Merge pull request #201 from lcgraham/v2_master

fixed type checking in postProcess and basicSampling, see issue #200

bet/calculateP/calculateP.py

@@ -29,8 +29,8 @@ def emulate_iid_lebesgue(domain, num_l_emulate, globalize=False):
     :returns: a set of samples for emulation
 
     """
-    num_l_emulate = (num_l_emulate/comm.size) + \
-            (comm.rank < num_l_emulate%comm.size)
+    num_l_emulate = int((num_l_emulate/comm.size) + \
+            (comm.rank < num_l_emulate%comm.size))
     lam_width = domain[:, 1] - domain[:, 0]
     lambda_emulate = lam_width*np.random.random((num_l_emulate,
         domain.shape[0]))+domain[:, 0]
@@ -50,10 +50,6 @@ def prob_emulated(discretization, globalize=True):
     ``num_l_emulate`` iid samples :math:`(\lambda_{emulate})`.
     This is added to the emulated input sample set object.
 
-    .. todo::
-        
-        @smattis the way this is written globalize does nothing
-
     :param discretization: An object containing the discretization information.
     :type class:`bet.sample.discretization`
     :param bool globalize: Makes local variables global.
@@ -86,6 +82,8 @@ def prob_emulated(discretization, globalize=True):
                         _probabilities[i]/Itemp_sum
 
     discretization._emulated_input_sample_set._probabilities_local = P
+    if globalize:
+        discretization._emulated_input_sample_set.local_to_global()
     pass
 
 def prob(discretization): 
@@ -159,7 +157,10 @@ def prob_mc(discretization):
     cvol = np.copy(vol)
     comm.Allreduce([vol, MPI.DOUBLE], [cvol, MPI.DOUBLE], op=MPI.SUM)
     vol = cvol
-    vol = vol/float(discretization._emulated_input_sample_set._values.shape[0])
+    num_l_emulate = discretization._emulated_input_sample_set.\
+            _values_local.shape[0]
+    num_l_emulate = comm.allreduce(num_l_emulate, op=MPI.SUM)
+    vol = vol/float(num_l_emulate)
     discretization._input_sample_set._volumes = vol
     discretization._input_sample_set.global_to_local()
 

bet/postProcess/plotDomains.py

@@ -52,7 +52,7 @@ def scatter_2D(sample_obj, sample_nos=None, color=None, p_ref=None, save=True,
     the other markers.
 
     :param sample_obj: contains samples to create scatter plot
-    :type sample_obj: :class:`~bet.sample.sample_set`
+    :type sample_obj: :class:`~bet.sample.sample_set_base`
     :param list sample_nos: indicies of the samples to plot
     :param color: values to color the samples by
     :type color: :class:`numpy.ndarray`
@@ -65,7 +65,7 @@ def scatter_2D(sample_obj, sample_nos=None, color=None, p_ref=None, save=True,
     :param string filename: filename to save the figure as
 
     """
-    if not isinstance(sample_obj, sample.sample_set):
+    if not isinstance(sample_obj, sample.sample_set_base):
         raise bad_object("Improper sample object")
     # check dimension of data to plot
     if sample_obj.get_dim() != 2:
@@ -119,7 +119,7 @@ def scatter_3D(sample_obj, sample_nos=None, color=None, p_ref=None, save=True,
     the other markers.
 
     :param sample_obj: Object containing the samples to plot
-    :type sample_obj: :class:`~bet.sample.sample_set`
+    :type sample_obj: :class:`~bet.sample.sample_set_base`
     :param list sample_nos: indicies of the samples to plot
     :param color: values to color the samples by
     :type color: :class:`numpy.ndarray`
@@ -133,7 +133,7 @@ def scatter_3D(sample_obj, sample_nos=None, color=None, p_ref=None, save=True,
     :param string filename: filename to save the figure as
 
