2023-06-22 Community Call

[pshriwise]

Attendees:

UW: @gonuke
ANL: @pshriwise
ORNL: Ahmad Ibrahim, Kyle Grammer
First Light Fusion: @shimwell, @rlbarker
Copenhagen Atomics: @ebknudsen
TUV: Lewis MacFarlane

[gonuke]

PyNE Updates:

• ORNL/SNS has been struggling through a PyNE installation
• UW team has been similarly updating the installation script and improving long term robustness

[KBGrammer]

The main thing we learned from that is that numpy must be later than 1.19.5 and earlier than 1.24.0. Numpy 1.20.0 is when they deprecated np.int and other similar types that were replaced by native python types. Numpy 1.24.0 (I think) is when those deprecations expired and no longer supported at all. I believe the fix requires a lot of code changes to replace np.int, np.int64, np.in32, np.bool, etc. with native python types.

Another issue we haven't solved is that we can't build nuc_data.h5 against our data directory because the following error appears:

~/$nuc_data_make --datapath ${DATAPATH} -m all -o testme.h5
Making nuc_data at testme.h5
skipping atomic mass data table creation; already exists.
skipping scattering lengths data table creation; already exists.
skipping ENSDF decay data table creation; already exists.
skipping simple XS data table creation; already exists.
Adding cinder data...
energy group boundaries.
neutron absorption cross sections.
Traceback (most recent call last):
File "${PYNEDIR}/bin/nuc_data_make", line 4, in <module>
__import__('pkg_resources').run_script('pyne==0.7.7', 'nuc_data_make')
File "${PYTHONDIR}/lib/python3.9/site-packages/pkg_resources/__init__.py", line 720, in run_script
self.require(requires)[0].run_script(script_name, ns)
File "${PYTHONDIR}/lib/python3.9/site-packages/pkg_resources/__init__.py", line 1559, in run_script
exec(code, namespace, namespace)
File "${PYNEDIR}/lib/python3.9/site-packages/pyne-0.7.7-py3.9.egg/EGG-INFO/scripts/nuc_data_make", line 6, in <module>
main()
File "${PYNEDIR}/lib/python3.9/site-packages/pyne-0.7.7-py3.9.egg/pyne/dbgen/nuc_data_make.py", line 205, in main
make_map[mo](args)
File "${PYNEDIR}/lib/python3.9/site-packages/pyne-0.7.7-py3.9.egg/pyne/dbgen/cinder.py", line 910, in make_cinder
make_mg_absorption(nuc_data, build_dir)
File "${PYNEDIR}/lib/python3.9/site-packages/pyne-0.7.7-py3.9.egg/pyne/dbgen/cinder.py", line 247, in make_mg_absorption
absorption_table = db.create_table(
File "${PYTHONDIR}/lib/python3.9/site-packages/tables/file.py", line 1022, in create_table
ptobj = Table(parentnode, name,
File "${PYTHONDIR}/lib/python3.9/site-packages/tables/table.py", line 807, in __init__
super().__init__(parentnode, name, new, filters, byteorder, _log,
File "${PYTHONDIR}/lib/python3.9/site-packages/tables/leaf.py", line 260, in __init__
super().__init__(parentnode, name, _log)
File "${PYTHONDIR}/lib/python3.9/site-packages/tables/node.py", line 249, in __init__
self._v_objectid = self._g_create()
File "${PYTHONDIR}/lib/python3.9/site-packages/tables/table.py", line 982, in _g_create
self._v_chunkshape = self._calc_chunkshape(
File "${PYTHONDIR}/lib/python3.9/site-packages/tables/leaf.py", line 321, in _calc_chunkshape
complib = self.filters.complib
AttributeError: 'numpy.int64' object has no attribute 'complib'
Closing remaining open files:testme.h5...done

[gonuke]

This appears to be a compatabilty issue between pytables and numpy.

[shimwell]

cadquery updates

two issues related to dagmc
CadQuery/cadquery#1346 round trip keeping track of volumes
CadQuery/cadquery#1243 gmsh integration

two pull requests related to dagmc
CadQuery/cadquery#1353 adds imprinting and volume tracking
CadQuery/cadquery#1343 introduces _address for in memory use

[pshriwise]

Shared surface discussion:

@ebknudsen made a couple of steps (leaps??) forward:

• the realization that using STLS files representing each surface (as opposed to volume) of the model allows us to potentially handle coincident parts
• discovered a way to insert shared surfaces into a .h5m file from CADQuery using the hashes of surfaces to determine whether or not the surface has already been meshed (tessellated). The surface's senses seem to be correct, but additional testing is underway.

The potential end result is that we would be able to represent shared/merged surfaces coming from CADQuery 🎉

Thanks for the relevant links above @shimwell!

[pshriwise]

Notes on improvements to the current overlap check tool: #797