This example shows you how to run a global optimisation with CMA-ES.
For a more elaborate example of an optimisation, see: basic optimisation example.
In [1]:
from __future__ import print_function
import pints
import pints.toy as toy
import numpy as np
import matplotlib.pyplot as pl
# Load a forward model
model = toy.LogisticModel()
# Create some toy data
real_parameters = [0.015, 500]
times = np.linspace(0, 1000, 1000)
values = model.simulate(real_parameters, times)
# Add noise
values += np.random.normal(0, 10, values.shape)
# Create an object with links to the model and time series
problem = pints.SingleOutputProblem(model, times, values)
# Select a score function
score = pints.SumOfSquaresError(problem)
# Select some boundaries
boundaries = pints.RectangularBoundaries([0, 400], [0.03, 600])
# Perform an optimization with boundaries and hints
x0 = 0.01, 450
sigma0 = [0.01, 100]
found_parameters, found_value = pints.optimise(
score,
x0,
sigma0,
boundaries,
method=pints.CMAES
)
# Show score of true solution
print('Score at true solution: ')
print(score(real_parameters))
# Compare parameters with original
print('Found solution: True parameters:' )
for k, x in enumerate(found_parameters):
print(pints.strfloat(x) + ' ' + pints.strfloat(real_parameters[k]))
# Show quality of fit
pl.figure()
pl.xlabel('Time')
pl.ylabel('Value')
pl.plot(times, values, label='Nosiy data')
pl.plot(times, problem.evaluate(found_parameters), label='Fit')
pl.legend()
pl.show()