def __init__(self, d=10):
f_bias = -460 # magic number from model specification
# input space is d values from -pi to pi
independents = tuple((IV(lo=-math.pi, hi=math.pi) for _ in xrange(d)))
# "q_ij, r_ij are integer random numbers in the range [-100, 100]"
q, r = _randint_matrix(d), _randint_matrix(d)
# "alpha... [is a vector] of random numbers in [-pi, pi]"
alpha = [random.uniform(-math.pi, math.pi) for _ in xrange(d)]
# 1D matrix of d sums
a = [
sum(q[i][j] * math.sin(alpha[j]) + r[i][j] * math.cos(alpha[j])
for j in xrange(d)) for i in xrange(d)
]
def b_sum(i):
# generate a function for b_i = sum(q_ij sin(x) + r_ij cos(x))
q_sin = lambda j, x: q[i][j] * math.sin(x)
r_cos = lambda j, x: r[i][j] * math.cos(x)
f = lambda x: sum(q_sin(j, x) + r_cos(j, x) for j in xrange(d))
return f
# generate 1D matrix of functions b_i
b = [base.memoize(b_sum(i)) for i in xrange(d)]
# and finally, here's the function to minimize
def f12(xs):
if len(xs) != d:
e = 'len of input vector to {0} must be {0}.d = {1}'.format(
self.__name__, d)
raise ValueError(e)
return sum((a[i] - b[i](x))**2 + f_bias for i, x in enumerate(xs))
super(Schwefel, self).__init__(independents=independents,
dependents=(f12, ))
def __init__(self, d=10):
f_bias = -460 # magic number from model specification
# input space is d values from -pi to pi
independents = tuple((IV(lo=-math.pi, hi=math.pi)
for _ in xrange(d)))
# "q_ij, r_ij are integer random numbers in the range [-100, 100]"
q, r = _randint_matrix(d), _randint_matrix(d)
# "alpha... [is a vector] of random numbers in [-pi, pi]"
alpha = [random.uniform(-math.pi, math.pi) for _ in xrange(d)]
# 1D matrix of d sums
a = [sum(q[i][j] * math.sin(alpha[j]) + r[i][j] * math.cos(alpha[j])
for j in xrange(d)) for i in xrange(d)]
def b_sum(i):
# generate a function for b_i = sum(q_ij sin(x) + r_ij cos(x))
q_sin = lambda j, x: q[i][j] * math.sin(x)
r_cos = lambda j, x: r[i][j] * math.cos(x)
f = lambda x: sum(q_sin(j, x) + r_cos(j, x) for j in xrange(d))
return f
# generate 1D matrix of functions b_i
b = [base.memoize(b_sum(i)) for i in xrange(d)]
# and finally, here's the function to minimize
def f12(xs):
if len(xs) != d:
e = 'len of input vector to {0} must be {0}.d = {1}'.format(
self.__name__, d)
raise ValueError(e)
return sum((a[i] - b[i](x))**2 + f_bias for i, x in enumerate(xs))
super(Schwefel, self).__init__(
independents=independents, dependents=(f12,))
def setUp(self): # noqa
self.mock = mock.MagicMock()
self.mock.method.__name__ = 'foo'
self.memo_mock = base.memoize(self.mock.method)
def setUp(self): # noqa
self.mock = mock.MagicMock()
self.mock.method.__name__ = 'foo'
self.memo_mock = base.memoize(self.mock.method)
