Specifying relationships between fit parameters about iminuit HOT 1 CLOSED

In Minuit itself, you can

• Fix parameters to constant values Minuit.fixed
• Constrain parameters to an interval Minuit.limits
• Use penalty-terms to soft-constrain a parameter with iminuit.cost.NormalConstraint
• Minuit.scipy further allows you to fit with arbitrary equality and inequality constraints, I think this is what you need

In a simple case like param2 = 2 * param1, you can reparametrize the model to implement the constraint.

def model1(x, param1, param2):
    return x * param1 + param2


def model2(x, param1):
    return model1(x, param1, 2*param1)

# proceed to fit model2 instead of model1

In more complicated scenarios where reparametrization is not obvious or possible, you can use Minuit.scipy with equality constraints.

from iminuit.