forcing fit parameters into a range

[Gelman, Laurent]

Hi everybody,

Is it possible, when using the fit function, not only to give the initial values for the parameters, but also to force these to stay within a given range, e.g. never get negative or positive... ?

Thanks for your help,

Best regards,

Laurent.


_____________________________________________
Laurent Gelman, PhD
Head of Facility for Advanced Imaging and Microscopy
(Light Microscopy)

Friedrich Miescher Institut
WRO 1066.2.16
Maulbeerstrasse 66
CH-4058 Basel
Fix: [phone]
Mobile: [phone]
Fax: [phone]

www.fmi.ch<http://www.fmi.ch/>
www.microscopynetwork.unibas.ch/<http://www.microscopynetwork.unibas.ch/>


--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Michael Schmid]

On Oct 28, 2013, at 12:57, Gelman, Laurent wrote:

> Hi everybody,
>
> Is it possible, when using the fit function, not only to give the initial values for the parameters, but also to force these to stay within a given range, e.g. never get negative or positive... ?

________________________________________________________________

Hi Laurent,

simply write your function such that it returns NaN (not a number) if it enters the forbidden range.
For this purpose, it's nice that the 'custom fit' function can also have a more complex form consisting of several statements.

Example:
  if (b < 0) y = NaN ; else y = a + b*x + c*x*x;

If you use one of the built-in fit functions, note that the fit procedure usually gets less efficient when converting it to a 'custom' function.  For the built-in functions, ImageJ splits off parameters that are mere factors and offsets and determines them by linear regression.  This is not done for custom functions.

Michael

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html