pyLaserPulse is a comprehensive simulation toolbox for modelling polarization-resolved, on-axis laser pulse propagation through nonlinear, dispersive, passive, and active optical fibre assemblies, stretchers, and compressors. pyLaserPulse was developed with rapid prototyping of pulsed fibre systems in mind, and includes: full integration of the Giles model [1] with the polarization-resolved generalized nonlinear Schrödinger equation; an extensive catalogue of preset models of off-the-shelf components, material reflectivites, Raman response functions and rare-earth spectroscopic data; base types for modelling custom components; and a flexible API.
pyLaserPulse can be used to build highly accurate models of any (quasi-) single-mode, solid-core fibre system producing output pulses with durations ranging from just a few optical cycles to multiple nanoseconds [2]. This includes fibre amplifiers, mode-locked fibre lasers, and supercontinuum generation.
- pyLaserPulse includes:
- Models of single and double clad step-index [3] and photonic crystal [4] passive and active fibres.
- Models of free-space and fibre-coupled components.
- A polarization-resolved generalised nonlinear Schrödinger equation (GNLSE) solver.
- Fourth-order Runge-Kutta interaction picture method for integration (RK4IP) [5].
- Conservation quantity error method adaptive step sizing [6].
- Full dispersion and propagation loss.
- Self-steepening, self-phase modulation, four-wave mixing (co- and cross-polarized), cross-phase modulation between polarization and wavelength components, and the full polarization-resolved Raman response [7].
- The Giles model for gain and amplified spontaneous emission (ASE) [1].
- Full integration of gain with the GNLSE for accurate modelling of highly nonlinear amplifiers.
- Full boundary value solver for co and counter propagating pump and ASE light as well as a single-pass solver for rapid estimation of amplifier performance.
- Core- and cladding-pumped fibre amplifiers.
- A choice of frequency-domain or hybrid time- and frequency-domain operators.
- Unlimited pump and ASE channels.
- Dispersion control.
- Quantum noise contributions including shot noise [11] and partition noise at partially reflective interfaces [12].
- A comprehensive plot library with full Matplotlib functionality.
pyLaserPulse uses python loops sparingly, relying instead on the numpy and scipy modules for most calculations. Simulation execution times are therefore reasonable, and a typical nonlinear amplifier simulation including the full GNLSE with co- and counter-propagating pump and ASE usually completes in less than 5 s on a modern laptop CPU. Models of systems comprising multiple amplifier stages, compression, and supercontinuum generation complete in a few minutes (although systems which have large stretching ratios, such as fibre CPAs, can require more time).
pyLaserPulse isn't on PyPI (yet), but it is easy to set up using the following steps:
1) Either download the code as a zip file, or install git and clone this repository using:
git clone https://github.com/jsfeehan/pyLaserPulse.git
2) Run the following command in the same directory into which pyLaserPulse has been downloaded/cloned:
python setup.py install
(Include the --user flag if you do not have admin or root privileges).
Manufacturers and suppliers mentioned in the example code are not affiliated with pyLaserPulse.
The example below shows how to model a simple Yb-doped fibre amplifier comprised of off-the-shelf components using the catalogue_components module. A hybrid isolator/WDM precedes a 1 m length of step-index Yb-doped fibre which is core pumped in the signal propagation direction at 916 nm. The amplifier is seeded with a 150 fs, sech2 input pulse with a peak power of 150 W and 30 dB polarization purity.
