Second Order Dierential Equations

Equivalent Forms; Initial Value Problems second order dierential equations take the form d2 y dy = f x, y, , dx2 dx where f (x, y, z ) is a continuous function of the variables x, y and z in some region R R3 and is continuously dierentiable with respect to y and z throughout that region.
dy , then we can write the above equation If we dene z = dx equivalently as a system of two (coupled) rst order equations:

In general,

dy = z; dx

dz = f (x, y, z ). dx

Written in this form and drawing on experience with rst order equations, we would expect to have to give initial values y (x0 ) = y0 , z (x0 ) = z0 ,

in order to specify an initial value problem. Going back to the second order equation, this means we need to give y (x0 ) = y0 , dy dy (x0 ) = z0 ; equivalently : (x0 ) = y1, dx dx

in order to specify an initial value problem for it. General Solutions We say that an expression y (x, c1 , c2 ) is a general solution for the second order dierential equation above if, as a function of x, it satises the dierential equation, and,
1

given initial values as above at a point x0 , with (x0 , y0 , y1) in the region R where f (x, y, z ) has the required properties, we can solve the equations y (x0 , c1 , c2 ) = y0, dy (x0 , c1 , c2 ) = y1 dx

for c1 and c2 to satisfy that initial value problem. Example 1 Consider the second order dierential equation d2 y 2 dy 2 = 2y 2 dx x dx x in the region x > 0, and let us try to nd solutions of the form y = xr . Substituting, we have r (r 1) xr2 = 2r xr2 2xr2 or Since xr2 is not identically zero, we need to solve the so-called indicial equation r 2 3r + 2 = 0 r = 2 or r = 1. So we conclude that y (x) = x and y (x) = x2 are solutions, which is easily checked out. Since the equation is linear with respect to y (if y and y are solutions, so is any linear combination of the two) it then makes sense to try a general solution of the form y (x, c1 , c2 ) = c1 x + c2 x2 . To satisfy initial conditions as specied above at x0 > 0 we need c 1 x0 + c 2 x0 2 = y 0 , c 1 + 2 c 2 x0 = y 1
2

xr2 (r 2 3r + 2) = 0.

a system of two linear algebraic equations. The determinant of the coecients of c1 and c2 on the left hand side is 2 x0 2 x0 2 = x0 2 = 0, so we conclude we can always obtain a unique solution of these equations. We conclude therefore that y (x, c1 , c2 ) = c1 x + c2 x2 is the general solution of the second order dierential d2 y 2 dy 2 equation dx 2 = x dx x2 y for x > 0. A General Method of Solution in the Autonomous Case If our second order dierential equation does not depend explicitly on x, which is technically referred to as the autonomous, i.e., self-governing, case, we can employ a transformation of variables to reduce the second order equation to two rst order dierential equations. Assuming the equation is dy d2 y = f y, , dx2 dx we let dy = z. dx Substituting this into the second derivative we have d2 y dz dz dy dz = = = z dx2 dx dy dx dy and thus the original second order equation becomes z dz = f (y, z ). dy

If we can solve this last equation to get z = z (y, c1 ), then the equation used to dene z becomes dy = z (y, c1 ) dx
3

which, being in separable form 1 dy = dx z (y, c1) can be integrated to give

1 d = x + c2 . z (, c1 )

Then, at least in principle, we can solve this to get y = y (x, c1 , c2 ). Success of the project ultimately depends on being able to get a closed form solution for the transformed second order equation, = f (y, z ), which has now become a rst order equation in z dz dy z and y . Example 2 We consider the autonomous second order dierential equation d2 y dy = y . dx2 dx Following the plan indicated above, we arrive at z dz dz = y z or = y. dy dy

This is solved immediately to obtain y 2 + c1 z = z (y, c1 ) = . 2 dy y 2 + c1 = . dx 2 There are three cases here, depending on whether c1 is 0, positive, or negative. Taking just the case where c1 is positive, we write c1 = c2 and we have c2 2 dy = dx. + y2
4

Then we have

Setting y = c w we have 2 1 dw = dx c 1 + w2 and we integrate to obtain 2 tan1 (w ) = x + c2 . c

c y = y (x, c, c2 ) = c tan( (x + c2 )). 2 The cases corresponding to c1 = 0 and c1 < 0 are handled in a similar way but dierent expressions are obtained. (Try those cases as an exercise.) Linear Second Order Dierential Equations It would be misleading to say that we can solve most second order dierential equations. In fact, the ones we can solve are rather special. Prominent among these are the linear second order dierential equations d2 y dy + p ( x ) + q (x) y = g (x), dx2 dx where p(x), q (x) and g (x) are known, continuous (or piecewise continuous) functions of x. Such a dierential equation is called homogeneous if g (x) 0 and inhomogeneous if g (x) 0, just as in the rst order case. For the present we consider only the homogeneous case d2 y dy + q (x) y = 0. + p(x) 2 dx dx A fundamental property of solutions of such a dierential equation is expressed in the following
5

