Electromagnetic fields
Electromagnetic Maxwell’s fields are propagated with a Finite-difference-time-domain (FDTD) method.
Note
Due to the choice of our unit system, the time step \(\Delta t\) is basically equivalent to the numerical speed of light, \(\hat{c}\), in the following discussion; therefore, when you see \(\hat{c}\) in the following formulas, you can mentally replace it by \(\Delta t\).
Grid staggering
Yee lattice is used for \(\mathbf{E}\) and \(\mathbf{B}\) fields such that they are staggered both in space and in time,
where \(\hat{\mathbf{E}}\) is located in the middle of the cell sides and \(\hat{\mathbf{B}}\) in the center of the cell faces. This makes it easy to compute the curl of the fields in the equations below.
The time staggering, in turn, increases the temporal order of the scheme since all the updates are leapfrog-like, \(x^{n+1} = x^n + v^{n+\frac{1}{2}} \Delta t\).
FDTD method
In the time domain we update \(\hat{\mathbf{E}}\) and \(\hat{\mathbf{B}}\) fields with discrete forms of Maxwell’s equations:
and
where \(\Delta[Q]_{t,\mathbf{x}}\) is the time differentiation or curl of a variable \(Q\) without the \(\Delta x\) or \(\Delta t\) denominator.
The only normalization factor entering these equations is the Courant velocity, \(\hat{c}\), since everything else is absorbed into \(B_0\) and \(J_0\). There is no need to solve the divergence equations if a charge-conserving scheme is used together with the Yee staggering of the fields.
Finite-difference solvers
Different FDTD solvers can be obtained by defining different discrete curl-operators, \(\nabla \times \mathbf{Q} \rightarrow \Delta[ \hat{\mathbf{Q}} ]_{\mathbf{x}}\), i.e., different finite-difference operators for the calculation of the derivative.
Note
TODO
Filtering
Note
TODO