Welcome to pyGDM¶
pyGDM is an open source python toolkit for electro-dynamical simulations. It is written in pure python and implements the Green dyadic method (GDM), a volume discretization technique, suited for single particle nano-optics simulations. pyGDM is based on simulation codes and theoretical models developed over the past 20 years by Christian Girard at CEMES (see e.g. Ch. Girard 2005 Rep. Prog. Phys. 68 1883), with contributions from G. Colas des Francs, A. Arbouet, R. Marty, C. Majorel, A. Patoux, Y. Brûlé and P.R. Wiecha. In contrast to most other coupled-dipole codes, pyGDM uses a generalized propagator, which allows to cost-efficiently treat large monochromatic problems such as angle-of-incidence scans or raster-scan simulations.
pyGDM includes tools to easily derive several physical quantities such as extinction, scattering and absorption cross-sections, far-field patterns, the electric and magnetic near-field and its multipole decomposition, the decay-rates / LDOS inside and in the vicinity of a structure, or the heat dissipated by a nanoparticle. pyGDM furthermore offers a toolkit for evolutionary optimization of nanoparticle geometries: The EO module allows to automatically design nanostructures which optimize optical properties such as a certain resonance wavelength, strong field enhancement or the direction of scattering.
You can download the code from the pypi repository or clone it from gitlab (see also the Overview section)
Please send comments, suggestions and report bugs by mail or via the gitlab issue tracker
Note
If you use pyGDM for your work, please cite at least one of our papers: (bibtex):
pyGDM documentation¶
Key features¶
Easy to install and easy to use. Since v1.1: pure python. Fully relying on freely available open-source python libraries (numpy, numba, scipy, matplotlib)
Fast: Performance-critical parts are accelerated and parallelized using numba, efficient numpy and scipy routines are used when possible. Spectra can be calculated via an MPI-parallelized routine
Enables electro-dynamical simulations including a substrate
Different illumination sources such as (oblique) plane wave, tightly focused vector beams or local dipolar emitter sources. Easy to implement own light sources in python
Efficient calculation of large monochromatic problems (raster-scan simulations, multiple polarizations / multiple angles of incidence)
Fast electron simulations (EELS, cathodoluminescence)
Provide tools to rapidly post-process the simulations and derive physical quantities such as
Many easy to use visualization tools, including animations of the time harmonic electro-magnetic fields
Evolutionary optimization of the nano-particle geometry with regards to specific optical properties
About the above animation¶
The animation at the top of this page was generated from the image
with the below script.
In the script the pyGDM function structures.image_to_struct is used to convert the bitmap pixel by pixel into a planar nano-structure: Every pixel darker than a specific threshold will be considered material (gold in the below example), all brighter pixels are part of the environment. The planar structure is then discretized on a cubic mesh using a pre-defined scaling and structure height. The time-harmonic electric field inside the gold-letter structure resulting from plane wave illumination is then calculated using pyGDM’s core.scatter and finally converted into a movie with visu.animate_vectorfield.
The structure is approximately 1100nm long, 300nm wide and 8nm high (one layer of meshpoints). It is made of gold, placed in vacuum and illuminated from your position towards the computer screen by a plane wave of 700nm wavelength, linearly polarized along 45°.
All other videos in this documentation are showing planar aluminum structures using a discretization of 5nm under 700nm wavelength plane wave illumination with -45° linear polarization. Aluminum has a far higher imaginary part of the dielectric function, which causes high losses due to absorption inside the metal. Due to these losses, the fields decay very quickly which looks “smoother”, which is why we use it in the here shown videos.
## --- load pyGDM
from pyGDM2 import structures
from pyGDM2 import materials
from pyGDM2 import fields
from pyGDM2 import core
from pyGDM2 import propagators
from pyGDM2 import tools
from pyGDM2 import visu
import matplotlib.pyplot as plt
#===============================================================
# Setup the simulation
#===============================================================
## --- structure
step = 8.0
geometry = structures.image_to_struct("pyGDM_logo_static.png",
useDarkPixel=1, threshold=100, H=1,
nm_per_pixel=1.*step, stepsize=step)
material = materials.gold()
struct = structures.struct(step, geometry, material)
## --- incident field
field_generator = fields.plane_wave
kwargs = dict(theta=45.0)
wavelengths = [700]
efield = fields.efield(field_generator, wavelengths=wavelengths,
kwargs=kwargs)
## --- vacuum environment
dyads = propagators.DyadsQuasistatic123(n1=1)
## --- simulation object
sim = core.simulation(struct, efield, dyads)
#===============================================================
# Run the simulation
#===============================================================
E = core.scatter(sim)
NF = tools.get_field_as_list_by_fieldindex(sim, 0)
#===============================================================
# create the field-animation
#===============================================================
## setup figure / axes
plt.figure(figsize=(6.0,2.5))
ax = plt.subplot(aspect='equal')
plt.axis('off')
plt.subplots_adjust(left=0, right=1, bottom=0,top=1)
## geometry
s = visu.structure(sim, scale=0.1, color='.75', show=0)
## field-animation
config_vectorfield = dict(cmin=0.5, cmap=plt.cm.Blues,
borders=5, vecwidth=0.8)
ani = visu.animate_vectorfield(NF, Nframes=100, scale=12,
kwargs=config_vectorfield,
ax=ax, show=False)
## save video to file
ani.save('pyGDM_logo.mp4', writer="ffmpeg",
codec='h264', bitrate=1500)
3D¶
Similarly, 3D visualizations of the fields can be animated using:
from pyGDM2 import visu3d
from mayavi import mlab
fig = mlab.figure( size=(600, 300), bgcolor=(1.0, 1.0, 1.0), fgcolor=(0.,0.,0.) )
## structure
visu3d.structure(sim, axis_labels=False, draw_substrate=False,
opacity=0.1, show=False)
## 3D field-animation
ani2 = visu3d.animate_vectorfield(NF, Nframes=100, scale=8,
draw_struct=False,
draw_substrate=False, substrate_size=1.1,
colormap='Blues', clim=[0.0, 0.5],
fig=fig, view=(85, -45, 350, (0,0,-15)),
ffmpeg_args="-b:v 1.5M -c:v libx264", mov_file="3D.mp4",
save_anim=True,
opacity=0.5)
which (depending on the model of course) will result in something like: