Adaptivity with moving grids
Adaptivity with moving grids
Adaptivity with moving grids
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
48 C. J. Budd, W. Huang and R. D. Russell<br />
These can then be solved using standard stiff ODE software, e.g., byusing<br />
an SDIRK (singly diagonally implicit Runge–Kutta) method. A simple such<br />
semi-discretization of (3.9) is given by<br />
(<br />
)<br />
ɛ Ẋ i − γ Ẋi+1 − 2Ẋi + Ẋi−1<br />
(∆ξ) 2 = E i (t), (3.11)<br />
where the equidistribution measure E i (t) is as given in (3.4). This leads<br />
(on inversion of the simple tri-diagonal system on the right-hand side of<br />
this equation) to a simple set of ODEs for the location of the mesh points.<br />
Alternatively, a simple full discretization of (3.9) for a mesh Xi<br />
n evaluated<br />
at the time t n = n∆t is proposed in Ceniceros and Hou (2001), and takes<br />
the form<br />
(<br />
ɛ<br />
X n+1<br />
i<br />
(<br />
ɛ<br />
− γ Xn+1 i+1 − 2Xn+1 i<br />
+ Xi−1<br />
n+1 )<br />
(∆ξ) 2 =<br />
Xi n − γ Xn i+1 − 2Xn i + Xi−1<br />
n )<br />
(∆ξ) 2 +∆tEi n . (3.12)<br />
These equations for the mesh can then be solved together <strong>with</strong> a suitable<br />
discretization of the underlying PDE, either simultaneously or alternately.<br />
We will give more details of this procedure later in this section in the context<br />
of <strong>moving</strong> meshes in higher dimensions, but we note at this stage that the<br />
simultaneous solution method is both possible and effective in such onedimensional<br />
problems.<br />
The method for evolving the mesh is typically implemented in two stages.<br />
(1) Starting from an initially uniform mesh in the physical space, we evolve<br />
this to equidistribute the monitor function at the initial time over Ω P .<br />
To do this we set M 0 (x) ≡ M(x, 0), <strong>with</strong> x ξ = a +(b − a)ξ, and<br />
solve (3.46) <strong>with</strong> M fixed to equal the function M 0 ,<strong>with</strong>ɛ =1for<br />
0 < t < T, where T is a fixed time. In this first calculation t is<br />
an artificial time during which the uniform mesh evolves toward an<br />
equidistributed mesh for which the right-hand side of (3.9) is zero.<br />
It follows from the earlier results that, provided M 0 > 0, such a mesh<br />
exists, and we show presently that it is stable. In this initial calculation<br />
the right-hand side of (3.9) is initially relatively large, and taking ɛ =1<br />
prevents the numerical calculation of the solution of the ODEs for<br />
the mesh point locations from being unnecessarily stiff. The value<br />
of T is chosen large enough to allow the mesh to relax toward the<br />
equidistributed state.<br />
(2) We then solve (3.9) <strong>with</strong> the true time-dependent monitor function<br />
M(x, t), <strong>with</strong> t now actual time. For this calculation we typically set<br />
ɛ =0.01. We show presently that the resulting mesh is then ɛ-close to a<br />
mesh which exactly equidistributes M(x, t), provided that M does not