MathGroup Archive 2004

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: ploting in high dimensions


On 7 Jul 2004, at 14:42, sean kim wrote:

> *This message was transferred with a trial version of CommuniGate(tm) 
> Pro*
>
> Hello group,
>
> I was wondering...
>
> let's say you have a 10 variable system.
>
> now i want to define the 10 axes for the variable like a spokes of 
> bicycle wheel.
>
> so you get 10 spokes emamanting out from the origin equally spaced.
>
> now plot the 10 variable on that space.
>
> is that doable or am i being too hopeful?
>
> If you Have to do something like this what would you all do?
>
> Thanks in advance for any thoughts or insights.
>
> Sean
>
>

In principle this can be done but to what purpose? You would get a 
phenomenally misleading picture, no better than projecting three 
dimensional solid onto a line.  There would all sorts of horrible 
intersections and singularities even in a representation of a smooth 
manifold. In my opinion this would be far worse than not having any 
picture at all. Actually, in some situations it is helpful to represent 
a number of parameters as one on a single axis, but you need a good 
insight into the situation to decide when this will be useful and which 
parameters should be groped together. But any such thing has to be very 
much tailored to a specific problem.

Andrzej Kozlowski


  • Prev by Date: RE : Parametric plot
  • Next by Date: Re: Normal distribtion
  • Previous by thread: ploting in high dimensions
  • Next by thread: RE: ploting in high dimensions