Multidimensional Visualizer

I spend a lot of time thinking about and working on multidimensional problems. This can be particularly daunting to visualize so I created a visual tool to allow for the exploration of multidimensional problems. First, let’s better explain what I mean by a multidimensional problems; so let’s look at an example. Let’s say my mood was tied to the time of day. Well I could easily construct a graph for someone to easily visualize that relationship where say my mood is low in the morning and the evening and is high in the middle of the day.

But my mood is also dependent on how hungry I am too… Well we could setup another axis and now my mood is defined by this surface described by the combination of time of day and hunger where my mood goes down as my hunger goes up.

But my mood is also dependent on the weather… Hmmm… Well I guess we could take that 3d surface and animate it so you can visualize how that surface changes over time and as we pass through time in the animation we start at bad weather and head towards good weather and we see the surface rise up.

*** well… it’s not animated really because I am lazy, but you imagine those surfaces animated moving up!

Buy my mood is also dependent on bank account balance… Uh-oh… we are going to have to start getting very creative! The point is that multidimensional problems; those systems that’s state is dependent on a number of simultaneous factors is just not easily visualized.

So the application I created exposes a technical interface that can plug-in to simulations. You just need to implement an interface from the library that allows for values to be retrieved for a given multidimensional input, the rest is taken care of. The application then lets you pan, zoom or rotate in any dimension and visualize a 2d slice of the multidimensional problem. Just like a cat scan allows you to see a slice of a 3d object, this allows you to take slices of objects of n dimensions from any dimensional angle.

I used this in aiding in solving an IBM problem where I needed to look for a local maximum. By seeing the contours I was able to use the pattern matching power of the human brain and a little exploration to root around in the multidimensional universe that the data creates and find the solution. I admittedly have not gotten the reuse I had hoped out of it, but I think it is very interesting tool to keep in my toolbox. I remember a real sense of exploring, like I had a magic little window that I could see through in any place and at any angle within an infinite multidimensional universe; it gives you a very tactile experience.