Cannot get_slice()

User Support
1

Hello!
  I am trying to get a tally slice for certain energy bin. For this
purpose I do the following:

sp3d = StatePoint('statepoint.1000.h5')
tally3d = sp3d.get_tally(scores=['fission'])
energy3d = tally3d.get_slice(filters=['energy'],filter_bins=[(1,20)])

which prints the following error message:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "build/bdist.linux-x86_64/egg/openmc/tallies.py", line 2363, in get_slice
  File "build/bdist.linux-x86_64/egg/openmc/tallies.py", line 1084, in get_values
  File "build/bdist.linux-x86_64/egg/openmc/tallies.py", line 937, in get_filter_indices
  File "build/bdist.linux-x86_64/egg/openmc/tallies.py", line 780, in get_filter_index
  File "build/bdist.linux-x86_64/egg/openmc/filter.py", line 389, in get_bin_index
TypeError: 'int' object has no attribute '__getitem__'

Is it my mistake or is it a mistake in OpenMC ? (I use OpenMC v.0.7.1)

  Thank you!
    Vladimir

2

Vladimir,

Thanks for bringing this up - the “Tally.get_slice(…)” routine should work for this use case, but the documentation is not quite correct. I will submit a patch to our code repository shortly to correct this.

In the meantime, try the following instead:

energy3d = tally3d.get_slice(filters=[‘energy’],filter_bins=[((1,20),)])

I believe the only issue is that you must put the filter bins as a “tuple of tuples” which is not correctly documented. Each tuple corresponds to all of the bins to slice for the corresponding filter in the “filters” parameter list. I recognize that this can be confusing and appears a bit clunky, but it is intended to allow one to slice a tally across multiple filters with a single function call.

For example, if one could simultaneously slice off a tally for cell filter bins 1 and 2 and energy filter bins (0, 0.625e-6) and (1e-3, 20.) with the following code:

simultaneous_slice = tally.get_slice(filters=[‘cell’, ‘energy’], filter_bins=[(1, 2), ((0, 0.625e-6), (1.e-3, 20.))])

Let me know if this helps. I apologize for the confusing documentation and will try to correct this shortly.

Best,
Will

3

Vladimir,

See this patch to the documentation to reflect the appropriate methodology for tally slicing with energy filters:

https://github.com/mit-crpg/openmc/pull/568/files

Best,
Will

4

Yes, this works. Thank you!
  However, this way it works suspiciously slowly. For example, if I do
a simple python slice like

    fis3d = tally3d.get_slice(scores=['fission'])
    energy_slice = fis3d.mean[start_val:end_val:]

it works more than 10 times faster. I am not sure about the correctness
of such approach though.

  Vladimir

5

Yes there is a lot more overhead involved with the “slice” method than a direct access into the underlying NumPy array. Tally slicing is useful if you wish to encapsulate the sliced data within a Tally object so that you can easily index the data by filter and/or nuclide and/or score which is not as easy to do with a direct access. In addition, you can use any of Tally class’ method with a sliced tally (pandas dataframes, tally arithmetic, etc.). But if you are concerned with performance then I recommend using direct access. Also, we can likely can greatly improve the performance of the Tally class methods -including “slice” - with further development, but these are such new features that we’ve had little time to do performance optimizations.