MathGroup Archive 1998

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

Search the Archive

Re: Re:Nonlinear fit to inverse power law data plot


Hi Debbie,

You can fit 
y=c*x^(-n) simply with a linear fit (Fit[]) if You use

{Log[x],Log[y]}

instead of the original data pairs {x,y}

Your model say that Log[y]==Log[c]-n*Log[x]

say Your data are named dataset
dataset=Table[{x,0.2*x^(-1.5)},{x,0.5,10,0.1}];

fitPara=Fit[dataset /. {x_?NumericQ,y_}:>{Log[x],Log[y]},{1,lnx},lnx] /.

    a_. +b_.*lnx :> {E^a,b}

will return a list of c and -n.

Hope that helps
  Jens


-----Original Message-----
From: jleddon <jleddon at cyberramp.net> To: mathgroup at smc.vnet.net
Subject: [mg14149] [mg14115] Re:Nonlinear fit to inverse power law data plot


>
>Hello,
>I have a set of data that when plotted, produces a curve that is
>definitely of a inverse-power-law form;
>
> y=constant*x^-n       where n may be a non-integer.
>
>I have tried the nonlinear fit function with model function forms,
>however, I have not been able to get a non-integer value for n.
>
>Does anyone have any ideas?
>
>Thanks in advance for your help!
>
>Regards,
>Debbie Leddon
>
>
>



  • Prev by Date: Re: 3D Object from Listplot?
  • Next by Date: Printing from Kernel
  • Previous by thread: Re:Nonlinear fit to inverse power law data plot
  • Next by thread: argument passing