21.04.2015 Views

The Computable Differential Equation Lecture ... - Bruce E. Shapiro

The Computable Differential Equation Lecture ... - Bruce E. Shapiro

The Computable Differential Equation Lecture ... - Bruce E. Shapiro

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

CHAPTER 5. RUNGE-KUTTA METHODS 109<br />

Hence absolute stability requires<br />

∣ y n ∣∣∣ ∣ =<br />

y n−1<br />

∣ 1 + h ( )∣ 6λ − 4hλ<br />

2<br />

4 h 2 λ 2 − 4hλ + 6 + 18λ ∣∣∣<br />

h 2 λ 2 (5.168)<br />

− 4hλ + 6<br />

=<br />

∣ 1 + 6z − z2<br />

z 2 − 4z + 6∣ (5.169)<br />

=<br />

6 + 2z<br />

∣z 2 − 4z + 6∣ ≤ 1 (5.170)<br />

where z = hλ. <strong>The</strong> region of absolute stability is shown in figure 5.6<br />

5.8 IRK Methods based on Collocation<br />

In the method of collocation, given an initial value problem that we have solved<br />

numerically on the interval [t 0 , t n ] we seek a ν th order polynomial P (t) such that<br />

P (t n ) = y n (5.171)<br />

P ′ (t n )(t n + c j h) = f(t n + c j h, P (t n + c j h)), j = 1, 2, . . . , ν (5.172)<br />

where the numbers c 1 , . . . , c n are called collocation parameters. Thus P satisfies<br />

the initial value problem defined by the numerical solution at t n . A collocation<br />

method finds such a polynomial P and sets y n+1 = P (t n+1 ).<br />

<strong>The</strong>orem 5.1. <strong>The</strong> Implicit Runge-Kutta Method<br />

c<br />

A<br />

b T (5.173)<br />

and the collocation method of equation 5.171 are identical if<br />

where<br />

a ji = 1<br />

q i (c i )<br />

b j = 1<br />

q j (c j )<br />

q j (t) =<br />

∫ cj<br />

0<br />

∫ 1<br />

ν∏<br />

0<br />

i=1,i≠j<br />

q i (s)ds (5.174)<br />

q j (s)ds (5.175)<br />

(t − c i ) (5.176)<br />

Proof. Define the ν th order polynomial<br />

r(t) =<br />

ν∑<br />

k=1<br />

q k ((t − t n )/h)<br />

w k (5.177)<br />

q k (c k )<br />

then<br />

r(t n + c j h) =<br />

ν∑<br />

k=1<br />

q k (c j )<br />

q k (c k ) w k =<br />

ν∑<br />

δ jk w k = w j (5.178)<br />

k=1<br />

c○2007, B.E.<strong>Shapiro</strong><br />

Last revised: May 23, 2007<br />

Math 582B, Spring 2007<br />

California State University Northridge

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!