BIF jO4F
: Chapter 10
LUTIONS OF ORDINARY
DIFFERENTIAL EQUATIONS
10.1 INTRODUCTION
Finding an analytical solution of an ordinary differential equation dy/dx = f(xy) subjected to an
initial condition y(xo) = yo means finding an expression for y in terms of x such that it satisfies the
differential equation as well as the initial condition, However, finding such expression is not always
possible for every differential equation and hence one is bound to go for some other method of
solution. To get numerical solutions of differential equations is one of stich possibilities.
Finding a numerical solution of an ordinary differential equation dy/dx = f(x,y) subjected to an
jon y(Xo) = yo means finding a set of values of y for different values of x, such that each
satisfies the differential equation as well as the initial condition, Differential equations such as
dilds = fixy), are known as differential equations of first order and first degree. There are various
methods like Pieard’s method, Euler’s method, Heun’s method and Runge-Kutta method for
solving numerically an ordinary differential equation.
10.2 PICARD METHOD
Fundamentals: Picard’s method for solving differential equation dy/dx = f(x,y) is a method of
successive approximations. This is based on three fundamentals:
¢ If bith sides of first order differential equation are multiplied by dx and the equation is
integrated with respect to x then we can get the solution of the differential equation.
¢ The variable y in the integrant of right hand side can be assigned a constant value as the
initial value of y, to make the integral solvable.
¢ An iterative procedure can be set, to get the modified values of y by using newly
calculated value of y in the integrant.
Derivation of the iterative formula: Consider that the solution of the first order and first degree
differential equation
dyldx = fly) an
is to be found at x, subject to the initial values xo and y(xo) =,yo. Integration of equation (1.1) gives
Sdy = Jf(x,y)dx306 __ Numerical Computattonal Methods
ice. y(x) — yo) = FPO y Dax 2)
Replacing y by yo on right hand side and simplifying we get
YX) = ¥(%o) + [F(R Y 2 dX 3)
‘The calculated y can be used in place of yo, and a new y can be obtained. This process can be repeated
until required convergence in y is achieved, the general formula for i" iterative step being
¥) = y60o) + Ff, y Oo} (14)
Merits and demerits of Picard method: This method is of historical importance but it is not in
use due to the following drawbacks.
1) It involves actual integration i.e, semi-numecie method, which is not suitable for
computer Implementation.
2). The integrant of the integral, in some cases, may have the form whose integral expression
can not be found.
3) Convergence may not always be possible or may be very slow.
‘BRVYSILEY Using Picard method find the solution cf dy/dx = x*+y* at x = 0.2 if y(0) = 2.
Answer: The solution of given differential equation can be written in integral form as
y6x) = y(0) + Toc? + y?)ax
‘The first iteration assumes y in the integrant to be y(C) = 2. Hence
YQ) = yO) +]oe +27)dx =2 arn
Using this value of y in the integrant we get the second iteration result
Y%(x) = y(0) + J [- +fevsnett}
x
=24f [x eave
0
At x= 0.2 we get the estimated value of y with the first two iterations as10 : Solutions of Ordinary Differential Equations 307
‘
2422). 40.2=2.8003
Y'0.2)
¥*0.2) =24440.2 +8402?
= 3.1668
10.3 TAYLOR SERIES METHOD
Fundamentals: The Taylor series method of solving ordinary differential equation of the type
dy/dx = f(xy), consists of finding values of y at successive steps of x with step size h. It uses two
ideas.
(i) The value of y at the end of (i+1)" step can be calculated by Taylor series
it - 2y| ay b3 aty|
Yier= yOxith) = y(x) +h] a. eal ws
(1) ‘The tharting point velues x, andy, ate requied forthe elite of the values xj.) and
{ies Felated to the end point of the step i+1, Here x, and y, are calculated as end point
values of the previous step. However for the first step the starting point values Xo , Yo are
known from initial condition,
Derivation of the formula: The Taylor series expression contains derivatives of y with respect to x
and they are again functions of x. Hence the formula must be simplified suitably to express it in terms
of partial derivatives. The Taylor series gives
Yier™ yOuth)
oy |
or yisi= vioo+ng|
' Fle,
But from the given differential equation
o
B= fxy) G2)
higher order derivatives of y can be obtained from one order lower derivatives of f(x,y). Thus use of
these in equation (3.1) provides
GA)
a
h” afl
se mel (3.3)
Yio “adele y, 3.3)
3
i+ AL ey + Teo oo
Since f(x,y) is @ function of x and y and depends on x, therefore, we can write
tee SE ow SE BEY oe 3.5
f' a. x ay dx fet fy f (3.5)