2 ODEs—Integrating Factors and Homogeneous Equations We begin with a slightly different type of equation: 2.1 Exact Equations These are ODEs whose general solution can be obtained by simply integrating both sides of the equation. Examples Find the general solutions of the following ODEs. a) dy dx b) d 2 dx (yx ) = 2 sin x, = 4x3 Z a) This is completely standard—we can rewrite as the integral equation Z dy = 2 sin(x)dx. Then integrating both sides, we obtain y = −2 cos x + c b) Here we do the same: Integrating both sides, we obtain Z 2 d(yx ) Z = 4x3 dx and hence yx2 = x4 + c. We can now divide through by x2 to obtain the general solution y = x2 + c . x2 Key Point The solution of the equation d f (x)y = g(x) dx is Z f (x)y = g(x) dx or y= Terminology: In the differential equation b) 1 f (x) Z g(x) dx c d (yx2 ) = 4x3 , the solution has two parts, since y = x2 + 2 . dx x 1 The part x2 is called the definite part. The part c (containing the arbitrary constant of integration) is x2 called the indefinite part. If we take the definite part of the solution, i.e. yd = x2 then d d d (yd · x2 ) = (x2 · x2 ) = (x4 ) = 4x3 . dx dx dx Hence yd = x2 is a solution of the original differential equation (b). Now consider the indefinite part of the solution, i.e. yi = c x2 , then d d d c 2 = · x (yi · x2 ) = (c) = 0. 2 dx dx x dx This works generally: 2.1.1 Useful fact: In solving an exact equation d f (x)y = g(x), the solution y has two parts: y = yd (x) + yi (x) where: dx 1. The definite part yd (x) which is a solution of the differential equation. 2. The indefinite part yi (x) which satisfies a simpler version of the differential equation in which the right-hand side is zero. Exercise: Solve the equation d (y sin x) = sinx dx and verify that the indefinite part of the solution satisfies the equation d dx (y sin Solution: Integrate both sides of the equation, y sin x = y = − cos x + C − cot x + Ccosec x. Hence yd (x) = − cot x and yi (x) = C cosec x. Now d d d (yi (x) sin x) = (cosec x sin x) = (C) = 0. dx dx dx 2 x) = 0. 2.1.2 Recognising exact equations d (yx2 ) = x2 is exact but if we expand the left-hand side of this equation using the dx product rule, we have that The equation b) d (yx2 ) dx = = dy d + y (x2 ) dx dx dy x2 + y(2x) dx x2 and so, x2 dy + 2xy = x2 . dx (2.1) Hence the equation (2.1) is exact but it is not in standard form. 2.1.3 Key point The equation f (x) dy + yf 0 (x) = g(x) is exact. It can be rewritten as: dx d (yf (x)) = g(x) dx Z so that yf (x) = 1 g(x) dx and y = f (x) Z g(x) dx. Example. Solve the equation e2x dy + 2e2x y = x2 . dx d 2x e = 2e2x . Therefore this is just the exact equation dx x3 gives e2x y = + c. Thus 3 x3 −2x y= e + ce−2x 3 Note that is the general solution. 3 d 2x dx (e y) = x2 . Integrating both sides 2.2 Integrating Factors Integrating factors can be used to transform certain ODEs which are not exact into exact ODEs. As an illustration, consider x3 dy + 4x2 y = x. dx As you may wish to check, this is not exact. However, if we multiply through by x, we get x4 dy + 4x3 y = x2 dx which can be rewritten as the exact equation d (yx4 ) = x2 . dx Thus we can solve this as before to get yx4 = Z d (yx4 )dx = dx Z x2 dx = x3 +c 3 or y= 1 c + 4. 