update: auto commit
This commit is contained in:
+1
-1
@@ -165,7 +165,7 @@
|
|||||||
\newcommand{\AD}[1]{\tensor{\mathcal{D}}{_{#1}}}
|
\newcommand{\AD}[1]{\tensor{\mathcal{D}}{_{#1}}}
|
||||||
\newcommand{\ADd}{\mathcal{D}}
|
\newcommand{\ADd}{\mathcal{D}}
|
||||||
\newcommand{\connection}[1]{\mathcal{#1}}
|
\newcommand{\connection}[1]{\mathcal{#1}}
|
||||||
\newcommand{\Partial}[1]{\tensor{\partial}{_{#1}}}%普通导数算符
|
\newcommand{\Partial}[1]{\tensor{\partial}{#1}}%普通导数算符
|
||||||
\newcommand{\tPartial}[1]{\tensor{\tilde{\partial}}{_{#1}}}
|
\newcommand{\tPartial}[1]{\tensor{\tilde{\partial}}{_{#1}}}
|
||||||
\newcommand{\Fd}[2][\tau]{\ensuremath{\frac{\mathrm{D_F}#2}{\dd{#1}}}}%费米导数
|
\newcommand{\Fd}[2][\tau]{\ensuremath{\frac{\mathrm{D_F}#2}{\dd{#1}}}}%费米导数
|
||||||
\newcommand{\Fdd}[2][\tau]{\ensuremath{\mathrm{D_F}#2/\dd{#1}}}%行内费米导数
|
\newcommand{\Fdd}[2][\tau]{\ensuremath{\mathrm{D_F}#2/\dd{#1}}}%行内费米导数
|
||||||
|
|||||||
+11
-11
@@ -566,14 +566,14 @@
|
|||||||
\subsection{First order reduction: wave equation as an example}
|
\subsection{First order reduction: wave equation as an example}
|
||||||
We can introduce auxiliary variables to reduce a second order PDE system to a first order one. For example, for the wave equation on flat spacetime:
|
We can introduce auxiliary variables to reduce a second order PDE system to a first order one. For example, for the wave equation on flat spacetime:
|
||||||
\begin{equation}
|
\begin{equation}
|
||||||
\Partial{t}^2 {\phi} - \tensor{\delta}{^i^j} \Partial{i} \Partial{j} \phi = 0,
|
\pt^2 {\phi} - \tensor{\delta}{^i^j} \Partial{_i} \Partial{_j} \phi = 0,
|
||||||
\end{equation}
|
\end{equation}
|
||||||
we can introduce $\pi := \Partial{t} \phi$ and $\tensor{\psi}{_i} := \Partial{_i} \phi$, and rewrite the wave equation as
|
we can introduce $\pi := \pt \phi$ and $\tensor{\psi}{_i} := \Partial{_i} \phi$, and rewrite the wave equation as
|
||||||
\begin{equation}
|
\begin{equation}
|
||||||
\begin{cases}
|
\begin{cases}
|
||||||
\Partial{t}{\phi} = \pi, \\
|
\pt{\phi} = \pi, \\
|
||||||
\Partial{t}{\pi} = \tensor{\delta}{^i^j} \Partial{i} \tensor{\psi}{_j},\\
|
\pt{\pi} = \tensor{\delta}{^i^j} \Partial{_i} \tensor{\psi}{_j},\\
|
||||||
\Partial{t}{\tensor{\psi}{_i}} = \Partial{i} \pi.
|
\pt{\tensor{\psi}{_i}} = \Partial{_i} \pi.
