LMFIT_SaveLoadModel.py

import numpy as np
import lmfit as lm
import matplotlib.pyplot as plt

def mySine(x, amp, freq, phi):
    return amp * np.sin(2*np.pi*freq*x - phi)

ModelFile = 'Sine_Model.sav'

Model1 = lm.Model(mySine)
Pars1  = Model1.make_params(amp=1., freq=0.25, phi=1.)
Pars1['amp'].max = 3.
Pars1['amp'].min = 0.
# Pars1['freq'].max = np.inf
Pars1['freq'].min = 0.
# Pars1['phi'].max = 2*np.pi
# Pars1['phi'].min = 0.

lm.model.save_model(Model1, ModelFile)

Model2 = lm.model.load_model(ModelFile, funcdefs={'mySine': mySine})

Pars2 = Model2.make_params(amp=2., freq=0.5, phi=-1.)

x  = np.linspace(0., 3*np.pi, 101)
dy = np.random.normal(0, 0.1, x.size)
y1 = mySine(x, **Pars1.valuesdict()) + dy
y2 = Model2.eval(Pars2, x=x) + dy

result1 = Model1.fit(y1, Pars1, x=x)
print(result1.fit_report())

result2 = Model2.fit(y2, Pars2, x=x)
print(result2.fit_report())

plt.plot(x, y1, 'bo')
plt.plot(x, result1.best_fit, 'r-')

plt.plot(x, y2, 'go')
plt.plot(x, result2.best_fit, 'r-')

plt.show()