def fitness_function(chromosome):
"""
Returns a tuple of fitnesses for the given chromosome.
The objectives used are:
overall least-squares fit.
number of times crossing the line y = 1
intial value y(x=0)
"""
ymesh = functional_form(chromosome, xmesh)
residual = chi_squared(ymesh, fit_data)
zero_crossings = count_zero_crossings(ymesh - 1)
return (1.0/(delta + residual),
1.0/(delta + (zero_crossings - fit_czc)**2),
1.0/(delta + (ymesh[0] - fit_data[0])**2))
x is where to evaluate the function
"""
return par[0] * numpy.exp(par[1] * x * numpy.sin(par[2] * x)) + par[3]
# Generate some noisy data to fit.
target_parameters = (1.5, 2.73, 5.15, 0.3)
data_size = 500
xmin = 0
xmax = 10
xmesh = numpy.linspace(xmin, xmax, data_size)
fit_data = functional_form(target_parameters, xmesh)\
+ numpy.random.normal(size=data_size)
fit_czc = count_zero_crossings(fit_data - 1)
# Here is a fitness function that evaluates individuals using multiple objectives
delta = 1e-15
def fitness_function(chromosome):
"""
Returns a tuple of fitnesses for the given chromosome.
The objectives used are:
overall least-squares fit.
number of times crossing the line y = 1
intial value y(x=0)
"""
ymesh = functional_form(chromosome, xmesh)
residual = chi_squared(ymesh, fit_data)
zero_crossings = count_zero_crossings(ymesh - 1)
