def test():
### Parameters or externals
info_ret = {'order': 2}
locs = np.random.random((1000, 2))
inttypes = [int, np.int32, np.int64]
# mainmapper1 = generate_random_relations(25, store='sparse')
# mainmapper2 = generate_random_relations(100, store='sparse')
# mainmapper3 = generate_random_relations(5000, store='sparse')
griddisc1 = GridSpatialDisc((5, 5), xlim=(0, 1), ylim=(0, 1))
griddisc2 = GridSpatialDisc((10, 10), xlim=(0, 1), ylim=(0, 1))
griddisc3 = GridSpatialDisc((50, 100), xlim=(0, 1), ylim=(0, 1))
### Testing utities
## util_spatial_relations
f = lambda x, y: x + y
sp_elements = np.random.random(10)
general_spatial_relation(sp_elements[0], sp_elements[0], f)
general_spatial_relations(sp_elements, f, simmetry=False)
general_spatial_relations(sp_elements, f, simmetry=True)
## format_out_relations
mainmapper1 = randint_sparse_matrix(0.8, (25, 25))
format_out_relations(mainmapper1, 'sparse')
format_out_relations(mainmapper1, 'network')
format_out_relations(mainmapper1, 'sp_relations')
lista = format_out_relations(mainmapper1, 'list')
u_regs = mainmapper1.data
## Element metrics
element_i, element_j = 54, 2
pars1 = {'periodic': 60}
pars2 = {}
unidimensional_periodic(element_i, element_j, pars=pars1)
unidimensional_periodic(element_i, element_j, pars=pars2)
unidimensional_periodic(element_j, element_i, pars=pars1)
unidimensional_periodic(element_j, element_i, pars=pars2)
measure_difference(element_i, element_j, pars=pars1)
measure_difference(element_i, element_j, pars=pars2)
measure_difference(element_j, element_i, pars=pars1)
measure_difference(element_j, element_i, pars=pars2)
def ensure_output(neighs, dists, mainmapper):
# print dists
assert(all([len(e.shape) == 2 for e in dists]))
assert(all([len(e) == 0 for e in dists if np.prod(e.shape) == 0]))
if mainmapper._out == 'indices':
# print neighs
correcness = []
for nei in neighs:
if len(nei):
correcness.append(all([type(e) in inttypes for e in nei]))
else:
correcness.append(nei.dtype in inttypes)
assert(correcness)
###########################################################################
### Massive combinatorial testing
# Possible parameters
relations, pars_rel, _data =\
compute_ContiguityRegionDistances(griddisc1, store='sparse')
pos_relations = [relations, relations.A,
format_out_relations(relations, 'network')]
pos_distanceorweighs, pos_sym = [True, False], [True, False]
pos_inputstypes, pos_outputtypes = [[None, 'indices', 'elements_id']]*2
pos_input_type = [None, 'general', 'integer', 'array', 'array1', 'array2',
'list', 'list_int', 'list_array']
pos_inputs = [[0], 0, 0, np.array([0]), np.array([0]), np.array([0]),
[0], [0], [np.array([0])]]
pos_data_in, pos_data = [[]]*2
possibles = [pos_relations, pos_distanceorweighs, pos_sym, pos_outputtypes,
pos_inputstypes, pos_input_type]
# Combinations
for p in product(*possibles):
mainmapper1 = RegionDistances(relations=p[0], distanceorweighs=p[1],
symmetric=p[2], output=p[3], input_=p[4],
input_type=p[5])
mainmapper1[slice(0, 1)]
# Define input
if p[5] is None:
if p[4] != 'indices':
neighs, dists = mainmapper1[mainmapper1.data[0]]
ensure_output(neighs, dists, mainmapper1)
neighs, dists = mainmapper1[0]
ensure_output(neighs, dists, mainmapper1)
neighs, dists = mainmapper1[np.array([-1])]
ensure_output(neighs, dists, mainmapper1)
if p[4] != 'elements_id':
neighs, dists = mainmapper1[0]
ensure_output(neighs, dists, mainmapper1)
try:
boolean = False
mainmapper1[-1]
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
else:
if p[5] == 'list':
# Get item
neighs, dists = mainmapper1[[0]]
ensure_output(neighs, dists, mainmapper1)
neighs, dists = mainmapper1[[np.array([0])]]
ensure_output(neighs, dists, mainmapper1)
try:
boolean = False
mainmapper1[[None]]
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
idxs = pos_inputs[pos_input_type.index(p[5])]
neighs, dists = mainmapper1[idxs]
ensure_output(neighs, dists, mainmapper1)
# Functions
mainmapper1.set_inout(p[5], p[4], p[3])
mainmapper1.transform(lambda x: x)
mainmapper1.data
mainmapper1.data_input
mainmapper1.data_output
mainmapper1.shape
