def test_multidim_stack_multidim(self):
# Multidim stack of arrays shape (4, 6)
arrays = np.array([[ConstantArray((), i) for i in range(6)] for
i in range(4)])
msg = 'multidimensional stacks not yet supported'
with self.assertRaisesRegexp(ValueError, msg):
ArrayStack.multidim_array_stack(arrays, (3, 2, 4))
def test_multidim_stack_multidim(self):
# Multidim stack of arrays shape (4, 6)
arrays = np.array([[ConstantArray((), i) for i in range(6)]
for i in range(4)])
msg = 'multidimensional stacks not yet supported'
with self.assertRaisesRegexp(ValueError, msg):
ArrayStack.multidim_array_stack(arrays, (3, 2, 4))
def test_order_incorrect_order(self):
# Specifying an unknown order.
array1 = fake_array(0)
array2 = fake_array(0)
with self.assertRaisesRegexp(TypeError, 'order not understood'):
ArrayStack.multidim_array_stack([array1, array2], (1, 2),
order='random')
def test_order_incorrect_order(self):
# Specifying an unknown order.
array1 = fake_array(0)
array2 = fake_array(0)
with self.assertRaisesRegexp(TypeError, 'order not understood'):
ArrayStack.multidim_array_stack([array1, array2], (1, 2),
order='random')
def test_stack_order_fortran_t1(self):
# Fortran index ordering
# 1D stack of arrays shape (6,)
res = ArrayStack.multidim_array_stack(self.arrays, (3, 2), order='F')
arr = np.array([i for i in range(6)])
target = np.reshape(arr, (3, 2), order='F')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_default(self):
# 1D stack of arrays shape (6,)
# Ensure that the default index ordering corresponds to C.
res = ArrayStack.multidim_array_stack(self.arrays, (3, 2))
arr = np.array([i for i in range(6)])
target = np.reshape(arr, (3, 2), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_c_list(self):
# 1D stack of arrays length 6
res = ArrayStack.multidim_array_stack(self.arrays.tolist(), (3, 2),
order='C')
arr = np.array([i for i in range(6)])
target = np.reshape(arr, (3, 2), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_c_numpy_array_t2(self):
# 1D stack of arrays shape (6,)
# alternate shape
res = ArrayStack.multidim_array_stack(self.arrays, (2, 3), order='C')
arr = np.array([i for i in range(6)])
target = np.reshape(arr, (2, 3), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_c_list(self):
# 1D stack of arrays length 6
res = ArrayStack.multidim_array_stack(self.arrays.tolist(), (3, 2),
order='C')
arr = np.array([i for i in range(6)])
target = np.reshape(arr, (3, 2), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_c_numpy_array_t2(self):
# 1D stack of arrays shape (6,)
# alternate shape
res = ArrayStack.multidim_array_stack(self.arrays, (2, 3), order='C')
arr = np.array([i for i in range(6)])
target = np.reshape(arr, (2, 3), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_default(self):
# 1D stack of arrays shape (6,)
# Ensure that the default index ordering corresponds to C.
res = ArrayStack.multidim_array_stack(self.arrays, (3, 2))
arr = np.array([i for i in range(6)])
target = np.reshape(arr, (3, 2), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_fortran_t1(self):
# Fortran index ordering
# 1D stack of arrays shape (6,)
res = ArrayStack.multidim_array_stack(self.arrays, (3, 2), order='F')
arr = np.array([i for i in range(6)])
target = np.reshape(arr, (3, 2), order='F')
self.assertTrue(np.array_equal(res.ndarray(), target))
def data(self):
if not self._structure_calculated:
self._calculate_structure()
if self._data_cache is None:
data_arrays = [f._data for f in self.fields]
self._data_cache = ArrayStack.multidim_array_stack(data_arrays, self.vector_dims_shape)
return self._data_cache
def test_stack_order_c_multidim(self):
# 1D stack of 6 arrays each (4, 5)
arrays = [NumpyArrayAdapter(np.arange(20).reshape(4, 5)) for
i in range(6)]
res = ArrayStack.multidim_array_stack(arrays, (2, 3), order='C')
arr = np.arange(20).reshape(4, 5) * np.ones((6, 4, 5))
target = np.reshape(arr, (2, 3, 4, 5), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_c_multidim(self):
# 1D stack of 6 arrays each (4, 5)
arrays = [NumpyArrayAdapter(np.arange(20).reshape(4, 5)) for
i in range(6)]
res = ArrayStack.multidim_array_stack(arrays, (2, 3), order='C')
arr = np.arange(20).reshape(4, 5) * np.ones((6, 4, 5))
target = np.reshape(arr, (2, 3, 4, 5), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_stack_order_fortran_t2(self):
# Fortran index ordering
# 1D stack of arrays shape (6,)
# alternate shape
res = ArrayStack.multidim_array_stack(self.arrays, (2, 3), order='F')
arr = np.arange(6)
target = np.reshape(arr, (2, 3), order='F')
self.assertTrue(np.array_equal(res.ndarray(), target))
def data(self):
if not self._structure_calculated:
self._calculate_structure()
if self._data_cache is None:
data_arrays = [f._data for f in self.fields]
self._data_cache = \
ArrayStack.multidim_array_stack(data_arrays,
self.vector_dims_shape)
return self._data_cache
def test_stack_order_c_numpy_array_t1(self):
# 1D stack of arrays shape (6,)
res = ArrayStack.multidim_array_stack(self.arrays, (3, 2), order='C')
arr = np.arange(6)
target = np.reshape(arr, (3, 2), order='C')
self.assertTrue(np.array_equal(res.ndarray(), target))
def test_incompatible_shape(self):
# 1D stack of arrays shape (6,)
# Specifying a stack shape that is not feasible.
msg = 'total size of new array must be unchanged'
with self.assertRaisesRegexp(ValueError, msg):
ArrayStack.multidim_array_stack(self.arrays, (3, 1), order='C')
def test_incompatible_shape(self):
# 1D stack of arrays shape (6,)
# Specifying a stack shape that is not feasible.
msg = 'total size of new array must be unchanged'
with self.assertRaisesRegexp(ValueError, msg):
ArrayStack.multidim_array_stack(self.arrays, (3, 1), order='C')