from pyLaserPulse import grid
from pyLaserPulse import pulse
from pyLaserPulse import optical_assemblies
from pyLaserPulse import single_plot_window
import pyLaserPulse.catalogue_components.active_fibres as af
import pyLaserPulse.catalogue_components.fibre_components as fc
#############################################
# Choose a directory for saving the data. #
# Leave as None if no data should be saved. #
#############################################
directory = None
############################################################
# Set time-frequency grid, pulse, and component parameters #
############################################################
# Time-frequency grid parameters
points = 2**9 # Number of grid points
central_wl = 1030e-9 # Central wavelength, m
max_wl = 1200e-9 # Maximum wavelength, m
# Laser pulse parameters
tau = 150e-15 # Pulse duration, s
P_peak = [150, .15] # [P_x, P_y], W
f_rep = 40e6 # Repetition frequency, Hz
shape = 'sech' # Can also take 'Gauss'
# isolator-WDM parameters
L_in = 0.2 # input fibre length, m
L_out = 0.2 # output fibre length, m
# Yb-fibre parameters
L = 1 # length, m
ase_points = 2**8 # number of points in pump & ASE grid
ase_wl_lims = [900e-9, max_wl] # wavelength limits for ASE grid
bounds = {'co_pump_power': 1, # co-pump power, W
'co_pump_wavelength': 916e-9, # co-pump wavelength, m
'co_pump_bandwidth': 1e-9, # co-pump bandwidth, m
'counter_pump_power': 0} # counter-pump power, W
##############################################################
# Instantiate the time-frequency grid, pulse, and components #
##############################################################
# Time-frequency grid defined using the grid module
g = grid.grid(points, central_wl, max_wl)
# pulse defined using the pulse module
p = pulse.pulse(tau, P_peak, shape, f_rep, g)
# Opneti isolator/WDM hybrid component from the catalogue_components module.
iso_wdm = fc.Opneti_PM_isolator_WDM_hybrid(g, L_in, L_out, g.lambda_c)
# Nufern PM-YSF-HI-HP defined using the catalogue_components module
ydf = af.Nufern_PM_YSF_HI_HP(g, L, p.repetition_rate, ase_points, ase_wl_lims,
bounds, time_domain_gain=True)
################################################################
# Use the optical_assemblies module for automatic inclusion of #
# coupling loss between components and for generating plots. #
################################################################
component_list = [iso_wdm, ydf]
amp = optical_assemblies.sm_fibre_amplifier(
g, component_list, plot=True, name='amp 1', high_res_sampling=100,
data_directory=directory, verbose=True)
######################
# Run the simulation #
######################
p = amp.simulate(p)
##########################################################
# Use the matplotlib_gallery module to display the plots #
##########################################################
if amp.plot:
plot_dicts = [amp.plot_dict]
single_plot_window.matplotlib_gallery.launch_plot(plot_dicts=plot_dicts)The optical_assemblies module used in the example above provides an easy way of turning a collection of components into an assembly with coupling losses inserted automatically between them, manages high-resolution field sampling options and data saving, prints information to the terminal, and also generates plots which can be displayed using the single_plot_window.matplotlib_gallery module. The latter includes a Qt-based gallery of plot thumbnails which, when clicked, are launched in a standard Matplotlib figure for full control over the plot visuals, axis limits, scaling, etc., as shown in the animation below.
All components have a verbose keyword argument, but this is overridden by the same keyword argument used when instantiating classes in the optical_assemblies module. When True, information about the progress of the simulation is printed to the terminal. The output for this example is given below. The amplifier name is printed, as well as the name of each component, the percentage propagation for the input and output fibres of each component, and the convergence of the boundary value solver for the active fibre.
Simulating amp 1
--------------------
Opneti_PM_isolator_WDM_hybrid
100.0 %
100.0 %
fibreToFibreCoupling
Nufern_PM_YSF_HI_HP
Convergence error (spectral density only):
55.38206546671114
12.000573541512274
3.46398001489616
0.9014605622333454
0.24244909411684906
0.06411436939453781
Convergence error (full field):
16.93329516050609
0.9818041820463349
0.19351727715062822
0.07243049202738491The code below models supercontinuum generation in PCF and compression of the spectrally-broadened pulses using a grating-based compressor. 100 fs, 5 kW seed pulses with a central wavelength of 1050 nm first pass through a free-space isolator (modelled using the base_components.component class) before being coupled into the PCF (modelled using the catalogue_components.passive_fibre.NKT_NL_1050_NEG_1 class). The spectrally broadened and chirped pulses then propagate through the compressor (modelled using the base_components.grating_compressor class). The optimize keyword argument is True when the compressor is instantiated, so the compressor will be optimized for maximum pulse peak power by adjusting the grating angle and separation (diffraction efficiency vs. angle of incidence is incorporated automatically). The verbose keyword argument of the compression optical assembly was also True, so the results of this optimization will be printed to the terminal.
