On Tue, 18 Mar 2008 15:02:47 GMT, "Nir via MathKB.com" <u41835@[EMAIL PROTECTED]
>
wrote:
>Hi,
>I would like to generate a 3D function of three variables based on
>experimental data f(x,y,z). For example, let assume that one measured the
>room temperature at 100 different location in the room.
>So, now we have 100x4 array thus, at each line we have x y z and T. My
goal
>is to fit a 3D function based on experimental data meaning to find
analytic
>expression that beast descried temperature in the room for any point
(T=f(X,Y,
>Z)). For example T=3.3X+4Y-5Z^2; the function can be from any kind
>(exponential, polynomial, sin, cos.)
>Can anyone tell me if it?s doable?.
Yes, it's doable. You probably want so look into multivariate spline
interpolation. One such method is Shepard's quadratic method and its
variations. You might start here:
http://****tal.acm.org/citation.cfm?id=45054.45055
Other papers by Barnhill may be useful.
--Lynn
|
