def test_new_identity(self):
# Valid dimensions and types
c = Covariance(2, ctypes.c_double)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(3, ctypes.c_double)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(2, ctypes.c_float)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(3, ctypes.c_float)
print('constructed matrix:\n', c.to_matrix())
def test_new_identity(self):
# Valid dimensions and types
c = Covariance(2, ctypes.c_double)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(3, ctypes.c_double)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(2, ctypes.c_float)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(3, ctypes.c_float)
print 'constructed matrix:\n', c.to_matrix()
def covariance(self, c):
"""
Set new landmark covariance
:type c: vital.types.Covariance
"""
expected_covar_type = Covariance.c_ptr_type(3, self._datatype)
# Convert covariance to correct type if necessary
if c.C_TYPE_PTR != expected_covar_type:
c = Covariance(3, self._datatype, c.to_matrix())
self._call_cfunc('vital_landmark_{}_set_covar'.format(self._tchar),
[self.C_TYPE_PTR, expected_covar_type], [self, c])
def test_new_matrix(self):
m = EigenArray(2, 2, dtype=numpy.double)
m[:] = 1.
c = Covariance(2, ctypes.c_double, m)
m_out = c.to_matrix()
print 'input matrix:\n', m
print 'output matrix:\n', m_out
numpy.testing.assert_array_equal(m_out, m)
# Type casting should be handled
m = EigenArray(2, 2, dtype=numpy.float32)
m[:] = 1.
c = Covariance(2, ctypes.c_double, m)
m_out = c.to_matrix()
print 'input matrix:\n', m
print 'output matrix:\n', m_out
numpy.testing.assert_array_equal(m_out, m)
# Any other numpy array of the correct shape should be acceptable
m = numpy.ndarray((2, 2))
m[:] = 3.
c = Covariance(2, ctypes.c_float, init_scalar_or_matrix=m)
m_out = c.to_matrix()
print 'input matrix:\n', m
print 'output matrix:\n', m_out
numpy.testing.assert_array_equal(m_out, m)
# Diagonally congruent values should be averages when initializing with
# matrix
m = numpy.eye(3, dtype=numpy.double)
m[0,2] = 2.
m_expected = m.copy()
m_expected[0,2] = 1.
m_expected[2,0] = 1.
c = Covariance(3, init_scalar_or_matrix=m)
m_out = c.to_matrix()
print 'input matrix:\n', m
print 'output matrix:\n', m_out
numpy.testing.assert_array_equal(m_out, m_expected)
def test_new_matrix(self):
m = EigenArray(2, 2, dtype=numpy.double)
m[:] = 1.
c = Covariance(2, ctypes.c_double, m)
m_out = c.to_matrix()
print('input matrix:\n', m)
print('output matrix:\n', m_out)
numpy.testing.assert_array_equal(m_out, m)
# Type casting should be handled
m = EigenArray(2, 2, dtype=numpy.float32)
m[:] = 1.
c = Covariance(2, ctypes.c_double, m)
m_out = c.to_matrix()
print('input matrix:\n', m)
print('output matrix:\n', m_out)
numpy.testing.assert_array_equal(m_out, m)
# Any other numpy array of the correct shape should be acceptable
m = numpy.ndarray((2, 2))
m[:] = 3.
c = Covariance(2, ctypes.c_float, init_scalar_or_matrix=m)
m_out = c.to_matrix()
print('input matrix:\n', m)
print('output matrix:\n', m_out)
numpy.testing.assert_array_equal(m_out, m)
# Diagonally congruent values should be averages when initializing with
# matrix
m = numpy.eye(3, dtype=numpy.double)
m[0, 2] = 2.
m_expected = m.copy()
m_expected[0, 2] = 1.
m_expected[2, 0] = 1.