3x x The function by which we multiply a given ODE in order to make it exact is called the integrating factor. In the above example x is the integrating factor. This works rather generally: 2.2.1 Fundamental Algorithm dy + b(x)y + c(x) = 0 can be transformed into an exact equation and dx then (hopefully!) solved. This requires the following steps: Any linear ODE of the form a(x) Step I: Divide by a(x) to get an equation of the form dy + f (x)y = g(x) dx Step II: Compute the integrating factor Z I.F. = exp f (x) dx Note that we do not need constants of integration—here the IF is any function of the form eF (x) where f (x) = F 0 (x). Step III: Multiply through by the IF to get eF (x) dy + f (x)eF (x) y = eF (x) g(x). dx d F (x) (e y) and so we can solve: dx Z Z d F (x) F (x) (e y) = eF (x) g(x). e y= dx Step IV: The LHS of this equation is simply 4 Example: Consider the example dy + 4x2 y = x dx dy 4 from before. This can be rewritten as + y = x−2 . Thus the integrating factor (I.F.) is dx x Z 4x−1 dx = exp(4 log x) = exp(log x4 ) = x4 . exp x3 Therefore, we multiply through by x4 , giving the exact ODE dy + 4x3 y = x2 ; x4 dx equivalently d dy (yx4 ) = x4 + 4x3 y = x2 dx dx as before. Example: Solve the ODE y 0 + 2y = sin x. Step 1 is already done! Step 2. Calculate the I.F. Z I.F. = exp 2 dx = e2x . Step 3. Multiply through by the I.F. and write the equation in exact form. e2x dy + 2e2x y = e2x sin x. dx Thus d 2x e y = e2x sin x. dx Step 4. Integrate both sides of the equation and find the general solution. This is a standard integration by parts, and I will let you check that 2x Z Z 2x e y= e ydx = 1 2 e2x sin x = · · · = − e2x cos x + e2x sin x + C. 5 5 dy + (sin x)y = cos2 x. dx dy Step 1. Rewrite the differential equation as + tan(x)y = cos x. dx Example: Solve the ODE cos x Step 2. Calculate the I.F. Z I.F. = exp tan(x) dx = exp(− ln(cos x)) = exp(ln(sec x)) = sec(x). 5 Step 3. Multiply through by the I.F. and write the equation in exact form. sec x dy + y sec x tan x = sec(x) cos x = 1 dx equivalently, d sec(x)y = 1 dx Step 4. Integrate both sides of the equation and find the general solution: Z d sec(x)y = Z dx which has solution sec(x)y = x + c or y = (x + c) cos(x). An electrical example: Suppose we are given an electrical circuit with a resistor and an inductance coil and a battery. R L + E - A basic fact from physics is that, if we close the circuit we get a current i flowing clockwise around the di circuit, with voltage drop across the resistor of iR and across the coil of L . Thus, dt E = iR + L di . dt [Note: These are not formulae you need to remember for this course.] Let’s put in some numbers; say E = 10, R = 8 and L = 2; with initial conditions of i = 0 at t = 0. Thus di 10 = 8i + 2 , or dt di + 4i = 5. dt 6 This can be solved either with an IF (which equals e4t ) or by separating variables and using (1.3.1) from page 6 of the Separation of Variables notes. Either way you find that i = 5/4 + Ce−4t . The initial conditions give 0 = 5/4 + C and so C = −5/4 and i= 5 5 −4t − e . 4 4 This has the graph with asymptote at i = 5/4 like those on page 7 of the Separation of Variables section. Now suppose the battery goes flat and we use the mains in place, say with voltage 2 cos(t). This means that we have to solve the equation 2 di + 8i = 2 cos(t) dt di + 4i = cos(t). dt or R This does now need an IF = exp( 4dt) = e4t . Thus d 4t di e i = e4t + 4e4t i = e4t cos(t). dt dt Z 1 4 4t e cos(t) + e4t sin(t) + C, where the integral on the RHS was 17 17 solved by integration by parts (twice). Solving for i we get 4t Therefore, e i = e4t cos(t) = i = 4 1 cos(t) + sin(t) + Ce−4t . 