gradient calculation and visualization issue for ansys 2D

[lordlzy]

Describe the bug
issue by using "import_mesh_ansys.ipynb" for 2D ansys rst

To Reproduce
run import_mesh_ansys.ipynb with a 2D ansys rst (my example is with 4 & 3 nodes per elements)

change line: grad1=pylife_mesh.gradient.gradient_of('mises')

code that triggers the bug

import pyvista as pv
pv.close_all()
pv.set_jupyter_backend('trame')

mpl.style.use('bmh')

grid = pv.UnstructuredGrid(*pylife_mesh.mesh.vtk_data())
plotter = pv.Plotter(window_size=[1920, 1080])
plotter.add_mesh(grid, scalars=pylife_mesh.groupby('element_id')['S1'].mean().to_numpy(),
show_edges=True, cmap='jet')
plotter.add_scalar_bar()
plotter.show(jupyter_backend='ipyvtklink')

Expected result
calculation of stress gradient
and visualization

Observed result

gradient calculation not able to be finished
& visualization error:

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
Cell In[12], line 8
      4 
      5 # import profile
      6 mpl.style.use('bmh')
      7 
----> 8 grid = pv.UnstructuredGrid(*pylife_mesh.mesh.vtk_data())
      9 plotter = pv.Plotter(window_size=[1920, 1080])
     10 plotter.add_mesh(grid, scalars=pylife_mesh.groupby('element_id')['S1'].mean().to_numpy(),
     11                 show_edges=True, cmap='jet')

File d:\PE0_Repo\pe0-repo\.venv\Lib\site-packages\pylife\mesh\meshsignal.py:260, in Mesh.vtk_data(self)
    257     selection = points.index.isin(nodes)
    258     return points[selection]
--> 260 connectivity, cell_types = calc_cells()
    261 points = first_order_points(connectivity)
    262 cells = cells_with_lengths(points.index, connectivity)

File d:\PE0_Repo\pe0-repo\.venv\Lib\site-packages\pylife\mesh\meshsignal.py:252, in Mesh.vtk_data.<locals>.calc_cells()
    249     choice = count == total_num
    250     connectivity[choice] = connectivity[choice].apply(lambda nds: nds[:first_order_num])
--> 252 return connectivity, count.apply(lambda c: element_types_dict[c][1]).to_numpy()

File d:\PE0_Repo\pe0-repo\.venv\Lib\site-packages\pandas\core\series.py:4943, in Series.apply(self, func, convert_dtype, args, by_row, **kwargs)
   4808 def apply(
   4809     self,
   4810     func: AggFuncType,
   (...)   4815     **kwargs,
   4816 ) -> DataFrame | Series:
   4817     """
   4818     Invoke function on values of Series.
   4819 
   (...)   4934     dtype: float64
   4935     """
   4936     return SeriesApply(
   4937         self,
   4938         func,
   4939         convert_dtype=convert_dtype,
   4940         by_row=by_row,
   4941         args=args,
   4942         kwargs=kwargs,
-> 4943     ).apply()

File d:\PE0_Repo\pe0-repo\.venv\Lib\site-packages\pandas\core\apply.py:1422, in SeriesApply.apply(self)
   1419     return self.apply_compat()
   1421 # self.func is Callable
-> 1422 return self.apply_standard()

File d:\PE0_Repo\pe0-repo\.venv\Lib\site-packages\pandas\core\apply.py:1502, in SeriesApply.apply_standard(self)
   1496 # row-wise access
   1497 # apply doesn't have a `na_action` keyword and for backward compat reasons
   1498 # we need to give `na_action="ignore"` for categorical data.
   1499 # TODO: remove the `na_action="ignore"` when that default has been changed in
   1500 #  Categorical (GH51645).
   1501 action = "ignore" if isinstance(obj.dtype, CategoricalDtype) else None
-> 1502 mapped = obj._map_values(
   1503     mapper=curried, na_action=action, convert=self.convert_dtype
   1504 )
   1506 if len(mapped) and isinstance(mapped[0], ABCSeries):
   1507     # GH#43986 Need to do list(mapped) in order to get treated as nested
   1508     #  See also GH#25959 regarding EA support
   1509     return obj._constructor_expanddim(list(mapped), index=obj.index)

File d:\PE0_Repo\pe0-repo\.venv\Lib\site-packages\pandas\core\base.py:925, in IndexOpsMixin._map_values(self, mapper, na_action, convert)
    922 if isinstance(arr, ExtensionArray):
    923     return arr.map(mapper, na_action=na_action)
--> 925 return algorithms.map_array(arr, mapper, na_action=na_action, convert=convert)

File d:\PE0_Repo\pe0-repo\.venv\Lib\site-packages\pandas\core\algorithms.py:1743, in map_array(arr, mapper, na_action, convert)
   1741 values = arr.astype(object, copy=False)
   1742 if na_action is None:
-> 1743     return lib.map_infer(values, mapper, convert=convert)
   1744 else:
   1745     return lib.map_infer_mask(
   1746         values, mapper, mask=isna(values).view(np.uint8), convert=convert
   1747     )

File pandas/_libs/lib.pyx:2999, in pandas._libs.lib.map_infer()
-> 2999 'Could not get source, probably due dynamically evaluated source code.'

File d:\PE0_Repo\pe0-repo\.venv\Lib\site-packages\pylife\mesh\meshsignal.py:252, in Mesh.vtk_data.<locals>.calc_cells.<locals>.<lambda>(c)
    249     choice = count == total_num
    250     connectivity[choice] = connectivity[choice].apply(lambda nds: nds[:first_order_num])
--> 252 return connectivity, count.apply(lambda c: element_types_dict[c][1]).to_numpy()

KeyError: 3

Environment (please complete the following information):

• OS: windows
• How installed uv
• Version [2.2.1]

Additional context
Add any other context about the problem here.

[johannes-mueller]

Thanks for bringing this up and sorry for the delay.

Maybe you can provide a minimal ANSYS file that triggers the problem? By minimal I mean, just a few (like 10) elements. Then we can debug this.

Maybe even better would be to provide the elements exported as CSV file.

Thanks

[lordlzy]

Due to some issues its failed to upload the file, I will try with other ways.
But, what I noticed as major issue already:

1. 2D with linear elements (4 nodes + 3 nodes) / quadratic elements (8 nodes + others) in these combinations, some nodes have (id 0 x y z NaN ). As soon as there are these nodes, error happens for visualization
2. gradient not yet clarified

And pyapdl is not supported anymore by ansys. This make its hard to debug for me and make further development or get ansys support.
-As possible alternative solution, is there any example to convert pylifemesh from VTK data structure or pydpf structure. Would be very grateful to have this.

[johannes-mueller]

@DKreuter maybe you can say something about the ANSYS support?

[DKreuter]

pydpf needs Ansys License afaik. pymapdl is legacy, but last release is from Jan 26, so there is still some work on it. I have tested several Ansys element types last week without any problems. @lordlzy : Could you share the element type (e.g. shell181)? Than we can create simple test examples and check the issue.