Adding 3D tally recomposition and 3D Domain Decomposition test

[alexandermote]

These changes are mainly improvements to #212, namely expanding the mesh tally decomposition to 3D domain decompositions. A new regression test, dd_cooper, has also been added to show and test this capability. Currently, this test passes on mean flux, but fails on standard deviation (see #231).

[ilhamv]

Thanks, Alex.

How did you generate the answer.h5 with which the test passes on the mean value but not the stdev?

[alexandermote]

Thanks, Alex.

How did you generate the answer.h5 with which the test passes on the mean value but not the stdev?

I ran the input with domain decomposition turned off. Everything else in the input should be the same.

[ilhamv]

I see. So the DD run does not reproduce the stdev of the non-DD run, right?

[ilhamv]

I see. So the DD run does not reproduce the stdev of the non-DD run, right?

... and have you made sure that the same issue only shows up in 3D? Does it also show up in 1D?

[ilhamv]

Would it be a good idea to make the dd_slab_reed to be tally-decomposed as well? (This would help in debugging as it gives a 1D problem variant)

[alexandermote]

dd_slab_reed is already tally-decomposed, as it’s an inherent part of domain decomposition. It passes all tests; I haven’t tested a 1D version of dd_cooper, however.

[ilhamv]

I see. We need to make sure if the issue is unique to 3D DD or not: Can you check if dd_slab_reed reproduces the non-DD run? I think you can do that by running dd_slab_reed manually without DD, saving the output as answer.h5, and then running the regression test normally. Thanks, Alex.

[alexandermote]

I see. We need to make sure if the issue is unique to 3D DD or not: Can you check if dd_slab_reed reproduces the non-DD run? I think you can do that by running dd_slab_reed manually without DD, saving the output as answer.h5, and then running the regression test normally. Thanks, Alex.

I ran the test as you described, and got the same issue. Seems like it is tied to DD in general and not 3D DD specifically. I also tested it without the tally recombination introduced in this update, and the issue still occurred, so it was not introduced by these changes.

[ilhamv]

The identified issue is not unique to this PR and will be tracked in #231.