Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.soft-sys.math.mathematica > #13046 > unrolled thread

Fitting multiple data sets to multiple functions with shared fit parameters

Started bydr DanW <dmaxwarren@gmail.com>
First post2013-01-07 05:35 +0000
Last post2013-01-08 04:05 +0000
Articles 2 — 2 participants

Back to article view | Back to comp.soft-sys.math.mathematica


Contents

  Fitting multiple data sets to multiple functions with shared fit parameters dr DanW <dmaxwarren@gmail.com> - 2013-01-07 05:35 +0000
    Re: Fitting multiple data sets to multiple functions with shared fit parameters danl@wolfram.com - 2013-01-08 04:05 +0000

#13046 — Fitting multiple data sets to multiple functions with shared fit parameters

Fromdr DanW <dmaxwarren@gmail.com>
Date2013-01-07 05:35 +0000
SubjectFitting multiple data sets to multiple functions with shared fit parameters
Message-ID<kcdmru$m0b$1@smc.vnet.net>
I am a an experienced user of Mathematica and am used to fitting complex valued functions to data sets, but only one function at a time.  I am attempting to determine the parameters in a model of a physical system (an acoustic transducer).  From the model, I can determine several transfer functions over frequency or the Laplace s, and I can also measure these transfer functions on a device.  This gives me, for example, three functions (which share parameters) and three data sets, and I want to find the parameters that fit all three functions to their respective data sets simultaneously.  The parameters in the model are real, but the functions and data are complex. Another complication is that the data sets have different units and scales, so I will need to normalize things so \chi^2 on all curves are weighted equally.

Is there a way to directly use the built in FindFit or NonlinearModelFit to solve this problem?  If not, can anyone suggest how I could construct my own method using NMinimize?  I would be very grateful if the group could point me toward an example where this as been done.

Regards,
Daniel

[toc] | [next] | [standalone]


#13056

Fromdanl@wolfram.com
Date2013-01-08 04:05 +0000
Message-ID<kcg5tn$rd7$1@smc.vnet.net>
In reply to#13046
On Sunday, January 6, 2013 11:35:58 PM UTC-6, dr DanW wrote:
> I am a an experienced user of Mathematica and am used to fitting complex valued functions to data sets, but only one function at a time.  I am attempting to determine the parameters in a model of a physical system (an acoustic transducer).  From the model, I can determine several transfer functions over frequency or the Laplace s, and I can also measure these transfer functions on a device.  This gives me, for example, three functions (which share parameters) and three data sets, and I want to find the parameters that fit all three functions to their respective data sets simultaneously.  The parameters in the model are real, but the functions and data are complex. Another complication is that the data sets have different units and scales, so I will need to normalize things so \chi^2 on all curves are weighted equally.
>
>
>
> Is there a way to directly use the built in FindFit or NonlinearModelFit to solve this problem?  If not, can anyone suggest how I could construct my own method using NMinimize?  I would be very grateful if the group could point me toward an example where this as been done.
>
>
>
> Regards,
>
> Daniel

The following links may help. I'm sure there are others as well; this has come up a number of times in comp.soft-sys.math.mathematica.

http://forums.wolfram.com/mathgroup/archive/2010/Apr/msg00280.html
http://forums.wolfram.com/mathgroup/archive/2012/May/msg00289.html

Daniel Lichtblau
Wolfram Research

[toc] | [prev] | [standalone]


Back to top | Article view | comp.soft-sys.math.mathematica


csiph-web