Boundary Conditions
Velocity
\(\vec{u}\) is calculated in the dynamo code using a poloidal toroidal decomposition:
\[
\vec{u} = \mathbf{\nabla} \times(\mathbb{T}\vec{r}) + \mathbf{\nabla} \times\mathbf{\nabla} \times(\mathbb{P}\vec{r}).
\]
No slip BCs:
\[
\mathbb{T}=0, \quad \mathbb{P}=0, \quad \partial_r\mathbb{P}=0.
\]
Stress-free
\[
\left(\partial_{r} - \,\frac1{r}\right)\mathbb{T} = 0,
\quad
\mathbb{P}=0, \quad \partial_{rr} \mathbb{P}=0.
\]
The condition \(P=0\) appears for both and corresponds to the no-penetration
condition \(u_r=0\).
Codensity/Temperature
Called codensity in the code, can act as a codensity term for single component convection, or temperature if separate temperature and compositional fields are required.
Fixed Temperature
Solving the conduction problem
\[
\frac{\partial T}{\partial r} = \frac{A}{r^2},
\]
gives
\[
T = \frac{-A}{r} + B.
\]
we set \(T|_{r_i}=1\) and \(T|_{r_o}=0\), giving a background conduction state of
\[
\frac{\partial{T}}{{\partial r}} = \frac{1}{r^2}\frac{r_ir_o}{r_i-r_o.}
\]
Fixed Flux
\[
{\frac{\partial T}{\partial r}\bigg|}_{r=r_i} = \text{d_qi}
\]
\[
{\frac{\partial T}{\partial r}\bigg|}_{r=r_o} = \text{d_qo}
\]
Composition
Solving the conduction problem
\[
\frac{\partial \xi}{\partial r} = \frac{A}{r^2},
\]
gives
\[
\xi = \frac{-A}{r} + B.
\]
we set \(\xi|_{r_i}=1\) and \(\xi|_{r_o}=0\), giving a background conduction state of
\[
\frac{\partial{\xi}}{{\partial r}} = \frac{1}{r^2}\frac{r_ir_o}{r_i-r_o.}
\]
Fixed Flux
\[
{\frac{\partial \xi}{\partial r}\bigg|}_{r=r_i} = \text{d_qi_comp}
\]
\[
{\frac{\partial \xi}{\partial r}\bigg\vert}_{r=r_o} = \text{d_qo_comp}
\]
Magnetic Field
For the decomposition
\[
\vec{B} = \mathbf{\nabla} \times(\mathcal{T}\vec{r}) + \mathbf{\nabla} \times\mathbf{\nabla} \times(\mathcal{P}\vec{r}).
\]
Insulating BCs
\[
{\mathcal T}=0, \quad
\left(\partial_{r}-\,\frac{l}{r}\right){\mathcal P}=0
\quad \text{ on }~~ r_i,
\]
\[
{\mathcal T}=0, \quad
\left(\partial_{r}+\,\frac{l+1}{r}\right){\mathcal P}=0
\quad \text{ on } r_o.
\]
Conducting inner Core
\[
\left[{\mathcal T}\right]=0,
\quad
\left[{\mathcal P}\right]=0,
\quad
\left[\partial_{r}{\mathcal P}\right]=0,
\quad
\nabla^2_1
\left[\partial_{r}{\mathcal T}\right] = \frac1{r}\,\hat{\vec{r}}\cdot\mathbf{\nabla} \times
\left(B_r\left[\vec{u}\right]\right)
\quad
\text{ on } r_i
\]
where \([\,\cdot\,]\) is the jump across the interface.
The last condition is valid for an inner core of the same magnetic
diffusivity, and for no-slip boundaries simplifies to
\([\partial_{r}{\mathcal T}]=0\).