Notes by David Groisser, Copyright c 1998
Partial Fractions and Laplace Transform Problems The general method for using the Laplace transform to solve a linear dierential equation L[y ] = g (with some initial conditions) is to (1) transform both sides of the equation, (2) solve for Y (s), then (3) invert the transform to nd y (t). The most dicult part is generally step (3), which often involves rewriting Y (s) using partial fractions in such a way that the result is a linear combination of terms that appear on the Laplace transform table. The examples presented in the textbook (Nagle & Sa) usually involve combining fractions between steps (1) and (2). While this approach is not wrong, often it results in more work than necessary, partly because a common denominator in Y (s) can have very high degree, and partly because the whole purpose of partial fractions is to un-combine fractions and rip apart common denominators. Below is an example of how to do a problem without rst combining fractions. To shorten this and other examples, it helps to know three frequently used simple partial fraction identities (which can easily be re-derived if you forget them): 1. 2. 3.
1 (x+a)(x+b) 1 x2 a2 x x2 a2
1 ( 1 ba x+a
1 ) x+b
= =
1 (xa)(x+a) x (xa)(x+a)
1 ( 1 2 a x a
1 ) x+ a
(this follows from previous line)
=1 ( 1 + 2 x a
1 ) x+ a
Example. Use Laplace transforms to solve the initial-value problem y 4y = 4t 8e2t , y (0) = 0, y (0) = 5. Method. Laplace-transforming this IVP gives s2 Y 5 4Y = 4/s2 8/(s + 2), so (s2 4)Y = 5 + and hence Y = s2 4 8 , 2 s s+2 (1)
5 4 8 + 2 2 2 . 4 s (s 4) (s 4)(s + 2)
One way to proceed is as in the textbook: combine fractions en route to equation (1), getting a 5th degree denominator, then set the resulting fraction equal to something of the form A/s + B/s2 + C/(s 2) + D/(s + 2) + E/(s + 2)2 , then multiply out, then solve for A, B, C, D, E . The alternative Im suggesting is to break expression (1) into three sub-expressions, and proceed as follows. Note that our basic identities easily handle two of these three sub-expressions, and for the third sub-expression we only have to deal with a cubic denominator instead of a 5th degree denominator. s2 5 1 5 1 1 =5 2 = ( ) 4 s 4 4 s2 s+2 using identity 2.
s2 (s2
4 1 1 1 =4 2 2 = 2 2 using identity 1 with x = s2 4) s (s 4) s 4 s 1 1 1 1 = ( ) 2 using identity 2. 4 s2 s+2 s 1
8 8 8 A B C = = = + + . (s2 4)(s + 2) ((s 2)(s + 2))(s + 2) (s 2)(s + 2)2 s 2 s + 2 (s + 2)2 Multiplying out, 8 = A(s + 2)2 + B (s 2)(s + 2) + C (s 2). Now use your favorite method to nd A, B, C . (My favorite is to expand out the righthand side and collect like powers of s, getting 8 = A(s2 + 4s + 4) + B (s2 4) + C (s 2) = s2 (A + B )+ s(4A + C )+(4A 4B 2C ), and then equate coecients of equal powers of s on the two sides of this equation, getting the simultaneous equations A + B = 0, 4A + C = 0, 4A 4B 2C = 8, which I then solve.) When done correctly, you wind up with A = 1/2, B = 1/2, C = 2, so (s2 1/2 1/2 2 8 = . 4)(s + 2) s 2 s + 2 (s + 2)2
Inserting these partial-fractions decompositions for our three sub-expressions into equation (1) and combining only terms with identical denominators, Y = 1 1 1 1 +2 , s2 s+2 (s + 2)2 s2
and hence, using the Laplace transform table, y (t) = e2t e2t + 2te2t t.