     """
-    if not isinstance(sample_obj, sample.sample_set):
+    if not isinstance(sample_obj, sample.sample_set_base):
         raise bad_object("Improper sample object")
     # check dimension of data to plot
     if sample_obj.get_dim() != 3:
@@ -191,7 +191,7 @@ def show_param(sample_disc, rho_D=None, p_ref=None, sample_nos=None,
 
     :param sample_disc: Object containing the samples to plot
     :type sample_disc: :class:`~bet.sample.discretization` 
-        or :class:`~bet.sample.sample_set`
+        or :class:`~bet.sample.sample_set_base`
     :param list sample_nos: sample numbers to plot
     :param rho_D: probability density function on D
     :type rho_D: callable function that takes a :class:`np.array` and returns a
@@ -217,7 +217,7 @@ def show_param(sample_disc, rho_D=None, p_ref=None, sample_nos=None,
     else:
         if isinstance(sample_disc, sample.discretization):
             sample_obj = sample_disc._input_sample_set
-        elif isinstance(sample_disc, sample.sample_set):
+        elif isinstance(sample_disc, sample.sample_set_base):
             sample_obj = sample_disc
         else:
             raise bad_object("Improper sample object")
@@ -271,7 +271,7 @@ def show_data(sample_obj, rho_D=None, Q_ref=None, sample_nos=None,
     the other markers.
 
     :param sample_obj: Object containing the samples to plot
-    :type sample_obj: :class:`~bet.sample.sample_set`
+    :type sample_obj: :class:`~bet.sample.sample_set_base`
     :param list sample_nos: sample numbers to plot
     :param rho_D: probability density on D
     :type rho_D: callable function that takes a :class:`np.array` and returns a
@@ -363,7 +363,7 @@ def show_data_domain_multi(sample_disc, Q_ref=None, Q_nums=None,
 
     :param sample_disc: Object containing the samples to plot
     :type sample_disc: :class:`~bet.sample.discretization` or 
-        :class:`~bet.sample.sample_set`
+        :class:`~bet.sample.sample_set_base`
     :param Q_ref: reference data value
     :type Q_ref: :class:`numpy.ndarray` of shape (M, mdim)
     :param list Q_nums: dimensions of the QoI to plot
@@ -472,7 +472,7 @@ def show_data_domain_2D(sample_disc, Q_ref=None, ref_markers=None,
 
     :param sample_disc: Object containing the samples to plot
     :type sample_disc: :class:`~bet.sample.discretization` 
-        or :class:`~bet.sample.sample_set`
+        or :class:`~bet.sample.sample_set_base`
     :param Q_ref: reference data value
     :type Q_ref: :class:`numpy.ndarray` of shape (M, 2)
     :param list ref_markers: list of marker types for :math:`Q_{ref}`
@@ -542,7 +542,7 @@ def scatter_param_multi(sample_obj, img_folder='figs/', showdim='all',
     Creates two-dimensional projections of scatter plots of samples.
 
     :param sample_obj: Object containing the samples to plot
-    :type sample_obj: :class:`~bet.sample.sample_set`
+    :type sample_obj: :class:`~bet.sample.sample_set_base`
     :param bool save: flag whether or not to save the figure
     :param bool interactive: flag whether or not to show the figure
     :param string img_folder: folder to save the plots to
@@ -551,7 +551,7 @@ def scatter_param_multi(sample_obj, img_folder='figs/', showdim='all',
     :type showdim: int or string
 
     """
-    if not isinstance(sample_obj, sample.sample_set):
+    if not isinstance(sample_obj, sample.sample_set_base):
         raise bad_object("Improper sample object")
 
     # If no specific coordinate number of choice is given set to be the first
@@ -622,7 +622,7 @@ def scatter2D_multi(sample_obj, color=None, p_ref=None, img_folder='figs/',
     markers.
 