c = Covariance(3, init_scalar_or_matrix=m)
m_out = c.to_matrix()
print('input matrix:\n', m)
print('output matrix:\n', m_out)
numpy.testing.assert_array_equal(m_out, m_expected)
def test_from_cptr(self):
# Create a new covariance from C function and create new python instance
# from that pointer
c_new_func = VitalObject.VITAL_LIB['vital_covariance_3d_new']
c_new_func.argtypes = [VitalErrorHandle.C_TYPE_PTR]
c_new_func.restype = Covariance.c_ptr_type(3, ctypes.c_double)
with VitalErrorHandle() as eh:
c_ptr = c_new_func(eh)
c = Covariance(N=3, c_type=ctypes.c_double, from_cptr=c_ptr)
nose.tools.assert_is(c.C_TYPE_PTR, Covariance.c_ptr_type(3, ctypes.c_double))
numpy.testing.assert_array_equal(c.to_matrix(), numpy.eye(3))
c_new_func = VitalObject.VITAL_LIB['vital_covariance_3f_new']
c_new_func.argtypes = [VitalErrorHandle.C_TYPE_PTR]
c_new_func.restype = Covariance.c_ptr_type(3, ctypes.c_float)
with VitalErrorHandle() as eh:
c_ptr = c_new_func(eh)
c = Covariance(N=3, c_type=ctypes.c_float, from_cptr=c_ptr)
nose.tools.assert_is(c.C_TYPE_PTR,Covariance.c_ptr_type(3, ctypes.c_float))
numpy.testing.assert_array_equal(c.to_matrix(), numpy.eye(3))
def test_from_cptr(self):
# Create a new covariance from C function and create new python instance
# from that pointer
c_new_func = VitalObject.VITAL_LIB['vital_covariance_3d_new']
c_new_func.argtypes = [VitalErrorHandle.C_TYPE_PTR]
c_new_func.restype = Covariance.c_ptr_type(3, ctypes.c_double)
with VitalErrorHandle() as eh:
c_ptr = c_new_func(eh)
c = Covariance(N=3, c_type=ctypes.c_double, from_cptr=c_ptr)
nose.tools.assert_is(c.C_TYPE_PTR,
Covariance.c_ptr_type(3, ctypes.c_double))
numpy.testing.assert_array_equal(c.to_matrix(), numpy.eye(3))
c_new_func = VitalObject.VITAL_LIB['vital_covariance_3f_new']
c_new_func.argtypes = [VitalErrorHandle.C_TYPE_PTR]
c_new_func.restype = Covariance.c_ptr_type(3, ctypes.c_float)
with VitalErrorHandle() as eh:
c_ptr = c_new_func(eh)
c = Covariance(N=3, c_type=ctypes.c_float, from_cptr=c_ptr)
nose.tools.assert_is(c.C_TYPE_PTR,
Covariance.c_ptr_type(3, ctypes.c_float))
numpy.testing.assert_array_equal(c.to_matrix(), numpy.eye(3))
def test_new_scalar(self):
c = Covariance(2, ctypes.c_double, 2.)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(3, ctypes.c_double, 2.)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(2, ctypes.c_float, 2.)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(3, ctypes.c_float, 2.)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(2, ctypes.c_double, 14.675)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(3, ctypes.c_double, 14.675)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(2, ctypes.c_float, 14.675)
print 'constructed matrix:\n', c.to_matrix()
c = Covariance(3, ctypes.c_float, 14.675)
print 'constructed matrix:\n', c.to_matrix()
def test_new_scalar(self):
c = Covariance(2, ctypes.c_double, 2.)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(3, ctypes.c_double, 2.)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(2, ctypes.c_float, 2.)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(3, ctypes.c_float, 2.)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(2, ctypes.c_double, 14.675)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(3, ctypes.c_double, 14.675)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(2, ctypes.c_float, 14.675)
print('constructed matrix:\n', c.to_matrix())
c = Covariance(3, ctypes.c_float, 14.675)
print('constructed matrix:\n', c.to_matrix())
