MathGroup Archive 2007

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

Search the Archive

Re: optimizing

  • To: mathgroup at smc.vnet.net
  • Subject: [mg74902] Re: optimizing
  • From: dh <dh at metrohm.ch>
  • Date: Tue, 10 Apr 2007 06:02:01 -0400 (EDT)
  • References: <evabpk$df6$1@smc.vnet.net>


Hi,

your model function has the form: Exp[-k t], no need to solve an DE. Now 

you have the model function and your data and you can e.g. use "FindFit" 

to get the obtimal k.

Daniel



kaiwan wrote:

> Hi! I have to find the best parameter "k" with witch function sd[k]->minimum. I do that with FindMinimum.  "sd" if a function that calculates the standard deviation from my experimental and model data. The model data is obtained from the mathematical model, first order differential equation...which I solve using NDsolve. The problem is that I cant make NDSolve work under FindMinimum. 

> 

> Something like this:

> First, I define function that hace to solve the equation c'[t] == -k c[t], where "k" is a variable. This gives me model data later on. I define it as a function cause I dont know the value of "k" yet. 

> 

> Then I define another function which calculates the standard deviation:

> sd[k_]:= function(model data, experimental data)

> 

> Still I have no concrete results. 

> 

> Then I use FindMiniumm for finding the "k" that gives the smallest Standard Deviation

> FindMinimum[sd[k], {k, 0.04}]

> 

> And her is the problem, I cant make this combination to work .... help please? Many thanks!

> 




  • Prev by Date: Re: small question
  • Next by Date: Re: Showing the points on a surface about a circle in the plane
  • Previous by thread: optimizing
  • Next by thread: .NET/Link Callback from a DLL