     :param sample_obj: Object containing the samples to plot
-    :type sample_obj: :class:`~bet.sample.sample_set`
+    :type sample_obj: :class:`~bet.sample.sample_set_base`
     :param color: values to color the ``samples`` by
     :type color: :class:`numpy.ndarray`
     :param string filename: filename to save the figure as
@@ -635,7 +635,7 @@ def scatter2D_multi(sample_obj, color=None, p_ref=None, img_folder='figs/',
     :type showdim: int or string
 
     """
-    if not isinstance(sample_obj, sample.sample_set):
+    if not isinstance(sample_obj, sample.sample_set_base):
         raise bad_object("Improper sample object")
     # If no specific coordinate number of choice is given set to be the first
     # coordinate direction.

bet/postProcess/plotP.py

@@ -23,16 +23,26 @@ class bad_object(Exception):
     Exception for when the wrong type of object is used.
     """
 
+class missing_attribute(Exception):
+    """
+    Exception for missing attribute.
+    """
+
 def calculate_1D_marginal_probs(sample_set, nbins=20):
 
     r"""
     This calculates every single marginal of the probability measure
     described by the probabilities within the sample_set object.
     If the sample_set object is a discretization object, we assume
-    that the probabilities to be plotted are from the input space.
+    that the probabilities to be plotted are from the input space on the
+    emulated samples
+    (``discretization._emulated_input_sample_set._probabilties_local``).
+
+    This assumes that the user has already run
+    :meth:`~bet.calculateP.calculateP.prob_emulated`.
 
     :param sample_set: Object containing samples and probabilities
-    :type sample_set: :class:`~bet.sample.sample_set` or 
+    :type sample_set: :class:`~bet.sample.sample_set_base` or 
         :class:`~bet.sample.discretization`
     :param nbins: Number of bins in each direction.
     :type nbins: :int or :class:`~numpy.ndarray` of shape (ndim,)
@@ -41,12 +51,21 @@ def calculate_1D_marginal_probs(sample_set, nbins=20):
 
     """
     if isinstance(sample_set, sample.discretization):
-        sample_obj = sample_set._input_sample_set
-    elif isinstance(sample_set, sample.sample_set):
+        sample_obj = sample_set._emulated_input_sample_set
+        if sample_obj is None:
+            raise missing_attribute("Missing emulated_input_sample_set")
+    elif isinstance(sample_set, sample.sample_set_base):
         sample_obj = sample_set
     else:
         raise bad_object("Improper sample object")
 
+    # Check for local probabilities
+    if sample_obj._probabilities_local is None:
+        if sample_obj.probabilities is None:
+            raise missing_attribute("Missing probabilities")
+        else:
+            sample_obj.global_to_local()
+
     # Make list of bins if only an integer is given
     if isinstance(nbins, int):
         nbins = nbins*np.ones(sample_obj.get_dim(), dtype=np.int)
@@ -61,8 +80,8 @@ def calculate_1D_marginal_probs(sample_set, nbins=20):
     # Calculate marginals
     marginals = {}
     for i in range(sample_obj.get_dim()):
-        [marg, _] = np.histogram(sample_obj.get_values()[:, i], bins=bins[i],
-                                 weights=sample_obj.get_probabilities())
+        [marg, _] = np.histogram(sample_obj.get_values_local()[:, i], bins=bins[i],
+                                 weights=sample_obj.get_probabilities_local())
         marg_temp = np.copy(marg)
         comm.Allreduce([marg, MPI.DOUBLE], [marg_temp, MPI.DOUBLE], op=MPI.SUM)
         marginals[i] = marg_temp
@@ -75,10 +94,16 @@ def calculate_2D_marginal_probs(sample_set, nbins=20):
     This calculates every pair of marginals (or joint in 2d case) of
     input probability measure defined on a rectangular grid.
     If the sample_set object is a discretization object, we assume
-    that the probabilities to be plotted are from the input space.
+    that the probabilities to be plotted are from the input space on the
+    emulated samples
+    (``discretization._emulated_input_sample_set._probabilties_local``).
+
+    This assumes that the user has already run
+    :meth:`~bet.calculateP.calculateP.prob_emulated`.
+
 