from pyLaserPulse import grid
from pyLaserPulse import pulse
from pyLaserPulse import base_components
from pyLaserPulse import data
from pyLaserPulse import optical_assemblies
from pyLaserPulse import single_plot_window
import pyLaserPulse.catalogue_components.passive_fibres as pf
#############################################
# Choose a directory for saving the data. #
# Leave as None if no data should be saved. #
#############################################
directory = None
############################################################
# Set time-frequency grid, pulse, and component parameters #
############################################################
# Time-frequency grid parameters
points = 2**14 # Number of grid points
central_wl = 1050e-9 # Central wavelength, m
max_wl = 8000e-9 # Maximum wavelength, m
# Laser pulse parameters
tau = 100e-15 # Pulse duration, s
P_peak = [5000, 25] # [P_x, P_y], W
f_rep = 40e6 # Repetition frequency, Hz
shape = 'Gauss' # Can also take 'sech'
# ANDi photonic crystal fibre parameters
L_beat = 1e-2 # polarization beat length (m)
L = 1 # length, m
# grating compressor parameters
loss = 0.04 # percent loss per grating reflection
transmission = 700e-9 # transmission bandwidth
coating = data.paths.materials.reflectivities.gold
epsilon = 1e-1 # Jones parameter for polarization mixing and phase
theta = 0 # Jones parameter for angle subtended by x-axis
crosstalk = 1e-3 # polarization crosstalk
beamsplitting = 0 # Useful for output couplers, etc.
l_mm = 600 # grating lines per mm
sep_initial = 1e-2 # initial guess for grating separation
angle_initial = 0.31 # initial guess for incidence angle, rad
##############################################################
# Instantiate the time-frequency grid, pulse, and components #
##############################################################
# Time-frequency grid defined using the grid module
g = grid.grid(points, central_wl, max_wl)
# pulse defined using the pulse module
p = pulse.pulse(tau, P_peak, shape, f_rep, g)
# isolator
iso = base_components.component(
0.2, 250e-9, g.lambda_c, epsilon, theta, 0, g, crosstalk, order=5)
# ANDi photonic crystal fibre - NKT NL-1050-NEG-1 - from catalogue_components
pcf = pf.NKT_NL_1050_NEG_1(g, L, 1e-6, L_beat)
# grating compressor defined using the base_components module
gc = base_components.grating_compressor(
loss, transmission, coating, g.lambda_c, epsilon, theta, beamsplitting,
crosstalk, sep_initial, angle_initial, l_mm, g, order=5, optimize=True)
################################################################
# Use the optical_assemblies module for automatic inclusion of #
# coupling loss between components and for generating plots. #
################################################################
scg_components = [iso, pcf]
scg = optical_assemblies.passive_assembly(
g, scg_components, 'scg', high_res_sampling=100,
plot=True, data_directory=directory, verbose=True)
compressor_components = [gc]
compression = optical_assemblies.passive_assembly(
g, compressor_components, 'compressor', plot=True,
data_directory=directory, verbose=True)
######################
# Run the simulation #
######################
p = scg.simulate(p)
p = compression.simulate(p)
##########################################################
# Use the matplotlib_gallery module to display the plots #
##########################################################
if scg.plot or compression.plot:
plot_dicts = [scg.plot_dict, compression.plot_dict]
single_plot_window.matplotlib_gallery.launch_plot(plot_dicts=plot_dicts)A few plots from this simulation are shown below. The development of the pulse and power spectral density over the length of ANDi PCF are shown in the top row (left and right, respectively), and the strongly-chirped pulse at the ANDi PCF output and the femtosecond pulse after the compressor are shown in the bottom row (left and right, respectively).
