t] = s22 +4 s2 +4 +4)2 3 1 t 2. a) 5 sin 5t b) 5 t2 e2t c) 3et 2e3t d) et cos 2t 2 e sin 2t e) 1 3 t2 + 6e2t f) 5et + 2e2t cos 3t 2 2 5e2t sin 3t s+3 3. a) The Laplace transform of the equation is s2 Y s + 3sY 3 + 2Y = 0 Y = s2 +3 = s+2 s+3 2 1 t 2t . The partial fraction decomposition is Y = s+1 s+2 . Thus y = 2e e . (s+1)(s+2)

b) The Laplace transform of the equation is s2 Y s 1 4sY +4+4Y = 0 Y = 1 1 The partial fraction decomposition is Y = s (s . Thus y = e2t te2t . 2 2)2

s3 s3 . s2 4s+4 (s2)2

Y (s2 6s + 5) = c) The Laplace transform of the equation is s2 Y + 1 6sY + 5Y = 2 s 2 s s 1 Y (s 1)(s 5) = 2 Y = s(s2 . The partial fraction decomposition is s s 5)(s1) 2 3 1 2 3 5t 1 t Y = 5s 20(s5) 4(s1) . Thus y = 5 20 e 4 e . 4. Note that the function on the right side of the equation can be represented as u5 (t)(t 5) u10 (t)(t 10). The Laplace transform makes the equation into s2 Y + 4Y = e5s s12 e10s s1 2. 1/4 1/4 1 1 5s 10s Thus Y = (e e ) s2 (s2 +4) . The fraction s2 (s2 +4) decomposes as s2 s2 +4 and its inverse 1 t 1 sin 2t. Thus, Laplace transform is f (t) = 4 8 y = L1 [Y ] = u5 (t)f (t 5) u10 (t)f (t 10) = u5 (t)( 1 (t 5) 1 sin 2(t 5)) u10 (t)( 1 (t 4 8 4 1 10) 8 sin 2(t 10)) = 0, 0 t < 5, 1 1 (t 5) 8 sin 2(t 5), 5 t < 10, = 5 14 1 8 sin 2(t 5) + 8 sin 2(t 10), t 10. 4 The solution is zero before 5. Between 5 and 10, it increases by oscillating about the line 1 (t 5). For t 10, the graph becomes one of 4 a simple harmonic oscillation (like shifted sine function) oscillating about 5/4. 5. The Laplace transform gives you s2 Y s + 4Y = e4s . Thus Y = cos 2t, t < 4, u4 (t) 1 sin 2(t 4 ) + cos 2t = 1 2 cos 2t + 2 sin 2(t 4 ) t 4.
e4 +s . s2 +4

Then y =

6. Let Y = L[y ]. Applying the Laplace transform to the equation y + 3y + 4y = (t 3) with e3s y (0) = y (0) = 0 produces s2 Y + 3sY + 4Y = e3s . From here Y = s2 +3 . Complete the s+4 3 1 2 2 +4 9 = denominator of F (s) = s2 +3s+4 to a sum of squares. s + 3s + 4 = s + 2s( 2 ) + 9 4 4

(s +

3 2 ) 2

7 . 4

Thus f (t) = L [F (s)] = L

3s

[ (s+ 31)2 + 7 2 4

]=

2 L1 [ (s+ 32)2 + 7 7 2 4

e y = L1 [Y ] = L1 [ s2 +3 ] = L1 [e3s F (s)] = u3 (t) f (t 3) = u3 (t) s+4

7 2 3t/2 e sin t. 2 7 7 2 3(t3)/2 e sin (t 3) 2 7

]=

y=

0,
2 3(t3)/2 e 7

sin

7 (t 2

3),

t < 3, t 3.

The object starts oscillating after the rst three second. It oscillates with a decreasing ampli2 3(t3)/2 tude given by e converging to 0. So the oscillations become negligible in time. 7 7. Let Y = L[y ] and F = L[f ]. The Laplace transform converts the equation to s2 Y 3s + 2 s 1 + F s21 = 3 s2s 1 +1 F 2 . 1 + 4Y = F Y (s2 + 4) = 3s 1 + F Y = 3 s2 +4 +4 +4 2 s2 +4 2 s2 +4 Applying the inverse Laplace transform, you obtain y = 3 cos 2t 1 sin 2t + 1 f (t) sin 2t = 2 2 1 1 t 3 cos 2t 2 sin 2t + 2 0 f ( ) sin 2(t )d. 8. The equation is y + t y = t. The Laplace transform gives you Y + and so y = sin t.
1 Y s2

1 . s2

Then Y =

1 s2 +1

1 1 9. The equation is y + y e2t = 1. Thus sY 1 + Y s+2 = 1 Y (s + s+2 ) = 1 +1 s s s(s+2)+1 (1+s)(s+2) (1+s)(s+2) 1+s s+2 Y s+2 = s Y = s(s2 +2s+1) = s(s+1)2 = s(s+1) . Find the partial fraction decomposition 1 to be Y = 2 s+1 y = 2 et . s

10. Let X = L[x] and Y = L[y ]. Taking L of both equations gives you sX 1 = X + Y and sY 2 = X Y. From the rst equation Y = sX + X 1. Plugging that in the second gives s+3 . you s(sX + X 1) 2 = X (sX + X 1) s2 X + 2sX + 2X = s + 3 X = s2 +2 s+2 s2 +3s+s+3s2 2s2 2s+1 Thus Y = = s2 +2s+2 . s2 +2s+2
+1+2 s+1 1 1 t Then x = L1 [X ] = L1 [ (ss+1) [ (s+1) cos t + 2et sin t and 2 +1 ] = L 2 +1 + 2 (s+1)2 +1 ] = e s+1 s+21 s+1 1 t y = L1 [Y ] = L1 [ s22 ] = L1 [ (2 ] = L1 [2 (s+1) cos t et sin t. 2 +1 (s+1)2 +1 ] = 2e +2s+2 s+1)2 +1

10