python

optimode — Python interface to OptiMode.jl

A Python package wrapping the OptiMode.jl differentiable electromagnetic mode solver via JuliaCall. It exposes the full pipeline — material dispersion models, sub-pixel dielectric smoothing, plane-wave Helmholtz eigensolves, mode analysis (group index, GVD, effective area, polarization, Kerr power corrections) and asynchronous SLURM batch sweeps — with Python/NumPy-native interfaces matching the Julia API.

Full interface documentation with usage examples: docs/python.md. The physics/mathematics of each component: docs/.

Installation

From a checkout of the OptiMode.jl repository (the package locates the Julia project automatically when used in-repo):

pip install -e python/          # or: pip install juliacall numpy && use python/ on sys.path

Requirements: Python ≥ 3.9 and a Julia ≥ 1.10 installation (found on PATH or in ~/.juliaup; override with PYTHON_JULIAPKG_EXE). The first import starts an embedded Julia runtime and loads the OptiMode packages (precompiling on first use). Outside a checkout, point OPTIMODE_JULIA_PROJECT at a Julia project where OptiMode is installed.

Quick start

import optimode as om

grid = om.Grid(4.0, 3.0, 128, 96)                       # 4×3 μm cell
mat_vals = om.f_eps_mats([om.Si3N4, om.SiO2])([1/1.55]) # (ε, ∂ωε, ∂²ωε) at ω = 1/λ
core = om.box([0, 0], [1.6, 0.8], 1)                    # Si₃N₄ core, SiO₂ background
sm = om.smooth_eps([core], mat_vals, (1, 2), grid)      # sub-pixel smoothing
eps_inv, deps, ddeps = om.inv_eps_slices(sm)

kmags, evecs = om.solve_k(1/1.55, eps_inv, grid, nev=2) # eigenmodes at fixed ω
neff = kmags[0] * 1.55
ng, gvd = om.ng_gvd(1/1.55, kmags[0], evecs[0], eps_inv, deps, ddeps, grid)

# Kerr: power-dependent effective index (n₂ from the material library)
n2_map = om.smooth_scalar([core], [om.kerr_n2(om.Si3N4), om.kerr_n2(om.SiO2)], (1, 2), grid)
res = om.solve_k_kerr(1/1.55, 1.0, eps_inv, deps, n2_map, grid)   # P = 1 W
dneff = (res["kmags"][0] - res["kmags_lin"][0]) * 1.55

A complete runnable example is in examples/si3n4_waveguide_kerr.py; tests in tests/test_optimode.py exercise the full API surface (run with pytest python/tests).

Units

Same as OptiMode.jl (c = 1): lengths/wavelengths in μm, frequencies ω = 1/λ in μm⁻¹, propagation constants k = n_eff/λ in μm⁻¹, powers in W, Kerr coefficients in μm²/W.