danielkotik / optical-beams-meep Goto Github PK

There is a related issue

• c4urself/bump2version#133 (comment)

Make a Sphinx documentation hosted on GitHub Pages.

• Check if intersphinx extension can be used to link to Meeps documentation (MkDocs), see https://gist.github.com/shimizukawa/3978232

Two step process:

1. Refactor the multiple optbeam.py files to a pure Python installable (via pip install -e .) package optbeam:

   optbeam
   ├── 2d/
   ├── 3d/
   └── __init__.py
   

2. Enable package build process for Cython using:

   python setup.py build_ext
   python setup.py install --user

   with

   optbeam
   ├── 2d/
   ├── 3d/
   ├── __init__.pxd
   └── __init__.py
   

   see:

   • https://levelup.gitconnected.com/how-to-deploy-a-cython-package-to-pypi-8217a6581f09
   • https://medium.com/swlh/distributing-python-packages-protected-with-cython-40fc29d84caf

• add repo folder structure
• introduce package optbeam and mention scripts folder
• explain necessity of Cython
• update citation section with link to correct DOI
• add link to GitHub Pages docs

There is a huge Python bottleneck due to the high number of necessary function calls of the integrand function in psi_spherical. We use Cython to tackle this problem. According to a527d2d we start our journey from an average execution time of roughly one second:

In [1]: import beamprofile                                                        

In [2]: psi = beamprofile.main()                                                  
Using MPI version 3.1, 1 processes

In [3]: %timeit psi()                                                             
Calculating inital field configuration. This will take some time...
1 s ± 58.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

Add support for incomplete Airy beams in 3d. This new class may inherit from Beam3dCartesian:

class IncAiry3d(Beam3dCartesian):
   ...

Compiler options are given by the -X flag like so

$ cythonize -3 -i -a -f -X cdivision=True boundscheck=False wraparound=False optbeam.py

Add NumPy style docstrings to optbeam library functions and classes.

Here is the output information

mpiexec -n 8 meep e_z=0 e_y=1 freq=5 LaguerreGauss3d.ctl

Using MPI version 3.1, 8 processes
command-line param: e_z=0
command-line param: e_y=1
command-line param: freq=5

start time: Fri May 10 09:57:04 2019
Gauss spectrum (cartesian): 0.6347225159862774
Gauss spectrum (spherical): 2.3981973818259543e-8

L-G spectrum (cartesian): 0.016907581994799106-0.006752644114666309i
L-G spectrum (spherical): 1.61036015580364e-24+2.6299177149278234e-8i

integrand (cartesian): 0.01237493696893364+0.0022382324370958263i
integrand (spherical): -8.908123979652118e-7+1.1204027172148073e-5i

psi (cartesian): 0.6060939710665376-1.7435922978423426i
psi (spherical): 0.5341009898668722-1.7119280296995554i

Specified variables and derived values:
n1: 1.0
n2: 1.54
chi: 45.0 [degree]
incl.: 45.0 [degree]
kw_0: 8
kr_w: 0
k_vac: 31.41592653589793
vortex charge: 2
Jones vector components: (e_z=0, e_y=1) ---> p-polarisation
degree of linear polarisation at pi/4: 0
degree of circular polarisation: 0

Initializing structure...
Working in 3D dimensions.
Computational cell is 5 x 5 x 4 with resolution 77
block, center = (2.5,-2.5,0)
size (1e+20,7.07107,1e+20)
axes (0.707107,0.707107,0), (-0.707107,0.707107,0), (0,0,1)
dielectric constant epsilon diagonal = (2.3716,2.3716,2.3716)
subpixel-averaging is 35.9121% done, 7.14225 s remaining
subpixel-averaging is 72.6905% done, 1.50484 s remaining
subpixel-averaging is 34.3034% done, 7.66376 s remaining
subpixel-averaging is 67.5178% done, 1.93616 s remaining
subpixel-averaging is 36.3081% done, 7.01777 s remaining
subpixel-averaging is 71.5025% done, 1.59486 s remaining
time for set_epsilon = 48.3139 s

Meep: using complex fields.

Meep is running but seems to stick here.

First complete the following tasks:

• API documentation (NumPy style docstrings)
• adding tests using the pytest framework
• #23

Declare spectrum amplitude functions (_f_Gauss, _f_Airy, _f_Gauss_cartesian, _f_Laguerre_Gauss_cartesian, _f_Gauss_spherical and _f_Laguerre_Gauss_spherical) as static methods.

Check if we can use ABC helper class and abstractmehod decorator along with Cython:

from abc import ABC, abstractmethod