Applying constraints to a Fit

I would like to constrain the value of one parameter depending on the value of one of the other parameters. For example I have 2 parameters a and b and I want a constraint such that:
b <= 2 * sqrt(a)
Can this be done?
Thanks in advance,


I have a similar question. For example, the two parameters(a,b) are contrained in equation function: b = f(a).
I know some principles(or methods, such as Lagrangian Multipliers Method…)
But i don’t know how to implement in Root(which class could I use? …)
Does anybody know how to apply these constraints in doing fitting?


ROOT does not provide algorithms for optimization with non trivial constraints. Simple constraints, like
A < parameter < B are supported and in some cases you can eliminate the constraints by a parameter transformation.
In some other cases you can express your constraints by adding a penalty term in the minimization function (your likelihood or least square function when fitting).
More complex situations are not supported, and you will have to implement your self or look what is available on the net. You find a review of the available algorithm under … ware_Guide