## Extreme cases
## Individual extreme cases
## Instantiation
pars_rel = {'symmetric': False}
relations = relations.A
data_in = list(np.arange(len(relations)))
wrond_data = np.random.random((100))
mainmapper3 = RegionDistances(relations=relations, _data=wrond_data,
data_in=data_in, **pars_rel)
data_in = np.arange(len(relations)).reshape((len(relations), 1))
mainmapper3 = RegionDistances(relations=relations, _data=wrond_data,
data_in=data_in, **pars_rel)
try:
boolean = False
wrond_data = np.random.random((100, 3, 4))
mainmapper3 = RegionDistances(relations=relations, _data=wrond_data,
data_in=data_in, **pars_rel)
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
try:
boolean = False
wrond_data = np.random.random((100, 3, 4))
sparse_rels = randint_sparse_matrix(0.8, (25, 25))
mainmapper3 = RegionDistances(relations=relations, _data=wrond_data,
data_in=data_in, **pars_rel)
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
try:
boolean = False
wrond_data = np.random.random((100, 3, 4))
mainmapper3 = RegionDistances(relations=relations, _data=wrond_data,
data_in=data_in, **pars_rel)
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
relations = np.random.random((20, 20))
try:
boolean = False
wrond_data = np.random.random((100, 3, 4))
mainmapper3 = RegionDistances(relations=relations, _data=wrond_data,
data_in=data_in, **pars_rel)
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
## Other cases
# Dummymap instantiation
regs0 = np.unique(np.random.randint(0, 1000, 200))
regs1 = regs0.reshape((len(regs0), 1))
regs2 = regs0.reshape((len(regs0), 1, 1))
pos_regs = [regs0, list(regs0), regs1]
possibles = [pos_regs, pos_input_type]
for p in product(*possibles):
dummymapper = DummyRegDistance(p[0], p[1])
# Get item
idxs = pos_inputs[pos_input_type.index(p[1])]
neighs, dists = dummymapper[idxs]
ensure_output(neighs, dists, dummymapper)
neighs, dists = dummymapper[slice(0, 1)]
ensure_output(neighs, dists, dummymapper)
## Functions
dummymapper.transform(lambda x: x)
dummymapper.data
dummymapper.data_input
dummymapper.data_output
dummymapper.shape
# Halting cases
try:
boolean = False
dummymapper = DummyRegDistance(regs2)
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
try:
boolean = False
dummymapper = DummyRegDistance(None)
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
try:
boolean = False
dummymapper[None]
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
try:
boolean = False
dummymapper[-1]
boolean = True
raise Exception("It has to halt here.")
except:
if boolean:
raise Exception("It has to halt here.")
###########################################################################
### Auxiliar parsing creation functions test
############################################
# Standarts
# * relations object
# * (main_relations_info, pars_rel)
# * (main_relations_info, pars_rel, _data)
# * (main_relations_info, pars_rel, _data, data_in)
#
## Main relations information
relations = np.random.random((100, 20))
_data = np.arange(20)
_data_input = np.arange(100)
relations_info = (relations, {})
relations_object = _relations_parsing_creation(relations_info)
assert(isinstance(relations_object, RegionDistances))
relations_info = (relations, {}, _data)
relations_object = _relations_parsing_creation(relations_info)
assert(isinstance(relations_object, RegionDistances))
relations_info = (relations, {}, _data, _data_input)
relations_object = _relations_parsing_creation(relations_info)
assert(isinstance(relations_object, RegionDistances))
relations_object = _relations_parsing_creation(relations_object)
assert(isinstance(relations_object, RegionDistances))
relations_object = _relations_parsing_creation(relations)
assert(isinstance(relations_object, RegionDistances))
###########################################################################
### Computers testing
#####################
## aux_regionmetrics
# Get regions activated
elements = griddisc1.get_regions_id()
get_regions4distances(griddisc1, elements=None, activated=None)
get_regions4distances(griddisc1, elements, activated=elements)
# Filter possible neighs
only_possible = np.unique(np.random.randint(0, 100, 50))
