Fit a scalar scale factor

Fit one bounded parameter to a small analytic forward model. This is the smallest complete inverse-problem setup: a callable, an observation, and one or more FitParameter objects.

The fitted scale is approximately 2.0 (dimensionless). Predictions and residuals have shape (3,). With these exact synthetic observations, the residuals approach zero.

converged (42 model evaluations)
objective = 0.000000
{'scale': 2.0}

import numpy as np

from PyMieSim import FitParameter, MeasuredData, fit_parameters

x = np.array([1.0, 2.0, 3.0])
observation = MeasuredData(values=np.array([2.0, 4.0, 6.0]))


def model(parameters):
    return parameters["scale"] * x


result = fit_parameters(
    model=model,
    observation=observation,
    parameters=[FitParameter(name="scale", initial=0.8, bounds=(0.0, 4.0))],
)

print(result.summary())
print(result.parameters)