Field propagator

JxBCold

Advances j.

What it solves

FEEC discretization of the following equations:

find $j \in H (curl)$ such that

\int_{Ω} \frac{1}{n _{0}} \frac{\partial j}{\partial t} \cdot F d x = \frac{1}{ε} \int_{Ω} \frac{1}{n _{0}} (j \times B_{0}) \cdot F d x \forall F \in H (curl),

time_discret: Crank-Nicolson (implicit mid-point), such that

M_{1/ n_{0}} (j^{n + 1} - j^{n}) = \frac{Δ t}{2} \frac{1}{ε} M_{B_{0} / n_{0}} (j^{n + 1} - j^{n}) .

The variables this step updates over a time step dt, and the discrete spaces each one accepts.

Name	Kind	Accepted spaces
`j`	FEECVariable	`Hcurl`

Set these on the propagator in a parameter file, for example JxBCold.Options(solver=...).

Option	Default	Choices	Description
`solver`	`pcg`	`pcgcg`	Symmetric iterative solver used for the linear current update.
`precond`	`MassMatrixPreconditioner`	`MassMatrixPreconditionerMassMatrixDiagonalPreconditionerNone`	Preconditioner applied to the weighted mass matrix `M1ninv`.
`solver_params`	`None`	SolverParameters	Iterative-solver controls. If `None`, defaults to `SolverParameters()`.

Each model splits its time step into a sequence of propagators. This one appears in 2 models, at the position shown.