[Python] Use `curve_fit` with `partial` using named parameters instead of positional parameters

I'm trying to use curve_fit on a function for which I want to freeze one or more parameters using partial.

E.g. this pattern, which is working:

from scipy.optimize import curve_fit
from functools import partial

# Irrelevant
xdata = [0]*5
ydata = xdata


def func(x, a, b):
return x + a + b

pfunc = partial(func, 2)
popt, _ = curve_fit(pfunc, xdata, ydata)
print(popt)
# > [-2.]


What I want to do is provide partial with the parameter(s) as named parameters, not positional. This however seems not to work reliably.

from scipy.optimize import curve_fit
from functools import partial

# Irrelevant
xdata = [0]*5
ydata = xdata


def func(x, a, b):
return x + a + b

pfunc = partial(func, b=2)
popt, _ = curve_fit(pfunc, xdata, ydata)
print(popt)
# > [-2.]

pfunc = partial(func, a=2)
popt, _ = curve_fit(pfunc, xdata, ydata)
print(popt)
# > raises ValueError: Unable to determine number of fit parameters.



Am I doing something wrong? Is there a way to get this to work? In my application, the func may take n>=3 of which I want to fix up to (n-1).

Continue reading...