How to use weights parameter

[isentropic]

I'm sort of confused with the norms after seeing multiple issues.
And it is not exactly clear from documentation:

If I have points with x and y with standard errors delta_y (standard deviations)
Should I
curvefit(f, x, y, 1/delta_y^2) or curvefit(f, x, y, 1/delta_y)

This confused me so many times and I wish this simple example would be included in the docs

[Magalame]

From the readme, w is the inverse covariance matrix. So if delta_y is the standard derviation, i.e. the square root of the covariance matrix (or vector in this case), you want 1/delta^2

[pkofod]

It is actually explained in the docs but they've not been generated for a long time. They will be updated when I tag a new version. https://julianlsolvers.github.io/LsqFit.jl/dev/tutorial/#Weighted-Least-Squares