The Computable Differential Equation Lecture ... - Bruce E. Shapiro
The Computable Differential Equation Lecture ... - Bruce E. Shapiro
The Computable Differential Equation Lecture ... - Bruce E. Shapiro
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
102 CHAPTER 5. RUNGE-KUTTA METHODS<br />
and Kutta’s third-order formula<br />
0 0<br />
1/2 1/2 0<br />
1 −1 2 0<br />
1/6 2/3 1/6<br />
(5.123)<br />
<strong>The</strong> classical Runge-Kutta method is the four-stage fourth-order method<br />
we have already met:<br />
0 0<br />
1/2 1/2 0<br />
1/2 0 1/2 0<br />
(5.124)<br />
1 0 0 1 0<br />
1/6 1/3 1/3 1/6<br />
<strong>The</strong> region of absolute stability for all ERK methods will always be bounded,<br />
and hence none of them can be A-stable. This is because the test equation will<br />
always produce<br />
y n = P (z)y n−1 (5.125)<br />
where P (z) is a polynomial. <strong>The</strong>re will always be sufficiently large values of z so that<br />
|P (z) > 1. Since none of these methods are A-stable, they are generally not good<br />
for stiff problems, as we have already seen for the Forward Euler Method (which is<br />
a one-stage IRK). Thus one commonly uses implicit RK methods (IRK) instead.<br />
5.5 Order Conditions<br />
Explicit Runge-Kutta (ERK) methods have an order that is at most equal to the<br />
number of stages; we have seen two, three, and four stage methods that have orders<br />
equal to the number of stages. However, for n > 4, it is not possible to find an ERK<br />
with an order that is equal to the number of stages. <strong>The</strong> best one can do is the<br />
following.<br />
Number of Stages 1 2 3 4-5 6 7-8 9-10 11 12 13-17<br />
Best Possible Order 1 2 3 4 5 6 7 8 9 10<br />
Unfortunately the proof of this statement is tedious - there is an explosion in the<br />
number of terms that must be compared as the Taylor order is increased. Interested<br />
students are referred to [7] for details.<br />
To determine the error in a carrying out a single step of a Runge-Kutta method,<br />
one must compare the successive terms in a Taylor series expansion of the exact and<br />
computed solutions. This calculation is quite tedious: “<strong>The</strong> reader is no asked to<br />
take a deep breath, take five sheets of reversed computer paper, remember the basic<br />
rules of differential calculus, and begin the following computations...” (see [10, page<br />
144]). <strong>The</strong> result is a set of order conditions that are necessary if a method is to<br />
have a given order; however, these conditions are not, in general, sufficient.<br />
Math 582B, Spring 2007<br />
California State University Northridge<br />
c○2007, B.E.<strong>Shapiro</strong><br />
Last revised: May 23, 2007