Create a Forum Topic by Logging In or Creating an Account

Fitting system of equation

Fri, 04/03/2015 - 05:32 pm

Hi all,

I am a low-level user of Igor; despite reading a lot of threads in the forum I was unable to find a solution for my problem so I will be very happy to get some advices.

Here is my issue; I have a set of experimental data, u, v, w, to fit in function of one variable, R, the same process characterizing their relationship, thanks to three parameters. So, a global fitting will do the job easily, having three functions for u(R), v(R) and w(R)... unfortunatelly, impossible as linear relationship are implied...

The best I can do is to express two of them in function of the variable and the last one, so w = fct(R, v) and u = fct(R, v). The last equation of the system allows me to obtain an non linear expression fct(R,v) = 0, for wich I can use Findroots to obtain v values during the fitting process for either u or w. I would thus be able to perform a global fitting on both u and w. But I cannot use the experimental v values I have to obtain a better fit...

My questions are thus (i) is there a better way to fit a system of equation? (ii) if not, how is it possible to extract the roots (v) used during the fitting process?

Thanks for help

johnweeks

I don't really follow what you are trying to describe, but this:

Quote:
The last equation of the system allows me to obtain an non linear expression fct(R,v) = 0, for wich I can use Findroots to obtain v values during the fitting process for either u or w.

suggests that you might be able to use implicit fitting:

DisplayHelpTopic "Fitting Implicit Functions"

To understand that section, you really need to start with the previous section:

DisplayHelpTopic "Errors in Variables: Orthogonal Distance Regression"

Copy those commands, paste them into Igor's command line, and press Enter.

John Weeks
WaveMetrics, Inc.
support@wavemetrics.com

April 6, 2015 at 10:13 am - Permalink