[cig-commits] r8605 - seismo/2D/SPECFEM2D/trunk

[walter at geodynamics.org]

Author: walter
Date: 2007-12-07 16:00:29 -0800 (Fri, 07 Dec 2007)
New Revision: 8605

Added:
   seismo/2D/SPECFEM2D/trunk/todo_list_please_dont_remove.txt
Removed:
   seismo/2D/SPECFEM2D/trunk/TODO_list
Log:
updated the todo list


Deleted: seismo/2D/SPECFEM2D/trunk/TODO_list
===================================================================
--- seismo/2D/SPECFEM2D/trunk/TODO_list	2007-11-21 17:11:04 UTC (rev 8604)
+++ seismo/2D/SPECFEM2D/trunk/TODO_list	2007-12-08 00:00:29 UTC (rev 8605)
@@ -1,13 +0,0 @@
-- splitting file part_unstruct.F90 in several files for clarity purpose.
-- improving compiling with SCOTCH (issue with header file scotchf.h which is Fortran77 legal). Having our own scotchf.h file (without the comments) is not wise.
-- comparing the different partitioning methods for METIS and SCOTCH, and finding a good default for SCOTCH.
-- partitioning using weights for load-balancing.
-- choosing a way to use assign_external_model?
-- adding comments.
-- checking for points with different normals for absorbing conditions, when the absorbing edges are not in the same elements (similar to what is done for the corners).
-- scripts for translating GID/CUBIT meshes into files for xmeshfem2D.
-- modifying scripts for UPPA cluster (when FS sync issues are solved and remote commands are available).
-- manual for unstructured meshes.
-- getting rid of constants_unstruct.h.
-- checking use of real or double precision. Gain in elapsed time is ok, but now we have to look for memory consumption.
-- checking output on stdout (for data that should be printed only once).

Copied: seismo/2D/SPECFEM2D/trunk/todo_list_please_dont_remove.txt (from rev 8604, seismo/2D/SPECFEM2D/trunk/TODO_list)
===================================================================
--- seismo/2D/SPECFEM2D/trunk/TODO_list	2007-11-21 17:11:04 UTC (rev 8604)
+++ seismo/2D/SPECFEM2D/trunk/todo_list_please_dont_remove.txt	2007-12-08 00:00:29 UTC (rev 8605)
@@ -0,0 +1,46 @@
+
+- splitting file part_unstruct.F90 in several files for clarity purpose.
+
+- improving compiling with SCOTCH (issue with header file scotchf.h which is Fortran77 legal). Having our own scotchf.h file (without the comments) is not wise.
+
+- comparing the different partitioning methods for METIS and SCOTCH, and finding a good default for SCOTCH.
+
+- partitioning using weights for load-balancing.
+
+- choosing a way to use assign_external_model?
+
+- adding comments.
+
+- checking for points with different normals for absorbing conditions, when the absorbing edges are not in the same elements (similar to what is done for the corners).
+
+- scripts for translating GID/CUBIT meshes into files for xmeshfem2D.
+
+- modifying scripts for UPPA cluster (when FS sync issues are solved and remote commands are available).
+
+- manual for unstructured meshes.
+
+- getting rid of constants_unstruct.h.
+
+- checking use of real or double precision. Gain in elapsed time is ok, but now we have to look for memory consumption.
+
+- checking output on stdout (for data that should be printed only once).
+
+- Hi Jeroen, Perfect. I think talking to Jean-Paul Ampuero would be useful
+as well because in Utrecht last year he had told us that
+he had implemented some nice 4th-order symplectic schemes
+in his version of SEM2D. Dimitri.
+Jeroen Tromp wrote:
+> Hi Dimitri:
+> This is one of the first things Tarje and I plan to work on after he arrives.
+> Jeroen
+> Dimitri Komatitsch wrote:
+>> Hi Jeroen,
+>> I think the last important thing that is missing
+>> in SPECFEM3D (and SPECFEM2D) is a fourth-order time scheme.
+>> I think both Jean-Paul Ampuero and Tarje Nissen-Meyer
+>> have worked on this, they will both be at Caltech
+>> soon therefore maybe they could take care of adding it?
+>> This would definitely increase the accuracy of very long
+>> simulations (e.g. multi-orbit surface waves).
+>> Dimitri.
+