def test_mass_matrix_inverse(self):
# get mesh
mesh = get_2p1d_mesh_for_testing(ypts=5, zpts=5)
spatial_methods = {
"macroscale": pybamm.FiniteVolume(),
"current collector": pybamm.ScikitFiniteElement(),
}
# create model
a = pybamm.Variable("a", domain="negative electrode")
b = pybamm.Variable("b", domain="current collector")
model = pybamm.BaseModel()
model.rhs = {a: pybamm.Laplacian(a), b: 4 * pybamm.Laplacian(b)}
model.initial_conditions = {a: pybamm.Scalar(3), b: pybamm.Scalar(10)}
model.boundary_conditions = {
a: {"left": (0, "Neumann"), "right": (0, "Neumann")},
b: {"negative tab": (0, "Neumann"), "positive tab": (0, "Neumann")},
}
model.variables = {"a": a, "b": b}
# create discretisation
disc = pybamm.Discretisation(mesh, spatial_methods)
disc.process_model(model)
# test that computing mass matrix block-by-block (as is done during
# discretisation) gives the correct result
# Note: inverse is more efficient in csc format
mass_inv = inv(csc_matrix(model.mass_matrix.entries))
np.testing.assert_equal(
model.mass_matrix_inv.entries.toarray(), mass_inv.toarray()
)
def test_evaluates_on_edges(self):
a = pybamm.StateVector(slice(0, 10), domain="test")
self.assertFalse(pybamm.Index(a, slice(1)).evaluates_on_edges("primary"))
self.assertFalse(pybamm.Laplacian(a).evaluates_on_edges("primary"))
self.assertFalse(pybamm.GradientSquared(a).evaluates_on_edges("primary"))
self.assertFalse(pybamm.BoundaryIntegral(a).evaluates_on_edges("primary"))
self.assertTrue(pybamm.Upwind(a).evaluates_on_edges("primary"))
self.assertTrue(pybamm.Downwind(a).evaluates_on_edges("primary"))
def test_evaluates_on_edges(self):
a = pybamm.StateVector(slice(0, 10))
self.assertFalse(a[1].evaluates_on_edges())
self.assertFalse(pybamm.Laplacian(a).evaluates_on_edges())