|
||||||
\end{cases}
|
\end{cases}
|
||||||
\end{equation}
|
\end{equation}
|
||||||
write $u^I = (\phi, \pi, \tensor{\psi}{_1}, \tensor{\psi}{_2}, \tensor{\psi}{_3})$, then the system can be written in the form of~\eqref{first-order-PDE} with
|
write $u^I = (\phi, \pi, \tensor{\psi}{_1}, \tensor{\psi}{_2}, \tensor{\psi}{_3})$, then the system can be written in the form of~\eqref{first-order-PDE} with
|
||||||
@@ -614,11 +614,11 @@
|
|||||||
\begin{frame}{First order reduction: wave equation as an example}{constraints during the reduction}
|
\begin{frame}{First order reduction: wave equation as an example}{constraints during the reduction}
|
||||||
During the first order reduction, we introduce new variables. A solution to the new system is a solution to the original system if and only if the new variables satisfy some constraints. For example, in the wave equation case, we have the constraints
|
During the first order reduction, we introduce new variables. A solution to the new system is a solution to the original system if and only if the new variables satisfy some constraints. For example, in the wave equation case, we have the constraints
|
||||||
\begin{equation}
|
\begin{equation}
|
||||||
\mathcal{C}_i \definedby \tensor{\psi}{_i} - \Partial{i} \phi = 0.
|
\mathcal{C}_i \definedby \tensor{\psi}{_i} - \Partial{_i} \phi = 0.
|
||||||
\end{equation}
|
\end{equation}
|
||||||
The evolution of the constraints is given by
|
The evolution of the constraints is given by
|
||||||
\begin{equation}
|
\begin{equation}
|
||||||
\Partial{t} \mathcal{C}_i = \Partial{t} \psi_i - \Partial{i} \Partial{t} \phi = \Partial{i} \pi - \Partial{i} \pi = 0,
|
\pt \mathcal{C}_i = \pt \psi_i - \Partial{_i} \pt \phi = \Partial{_i} \pi - \Partial{_i} \pi = 0,
|
||||||
\end{equation}
|
\end{equation}
|
||||||
which means that if the constraints are satisfied initially, they will be satisfied for all time. However, in numerical simulations, there will always be some constraint violation due to numerical errors, and thus it's better to add some constraint damping terms to the evolution equations to suppress the growth of constraint violation.
|
which means that if the constraints are satisfied initially, they will be satisfied for all time. However, in numerical simulations, there will always be some constraint violation due to numerical errors, and thus it's better to add some constraint damping terms to the evolution equations to suppress the growth of constraint violation.
|
||||||
\end{frame}
|
\end{frame}
|
||||||
@@ -627,14 +627,14 @@
|
|||||||
For example, we can add a constraint damping term $-\gamma \mathcal{C}_i$ to the evolution equation of $\tensor{\psi}{_i}$,
|
For example, we can add a constraint damping term $-\gamma \mathcal{C}_i$ to the evolution equation of $\tensor{\psi}{_i}$,
|
||||||
\begin{equation}
|
\begin{equation}
|
||||||
\begin{cases}
|
\begin{cases}
|
||||||
\Partial{t}{\phi} = \pi, \\
|
\pt{\phi} = \pi, \\
|
||||||
\Partial{t}{\pi} = \tensor{\delta}{^i^j} \Partial{i} \tensor{\psi}{_j},\\
|
\pt{\pi} = \tensor{\delta}{^i^j} \Partial{_i} \tensor{\psi}{_j},\\
|
||||||
\Partial{t}{\tensor{\psi}{_i}} = \Partial{i} \pi - \gamma \left( \tensor{\psi}{_i} - \Partial{i} \phi \right),
|
\pt{\tensor{\psi}{_i}} = \Partial{_i} \pi - \gamma \left( \tensor{\psi}{_i} - \Partial{_i} \phi \right),
|
||||||
\end{cases}
|
\end{cases}
|
||||||
\end{equation}
|
\end{equation}
|
||||||
where $\gamma > 0$ is a constant. Then the evolution of the constraints becomes
|
where $\gamma > 0$ is a constant. Then the evolution of the constraints becomes
|
||||||
\begin{equation}
|
\begin{equation}
|
||||||
\Partial{t} \mathcal{C}_i = -\gamma \mathcal{C}_i,
|
\pt \mathcal{C}_i = -\gamma \mathcal{C}_i,
|
||||||
\end{equation}
|
\end{equation}
|
||||||
which means that the constraint violation will decay exponentially with time, and thus the system is more stable for numerical simulations.
|
which means that the constraint violation will decay exponentially with time, and thus the system is more stable for numerical simulations.
|
||||||
\end{frame}
|
\end{frame}
|
||||||
|
|||||||
Reference in New Issue
Block a user