Thread Subject: multi dimensional contour plotting

Hi everyone,
I've searched around on google and matlab forums, but i haven't found an adequate solution to my situation.

I am exploring phase space of 3 parameters (A, B, C) in a non linear model, and tracking a response variable (N).

I set A=some value,
then scan the plausible ranges for B & C and create a contourf plot.

I've done this over a range of plausible values for parameter A, but i'd like to put everything into 1 figure.

I was thinking B and C could be on the X and Y axes, and A could be on the Z axis. Then for levels of Z i could insert the aforementioned contourf plots.

Any input would help!
Thanks in advance,
Gurbir

Gurbir,

What you're looking for is an isosurface.

Check out the isosurface commands built into matlab - I'd recommend (certainly to start with) putting A,B and C on the X, Y, and Z axes. then, plot isosurfaces of your response data at different values - after a while, you'll become familiar with what isosurfaces are best.

If you're ultra-keen, you could plot multiple isosurfaces at the same time, and set them to be semi-transparent so you can compare!

Hints:
1. Use meshgrid or ndgrid to generate 3D coordinates arrays (X,Y,Z).

2. Generate an array R (same size as X,Y and Z) containing the value of the response at this combination of X,Y and Z. e.g. R(1,1,1) = response_function(X(1,1,1), Y(1,1,1), Z(1,1,1));

3. You can achieve (2) either by evaluating the response at each combination of x,y,z (could be slow, depending on the response function) or by determining the response at various scattered X,Y,Z coordinates then using griddatan to interpolate the scattered data to a regular grid.

Hope this helps!

Tom C