MathGroup Archive 2009

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

Search the Archive

Re: Fitting problem

  • To: mathgroup at smc.vnet.net
  • Subject: [mg95536] Re: Fitting problem
  • From: dh <dh at metrohm.com>
  • Date: Wed, 21 Jan 2009 06:44:32 -0500 (EST)
  • References: <gl4a9k$gk9$1@smc.vnet.net>


Hi,

here is an attempt to tackle your problem:



We have one independent variable, time: t.

we have datasets

d1:{{t1,d11},{t2,d12}..

d2:{{t1,d21},{t2,d22}..

..

we have models with parameters p (not all models may need all parameters):

mod1[t,p1,p2..]

mod22[t,p1,p2..]

...

we define data interpolation functions that we will use to define an 

error function:

int1:=Interpolation[d1]

int2:=Interpolation[d2]

...

define an error function:

err[t_,p1_,..]:= (f1[t]-mod1[t,p1,..])^2 + (f2[t]-mod2[t,p1,..])^2+...

we define artificial zero data:

artdat={{t1,0}{t2,0},..}

Finally we fit err against the zero data:

FindFit[artdat,err[t,p1,..],{p1,p2,..},t]



hope this helps, Daniel



Ktota wrote:

> Dear All ;),

> 

> i have the following situation:

> 

> 5 mathematical models describing the formation of different species.

> The models  (so the species) are interdependent, means.. they all have

> the same parameters but some of the have also other additional

> parameter. Example: Model 1 has a parameter "lifetime1"... model 2

> will include this parameter "liftime1" and a new parameter

> "lifetime2".

> 

> My trouble is now to find best fit values for the parameters in all

> the models at the same time. I don't want to find a best fit for model

> 1 and then find out when doing best fit for model 2 that model 2 won't

> work with the parameter values found in the first run with model 1. So

> the fitting must consider all the models to find best fit values for

> the parameter.

> 

> I wonder if this achievable with mathematica or if i would be better

> of with other software

> 

> thank you for your help

> 

>  Ktota

> 

> 




  • Prev by Date: Re: Re: Which editor do you use for math articles
  • Next by Date: Re: bug in LinearProgramming
  • Previous by thread: Re: slots with 2 indexed array?
  • Next by thread: Re: Fitting problem