Example 6.
17 'Given the initial value problem
u' = -2 tu 2 , u(0) =1
estimate u (0.4) using (i) modified Euler-Cauch y method, and (ii) Heun method, with h = 0.2. Compare
the results with the exact solution u (t) = 1/(1 + t ).
2
(i) The modified Euler-Cauc hy method is given by
- 1 =u-+K
u J+ J 2
where
and
For j = 0, we have
t0 = 0, u0 = I, K 1 = 0, = -0.4 (0.1) (1)2 = -0.04.
K2
u (0.2) = u 1 = u 0 + K 2 = 1 - 0.04 = 0.96.
For j = 1, we have
!1 = 0.2, U1 = 0.96, K1 = -0.4 (0.2) (0.96)2 = -0.073728.
2
K 2 = - 0.4 (0.2 + 0.1) (0. 96 - 0.036864) = - 0.102262.
u(0.4) = u 2 = u 1 + K 2 = 0.96 - 0.102262 = 0.857738.
(ii) The Heun method is given by
1
uj+t = uj + 2 (Ki + K2)
where K 1 = hf(ti' u j) = - 0.4 tj u;
and K2 = hf(tj + h, uj + Ki) = -0.4(tj + 0.2) (uj + Ki/-
For j = 0, we have
t0 = 0, u 0 = I, Ki = o, K2 = -0.4 (0.2) (1)2 = - 0.08.
1 1
u (0.2) = u 1 = u0 + (K1 + K 2 ) = I + (- 0.08) = 0.96.
2 2
For j = I, we have
ti = 0.2, U1 = 0.96, K1 = -0.4(0.2) (0.96)2= - 0.073728.
K2 = - 0.4 (0.2 + 0.2) (0.96 - 0.073728)2 = - 0.125676.
u (0 .4 ) = u2 = u1 + _!_ (K + K ) = 0 . 9 6 + 1 (- 0 .0 7 3 7 2 8 25676
2 I 2 - O. l
2 '
= 0 .8 6 0 2 9 8 .
T he ex ac t so lu ti on is
u (0 .2 ) = 0 .9 6 1 5 3 8 , u
T he ab so lu te er ro rs in (0 .4 ) = 0 .8 6 2 0 6 9 •
th e nu m er ic al so lu ti o n
M od if ie d E ul er -C au ch
y m et ho d: e (0 .2 ) == s ar e:
H eu n m et h o d : e (0 .2 )
0 .0 0 1 5 3 8 , £ (0 .4 )
== 0 .0 0 1 5 3 8 , e (0 .4 )
= 0 .0 04 3 3 1 ·
E xa m o le Ii 1R ~
= 0 .0 0 1 7 7 1 .
Section 14.2 Finite-Difference Method 599
The General Linear BVP
We now consider the general linear two-point boundary-value problem
= p(x)y' + q(x)y + r(x), a~ x b
y"
with boundary conditions y(a) = a, y(b) = /3. To solve this problem using finite
differences, we divide the interval [a, b] into n subintervals, so that h = (b - a)/n.
To approximate the function y(x) at the points x 1 = a+h, ... , Xn-l = a+ (n- l)h,
we use the central difference formulas from Chapter 11:
y
"( ) ~~ Yi+l -
Xi
2yi
h2
+ Yi-1
' y
'(
Xi
) ' Yi+l - Yi-1
2h
Substituting these expressions into the BVP and writing p(xi) as Pi, q(xi) as
q;, and r(xi) as r; gives
Yi+l - 2yi + Yi-1 Yi+l - Yi-1
h2 = Pi 2h + q.y. + Ti
Further algebraic simplification leads to a tridiagonal system for the unknowns
y 1 , . . . , Yn-1 - that is, for i = 1, . . . , n - 1,
(2 + h Pi)Y;-1 - (4 + 2h2 q;)Yi + (2 - h p;)Yi+l = 2h2 ri
where y0 = y(a) = a and Yn = y(b) = /3.
Using the boundary conditions and simplifying gives:
first (for i = 1),
-(4 + 2h2 qi)y 1 + (2 - hpi)y2 = 2h2r1 - o(2 + hp1)
t hen (for i = 2, . .. ,n - 2) ,
(2 + hp.~Yi-1 - (4 + 2h 2 qi)Yi + (2 - hp;)Yi+l = 2h2r;
and , finally (for i = n- 1),
(2 + hPn-l)Yn-2 - (4 + 2h2 qn-dYn-1 = 2h2 Tn-1 - /3(2 - hPn-1)
The system in matrix form is Ay = b, with A
-(4 + 2h 2 qi) (2 - hpi)
(2 + hP2) -(4 + 2h 2q2)
(2 + hp.) (2 - hp.)
-(4 + 2h 2qn-2) (2 - hPn-2)
(2 + hPn-1) -(4 + 2h 2 Qn-d
L
EXAMPLE 14.5 A Finite-Difference Problem
Use the finite-difference method to solve the problem
y
11
= y +X(X - 4), 0$ X $ 4
with y(O) = y(4) = 0 and n = 4 subintervals. The finite-difference method will find
an approximate solution at the points x 1 = 1, x 2 = 2, and X3 = 3. Using the central
difference formula for the second derivative, we find that the differential equation
becomes the system
Yi+ 1 2yi + Yi-1 = y,· + X·, (X ,· - 4) ,
YII ( X·, ) -
:=::::; - - - - - : h
2
--::----
i = 1, 2, 3
For this example, h = 1. In writing out the system of algebraic equations, we make
use of t}:te fact that at i = 1; yg = 0 (from the boundary condition at x = 0), and
similarly, at i = 3, y4 = 0. Substituting in the values of x 1 , x 2 , ·and x we obtain
3
Y2 - 2y1 + 0 = Y1 + 1(1 -
4}
Y3 - 2y2 + Y1 = Y2 + 2(2 - 4)
0 - 2y3 + Y2 = y3 + 3(3 - 4)
Combining like terms and simplifying gives
-3y1 + Y2 =-3
Y1 - 3y2 +y3 =-4
+ Y2 -3y3 =-3
Solving, we find that Y1 = 13/7, Y2 18/7, and y3 = 13/7. =
We note for comparison that the exact solution of this problem is
2(1 - e4 ) 2(1 - e- 4 ) :i:
y = --:-----:-e
e-4 - e4
-:i:
- -e-4 -e - x2
~ -- :e4 + 4x - 2
At x = 1, the exact solution is 1.8341 (to four decimal places) ; the finite-difference
solution is Y1 = 13/7 = 1.8571.
+X T in( m)
h: 0-2S
) 3 ) =o,
J. y ) Jl»)D No:
- ?
(24) i)= o
-(u2)***i
m a
0 tAS3*1
1 o 3S Pao
(
al44)
(44)
4-
130
ol&) ol4') o4)
30
SL
120
(4)-°)