17 17 4 Finally, solving for C from i(0) = 0 gives C = − 17 and so, finally, i = 4 1 4 cos(t) + sin(t) − e−4t . 17 17 17 Example: Even quite innocent-looking equations can lead to impossible integrals. For example the R 2 equation y 0 + xy = 1 leads to the IF = exp( xdx) = ex /2 and hence to the equation ex 2 /2 Z y= ex 2 /2 dx, which has no closed form. (In fact, you can always solve this sort of equation by writing ex series in x, but that is another story—and another course.) 7 2 /2 as a power 3 Homogenous Equations These are differential equations of the form: dy = f (x, y) dx where the function f (·, ·) satisfies f (λx, λy) = f (x, y), for any λ. Remark: Be careful about looking up homogeneous equations in the literature since this has more than one meaning. Examples of homogeneous equations: a) f (x, y) = 1 + y x since f (tx, ty) = 1 + x tx ty tx y x =1+ = f (x, y). x b) f (x, y) = e y since f (tx, ty) = e ty = e y = f (x, y). c) f (x, y) = 3.1 x y + y x since f (tx, ty) = tx ty + ty tx = x y + y x = f (x, y). Basic Technique for Solving Homogeneous Equations. To solve homogeneous equations, we do a substitution: z= y (or y = zx) x and use the rule dy d dz = (zx) = x + z. dx dx dx We illustrate how this is done with an example. Example: Solve the ODE x dy = x + y. dx Step 1. Divide through by x to get a homogenous equation: Step 2. Substitute y = zx to get d dx (xz) dy dx = 1 + xy . = 1 + z. The Key Step 3. Expand the LHS of the equation using the product rule, x dz d + z (x) = 1 + z dx dx 8 or x dz + z = 1 + z. dx Step 4. Hopefully (!) this can now be solved by one of the earlier techniques. In this case we can cancel dz the “z’s” to get x = 1. This can then be solved by separating variables: dx Z Z dz 1 dx = and so dz = . dx x x This has solution z = ln x + C. Step 5. Finally substitute back z = y/x; thus y = ln(x) + C x or y = x ln(x) + Cx. Lets do this with some more examples. Example. Solve the ODE x dy = y + xey/x dx Solution: Divide by x to get y = 1 at x = 1. y dy = + ey/x . dx x This is homogeneous, so we set z = y x x Cancel the z to get x subject to or y = zx to get dz d(zx) +z = = z + ez . dx dx dz = ez . This can now be solved by separating variables: dx Z Z −z e dz = x−1 dx, with solution −e−z = ln|x| + C. Substituting back z = y/x we get −ey/x = ln |x| + C. Plugging in y = 1 at x = 1 gives −e−1 = 0 + C and so C = −e−1 . Thus, finally, −ey/x = ln |x| − e−1 . If you like you can solve for y by taking logs, but the answer is still not very elegant. 9 Example. Solve the ODE dy (y + x)y =− −2 dx x(x − y) Solution: This is not obviously homogeneous, but we make it so by dividing the top and bottom of the fraction on RHS by x2 : dy =− dx y x + 1 xy −2 1 − xy Substitute in y = zx x dz +z dx x dz dx (z + 1)z −2 (1 − z) 2 z +z = − −z−2 1−z = − We can now simplify the RHS: x dz −z 2 − z − z(1 − z) − 2(1 − z) −z 2 − z − z + z 2 − 2 + 2z 2 2 = = = − = . dx 1−z 1−z 1−z z−1 Thus we can separate variables: z−1 2 dz = 1 dx x or z 1 − 2 2 dz = 1 dx x Integrate both sides of the equation to get z2 z − = ln x + c 4 2 Finally we substitute back z = y x to get y2 y − = ln x + c. 2 4x 2x With a bit of an effort you can write this as y = g(x) for some function g(x) but I really do not think it is worth the bother. 10