Re: [ET Trac] [Einstein Toolkit] #877: EinsteinExact arrangement should be added to the Einstein Toolkit

#877: EinsteinExact arrangement should be added to the Einstein Toolkit ------------------------------------+--------------------------------------- Reporter: hinder | Owner: Type: task | Status: new Priority: major | Milestone: ET_2012_05 Component: EinsteinToolkit thorn | Version: Resolution: | Keywords: ------------------------------------+---------------------------------------

Comment (by barry.wardell):

Thanks for the comments!

Replying to [comment:9 bmundim]:

1. When running the test suites I got a message that it is not runnable

for EinsteinExact:

No runnable testsuites in arrangements: CactusBase CactusConnect

CactusDoc CactusElliptic CactusIO CactusPUGH EinsteinBase EinsteinEOS EinsteinUtils ExternalLibraries TAT KrancNumericalTools EinsteinExact

However EinsteinExact does have a test suite at

EinsteinExact/EinsteinExact_Test/test/ShiftedGaugeWave.par

EinsteinExact_Test is correctly symbolically linked in the arrangements

directory. Any ideas what could be going wrong here?

The tests require 1 process to run. Can you check that you are running with 1 MPI process?

2. When I ran the ShiftedGaugeWave.par test case I found a bit

suspicious that the values output were exactly one or zero for gxx, kxx, alp or betax. Was this intentional? Doesn't it make the test too trivial?

3. Note too that ADMBase::dtalp and ADMBase::dtbetax are not output at

all despite enlisted to do so. I guess this is happening because initial_dtlapse and initial_dtshift default to "none". So we need to

set those parameters in the test par file. In any case, I added the

following parameters to the par file and I got an output with non trivial values:

ADMBase::initial_data = ShiftedGaugeWave ADMBase::initial_shift = ShiftedGaugeWave ADMBase::initial_lapse = ShiftedGaugeWave ADMBase::initial_dtshift = ShiftedGaugeWave ADMBase::initial_dtlapse = ShiftedGaugeWave

These were the problems reported by Peter. Although I did fix them a few days ago (and I think Peter tested the fix), I somehow forgot to merge the fix into the master branch. I've now merged them into master so if you pull you should have something which resolves both of these problems.

4. In the documentation it says that the code ignores initial_shift,

initial_lapse, initial_dtshift and initial_dtlapse. Per example above, I don't think this is true.

It ignores initial_shift, initial_lapse, initial_dtshift and initial_dtlapse. The setting of the initial data for all of the ADMBase variables is determined only by the initial_data parameter.

5. There are a few instances of "CCTK_INT ierr = 0;"

declarations/initializations that are not used afterwards. That could be taken out to avoid compiler warning complaints.

I guess this must be a Kranc related thing? There was a separate discussion about fixing these things in Kranc, so once that is resolved the problem should also go away here.