A research-grade solver for nonlinear pulse propagation. Pick an example, move the sliders, and watch solitons break up, spectra explode into a supercontinuum, and pulses shift colour, all computed on our server in seconds.
Background: spectral evolution of the Dudley et al. (2006) supercontinuum benchmark, computed with this solver. Wavelength left to right, fiber length bottom to top.
Our free server sleeps when idle, so the very first run after a quiet spell can take up to a minute. After that, runs take seconds and repeated settings are instant.
A simulator is only useful if you can trust it. These examples recreate published experiments and simulations; each card shows our own computed result, and you can rerun it yourself.
A 28 fs, 10 kW pulse at 835 nm in 15 cm of photonic crystal fiber. The soliton fissions within the first centimetre; a Raman soliton walks to the red while dispersive waves form in the blue.
output 514 – 1251 nm at −40 dB Run this example →
An all-normal-dispersion fiber pumped at 790 nm with 8 nJ. No solitons form: self-phase modulation and optical wave breaking give a smooth, coherent spectrum.
output 596 – 1312 nm at −40 dB Run this example →
A chalcogenide-glass core inside silica, pumped at 1550 nm just above its 1513 nm zero-dispersion wavelength. The high nonlinearity pushes the spectrum beyond 2 µm.
output 1104 – 2078 nm at −40 dB Run this example →
A second-order soliton with third-, fourth- and fifth-order dispersion. Changing the dispersion terms decides whether the soliton radiates in the blue, the red, or both.
dispersive waves at 734 nm and 1038 nm Run this example →
For each of the four dispersion settings in Roy et al., the top panel shows the computed spectral evolution, the middle the output spectrum, and the bottom the phase-matching curve between the soliton and linear waves.
The dashed lines mark where that curve crosses zero. The spectral peaks of the emitted dispersive waves fall on those lines, an independent check that the solver gets the physics, not just the picture, right.
Every run integrates the generalized nonlinear Schrödinger equation, the standard model for ultrashort pulses in fibers and waveguides, with full dispersion and a delayed Raman response.
Taylor coefficients to any order or a full tabulated β(ω), so measured fiber data and engineered waveguides go in directly.
Instantaneous Kerr effect, frequency-dependent γ(ω) with self-steepening, and a delayed Raman response for silica or chalcogenide glass.
Blow–Wood silica, the Lin–Agrawal model with boson peak, As₂S₃ chalcogenide, or an instantaneous response to switch Raman off.
A split-step integrator estimates its own local error at each step and adjusts the step length to keep it below a set tolerance, from gentle propagation to soliton fission.
Dispersion and nonlinearity can change along the fiber from mode tables, with a correction that keeps photon number conserved through a taper.
Spectral and temporal evolution maps, output spectrograms, spectral width, peak power, pulse width, energy and photon-number conservation for every run.
Short explanations of what you see in the simulator. Each one links to an example you can run and modify.
Different colours travel at different speeds. In the normal regime red light is faster than blue; in the anomalous regime it is the other way round. The group-velocity dispersion β₂ sets the dispersion length.
LD = T₀² / |β₂|See dispersion at work →With anomalous dispersion, the Kerr effect can exactly cancel dispersive spreading. The soliton number N compares the two; an N-th order soliton breathes and recovers every soliton period z₀ = (π/2) LD.
N² = γP₀T₀² / |β₂|Run a higher-order soliton →Perturbations such as higher-order dispersion and Raman scattering break a higher-order soliton into N fundamental solitons after roughly LD/N. This is the engine of most supercontinuum sources.
Lfiss ≈ LD / NWatch a soliton fission →Intrapulse Raman scattering transfers energy from the blue to the red side of a soliton's own spectrum, so the soliton continuously moves to longer wavelengths. Shorter pulses shift much faster, roughly as T₀⁻⁴.
dω/dz ∝ −T₀⁻⁴See a Raman soliton →Near a zero-dispersion wavelength, a soliton can phase-match to linear waves on the other side of it and shed energy there as a narrow spectral peak, sometimes called optical Cherenkov radiation.
β(ω) − β(ωs) − β₁(ω − ωs) ≈ γPs/2Generate dispersive waves →In all-normal-dispersion fibers no solitons form. Self-phase modulation broadens the pulse and optical wave breaking fills in the spectrum, giving smooth, highly coherent continua for imaging and pulse compression.
coherent · flat · soliton-freeRun an ANDi fiber →The public lab uses fixed fibers and short runs. The full solver handles your own fibers, waveguides and parameter studies.
The complete Python solver on your own machine, without the limits of the public lab.
We model your design question and deliver figures, data and a written summary.
Hands-on nonlinear fiber optics for research groups, companies and courses.
Yes. All examples run free of charge. To keep the service available to everyone, parameters are limited to short runs and there is a fair-use limit on new runs per hour.
On our server. Your browser sends the example and parameter values and receives the figures and numbers back. Nothing is installed on your computer.
The solver reproduces the spectral features of published results, including Dudley et al. (2006), Heidt et al. (2011), Granzow et al. (2011) and Roy et al. (2009). Every run also reports photon-number conservation, typically within 0.2 %.
Yes. Please credit “Nonlinear Fiber Lab, optocpo (lab.optocpo.com)”.
The lab runs on a free server that sleeps after a period without visitors. Waking it up takes up to a minute; after that, runs take seconds, and settings someone has already run are returned instantly.
The public examples use fixed fibers. For your own dispersion data, geometries or longer runs, contact us about a solver licence or a custom simulation study.