MA/CS 371 - Lab 9

MA/CS 371 - Lab 9
Comparing Three Different Methods on an ODE

Section 1: Introduction

In this lab you will use Euler's Method, the second-order Runge-Kutta method, and the classical fourth-order Runge-Kutta method to calculate approximate solutions to a given ODE.

If you have not already done so, create a directory for this lab and copy the lab 9 files into it. If you start off in your home directory you can do the following:

	mkdir ~/cs371/lab9
	cd ~/cs371/lab9
	cp ~cs371/lab9/* .

and (as always) don't forget to type that last period.

Section 2: Topics for Lecture

No new MATLAB tricks are required for this lab so there will be no lecture.

Section 3: The ODE

The ODE we will be solving in this lab is:

	y' = 2y - 2t + 1
	y(0) = 2

Normally we use an ODE solver when we can't solve the equation analytically. In this case, since we're testing various methods, we've chosen a function whose answer is known. In this case, you should check that the function

	y = t + 2*exp(-2*t)

is the correct solution.

I have supplied the one-line function func.m which you can call as follows:

	y_prime = func(y,t)

so that when you write your ODE solvers you can have them call func.m rather than have a hard coded function inside them.

Section 4: The Three Methods

Write three MATLAB functions: euler_meth.m, rk2.m, and rk4.m. Each should take four input parameters: the left and right endpoints of the interval, the number of steps, and the value of the function at the left endpoint. Each function should output a vector of approximate function values at the indicated points. So for example after the call

	values = euler_meth(0,2,21,0);

the output variable "values" should be a 21-element row vector with the approximate function values at t = [0 0.1 0.2 ... 2].

Pseudocode for euler_meth is available on page 328 of the text and rk4 is on pp. 339-340. The text does not include pseudocode for the second-order Runge-Kutta method but you should be able to work it out from the description on pp. 337-338 and by similarity with the other programs.

NOTE: You will have to restructure the pseudocode a little bit in order to return the answer as a vector.

Section 5: Running your programs

Write a script file ode_methods.m which will do the following:

Run each ODE solver (euler_meth, rk2, and rk4) on the function in the range [0,2] with 5, 21, and 101 points (4, 20, and 100 intervals).
Produce a table showing how many flops each method requires with each number of intervals.
Produce three graphs - one for the 5-point case, one for the 21-point case, and one for the 101-point case - where each graph plots:
- the true solution as a solid line
- the euler_meth solution as a dotted line (plot symbol ":")
- the rk2 solution as a dashed line (plot symbol "--")
- the rk4 solution as a dot-dash line (plot symbol "-.")
Make sure that each graph has an appropriate title.

Section 6: Turning In Your Work

When your euler_meth.m, rk2.m, rk4.m, and ode_methods.m files are ready, type

	
	~cs371/submit lab9

These programs will be due by 5:45pm on Thursday, April 18.

This lab's question.