MathGroup Archive 2004

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

Search the Archive

Re: 3D data set

  • To: mathgroup at smc.vnet.net
  • Subject: [mg50980] Re: 3D data set
  • From: koopman at sfu.ca (Ray Koopman)
  • Date: Thu, 30 Sep 2004 04:52:40 -0400 (EDT)
  • References: <cjdq0q$ar3$1@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Giovanni Bellesia <giovanni.bellesia at ucd.ie> wrote in message 
news:<cjdq0q$ar3$1 at smc.vnet.net>...
> I have a general question regarding a topic which is not completely
> new to the forum.
> I have a 3D data set (from a Monte Carlo simulation) which are
> supposed to lay approximately on a helix.
> Does anybody knows a clear and efficient procedure to fit these
> points to a regular, circular helix.
> I read something about this in a message dated may 2004 by D.L.
> but I wasn't able to download the related files

The thread you refer to was concerned with fitting a cylinder, which
is a reasonable first step in fitting a helix. But before putting
a lot of time into developing an automatic fitter, have you tried
looking at your data? How many points do you have? How much do you
know about them? What should the cylinder/helix look like -- long &
thin, or short & thick? How uniformly should the points be distributed
along the 'time' axis? How many periods should there be? Where should
the axis of the helix be, and what should its radius be? How much
noise is there, relative to the length & radius of the helix? Is there
more noise is some directions than others? If you were to 'look down
the tube', could you see through to the end, or would noise make it
look like a clogged artery? A general routine, that goes in blind to
the data, will have to answer these questions for itself and will be
much harder to develop than one that knows something about the data.


  • Prev by Date: Re: Error with NDSolve
  • Next by Date: Re: unevaluated, hold, holdform
  • Previous by thread: 3D data set
  • Next by thread: "Evaluate from top down to specific cell" Command (interface suggestion to Wolfram)