     :param sample_set: Object containing samples and probabilities
-    :type sample_set: :class:`~bet.sample.sample_set` 
+    :type sample_set: :class:`~bet.sample.sample_set_base` 
         or :class:`~bet.sample.discretization`
     :param nbins: Number of bins in each direction.
     :type nbins: :int or :class:`~numpy.ndarray` of shape (ndim,)
@@ -87,12 +112,21 @@ def calculate_2D_marginal_probs(sample_set, nbins=20):
 
     """
     if isinstance(sample_set, sample.discretization):
-        sample_obj = sample_set._input_sample_set
-    elif isinstance(sample_set, sample.sample_set):
+        sample_obj = sample_set._emulated_input_sample_set
+        if sample_obj is None:
+            raise missing_attribute("Missing emulated_input_sample_set")
+    elif isinstance(sample_set, sample.sample_set_base):
         sample_obj = sample_set
     else:
         raise bad_object("Improper sample object")
 
+    # Check for local probabilities
+    if sample_obj._probabilities_local is None:
+        if sample_obj.probabilities is None:
+            raise missing_attribute("Missing probabilities")
+        else:
+            sample_obj.global_to_local()
+
     if sample_obj.get_dim() < 2:
         raise dim_not_matching("Incompatible dimensions of sample set"
                                " for plotting")
@@ -112,9 +146,9 @@ def calculate_2D_marginal_probs(sample_set, nbins=20):
     marginals = {}
     for i in range(sample_obj.get_dim()):
         for j in range(i+1, sample_obj.get_dim()):
-            (marg, _) = np.histogramdd(sample_obj.get_values()[:, [i, j]],
+            (marg, _) = np.histogramdd(sample_obj.get_values_local()[:, [i, j]],
                     bins=[bins[i], bins[j]],
-                    weights=sample_obj.get_probabilities())
+                    weights=sample_obj.get_probabilities_local())
             marg = np.ascontiguousarray(marg)
             marg_temp = np.copy(marg)
             comm.Allreduce([marg, MPI.DOUBLE], [marg_temp, MPI.DOUBLE],
@@ -139,7 +173,7 @@ def plot_1D_marginal_probs(marginals, bins, sample_set,
         calculating marginals
     :type bins: :class:`~numpy.ndarray` of shape (nbins+1,)
     :param sample_set: Object containing samples and probabilities
-    :type sample_set: :class:`~bet.sample.sample_set` 
+    :type sample_set: :class:`~bet.sample.sample_set_base` 
         or :class:`~bet.sample.discretization`
     :param filename: Prefix for output files.
     :type filename: str
@@ -153,7 +187,7 @@ def plot_1D_marginal_probs(marginals, bins, sample_set,
     """
     if isinstance(sample_set, sample.discretization):
         sample_obj = sample_set._input_sample_set
-    elif isinstance(sample_set, sample.sample_set):
+    elif isinstance(sample_set, sample.sample_set_base):
         sample_obj = sample_set
     else:
         raise bad_object("Improper sample object")
@@ -202,7 +236,7 @@ def plot_2D_marginal_probs(marginals, bins, sample_set,
     :param bins: Endpoints of bins used in calculating marginals
     :type bins: :class:`~numpy.ndarray` of shape (nbins+1,2)
     :param sample_set: Object containing samples and probabilities
-    :type sample_set: :class:`~bet.sample.sample_set` 
+    :type sample_set: :class:`~bet.sample.sample_set_base` 
         or :class:`~bet.sample.discretization`
     :param filename: Prefix for output files.
     :type filename: str
@@ -216,7 +250,7 @@ def plot_2D_marginal_probs(marginals, bins, sample_set,
     """
     if isinstance(sample_set, sample.discretization):
         sample_obj = sample_set._input_sample_set
-    elif isinstance(sample_set, sample.sample_set):
+    elif isinstance(sample_set, sample.sample_set_base):
         sample_obj = sample_set
     else:
         raise bad_object("Improper sample object")

bet/postProcess/postTools.py

@@ -28,7 +28,7 @@ def sort_by_rho(sample_set):
     are also sorted.
 
