where $\partial\Omega_-$ describes the inflow portion of the boundary and is
formally defined by
@f[
- \partial\Omega_-
+ \partial\Omega_-
=
\{{\mathbf x}\in \partial\Omega: \beta\cdot{\mathbf n}({\mathbf x}) < 0\},
@f]
As it is stated, the transport equation is not stably solvable using
the standard finite element method, however. The problem is that
solutions to this equation possess only insufficient regularity
-orthogonal to the transport direction: while they are smooth parallel
+orthogonal to the transport direction: while they are smooth %parallel
to $\beta$, they may be discontinuous perpendicular to this
direction. These discontinuities lead to numerical instabilities that
make a stable solution by a straight-forward discretization
with basis functions $\varphi_i,\varphi_j$. However, this is a
pitfall that happens to every numerical analyst at least once
(including the author): we have here expanded the solution
-$u_h = u_i \varphi_i$, but if we do so, we will have to solve the
-problem
+$u_h = u_i \varphi_i$, but if we do so, we will have to solve the
+problem
@f[
{\mathbf u}^T A = {\mathbf f}^T,
@f]
the following domain and functions (in $d=2$ space dimensions):
@f{eqnarray*}
\Omega &=& [-1,1]^d \\
- \beta({\mathbf x})
+ \beta({\mathbf x})
&=&
\left(
- \begin{array}{c}1 \\ 1+\frac 45 \sin(8\pi x)\end{array}
+ \begin{array}{c}1 \\ 1+\frac 45 \sin(8\pi x)\end{array}
\right),
\\
f({\mathbf x})
&=&
\left\{
\begin{array}{ll}
- \frac 1{10 s^d} &
+ \frac 1{10 s^d} &
\mathrm{for}\ |{\mathbf x}-{\mathbf x}_0|<s, \\
0 & \mathrm{else},
\end{array}
\right.
\qquad\qquad
- {\mathbf x}_0
+ {\mathbf x}_0
=
\left(
\begin{array}{c} -\frac 34 \\ -\frac 34\end{array}
boundary conditions a blob in the lower left corner, which is then
transported along.
<li> The inflow boundary conditions impose a weighted sinusoidal
-structure that is transported along with the flow field. Since
-$|{\mathbf x}|\ge 1$ on the boundary, the weighting term never gets very large.
+structure that is transported along with the flow field. Since
+$|{\mathbf x}|\ge 1$ on the boundary, the weighting term never gets very large.
</ol>
derivatives, weighted by a power of the cell size that is adjusted to
the linear elements assumed to be in use here:
@f[
- \eta_K \approx
+ \eta_K \approx
C h \| \nabla^2 u \|_K,
@f]
which itself is related to the error size in the energy norm.
\frac{u(K') - u(K)}{|{\mathbf y}_{KK'}|},
@f]
where $u(K)$ and $u(K')$ denote $u$ evaluated at the centers of the
-respective cells. We now multiply the above approximation by
+respective cells. We now multiply the above approximation by
${\mathbf y}_{KK'}/|{\mathbf y}_{KK'}|$ and sum over all neighbors $K'$ of $K$:
@f[
\underbrace{
@f]
which is inspired by the following (not rigorous) argument:
@f{eqnarray*}
- \|u-u_h\|^2_{L_2}
+ \|u-u_h\|^2_{L_2}
&\le&
C h^2 \|\nabla u\|^2_{L_2}
\\