- removed comment
WaveToy2D outputs what looks like unitiialized data
Issue #1541
closed
the test test_WaveToy2D in WaveToy2DF77 in the CactusExamples arrangement fails for me (with a [mostly] clean checkout):
WaveToy2DF77: test_WaveToy2D
Failure: 7 files compared, 6 differ, 6 differ significantly
attached are diff log and one of the data files. This happens with current trunk.
It looks to me as if I am seeing poison.
Keyword: WaveToy2DF77
Comments (3)
-
-
reporter
- changed status to resolved
- removed comment
This was caused by a bug in Boundary that was fixed in rev 327 that prevented Boundary from applying the boundary condition for Copy, Flat, Scalar or Static boundary conditions for 1d and 2d evolutions (which excludes all Carpet based runs),
-
reporter
- edited description
- changed status to closed
- Log in to comment
- Assignee
- –
- Type
- Priority
- Status
- closed
- Component
- Cactus
- Milestone
- –
- Version
- development version
- Votes
- 0
- Watchers
- 0
The initial data do not set up the boundaries of the past time level. This should not matter, though, since these values are not used.
The solution is wrong near the boundary. The "poison" values are about 1e178, much larger than poison. This simulation uses PUGH, which (I believe) does not poison grid functions by default.