Python lmfit：模型对象尝试使用“ self”作为自由变量

Question

I am using the Python module "lmfit" to fit a function to some data. The variables should have constraints and I want to implement the fitting procedure into an existing framework. That is why the fit function has to be a class method. Therefore the first argument of the function should be "self".

 import lmfit class foo(object): def __init__(self): self.x_data = range(10) self.y_data = range(10) + 2 def fit_func(self, x, a, b): return a * x + b def fit_with_lmfit(self): model = lmfit.Model(self.fit_func, independent_vars=["x"]) model.set_param_hint("a", 1) model.set_param_hint("b", 2) results = model.fit(self.y_data, x=self.x_data) return results 

But now lmfit interprets "self" as free variable for the fit alogrithm. I tried adding a parameter hint:

model.set_param_hint("self", self, vary=False)

But that results in the following error:

TypeError: fit_func() got multiple values for keyword argument 'self'

I know that that curve_fit() from scipy.optimize does not has any problems with class methods and neither should leastsq() . But I would like to use lmfit since it offers a nice interface to implement constraints for some variables.

On could also define fit_function() inside fit_with_lmfit() but I need to have access to it from outside the class as well and returning the function alongside results seems messy.

So my question would be: is there any way to let lmfit ignore certain variables of a fit function or better yet ignore "self" as positional argument?

Answer 1

Since fit_func is not using self, make it a static method:

@staticmethod
def fit_func(x, a, b):
    return a * x + b

This way you don't have to worry about self .