neighs = [np.unique(np.random.randint(0, 100, 6)) for i in range(4)]
dists = [np.random.random(len(neighs[i])) for i in range(4)]
filter_possible_neighs(only_possible, neighs, dists)
filter_possible_neighs(only_possible, neighs, None)
# TODO: Sync with other classes as sp_desc_models
lista = [[0, 1, 2, 3], [0, 2, 3, 5], [1, 1, 1, 1]]
u_regs = np.arange(25)
regions_id = np.arange(25)
elements_i = np.arange(25)
element_labels = np.arange(25)
discretizor = GridSpatialDisc((5, 5), xlim=(0, 1), ylim=(0, 1))
locs = np.random.random((100, 2))
retriever = KRetriever
info_ret = np.ones(100)*4
descriptormodel = DummyDescriptor()
sp_descriptor = discretizor, locs, retriever, info_ret, descriptormodel
sparse_from_listaregneighs(lista, u_regs, symmetric=True)
sparse_from_listaregneighs(lista, u_regs, symmetric=False)
ret_selfdists = KRetriever(locs, 4, ifdistance=True)
compute_selfdistances(ret_selfdists, np.arange(100), typeoutput='network',
symmetric=True)
compute_selfdistances(ret_selfdists, np.arange(100), typeoutput='sparse',
symmetric=True)
compute_selfdistances(ret_selfdists, np.arange(100), typeoutput='matrix',
symmetric=True)
# create_sp_descriptor_points_regs(sp_descriptor, regions_id, elements_i)
# create_sp_descriptor_regionlocs(sp_descriptor, regions_id, elements_i)
## Compute Avg distance
locs = np.random.random((100, 2))
sp_descriptor = (griddisc1, locs), (KRetriever, {'info_ret': 5}), None
relations, pars_rel, _data =\
compute_AvgDistanceRegions(sp_descriptor, store='network')
regdists = RegionDistances(relations=relations, _data=_data, **pars_rel)
relations, pars_rel, _data =\
compute_AvgDistanceRegions(sp_descriptor, store='matrix')
regdists = RegionDistances(relations=relations, _data=_data, **pars_rel)
relations, pars_rel, _data =\
compute_AvgDistanceRegions(sp_descriptor, store='sparse')
regdists = RegionDistances(relations=relations, _data=_data, **pars_rel)
# Region spatial relations
# For future (TODO)
### RegionDistances Computers
griddisc1 = GridSpatialDisc((5, 5), xlim=(0, 1), ylim=(0, 1))
## Compute Contiguity
relations, pars_rel, _data =\
compute_ContiguityRegionDistances(griddisc1, store='matrix')
relations, pars_rel, _data =\
compute_ContiguityRegionDistances(griddisc1, store='network')
relations, pars_rel, _data =\
compute_ContiguityRegionDistances(griddisc1, store='sparse')
mainmapper1 = RegionDistances(relations=relations, _data=None,
**pars_rel)
neighs, dists = mainmapper1.retrieve_neighs([0])
assert(len(neighs) == len(dists))
assert(len(neighs) == 1)
neighs, dists = mainmapper1.retrieve_neighs([0, 1])
assert(len(neighs) == len(dists))
assert(len(neighs) == 2)
## Compute CenterLocs
sp_descriptor = griddisc1, None, None
## TODO: pdist problem
relations, pars_rel, _data =\
compute_CenterLocsRegionDistances(sp_descriptor, store='network',
elements=None, symmetric=True,
activated=None)
relations, pars_rel, _data =\
compute_CenterLocsRegionDistances(sp_descriptor, store='matrix',
elements=None, symmetric=True,
activated=None)
relations, pars_rel, _data =\
compute_CenterLocsRegionDistances(sp_descriptor, store='sparse',
elements=None, symmetric=True,
activated=None)
sp_descriptor = griddisc1, (KRetriever, {'info_ret': 2}), None
### TODO: Descriptormodel
relations, pars_rel, _data =\
compute_CenterLocsRegionDistances(sp_descriptor, store='sparse',
elements=None, symmetric=True,
activated=None)
## Retriever tuple
## Retriever object
## Spdesc
## Compute PointsNeighsIntersection
## Aux_regionmetrics
#sparse_from_listaregneighs(lista, u_regs, symmetric)
###########################################################################
### Relative positioner testing
###############################
n_el, n_dim = 5, 2
elements_i = np.random.random((n_el, n_dim))
elements_neighs = []
for i in range(n_el):
aux_neigh = np.random.random((np.random.randint(1, 4), n_dim))
elements_neighs.append(aux_neigh)
rel_pos = BaseRelativePositioner(metric_distances)
rel_pos.compute(elements_i, elements_neighs)
rel_pos = BaseRelativePositioner(diff_vectors)
rel_pos.compute(elements_i, elements_neighs)
