Scribd
Upload
8 views1 page

HW 1

The document outlines Homework 1 for AOE 5404 Numerical Methods, due on January 29, 2025, which includes problems on floating-point representation, iterative methods, and root-finding techniques. Students are required to apply various numerical methods such as bisection, false-position, Newton's method, and secant method to find roots of a given function and analyze errors. Additionally, there is a theoretical problem regarding the properties of the bisection method's error calculations.

Uploaded by

celin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
8 views1 page

HW 1

The document outlines Homework 1 for AOE 5404 Numerical Methods, due on January 29, 2025, which includes problems on floating-point representation, iterative methods, and root-finding techniques. Students are required to apply various numerical methods such as bisection, false-position, Newton's method, and secant method to find roots of a given function and analyze errors. Additionally, there is a theoretical problem regarding the properties of the bisection method's error calculations.

Uploaded by

celin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

AOE 5404 Numerical Methods for AOE

Homework 1
Due on Jan. 29, 2025

Read Chapters 5.1, 5.4, and 5.5 of Heath.

Problem 1 [3 pts]: Floating-point representation of numbers, and round-off error (Ref: Heath Chap. 1.3)
(1) Using any programming language/platform, check if the following equalities hold true. If not, explain why.
a. 𝑎𝑎 + 𝑏𝑏 + 𝑐𝑐 = 𝑎𝑎 + 𝑐𝑐 + 𝑏𝑏, where 𝑎𝑎 = 1, 𝑏𝑏 = 10−16 , 𝑐𝑐 = −1.
b. 0.3 + 0.6 = 0.9 (Check if it holds exactly, i.e., whether “0.3 + 0.6 == 0.9” is true or false.)
Hint: In Matlab, you can convert a number (“x”) to a character string using “num2str(x, 18)” to see all its digits.
(2) The following program is supposed to print “Launch!” on the screen at 𝑡𝑡 = 𝜋𝜋/2 (s) after the program starts.
(Suppose this is the optimal time to launch a rocket…)
Matlab: Python:
tic; import time, math
while true start_time = time.time()
elapsed_time = toc; while True:
if(elapsed_time == pi/2) elapsed_time = time.time() - start_time
disp("Launch!"); if elapsed_time == math.pi/2:
break; print("Launch!")
end break
end
Does this program do what it is supposed to do? If not, what could be done to fix the problem?

Problem 2 [5 pts]: For an iterative method, we define true relative error, 𝜖𝜖𝑡𝑡 , and approximate relative error, 𝜖𝜖𝑎𝑎 , at iteration
𝑘𝑘 (𝑘𝑘 = 0, 1, 2, … ) as
|𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 − 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠| 𝑥𝑥𝑘𝑘+1 − 𝑥𝑥 ∗
𝜀𝜀𝑡𝑡 = =� �,
|𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠| 𝑥𝑥 ∗
|𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 − 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎| 𝑥𝑥𝑘𝑘+1 − 𝑥𝑥𝑘𝑘
𝜀𝜀𝑎𝑎 = =� �.
|𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎| 𝑥𝑥𝑘𝑘+1
Consider function 𝑓𝑓(𝑥𝑥) = 𝑥𝑥𝑒𝑒 𝑥𝑥 − 2cos 𝑥𝑥.
(1) Apply the bisection method to find the root of 𝑓𝑓(𝑥𝑥) = 0 within interval (0, 1). Use (0, 1) as the initial bracketing
interval. Iterate until 𝜀𝜀𝑎𝑎 is less than 10−6. Show 𝜀𝜀𝑡𝑡 at each iteration — that is, plot 𝜖𝜖𝑡𝑡 vs. iteration number. (To get
the “true” solution, you may run the bisection method until 𝜀𝜀𝑎𝑎 is very small, say, less than 10−8.)
(2) Repeat (1) using the false-position (linear interpolation) method.
(3) For the same equation, 𝑓𝑓(𝑥𝑥) = 0, apply Newton’s method with an initial guess of 𝑥𝑥0 = 1, and the same stopping
criterion 𝜀𝜀𝑎𝑎 < 10−6. Again, plot 𝜀𝜀𝑡𝑡 vs. iteration number. Then, try initial guess 𝑥𝑥0 = −0.5, and explain your
result.
(4) Solve the same equation using the secant method. Try initial guess 𝑥𝑥0 = 0.0 and 𝑥𝑥1 = 1.0. Use the same stopping
criterion 𝜀𝜀𝑎𝑎 < 10−6.

Problem 3 [2 pts]: Prove that using the bisection method, the true error, |𝑥𝑥𝑘𝑘+1 − 𝑥𝑥 ∗ |, is always less than or equal to the
approximate error, |𝑥𝑥𝑘𝑘+1 − 𝑥𝑥𝑘𝑘 |, at any iteration.
Note: In practice, it is usually impossible to calculate the true error, because the true solution 𝑥𝑥 ∗ is unknown. Therefore,
what you have proven is a very nice feature of the bisection method.

You might also like