A problem in official website 'Least-Squares (Model Fitting) Algorithms'

Under 'Levenberg-Marquardt Method', it says 'Consequently, for some sufficiently large λk, the term F(xk + dk) < F(xk) holds true'. 'F(xk + dk)<F(xk)' should be corrected to 'norm(F(xk + dk))<norm(F(xk))' . Right?

[Steven Lord]

Searching the online documentation for 'Levenberg-Marquardt Method' found 73 hits in the documentation itself, with additional hits in the Examples, Functions, Blocks, Apps, Videos, Answers, and File Exchange categories. To which of those pages does your feedback relate? Please post a link to the page in question.

[Zhanchun Yang]

https://uk.mathworks.com/help/optim/ug/least-squares-model-fitting-algorithms.html?requestedDomain=uk

And the same issue on pages of other languages, e.g．Chinese．

https://ww2.mathworks.cn/help/optim/ug/least-squares-model-fitting-algorithms.html

Thanks！

Answer 1

[Alan Weiss]

Thank you for pointing out this long-standing error in the documentation. I will fix the error as soon as practicable.

Alan Weiss

MATLAB mathematical toolbox documentation

Answer 2

[Bruno Luong]

I think more like

... f(xk + dk) < f(xk) holds true