martinpdes / pymiesim Goto Github PK

Unittesting doesn't run without error for the self-hosted arm64 workflow.
It looks like cibuildwheel use the command pip install, and I cannot make an alias of pip (looks like it's a known bug of GH-actions).

Hi,

First, thanks so much for this great package. Very impressive work!

While trying to get myself more acquainted with the package I got a bit confused about the use of shell_width vs shell_diameter in a CoreShell particle.

shell_width is used in a CoreShell particle (see) and shell_diameter is used in a CoreShellSet (see).
I assume both refer to the thickness of the shell? However, when I look at one of your examples you set the shell_diameter to 800 nm and the core_diameter to 400 - 600 nm, suggesting shell_diameter does not refer to the thickness of the shell but to the diameter of the entire particle.

So could you clarify for me if shell_width and shell_diameter refer to the same thing and if so, if they mean the thickness of the shell?

Contrary to all other angles (gamma_offset, phi_offset, source polarization_value) which are to be given in degrees, the detector's polarization_filter is to be given in rad. This is not documented and also the default plots show it as if were degrees. I'd suggest to changing to degrees for consitency with the other angles.

The Experiment class cannot compute the detector coupling with the OpenMP parallel for because of the ndarray which, allegedly, would need to get hand on the GIL of the python interpreter. Easiest fix-up would be to just get rid of ndarray and work only with the equivalent std::vector.

Hello,
I went through the PyMieSim and it is a great job, congrats!

I installed PyMieSim after cloning the repository locally. I try to follow the SPF computation example, but it fails. See below the error message:

❯ python PyMieSim/mesh.py
Traceback (most recent call last):
  File "/home/stathaki/EPFL/Simulations/PyMieSim/PyMieSim/mesh.py", line 7, in <module>
    from PyMieSim.binary.Fibonacci import FIBONACCI
ModuleNotFoundError: No module named 'PyMieSim.binary.Fibonacci'

Also from inside an ipython session:

In [1]: from PyMieSim.scatterer import Sphere
---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
Cell In[1], line 1
----> 1 from PyMieSim.scatterer import Sphere

File ~/EPFL/Simulations/PyMieSim/.direnv/python-3.10.12/lib/python3.10/site-packages/PyMieSim/scatterer.py:10
      7 from tabulate import tabulate
      9 from PyOptik import ExpData
---> 10 from PyMieSim.mesh import FibonacciMesh
     11 from PyMieSim.source import PlaneWave
     12 from PyMieSim.representations import S1S2, FarField, Stokes, SPF, Footprint

File ~/EPFL/Simulations/PyMieSim/.direnv/python-3.10.12/lib/python3.10/site-packages/PyMieSim/mesh.py:7
      4 import numpy
      5 from dataclasses import dataclass
----> 7 from PyMieSim.binary.Fibonacci import FIBONACCI
      8 from MPSPlots.render3D import SceneList as SceneList3D
     11 @dataclass
     12 class FibonacciMesh():

ModuleNotFoundError: No module named 'PyMieSim.binary.Fibonacci'

What am I doing wrong?

Thank you

It seems that the code produce negative Qabs in some conditions.
MWE:

import numpy as np
from PyMieSim            import Material
from PyMieSim.Scatterer  import Sphere
from PyMieSim.Source     import PlaneWave
from PyMieSim.Detector   import Photodiode
from PyMieSim.Experiment import ScatSet, SourceSet, Setup, DetectorSet

scatKwargs   = { 'Diameter' : np.geomspace(6.36e-09, 10000e-9, 1500),
                 'Material' : [Material('Silver')],
                 'nMedium'  : [1] }

sourceKwargs = { 'Wavelength'   : [400e-9],
                 'Polarization' : [0]}

scatSet   = ScatSet(Scatterer = Sphere,  kwargs = scatKwargs )

sourceSet = SourceSet(Source = PlaneWave, kwargs = sourceKwargs )

Experiment = Setup(ScattererSet = scatSet,
                   SourceSet    = sourceSet)

Data = Experiment.Get(Input=['Qsca', 'Qabs'])

Data.Plot(x='diameter', Scale='log', Groupby='type')

There appears to be a runtime linking error with the windows wheel build on GH-actions: windows-latest.
This is closely related to MinGW not being in the path.
What I already tried:
- statically link winpthread (unsuccessfull)
- adding Mingw library directory to path (unsuccessfull)
- removing OpenMP::OpenMP_CXX target link from CmakeList for windows build only (successfull)

I am leaving it this way for the moment, but it is on the to-do list!