[docs]
class MaxwellWeakAmpere(Propagator):
r"""FEEC discretization of the following equations:
find :math:`\mathbf E \in H(\textnormal{curl})` and :math:`\mathbf B \in H(\textnormal{div})` such that
.. math::
\int_{\Omega} \frac{\partial \mathbf E}{\partial t} \cdot \mathbf F \, \textrm d \mathbf x - \int_{\Omega} \mathbf B \cdot \nabla \times \mathbf F \,\textrm d \mathbf x &= 0\,, \qquad \forall \, \mathbf F \in H(\textnormal{curl})
\\[2mm]
\frac{\partial \mathbf B}{\partial t} + \nabla\times\mathbf E &= 0\,.
Time discretization:
- implicit: Crank-Nicolson (implicit mid-point)
- explicit: explicit RK methods from ButcherTableau
System size reduction via :class:`~struphy.linear_algebra.schur_solver.SchurSolver`.
"""
[docs]
class Variables:
"""Container for Maxwell equation variables.
Attributes
----------
e : FEECVariable
Electric field in H(curl) space.
b : FEECVariable
Magnetic field in H(div) space.
"""
def __init__(self):
self._e: FEECVariable = None
self._b: FEECVariable = None
@property
def e(self) -> FEECVariable:
return self._e
@e.setter
def e(self, new):
assert isinstance(new, FEECVariable)
assert new.space == "Hcurl"
self._e = new
@property
def b(self) -> FEECVariable:
return self._b
@b.setter
def b(self, new):
assert isinstance(new, FEECVariable)
assert new.space == "Hdiv"
self._b = new
def __init__(self):
self.variables = self.Variables()
[docs]
@dataclass(repr=False)
class Options(OptionsBase):
"""Configuration options for :class:`MaxwellWeakAmpere`.
Parameters
----------
algo : {"implicit", "explicit"}, default="implicit"
Time stepping scheme to use for the Maxwell equations.
- ``"implicit"``: Crank-Nicolson (implicit mid-point) scheme.
- ``"explicit"``: explicit Runge-Kutta methods from ``ButcherTableau``.
solver : LiteralOptions.OptsSymmSolver, default="pcg"
Name of the symmetric iterative solver passed to
:func:`psydac.linalg.solvers.inverse`.
precond : LiteralOptions.OptsMassPrecond, default="MassMatrixPreconditioner"
Name of the preconditioner configuration.
Currently used only for implicit time stepping.
solver_params : SolverParameters, default=None
Iterative-solver controls (for example ``tol``, ``maxiter``,
``verbose``, ``info``, ``recycle``).
If ``None``, defaults to ``SolverParameters()``.
butcher : ButcherTableau, default=None
Butcher tableau for explicit Runge-Kutta methods.
Only used when ``algo="explicit"``.
If ``None``, defaults to ``ButcherTableau()``.
Notes
-----
System size reduction is performed via :class:`SchurSolver` for
implicit time stepping to eliminate the magnetic field and reduce
the system to the electric field equation.
"""
# specific literals
OptsAlgo = Literal["implicit", "explicit"]
# propagator options
algo: OptsAlgo = "implicit"
solver: LiteralOptions.OptsSymmSolver = "pcg"
precond: LiteralOptions.OptsMassPrecond = "MassMatrixPreconditioner"
solver_params: SolverParameters = None
butcher: ButcherTableau = None
def __post_init__(self):
# checks
check_option(self.algo, self.OptsAlgo)
check_option(self.solver, LiteralOptions.OptsSymmSolver)
check_option(self.precond, LiteralOptions.OptsMassPrecond)
# defaults
if self.solver_params is None:
self.solver_params = SolverParameters()
if self.algo == "explicit" and self.butcher is None:
self.butcher = ButcherTableau()
@property
def options(self) -> Options:
if not hasattr(self, "_options"):
self._options = self.Options()
return self._options
@options.setter
def options(self, new):
assert isinstance(new, self.Options)
self._options = new
logger.info(f"\nNew options for propagator '{self.__class__.__name__}':\n{self._options}")
@profile
def allocate(self):
# obtain needed matrices
M1 = self.mass_ops.M1
M2 = self.mass_ops.M2
curl = self.derham.curl
# Preconditioner for M1 + ...
if self.options.precond is None:
pc = None
else:
pc_class = getattr(preconditioner, self.options.precond)
pc = pc_class(M1)
if self.options.algo == "implicit":
self._info = self.options.solver_params.info
# Define block matrix [[A B], [C I]] (without time step size dt in the diagonals)
_A = M1
# no dt
self._B = -1 / 2 * curl.T @ M2
self._C = 1 / 2 * curl
# Instantiate Schur solver (constant in this case)
_BC = self._B @ self._C
self._schur_solver = SchurSolver(
_A,
_BC,
self.options.solver,
precond=pc,
solver_params=self.options.solver_params,
)
# pre-allocate arrays
self._byn = self._B.codomain.zeros()
else:
self._info = False
# define vector field
M1_inv = inverse(
M1,
self.options.solver,
pc=pc,
tol=self.options.solver_params.tol,
maxiter=self.options.solver_params.maxiter,
verbose=self.options.solver_params.verbose,
)
weak_curl = M1_inv @ curl.T @ M2
# allocate output of vector field
out1 = self.variables.e.spline.vector.space.zeros()
out2 = self.variables.b.spline.vector.space.zeros()
def f1(t, y1, y2, out: BlockVector = out1):
weak_curl.dot(y2, out=out)
out.update_ghost_regions()
return out
def f2(t, y1, y2, out: BlockVector = out2):
curl.dot(y1, out=out)
out *= -1.0
out.update_ghost_regions()
return out
vector_field = {self.variables.e.spline.vector: f1, self.variables.b.spline.vector: f2}
self._ode_solver = ODEsolverFEEC(vector_field, butcher=self.options.butcher)
# allocate place-holder vectors to avoid temporary array allocations in __call__
self._e_tmp1 = self.variables.e.spline.vector.space.zeros()
self._e_tmp2 = self.variables.e.spline.vector.space.zeros()
self._b_tmp1 = self.variables.b.spline.vector.space.zeros()
@profile
def __call__(self, dt):
# current FE coeffs
en = self.variables.e.spline.vector
bn = self.variables.b.spline.vector
if self.options.algo == "implicit":
# solve for new e coeffs
self._B.dot(bn, out=self._byn)
en1, info = self._schur_solver(en, self._byn, dt, out=self._e_tmp1)
# new b coeffs
_e = en.copy(out=self._e_tmp2)
_e += en1
bn1 = self._C.dot(_e, out=self._b_tmp1)
bn1 *= -dt
bn1 += bn
diffs = self.update_feec_variables(e=en1, b=bn1)
else:
self._ode_solver(0.0, dt)
if self._info and MPI.COMM_WORLD.Get_rank() == 0:
if self.options.algo == "implicit":
logger.info(f"Status for Maxwell: {info['success']}")
logger.info(f"Iterations for Maxwell: {info['niter']}")
logger.info(f"Maxdiff e for Maxwell: {diffs['e']}")
logger.info(f"Maxdiff b for Maxwell: {diffs['b']}")
logger.info("")
