Skip to content
wojdyr edited this page Apr 9, 2013 · 2 revisions

Why Fityk and Excel (or other program) give different results?

Fityk is using weighted least squares and the the default weights of points are not equal. To get the same results as in Excel or gnuplot, set all weight to 1: S=1. To make not-weighted fitting default, add default_sigma=one to the init file. (Originally, the program was written to analyse powder diffraction data. In this, and in many other fields, residuals are usually weighted by 1/√y).

How can i constrain a particular fitting parameter to always stay positive?

See InequalityConstraints.

Clone this wiki locally