Description
Final Project Mandelbrot Set Display
Introduction In this assignment we will use the OpenGL graphics API to compute and display a visual image of the
Mandelbrot Set. Once the set is displayed, we will use the mouse to select a square region from the displayed set, and
recompute the set using only the selected region of the complex plane. Details of the math to compute the Mandelbrot
Setis given in the paragraphs below.
The Mandelbrot Set The Mandelbrot Set is defined as the set of points in the complex plane that satisfy
M = {c ∈ C| limn→∞ Zn 6= ∞}
where
M is the set of all complex numbers in the Mandelbrot Set.
C is the set of all complex numbers in the complex plane,
Z0 = c,
Zn+1 = Z
2
n + c
From this description it appears we have to iterate n over all values from 0 to ∞, which would take quite a bit of
computation. Luckily, we can make two simplifying assumptions.
1. If the magnitude of any Zn is greater than 2.0, then it can be proved that Z will eventually reach infinity, so if
we ever get a Zn for which the magnitude is greater then 2.0 we can stop iterating and claim that the point c is
not in the Mandelbrot Set.
2. If we have iterated 2,000 times and still not found a Zn with a magnitude greater than 2.0, we can again stop
iterating, but this time claim the c is in the Mandelbrot Set. The value 2000 was chosen arbitrarily but seems to
work pretty well for this assignment.
Displaying the Mandelbrot Set Start by creating a display window of 512 by 512 pixels. Then compute and display
the Mandelbrot Set. Unfortunately, it appears in the above discussion of the math to compute the Mandelbrot Set, that
we have to iterate all possible values of c ∈ C. Again we are lucky that we can limit the possible range of the c value
as follows:
(−2.0, −1.2) < c < (1.0, 1.8)
This still seems to be an infinite number of possible c values. To again avoid infinite processing, simply select 512
discrete c values in both the real and imaginary ranges above, resulting in 512 * 512 total unique c values. For each c,
simply iterate the Z values as shown above. It the c is in the Mandelbrot Set, display a black pixel at the appropriate
point in the display window. If the point is not in the Mandelbrot Set(ie you found a Zn 2.0), display the pixel in a
color of your choosing. However, the chosen color must be a function of the number of iterations taken to find that Z
will be infinite. For example, if on the 1999th iteration you found that the magnitude of Z was greater than 2.0, then
the color painted must be the same for all c values that required 1999 iterations. Your result in this case will be similar
to that below.
USE 16 separate threads to compute the Mandelbrot Set.
1
Graduate Students Only. Next, enable the mouse clicks and movements in OpenGL and allow the user to select
a square region in the displayed Mandelbrot Set image. When the region is selected (ie. the mouse button released),
recalculate the Mandelbrot Set using the minimum and maximum c range based on the selected region. Limit the
selected region to be square, not rectangular. Also, again for grad students only, maintain a history of the displayed
sets and allow for a keypress of b to return to the previous set.
Copying the Project Skeletons
1. Copy the files from the ECE6122 user account using the following command:
/usr/bin/rsync -avu /nethome/ECE6122/MBSet .
Be sure to notice the period at the end of the above commands.
2. Change your working directory to MBSet
cd MBSet
3. Copy the provided MBSet-skeleton.cc to MBSet.cc as follows:
cp MBSet-skeleton.cc MBSet.cc
4. Build the binaries using the provided Makefile.
5. Run your program normally:
./MBSet
This should open an X-window on your laptop with the proper Mandelbrot set display.
2
USing CUDA You may want to use CUDA and GPU’s for this project to add a skillset to your resume; I will grade
the CUDA submissions for a small number of extra credit points. Be aware that there are not very many jinx systems
with GPU’s so you might have trouble getting access to one.
1. Log in to the cluster using jinx-login.cc.gatech.edu.
2. Use the qsub command to gain access to one of the jinx systems with attached GPU card
qsub -I -q class -l nodes=1:gpu
3. After a hopefully short wait you will get access to one of the GPU-Enabled systems, and you can edit and
compile your program as needed. Entering exit will exit the GPU node back to jinx-login.
4. There is a Makefile and skeleton code in /nethome/ECE6122/MBSet-Cuda that you can copy as usual.
Turning in your Project. Use the normal turnin-ece6122 procedure as always.
If you implemented a CUDA version then turn in two different solutions. First the MBSet as above.
In addition also turn in the CUDA version from a directory called:
MBSet-Cuda
You will have to create this directory and put all MBSet code there including the CUDA code.
3
