def setUp(self):
self.set_new = samp.rectangle_sample_set(dim=2)
self.set_new.set_domain(np.array([[0.0, 1.0], [0.0, 1.0]]))
self.set_new.setup(maxes=[[0.75, 0.75]], mins=[[0.25, 0.25]])
self.set_old = samp.cartesian_sample_set(dim=2)
self.set_old.set_domain(np.array([[0.0, 1.0], [0.0, 1.0]]))
self.set_old.setup([np.linspace(0, 1, 21), np.linspace(0, 1, 21)])
num_old = self.set_old.check_num()
probs = np.zeros((num_old,))
probs[0:-1] = 1.0/float(num_old-1)
self.set_old.set_probabilities(probs)
self.set_em = samp.cartesian_sample_set(dim=2)
self.set_em.set_domain(np.array([[0.0, 1.0], [0.0, 1.0]]))
self.set_em.setup([np.linspace(0, 1, 101), np.linspace(0, 1, 101)])
def setUp(self):
self.set_new = samp.rectangle_sample_set(dim=2)
self.set_new.set_domain(np.array([[0.0, 1.0], [0.0, 1.0]]))
self.set_new.setup(maxes=[[0.75, 0.75]], mins=[[0.25, 0.25]])
self.set_old = samp.cartesian_sample_set(dim=2)
self.set_old.set_domain(np.array([[0.0, 1.0], [0.0, 1.0]]))
self.set_old.setup([np.linspace(0, 1, 21), np.linspace(0, 1, 21)])
num_old = self.set_old.check_num()
probs = np.zeros((num_old, ))
probs[0:-1] = 1.0 / float(num_old - 1)
self.set_old.set_probabilities(probs)
self.set_em = samp.cartesian_sample_set(dim=2)
self.set_em.set_domain(np.array([[0.0, 1.0], [0.0, 1.0]]))
self.set_em.setup([np.linspace(0, 1, 101), np.linspace(0, 1, 101)])
def regular_partition_uniform_distribution_rectangle_size(data_set, Q_ref=None,
rect_size=None,
cells_per_dimension=1):
r"""
Creates a simple function approximation of :math:`\rho_{\mathcal{D},M}`
where :math:`\rho_{\mathcal{D},M}` is a uniform probability density
centered at ``Q_ref`` (or the ``reference_value``
of a sample set. If ``Q_ref`` is not given the reference value is used)
with ``rect_size`` of the width of a hyperrectangle.
Since rho_D is a uniform distribution on a hyperrectanlge we can represent
it exactly.
:param rect_size: The size used to determine the width of the uniform
distribution
:type rect_size: double or list
:param data_set: Sample set that the probability measure is defined for.
:type data_set: :class:`~bet.sample.discretization` or
:class:`~bet.sample.sample_set` or :class:`~numpy.ndarray`
:param Q_ref: :math:`Q(\lambda_{reference})`
:type Q_ref: :class:`~numpy.ndarray` of size (mdim,)
:param list cells_per_dimension: number of cells per dimension.
:rtype: :class:`~bet.sample.rectangle_sample_set`
:returns: sample_set object defining simple function approximation
"""
(num, dim, values, Q_ref) = check_inputs(data_set, Q_ref)
data = values
if rect_size is None:
raise wrong_argument_type("Missing rectangle size.")
elif not isinstance(rect_size, collections.Iterable):
rect_size = rect_size * np.ones((dim,))
if np.any(np.less_equal(rect_size, 0)):
msg = 'rect_size must be greater than 0'
raise wrong_argument_type(msg)
if not isinstance(cells_per_dimension, collections.Iterable):
cells_per_dimension = np.ones((dim,)) * cells_per_dimension
maxes = [Q_ref + 0.5*np.array(rect_size)]
mins = [Q_ref - 0.5*np.array(rect_size)]
xi = []
for i in xrange(dim):
xi.append(np.linspace(mins[0][i], maxes[0][i],
cells_per_dimension[i] + 1))
s_set = samp.cartesian_sample_set(dim)
s_set.setup(xi)
domain = np.zeros((dim, 2))
domain[:, 0] = mins[0]
domain[:, 1] = maxes[0]
s_set.set_domain(domain)
s_set.exact_volume_lebesgue()
vol = np.sum(s_set._volumes[0:-1])
prob = np.zeros(s_set._volumes.shape)
prob[0:-1] = s_set._volumes[0:-1]/vol
s_set.set_probabilities(prob)
if isinstance(data_set, samp.discretization):
data_set._output_probability_set = s_set
return s_set