Proposition If y1(x) and y2(x) are any two solutions of the above linear homogeneous second order o.d.e., then for any constants c1 and c2 , the linear combination y (x) = c1 y1(x) + c2 y2(x) is also a solution. The proof is immediate when we use d dy1 dy2 (x) + c2 c1 y1(x) + c2 y2(x) = c1 dx dx dx and the similar relationship for the second derivative of the linear combination. Constant Coecient Linear Homogeneous Second Order Dierential Equations These are dierential equations of the form d2 y dy + p + qy = 0 dx2 dx where p and q are constants. These dierential equations are particularly easy to solve. Since in the linear homogeneous rst order constant coecient case dy + py = 0 dx we have the exponential solution y (x) = epx it makes sense to guess that the corresponding second order equation might also have solutions of the form y (x) = erx . Substituting this proposed solution form into the dierential equation we have r 2 erx + p r erx + q erx = erx (r 2 + p r + q ) = 0. The exponential is never equal to zero so we must have r 2 + p r + q = 0, an algebraic equation which is called the characteristic equation for the dierential equation in question. Use of the quadratic
6

formula gives the two roots r = p 2 p 4q ; 2

we will refer to the two roots thus obtained as r1 and r2 . Then, retracing our steps, we can see that with y (x) = eri x , i = 1, 2, we obtain d2 y dy + p + q y = 0 = eri x (ri 2 + p ri + q ) = 0, i = 1, 2, 2 dx dx so that these are, indeed, solutions. Then, from the Proposition we have solutions y (x, c1 , c2 ) = c1 er1 x + c2 er2 x . It is natural to suspect that this might be the general solution. To check on this we suppose that we want to solve y (x0, c1 , c2 ) = y0 . d y ( x , c , c ) = y 0 1 2 1 dx Substituting y (x, c1 , c2 ) = c1 er1 x + c2 er2 x we have c 1 e r 1 x0 + c 2 e r 2 x0 = y 0 . c 1 r 1 e r 1 x0 + c 2 r 2 e r 2 x0 = y 1 The coecients of the unknowns c1 and c2 form the array e r 1 x0 r 1 e r 1 x0 This array has determinant e(r1+r2 )x0 (r2 r1 ) which is dierent from 0 if r1 = r2 , i.e., if the roots of the characteristic equation are distinct. Thus we have the form of the general solution in this case. We postpone consideration of
7

e r 2 x0 . r 2 e r 2 x0

the case r1 = r2 , wherein r 2 + p r + q = (r r1 )2 until a later lecture. Example 3 Find the solution of the initial value problem y (0) = 0, y (0) = 1.

d2 y dy 7 + 12 y = 0, dt2 dt Solution

In this case the characteristic equation is r 2 7 r + 12 = (r 3)(r 4) = 0,

yielding the roots r1 = 3, r2 = 4. Thus the general solution is y (t, c1 , c2 ) = c1 e3t + c2 e4t . Its derivative is y (t, c1 , c2 ) = 3 c1 e3t + 4 c2 e4t. At t = 0 we require y (0, c1 , c2 ) = c1 e30 + c2 e40 = c1 + c2 = 0, y (0, c1 , c2 ) = 3 c1 e30 + 4 c2 e40 = 3 c1 + 4 c2 = 1. Multiplying the rst equation by 3 and subtracting the result from the second equation we obtain c2 = 1. Then the rst equation gives c1 = 1 and the desired solution is y (t) = e3t + e4t. Example 4 Find the general solution of

d2 y dy 6 2 19 + 15 y = 0. dt dt Here the characteristic equation is 6 r 2 19 r + 15 = 0.

The roots are not immediately obvious. We recall the quadratic formula for nding the roots of a quadratic equation a r 2 + b r + c = 0: b b2 4 a c r = . 2a Applying this formula to our case we obtain the roots 19 (19)2 4 6 15 19 361 360 r = = , 26 12 which yields the roots r1 = 20 5 18 3 = ; r2 = = . 12 3 12 2

Thus the general solution is y (t, c1, c2 ) = c1 e5t/3 + c2 e3t/2. Example 5 Two blocks of material are placed in contact with each other on a large stone table whose temperature is maintained at 0 . The heat transfer coecient between the two blocks and between the blocks and the table is k > 0. Describe the evolution of the temperature T1 (t) in the rst block. Solution Applying the laws of heat conduction we obtained a coupled system of two rst order equations: dT1 = k (T1 T2 ) k (T1 0) = 2 k T1 + k T2 ; (I) dt dT2 = k (T2 T1 ) k (T2 0) = 2 k T2 + k T1 . (II) dt We dierentiate the equation (I) with respect to t: d2 T1 dt1 dT2 = 2 k + k . (III) dt2 dt dt
9

We substitute equations (I) and (II) into (III) and obtain d2 T1 = 2k ( 2k T1 + k T2 ) + k ( 2k T2 + k T1 ) dt2 To eliminate T2 in equation (IV) we multiply equation (I) by 4 k to obtain dT1 4k = 8 k 2 T1 + 4 k 2 T2 . (V) dt Then adding equations (IV) and (V) we obtain d2 T1 dT1 + 4 k = 3 k 2 T1 , 2 dt dt or dT1 d2 T1 + 4 k + 3 k 2 T1 = 0. 2 dt dt Here the characteristic equation is r 2 + 4k r + 3 k 2 = (r + k )(r + 3k ) = 0 with the roots Accordingly, th general solution is r1 = k, r2 = 3k. = 5 k 2 T1 4 k 2 T2 . (IV)

T1 (t, c1 , c2 ) = c1 ekt + c2 e3kt . Follow-up Question compute T1 (t). If k = 1, T1 (0) = 10, T2 (0) = 30,
dT1 dt (0)

Solution The initial condition for use of equation (I) above:

is obtained with the

dT1 (0) = 2k T1 (0) + k T2 (0) = 20 + 30 = 10. dt

Thus, evaluating the given solution form and its derivative at t = 0, c1 e0 + c2 e30 = 10; c1 e0 3 c2 e30 = 10. From this we obtain c2 = 10; c1 = 20 and the solution is seen to be T1 (t) = 20 et 10 e3t , dT1 (t) = 30 e3t 20 e2t . dt

Because the second block is relatively warm, the rst block initially warms up before ultimately cooling toward the temperature of the table.

11