Matlab ToDo

[gusl]

The research is looking quite promising. Mary's and Scott's tips were very useful for making my surface plot.

To do:

Find isosurfaces for my function of 4 variables. I can't visualize the 4 variables at once, but I'm ok visualizing 3 variables at a time, with e.g., an onion-peel visualization (each surface corresponds to a value of the function). I don't know if this is standard.

If it isn't, I'll need to write code for computing tangent vectors, generate my own mesh, etc. I really don't want to go down that path.

Also, at some point, I'll want to do some optimization under linear constraints.

Comments

[infryq]

you *could* attempt a 3D contour plot, using surfaces like you said.

You might also try using plot3 to plot a 3D grid, letting the color of each point represent the value of the function at that point.

You might also try plotting a 3D vector space (quiver?), letting all vectors point in the same direction and letting their magnitude be your 4th dimension.

[gustavolacerda.livejournal.com]

Thanks.

Your last idea seems to make no sense. Remember that I want to see the value of the function at each point.

[gustavolacerda.livejournal.com]

oh, would you get rainbow-colored arrows for each point in 3D space?

[infryq]

If you combined 2 and 3, sure.

Otherwise, 3 is just like 2 -- you don't get contour plots, so you're only going to be able to see very general shapes in the values of the function along the 4th axis. It's just that instead of colors in the visible spectrum, you're using arrow lengths between some min and max. 'quivver' plots a vector field; I imagine there's an example in the matlab help for it.