MathGroup Archive 2005

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

Search the Archive

Minimax Approximation

  • To: mathgroup at
  • Subject: [mg60738] Minimax Approximation
  • From: dh <dh at>
  • Date: Mon, 26 Sep 2005 04:05:42 -0400 (EDT)
  • Sender: owner-wri-mathgroup at

in numerical math, a common job is to fit some function with parameters 
"as well as possible" to either a given function or to a set of data 
points. "As well as possible" is mostly chosen to  mean "least square", 
that is the sum of squares of the error is minimized. However, there are 
cases wher one would like to minimize the largest error. This 
approximation scheme is called Minimax or L-infinity approximation. 
Unfortunately, Mathematica offers only a very restricted version for 
Minimax, that only works for rational function that are not zero.
Well, does anybody known a Mathematica implementation of a more useful general 
Minimax routine?

  • Prev by Date: Re: errors on parameters from NonlinearRegress
  • Next by Date: Interactive surface manipulation with GUIkit
  • Previous by thread: Re: Is this possible? Residue computation leads to complexInfinity?
  • Next by thread: Interactive surface manipulation with GUIkit