max_iter not being obeyed?

[phajy]

When trying a test fit with very few iterations, e.g.,

result = fit(prob, LevenbergMarquadt(), max_iter=2)

the fit just keeps going. Not sure why. It looks to me like max_iter gets passed as maxIter to LsqFit.curve_fit and then on to levenberg_marquardt.

[phajy]

As a corollary it would be useful is this worked with max_iter=0 or we could define a "do nothing" fitting algorithm for code testing (would save having to write a different few lines to evaluate the model when testing). This would basically be the same as function fit(config::FittingConfig, alg::LevenbergMarquadt; etc. just without calling _lsq_fit.

[fjebaker]

I can't reproduce this. When I pass max_iter=1 for example, it performs only a single iteration of the algorithm. Do you have some example code that reproduces what you were experiencing?

[phajy]

I'll try to prepare a minimal (not) working example. Although perhaps it is the difference between 0 and 1? Watch this space...