MathGroup Archive 2006

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

Search the Archive

Re: Non Linear fit problem

  • To: mathgroup at smc.vnet.net
  • Subject: [mg67868] Re: Non Linear fit problem
  • From: dh <dh at metrohm.ch>
  • Date: Tue, 11 Jul 2006 05:58:33 -0400 (EDT)
  • References: <e8nst3$k60$1@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Hi Raj,
I am not sure that I understand what you are saying.
You have 2 time series whose amplitudes are proportional with a factor 
Erf[..] but whose time behaviour have a shift?
The shift is best computed using Correlation.
If you know the shift, you can then shift one of the series and fit the 
factor.
Daniel

Raj wrote:
> I have a problem of fitting two temperature lists to an error
> function. The physical description of the problem is as follows:
> 
> t1 is a list of temperatures on the surface of the soil and t2 is the
> list of temperatures at a certain depth (say x).
> We know that the dependence of t2 on t1 is as
> 
> t2 = t1 Erfc[x/(2 sqrt(k)]
> 
>  where x is the depth at which t2 is measured and k is the constant of
> diffusivity.
> 
> When I plot the two temperatures, I can see that t2(temperature at a
> depth) lags behind t1 by a certain time.
>  My problem is to find the temperature at any depth given the
> temperature at the surface (t1). Is there a way in Mathematica to find
> k such that t2 and t1 have a close fit?
> 
> 
> Thanks,
> 
> Raj
> 


  • Prev by Date: 3D plot of globe (map of world)?
  • Next by Date: Re: Redundant numerical input
  • Previous by thread: Re: Non Linear fit problem
  • Next by thread: A mistake by Mathematica?