     :param sample_set: Object containing samples and probabilities
-    :type sample_set: :class:`~bet.sample.sample_set` or
+    :type sample_set: :class:`~bet.sample.sample_set_base` or
         :class:`~bet.sample.discretization` 
     :param indices: sorting indices
     :type indices: :class:`numpy.ndarray` of shape (num_samples,)
@@ -45,7 +45,7 @@ def sort_by_rho(sample_set):
         P_samples = sample_set._input_sample_set.get_probabilities()
         lam_vol = sample_set._input_sample_set.get_volumes()
         data = sample_set._output_sample_set.get_values()
-    elif isinstance(sample_set, sample.sample_set):
+    elif isinstance(sample_set, sample.sample_set_base):
         samples = sample_set.get_values()
         P_samples = sample_set.get_probabilities()
         lam_vol = sample_set.get_volumes()
@@ -95,7 +95,7 @@ def sample_prob(percentile, sample_set, sort=True, descending=False):
     :param percentile: ratio of highest probability samples to select
     :type percentile: float
     :param sample_set: Object containing samples and probabilities
-    :type sample_set: :class:`~bet.sample.sample_set` or 
+    :type sample_set: :class:`~bet.sample.sample_set_base` or 
         :class:`~bet.sample.discretization`
     :type indices: :class:`numpy.ndarray` of shape (num_samples,)
     :param indices: sorting indices
@@ -114,7 +114,7 @@ def sample_prob(percentile, sample_set, sort=True, descending=False):
         P_samples = sample_set._input_sample_set.get_probabilities()
         lam_vol = sample_set._input_sample_set.get_volumes()
         data = sample_set._output_sample_set.get_values()
-    elif isinstance(sample_set, sample.sample_set):
+    elif isinstance(sample_set, sample.sample_set_base):
         samples = sample_set.get_values()
         P_samples = sample_set.get_probabilities()
         lam_vol = sample_set.get_volumes()
@@ -129,7 +129,7 @@ def sample_prob(percentile, sample_set, sort=True, descending=False):
             P_samples = sample_set._input_sample_set.get_probabilities()
             lam_vol = sample_set._input_sample_set.get_volumes()
             data = sample_set._output_sample_set.get_values()
-        elif isinstance(sample_set, sample.sample_set):
+        elif isinstance(sample_set, sample.sample_set_base):
             samples = sample_set.get_values()
             P_samples = sample_set.get_probabilities()
             lam_vol = sample_set.get_volumes()
@@ -182,7 +182,7 @@ def sample_highest_prob(top_percentile, sample_set, sort=True):
     :param top_percentile: ratio of highest probability samples to select
     :type top_percentile: float
     :param sample_set: Object containing samples and probabilities
-    :type sample_set: :class:`~bet.sample.sample_set` 
+    :type sample_set: :class:`~bet.sample.sample_set_base` 
         or :class:`~bet.sample.discretization`
     :type indices: :class:`numpy.ndarray` of shape (num_samples,)
     :param indices: sorting indices
@@ -208,7 +208,7 @@ def sample_lowest_prob(bottom_percentile, sample_set, sort=True):
     :param top_percentile: ratio of highest probability samples to select
     :type top_percentile: float
     :param sample_set: Object containing samples and probabilities
-    :type sample_set: :class:`~bet.sample.sample_set` 
+    :type sample_set: :class:`~bet.sample.sample_set_base` 
         or :class:`~bet.sample.discretization`
     :type indices: :class:`numpy.ndarray` of shape (num_samples,)
     :param indices: sorting indices of unsorted ``P_samples``

bet/sample.py

@@ -610,6 +610,13 @@ def estimate_volume(self, n_mc_points=int(1E4)):
         Calculate the volume faction of cells approximately using Monte
         Carlo integration. 
 
+        .. todo::
+
+           This currently presumes a uniform Lesbegue measure on the
+           ``domain``. Currently the way this is written
+           ``emulated_input_sample_set`` is NOT used to calculate the volume.
+           This should at least be an option. 
+
         :param int n_mc_points: If estimate is True, number of MC points to use
         """
         num = self.check_num()