User: rhaas Date: 2012/08/09 01:10 AM

Modified: /trunk/ schedule.ccl /trunk/par/ balsara1d.par /trunk/src/ GRHydro_ShockTubeM.F90

Log: Introduce a better treatment of the shocktube boundaries.

* The 2d diagonal boundary conditions are correctly applied when using multiprocessors.

* The ymin (ymax) surfaces have their points in the region indsum < minsum (indsum > maxsum) fixed after a first sync operation. A second sync guarantees the neighbour grid receives the correct value of the boundary on those regions.

* Mirror symmetries applied at the faces intercepting the upper right corner assumed cubic local domain. However, rectangular local domains are common after domain decomposition. This commit fixes the indexes applying this symmetry for these cases.

* Upper right corner/edge/surface condition should use sten instead of stenp1. Thanks for catching this Josh!

* Attempt at a 3d diagonal boundary treatment (Josh's commit)

File Changes:

Directory: /trunk/par/ ======================

File [modified]: balsara1d.par Delta lines: <Unavailable>

Directory: /trunk/src/ ======================

File [modified]: GRHydro_ShockTubeM.F90 Delta lines: <Unavailable>

Directory: /trunk/ ==================

File [modified]: schedule.ccl Delta lines: <Unavailable>