Python and C++ split step method simulation library for modeling propagation in nonlinear optical processes and for Gaussian quantum optics.
The code currently supports ꭓ⁽²⁾ and ꭓ⁽³⁾ processes, with (3+1)-dimensional propagation. This includes dispersion/diffraction and nonlinear effects, for arbitrary pump and signal shapes. It is straightforward to add differential equations with an arbitrary number of signal and pump modes.
The nonlinearmedium module contains the classes for simulating optical propagation based on given material and optical-field parameters.
The simulations solve the dimensionless propagation equations.
Green's functions can be calculated with linear equations.
nonlinearmedium is a compiled C++ library meant to be imported and used in Python.
To compile, it requires Pybind11 for Python binding and Eigen for vectorized operations.
It will also compile with fftw if it is found.
The program is written using the curiously recurring template pattern (CRTP) to efficiently implement as many differential equation solvers as one can dream of.
The solvers are implemented and described in the solvers/ directory, and registered in solvers/RegisteredSolvers.hpp.
There is a repository containing a collection of Jupyter notebooks that may serve as useful examples, test for correct behavior, or reproduce some published results.
The classical, poling and multimode Python modules provide functions for configuring simulations and analysis routines, for example:
- Converting to dimensionless quantities,
- Generating poling patterns,
- Calculating covariance matrices and related quantities.
The decompositions module is borrowed from Strawberry Fields with minor improvements and provides good implementations of matrix decompositions.
The materials library is built using Symbolic Python (Symengine or SymPy), and can evaluate Sellmeier equations as a function of wavelength and temperature.
This allows easy calculation of the index, group index, group velocity dispersion, or wavenumber β coefficients up to 3rd order
(ie k = β₀ + β₁ Δω + β₂ Δω² + β₃ Δω³ …).
It is straightforward to add materials: only a Sellmeier equation is required, Python decorators take care of the rest.
To compile nonlinearmedium, create a build directory and, from inside, run cmake .. -DCMAKE_BUILD_TYPE=Release and make -jX (replacing X with some number of compilation cores, eg -j4).
The binary will be created in the main directory, and can be imported like a regular Python module.
On some systems, the compiled binary file name contains additional text, and must be renamed in order for Python to recognize the library as named.