This information regarding the compressor optimization and the optimized compressor setup is as follows:
Optimizing the compressor
-------------------------
Convergence reached: True
Optimization info.: ['requested number of basinhopping iterations completed successfully']
Number of optimization iterations: 10
Pulse compression data
----------------------
Grating separation: 4.998 mm
Incident angle: 24.528 degrees.
Pulse peak power with respect to peak power of transform limit:
Before compressor: .98 %
After compressor: 39.64 %Documentation is still in progress, but a detailed module reference can be found at https://pylaserpulse.readthedocs.io. The docstrings can also be accessed in the usual way (e.g., help(pyLaserPulse.grid.grid)). There are also example scripts which show how to model active and passive fibre systems and give a general overview of how to use pyLaserPulse (see pyLaserPulse/examples).
This repository is for the active development of pyLaserPulse. If you would like to contribute a new feature or bugfix, or to report an issue, please submit your contribution to this repository using GitHub's pull request and issue tracking features.
For anything else, please send an email to pylaserpulse@outlook.com.
pyLaserPulse has benefitted from information shared freely and generously by others who have published open-source code and research findings, contributed to discussions, and provided feedback. Because of this, pyLaserPulse has been released under a GPLv3 license in the hope that it will be useful for the ultrafast laser community. This license does not allow for the inclusion of a clause stipulating that this project must be cited, but do please consider adding the following citation if pyLaserPulse has been useful for your work:
Feehan, J. S. (2022). pyLaserPulse (Version 0.0.0) [Computer software]. https://github.com/jsfeehan/pyLaserPulse
By doing so, you will be encouraging others to get involved in this project, which will result in better, more feature-rich (and free!) software.
| [1] | (1, 2) C R Giles and E Desurvire, "Modeling erbium-doped fiber amplifiers", Journal of Lightwave Technology 9(2), pp 271-283 (1991) |
| [2] | Longer pulses can be modelled using pyLaserPulse but if a wide wavelength span is also required then you will need to use a large number of grid points, which will increase the execution time. For nanosecond-pulse to continuous-wave fibre laser and amplifier simulations requiring a broad wavelength grid, there is also the pyFiberAmp library (rate equation & boundary value solver with Raman; no GNLSE). |
| [3] | D Gloge, "Weakly guiding fibers", Applied Optics 10(10), pp 2252-2258 (1971) |
| [4] | K Saitoh and M Koshiba, "Empirical relations for simple design of photonic crystal fibers", Optics Express 13(1), pp 267-274 (2005) |
| [5] | J Hult, "A fourth-order Runge-Kutta in the interaction picture method for simulating supercontinuum generation in optical fibers", Journal of Lightwave Technology 25(12), pp. 3770-3775 (2007) and https://freeopticsproject.org |
| [6] | A M Heidt, "Efficient adaptive step size method for the simulation of supercontinuum generation in optical fibers", Journal of Lightwave Technology 27(18), pp. 3984-3991 (2009) and https://freeopticsproject.org |
| [7] | S Trillo and S Wabnitz, "Parametric and Raman amplification in birefringent fibers", Journal of the Optical Society of America B 9(7), pp. 1061-1082 (1992) |
| [8] | R L Fork, C H Brito Cruz, P C Becker, C V Shank, "Compression of optical pulses to six femtoseconds by using cubic phase compensation", Optics Letters 12(7), pp 483-485 (1987) |
| [9] | F Kienle, "Advanced high-power optical parametric oscillators synchronously pumped by ultrafast fibre-based sources", PhD Thesis, University of Southampton, 2012 (see page 37). |
| [10] | R Casini and P G Nelson, "On the intensity distribution function of blazed reflective diffraction gratings", Journal of the Optical Society of America A 31(10), pp 2179-2184 (2014) |
| [11] | P Drummond and J F Corney, "Quantum noise in optical fibers. I. Stochastic equations", Journal of the Optical Society of America B 18(2), pp 139-152 (2001) |
| [12] | B Huttner and Y Ben-Aryeh, "Influence of a beam splitter on photon statistics", Physical Review A 38(1), pp 204-211 (1988) |