Skip to content

Commit 0a86b34

Browse files
committed
deploy: 473b4f0
1 parent e31427a commit 0a86b34

5 files changed

Lines changed: 15 additions & 15 deletions

File tree

.doctrees/6_inputdeck.doctree

1.59 KB
Binary file not shown.

.doctrees/environment.pickle

0 Bytes
Binary file not shown.

6_inputdeck.html

Lines changed: 7 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -191,22 +191,22 @@ <h2>Numerical Methods<a class="headerlink" href="#numerical-methods" title="Link
191191
<tr class="row-even"><td><p><code class="docutils literal notranslate"><span class="pre">wall_treatment</span></code></p></td>
192192
<td><p>Wall-treatment strategy</p></td>
193193
<td><p>(<code class="docutils literal notranslate"><span class="pre">clamp</span></code>)</p></td>
194-
<td></td>
194+
<td><p>Out-of-bounds points take nearest velocity. Applicable to subdomains from larger ones.</p></td>
195195
</tr>
196196
<tr class="row-odd"><td></td>
197197
<td></td>
198198
<td><p><code class="docutils literal notranslate"><span class="pre">dirichlet</span></code></p></td>
199-
<td></td>
199+
<td><p>Simple and fast BC treatment, zeroing the boundary values. Introduces artifical shear.</p></td>
200200
</tr>
201201
<tr class="row-even"><td></td>
202202
<td></td>
203203
<td><p><code class="docutils literal notranslate"><span class="pre">reflect</span></code></p></td>
204-
<td></td>
204+
<td><p>Reflect the particles back to the computational domain. Likely a non-shear wall BC.</p></td>
205205
</tr>
206206
<tr class="row-odd"><td></td>
207207
<td></td>
208208
<td><p><code class="docutils literal notranslate"><span class="pre">periodic</span></code></p></td>
209-
<td></td>
209+
<td><p>Wrap around to the opposite side. Applicable to fully-developed subdomains and periodic flows.</p></td>
210210
</tr>
211211
<tr class="row-even"><td><p><code class="docutils literal notranslate"><span class="pre">advc_method</span></code></p></td>
212212
<td><p>Lagrangian advection scheme</p></td>
@@ -236,7 +236,7 @@ <h2>Numerical Methods<a class="headerlink" href="#numerical-methods" title="Link
236236
<tr class="row-odd"><td></td>
237237
<td></td>
238238
<td><p>(<code class="docutils literal notranslate"><span class="pre">tricubic</span></code>)</p></td>
239-
<td><p>Catmull–Rom 3D convolution kernel.</p></td>
239+
<td><p>Catmull–Rom 3D convolution kernel for tricubic interpolation.</p></td>
240240
</tr>
241241
<tr class="row-even"><td></td>
242242
<td></td>
@@ -286,8 +286,8 @@ <h2>Numerical Methods<a class="headerlink" href="#numerical-methods" title="Link
286286
</tbody>
287287
</table>
288288
<p>As for your reference, and configured as defaults, the <em>Berkeley LCS Tutorials</em> used <code class="docutils literal notranslate"><span class="pre">RK4</span></code> for advection.
289-
The velocity fields were interpolated by <code class="docutils literal notranslate"><span class="pre">tricubic-FL</span></code>, originating from <span id="id2">[Lekien2005]</span>, which has higher performance from
290-
Although non detailed, <code class="docutils literal notranslate"><span class="pre">grad_order=2</span></code> was employed by them from the equation, supposing the mesh is sufficiently refined.
289+
The velocity fields were interpolated by <code class="docutils literal notranslate"><span class="pre">tricubic-FL</span></code>, originating from <span id="id2">[Lekien2005]</span>, which has higher performance by solving a 64*64 linear system using the function values, gradients, and mixed partial derivatives at its eight corners, which is in future plan for <code class="docutils literal notranslate"><span class="pre">Py3DFTLE</span></code> with high priority.
290+
Although not detailed, <code class="docutils literal notranslate"><span class="pre">grad_order=2</span></code> was employed by them from the equation, supposing the mesh is sufficiently refined.
291291
Please always notice that, although providing much better numerical precision and looks cool in papers, high-order methods could be resource-consuming, even hundreds of times.</p>
292292
</section>
293293
<section id="dynamic-lcs-window">

_sources/6_inputdeck.rst.txt

Lines changed: 7 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -47,13 +47,13 @@ Numerical Methods
4747
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
4848
| | | (``1``) | Compute in both side (both repelling and attracting structures). |
4949
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
50-
| ``wall_treatment``| Wall-treatment strategy | (``clamp``) | |
50+
| ``wall_treatment``| Wall-treatment strategy | (``clamp``) | Out-of-bounds points take nearest velocity. Applicable to subdomains from larger ones. |
5151
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
52-
| | | ``dirichlet`` | |
52+
| | | ``dirichlet`` | Simple and fast BC treatment, zeroing the boundary values. Introduces artifical shear. |
5353
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
54-
| | | ``reflect`` | |
54+
| | | ``reflect`` | Reflect the particles back to the computational domain. Likely a non-shear wall BC. |
5555
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
56-
| | | ``periodic`` | |
56+
| | | ``periodic`` | Wrap around to the opposite side. Applicable to fully-developed subdomains and periodic flows. |
5757
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
5858
| ``advc_method`` | Lagrangian advection scheme | (``Euler``) | Simple and fast Euler stepping with :math:`O(h^1)` error. |
5959
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
@@ -65,7 +65,7 @@ Numerical Methods
6565
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
6666
| ``interp_method`` | Interpolation method for velocity fields | ``trilinear`` | Simple and fast triliear interpolation method. Very fast on GPUs. |
6767
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
68-
| | | (``tricubic``) | Catmull–Rom 3D convolution kernel. |
68+
| | | (``tricubic``) | Catmull–Rom 3D convolution kernel for tricubic interpolation. |
6969
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
7070
| | | ``hermite`` | |
7171
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
@@ -87,8 +87,8 @@ Numerical Methods
8787
+-------------------+------------------------------------------------------+-------------------------+---------------------------------------------------------------------------------------------------------+
8888

8989
As for your reference, and configured as defaults, the *Berkeley LCS Tutorials* used ``RK4`` for advection.
90-
The velocity fields were interpolated by ``tricubic-FL``, originating from [Lekien2005]_, which has higher performance from
91-
Although non detailed, ``grad_order=2`` was employed by them from the equation, supposing the mesh is sufficiently refined.
90+
The velocity fields were interpolated by ``tricubic-FL``, originating from [Lekien2005]_, which has higher performance by solving a 64*64 linear system using the function values, gradients, and mixed partial derivatives at its eight corners, which is in future plan for ``Py3DFTLE`` with high priority.
91+
Although not detailed, ``grad_order=2`` was employed by them from the equation, supposing the mesh is sufficiently refined.
9292
Please always notice that, although providing much better numerical precision and looks cool in papers, high-order methods could be resource-consuming, even hundreds of times.
9393

9494
Dynamic LCS Window

0 commit comments

Comments
 (0)