[cig-commits] r18707 - seismo/3D/FAULT_SOURCE

[ampuero at geodynamics.org]

Author: ampuero
Date: 2011-07-07 05:55:56 -0700 (Thu, 07 Jul 2011)
New Revision: 18707

Modified:
   seismo/3D/FAULT_SOURCE/TO_DO
Log:
outline plan for fault parallelization in TO_DO

Modified: seismo/3D/FAULT_SOURCE/TO_DO
===================================================================
--- seismo/3D/FAULT_SOURCE/TO_DO	2011-07-07 00:08:25 UTC (rev 18706)
+++ seismo/3D/FAULT_SOURCE/TO_DO	2011-07-07 12:55:56 UTC (rev 18707)
@@ -1,8 +1,22 @@
-+ parallelize the fault
++ parallelize the fault:
+	- in fault_scotch.f90:
+		. reorder the fault element lists so that ispec1 and ispec2
+		  are facing each other: for each face ispec1 search the 
+		  face ispec2 with nearest center point
+		. for each fault, for each pair (ispec1,ispec2) put both 
+		  elements in the processor with lowest rank among the two
+		. repeat the loop above to handle triple junctions
+	- in fault_solver.f90:
+		. assemble B and the fault normal vectors during initialization
+		. assemble internal forces before and after fault solver
+		. parallelize outputs
+
 + close fault edges in CUBIT
 + mesh coarsening in CUBIT
+
 + add snapshot outputs to kinematic solver
 + test kinematic solver
 + factor fault solvers
+
 + eliminate fault edge nodes (sticky, same ibulk) ?