Simple module to find numerically the propagation modes and their corresponding propagation constants of multimode fibers of arbitrary index profiles.
pyMMF is a simple module that allows finding the propagating modes of multimode fibers with arbitrary index profiles and simulates the transmission matrix for a given length. The solver can also take into account the curvature of the fiber (experimental). This code is not designed to compete with commercially available software in term of accuracy of the mode profiles/propagation constants or speed, it aims at quickly simulating realistic transmission matrices of short sections of fiber.
If the code was helpful to your work, please consider citing it:
Test pyMMF on Replicate
Download the file and execute the following command.
python setup.py installThis code is written and maintained by S. M. Popoff
I thank contributions from Pavel Gostev vongostev/pyMMF:
- Semianalytical solver has parallelized by joblib, thanks to which its performance has increased dramatically on thick fibers.
- Stability of fast radial solver increased, specifically on thick fibers and small wavelengths.
pyMMF proposes different solvers to find the propagation constants and mode profiles of multimode optical fibers.
They solve the the transverse part of the scalar propagation equation.
Ideal step-index fibers allow anlytical dispersion relations and mode profile expressions. This solver numericaly solves this relation dispersion and compute the modes using the analytical formula of the modes. It is only valid for ideal step-index fibers.
Use solver.solve(mode = 'SI', ...)
Solver for fibers with an axisymmetric index profile defined by a radial function, e.g. graded index fibers. It solves the 1D problem using the finite difference recursive scheme for Riccati's equations. It allows finding accurately and quickly the mode profiles and propagation constants for fibers when the index profiles only depends on the radial coordinate.
More details here:
Use solver.solve(mode = 'radial', ...)
It finds the modes by numerically finding the eigenvalues of the transverse operator represented as a large but sparse matrix on a square mesh. The eigenvectors represent the mode profiles and the eigenvalues give the corresponding propagation constants. The solver needs to know how many modes you want to compute, if the number set is higher than the number of propagationg modes, it will only returns the propagating modes. This solver is slower and requires finer discretisations compared to the radial solver, but it allows using arbitrary, and in particular non-axisymmetric, index profiles. It also allows introducing bending to the fiber and finding the modes of the perturbed fiber.
More detailed explanations can be found is this two-part tutorial:
Use solver.solve(mode = 'eig', ...)
Find the propagation constants of parabolic GRIN multimode fibers under the WKB (Wentzel–Kramers–Brillouin) approximation [1]_. This approximation leads to inaccurate results for modes close to the cutoff, which can be a significant proportion of the modes for typical fibers. It is provided only for comparison.
import pyMMF
import numpy as np
import matplotlib.pyplot as pltWe first set the parameters of the fiber we want to simulate.
NA = 0.275
radius = 7 # in microns
areaSize = 2.5*radius # calculate the field on an area larger than the diameter of the fiber
npoints = 2**7 # resolution of the window
n1 = 1.45
wl = 0.6328 # wavelength in microns
curvature = NoneWe first create the fiber object
profile = pyMMF.IndexProfile(npoints = npoints, areaSize = areaSize)We use the helper function that generates a parabolic index profile:
profile.initParabolicGRIN(n1=n1, a=radius, NA=NA)We then give the profile and the wavelength to the solver
solver = pyMMF.propagationModeSolver()
solver.setIndexProfile(profile)
solver.setWL(wl)The solver needs to know how many modes you want to compute. We estimate the number of modes of a GRIN multimode fiber.
NmodesMax = pyMMF.estimateNumModesGRIN(wl,radius,NA)To be safe, we ask for a bit more than the estimated number of modes previously calculated.
modes = solver.solve(nmodesMax=NmodesMax+10,
boundary = 'close',
mode = 'eig',
curvature = curvature)modes = solver.solve(mode = 'radial')Ask for the number of propagating modes found by the solver (other modes are discarded).
Nmodes = modes.numberDisplay the profile of a mode
m = 10
plt.figure()
plt.subplot(121)
plt.imshow(np.real(modes.profiles[m]).reshape([npoints]*2))
plt.subplot(122)
plt.imshow(np.imag(modes.profiles[m]).reshape([npoints]*2))Other examples are provided as notebooks in the example folder.
- solve issue with optimized (scipy bisect) radial solver (see PR #8)
- switch radial solvers:
radialcorresponds now to the corrected optimized radial solver using scipy for bisect search,radial_legacyis the old one - Store radial and azimuthal functions of the modes in the
radialsolver inmodes0.data[<ind_mode>]['radial_func']andmodes0.data[<ind_mode>]['azimuthal_func'], can be used to apply to your mesh, e.g.:
modes = solver.solve(mode='radial_test', ...)
X, Y = np.meshgrid(...)
TH = np.arctan2(Y, X)
R = np.sqrt(X**2 + Y**2)
ind_mode = 0
psi_r = modes.data[ind_mode]['radial_func'](R)
psi_theta = modes.data[ind_mode]['azimuthal_func'](TH)
plt.figure()
plt.imshow(np.real(R*TH))- in the radial solver, argument
min_radius_bcis now in units of wavelength, defaults to 4.
- Radial solver performance improvements (Pavel Gostev)
- Semi-analytical solver performance improvements (Pavel Gostev)
- Improved documentation
- Add Jupyter notebook examples
- added Ricatti solver for axisymmetric index profiles
- First public version