JUST THE MATHS UNIT NUMBER NUMERICAL MATHEMATICS 6 (Numerical solution) of (ordinary differential equations (A)) A.J.Hobson

Transcription

1 JUST THE MATHS UNIT NUMBER 17.6 NUMERICAL MATHEMATICS 6 (Numerical solution) of (ordinary differential equations (A)) by A.J.Hobson Euler s unmodified method Euler s modified method Exercises Answers to exercises

2 UNIT NUMERICAL MATHEMATICS 6 NUMERICAL SOLUTION OF ORDINARY DIFFERENTIAL EQUATIONS (A) EULER S UNMODIFIED METHOD Every first order ordinary differential equation can be written in the form dx = f(x, y); and, if it is given that y = y 0 when x = x 0, then the solution for y in terms of x represents some curve through the point P 0 (x 0, y 0 ). Suppose that we wish to find the solution for y at x = d, where d > x 0. We sub-divide the interval from x = x 0 to x = d into n equal parts of width, δx. y O P 0 x0 x d Letting x 1, x 2, x 3,... be the points of subdivision, we have x 1 = x 0 + δx, x 2 = x 0 + 2δx, x 3 = x 0 + 3δx, d = x n = x 0 + nδx. If y 1, y 2, y 3,... are the y co-ordinates of x 1, x 2, x 3, we are required to find y n. 1

3 From elementary calculus, the increase in y, when x increases by δx, is given approximately by δx; and since, in our case, = f(x, y), we have dx dx y 1 = y 0 + f(x 0, y 0 )δx, y 2 = y 1 + f(x 1, y 1 )δx, y 3 = y 2 + f(x 2, y 2 )δx, y n = y n 1 + f(x n 1, y n 1 )δx, each stage using the previously calculated y value. Note: The method will be the same if d < x 0, except that δx will be negative. In general, each intermediate value of y is given by the formula y i+1 = y i + f(x i, y i )δx. EXAMPLE Use Euler s method with 5 sub-intervals to continue, to x = 0.5, the solution of the differential equation, dx = xy, given that y = 1 when x = 0; (that is, y(0) = 1). Solution i x i y i f(x i, y i ) y i+1 = y i + f(x i, y i )δx Accuracy The differential equation in the above example is simple to solve by an elementary method, 2

4 such as separation of the variables. It is therefore useful to compare the exact result so obtained with the approximation which comes from Eulers method. Therefore that is, y = xdx. ln y = x2 2 + C; y = Ae x2 2. At x = 0, we are told that y = 1 and, hence, A = 1, giving y = e x2 2. But a table of values of x against y in the previous interval reveals the following: x e x2 2 There is thus an error in our approximate value of , which is about 2.6%. Attempts to determine y for values of x which are greater than 0.5 would result in a very rapid growth of error EULER S MODIFIED METHOD In the previous method, we used the gradient to the solution curve at the point P 0 in order to find an approximate position for the point P 1, and so on up to P n. But the approximation turns out to be much better if, instead, we use the average of the two gradients at P 0 and P 1 for which we need use only x 0, y 0 and δx in order to calculate approximately. 3

5 y O P 0 x P 1 x 0 x 1 The gradient, m 0, at P 0, is given by m 0 = f(x 0, y 0 ). The gradient, m 1, at P 1, is given approximately by where δy 0 = f(x 0, y 0 )δx. m 1 = f(x 0 + δx, y 0 + δy 0 ), Note: We cannot call y 0 + δy 0 by the name y 1, as we did with the unmodified method, because this label is now reserved for the new and better approximation at x = x 0 + δx. The average gradient, between P 0 and P 1, is given by m 0 = 1 2 (m 0 + m 1 ). Hence, our approximation to y at the point P 1 is given by y 1 = y 0 + m 0δx. Similarly, we proceed from y 1 to y 2, and so on until we reach y n. In general, the intermediate values of y are given by y i+1 = y i + m i δx. 4

6 EXAMPLE Solve the example in the previous section using Euler s Modified method. Solution i x i y i m i = δy i = m i+1 = m i = y i+1 = 1 f(x i, y i) f(x i, y i)δx f(x i + δx, y i + δy i) 2 (mi + mi+1) yi + m i δx EXERCISES 1. (a) Taking intervals δx = 0.2, use Euler s unmodified method to determine y(1), given that dx + y = 0, and that y(0) = 1 Compare your solution with the exact solution given by y = e x. (b) Taking intervals δx = 0.1, use Euler s unmodified method to determine y(1), given that dx = x2 + y x and that y(0.5) = 0.5. Compare your solution with the exact solution given by y = x 2 + x 2. (c) Taking intervals δx = 0.2, use Euler s unmodified method to determine y(1), given that dx = y + e x, and that y(0) = 0. Compare your solution with the exact solution given by y = sinh x. 5

7 (d) Given that y(1) = 2, use Euler s unmodified method to continue the solution of the differential equation, dx = x2 + y 2, to obtain values of y for values of x from x = 1 to x = 1.5, in steps of Repeat all parts of question 1 using Euler s modified method ANSWERS TO EXERCISES 1. (a) 0.33, 0.37, 11% low; (b) % low; (c) 1.113, 1.175, 5% low; (d) , 2.200, , (a) 0.371, 0.368, 0.8% high; (b) 1.495, 1.50, 0.33% low; (c) 1.175, accurate to three decimal places; (d) 2.000, 2.216, 2.465, 2.751, 3.079, Note: In questions 1(d) and 2(d), the actual values are 2.000, 2.245, 2.496, 2.784, and 3.489, from the exact solution of the differential equation. 6