def __init__(self, copula_spec):
"""Init class."""
version = distribute_copula_spec(copula_spec, 'version')
self.attr = dict()
self.attr['version'] = version
# Assign correct monitoring and attribute checks function.
if version in ['scaled_archimedean']:
self._logging = partial(log_scaled_archimedean, self)
self._check_attributes = partial(check_attributes_scaled_archimedean, self)
elif version in ['nonstationary']:
self._logging = partial(log_nonstationary, self)
self._check_attributes = partial(check_attributes_nonstationary, self)
elif version in ['warmglow']:
self._logging = partial(log_warmglow, self)
self._check_attributes = partial(check_attributes_warmglow, self)
else:
raise NotImplementedError
# Handle other attributes
if version in ['scaled_archimedean']:
args = ['r', 'bounds', 'delta', 'u', 'generating_function', 'a', 'b', 'marginals']
r, bounds, delta, u, generating_function, a, b, marginals = \
distribute_copula_spec(copula_spec, *args)
marginal_utility = []
for i, marginal in enumerate(marginals):
if marginal == 'power':
marginal_utility += [PowerCls(r[i], a, b, bounds[i])]
elif marginal == 'exponential':
marginal_utility += [ExponentialCls(r[i], a, b, bounds[i])]
self.attr['x_uniattribute_utility'] = marginal_utility[0]
self.attr['y_uniattribute_utility'] = marginal_utility[1]
self.attr['bounds'] = bounds
self.attr['delta'] = delta
self.attr['u_1'] = u[0]
self.attr['u_2'] = u[1]
copula = ScaledArchimedeanCls(generating_function, u[0], u[1], delta)
elif version in ['nonstationary', 'warmglow']:
args = ['alpha', 'beta', 'gamma', 'discount_factors', 'y_scale',
'unrestricted_weights', 'discounting']
alpha, beta, gamma, discount_factors, y_scale, unrestricted_weights, discounting = \
distribute_copula_spec(copula_spec, *args)
self.attr['unrestricted_weights'] = unrestricted_weights
self.attr['discount_factors'] = discount_factors
self.attr['discounting'] = discounting
self.attr['y_scale'] = y_scale
self.attr['alpha'] = alpha
self.attr['gamma'] = gamma
self.attr['beta'] = beta
if version in ['nonstationary']:
copula = NonstationaryUtilCls(
alpha, beta, gamma, discount_factors, y_scale, unrestricted_weights,
discounting)
elif version in ['warmglow']:
# Handle version-specifc parameters
args = ['warmglow_type']
warmglow_model = distribute_copula_spec(copula_spec, *args)
copula = WarmglowUtilCls(
alpha, beta, gamma, discount_factors, y_scale, unrestricted_weights,
discounting, warmglow_model)
else:
raise NotImplementedError
else:
raise NotImplementedError
self.attr['copula'] = copula
self._check_attributes()
self._logging()
np.random.seed(123)
for _ in range(10):
x, y, is_normalized, copula_spec = generate_random_request(
{'version': 'nonstationary'})
# copula_spec['marginals'] = ['exponential', 'exponential']
# copula_spec['r'] = [-5, -5]
# copula_spec['bounds'] = [500, 500]
# print(is_normalized, 'out')
copula = UtilityCopulaCls(copula_spec)
util = copula.evaluate(x, y, is_normalized)
alpha, beta, gamma, y_scale, version = \
distribute_copula_spec(copula_spec, 'alpha', 'beta', 'gamma', 'y_scale', 'version')
print('version: {}'.format(version))
print('alpha: {0:.2f}, beta: {1:.2f}, gamma: {2:.2f}, y_scale: {3:.2f}'.
format(alpha, beta, gamma, y_scale))
print('x: {0:.2f}, y: {1:.2f}, utility: {2:.2f}.'.format(x, y, util))
# Don't expect monotonicity here
print(copula.evaluate(x, y))
print(copula.evaluate(x, y, t=0))
print(copula.evaluate(x, y, t=1))
print(copula.evaluate(x, y, t=3))
print(copula.evaluate(x, y, t=6))
print(copula.evaluate(x, y, t=12))
print(copula.evaluate(x, y, t=24))