How can I adjust the behavior of 'levenberg-marquardt' algorithm in fsolve?

I am solving equation numerically by using fsolve. When I used 'trust-region-dogleg' algorithm, the numerical result changes a lot when I adjust the value of 'FunctionTolerance' and 'OptimalityTolerance' in the optimoptions function.

Then, I tried to use the 'levenberg-marquardt' algorithm. But the numerical result did not change even if I set 'FunctionTolerance' and 'OptimalityTolerance' to some crazy values.

Therefore, I would like to ask which options in the optimoptions function should I set so that I can play with different results with the 'levenberg-marquardt' algorithm? In other words, which parameters are the 'levenberg-marquardt' algorithm more sensitive to?

Thanks in advance.

Answer 1

[Matt J]

There is no general pattern of sensitivity that you will see with either of the algorithms to any particular tolerance parameter. Whatever pattern you are seeing in your experiments is purely due to the specifics of your objective function, and maybe also your initial guess.