Attempt to Plot a Coloured Surface to Visualize 4D data.

Hello, I'm attempting to visualize some 4D data in order to understand the toplogy of an optimization problem that I've been set.

I currently have a large matrix, that is size 3600x4. The first 3 values are randomly generated "gains" for a control system, and the last is the "fitness" of the gains determined by running the gains through a dynamic simulation and determining how close those gains get us to a prescribed demand.

Now what I want is to plot out a surface that covers the three gains say [K1, K2, K3] = data(:,1:3); and determine their colour based on the value of the fitness F, with say, F = 0 being red, and F = 1 being green (F is between 0 and 1 not inclusive.)

The code below generates a nearly identical data set.

x = 20*rand(3600,1);
y = 20*rand(3600,1);
z = 20*rand(3600,1);
F = rand(3600,1);
data = [x,y,z,F];

Where for my own eyes I usually follow this with a

topologydata = sortrows(data,size(data,2));

So that I can see if the best fitness is within reasonable bounds. However, you can leave it unsorted if the method that you use prefers or doesn't need it. Realistically, if I can get the plot out the way I'm visualizing it, then I won't need to sort it.

[darova]

What about slice or isosurface?

[Ayden Clay]

Either of those would absolutely do the trick, and I've since figured out my mistake. Thank you for your suggestions.

Answer 1

[Ayden Clay]

As the comments have mentioned, slice/isosurface/scatter3/etc. will all do the trick, you can input the three dimensions as the gains and then use the fitness to determine the colour using something like:

m = size(topologydata,1);
green = [0,1,0];
red = [1,0,0];
colors_p = [linspace(red(1),green(1),m)', linspace(red(2),green(2),m)', linspace(red(3),green(3),m)'];

Generates the red -> green colour map.