oxiphoton — Pure Rust Computational Photonics Framework

A comprehensive, pure-Rust computational photonics and optical simulation framework covering FDTD, BPM, mode solving, silicon photonics, fiber optics, quantum photonics, and more — with zero C/Fortran dependencies.

Overview

oxiphoton is a full-stack photonics simulation library written entirely in Rust. It provides production-grade implementations of the core algorithms used in modern photonic design and research: Finite-Difference Time-Domain (FDTD) in 2D and 3D, the Beam Propagation Method (BPM), Transfer Matrix / S-matrix methods, Rigorous Coupled-Wave Analysis (RCWA), finite-difference and FEM mode solvers, and geometric ray tracing — alongside a rich material database, type-safe unit system, and device library for silicon photonics.

The library is organized into 73 public modules spanning over 446 Rust source files (~124 K lines of implementation). It is part of the COOLJAPAN ecosystem and depends only on pure-Rust crates: linear algebra via oxiblas, FFT via oxifft, and standard scientific helpers. There are no C, Fortran, or unsafe-heavy native bindings in the default build.

With 3,908 passing tests and 11 runnable examples, oxiphoton is designed for correctness and ergonomics. The API is strongly typed throughout — wavelengths, frequencies, refractive indices, field components, and geometric quantities are all distinct types with compile-time unit safety and automatic conversions.

Status (v0.1.0 — 2026-03-09)

Key Features

• FDTD 2D/3D — Yee-cell engine, CPML absorbing boundaries, Bloch periodic BCs, dispersive media (Drude, Lorentz ADE), nonlinear FDTD (Kerr, Raman, SHG), anisotropic and thermal extensions, sweep infrastructure, field/flux/DFT/far-field monitors
• Frequency-domain solvers — Transfer Matrix Method (TMM), S-matrix eigenmode expansion, RCWA for gratings, finite-difference BPM
• Mode solvers — 2D finite-difference, FEM, coupled-mode theory, effective-index method
• Silicon photonics device library — waveguides (single-mode, multimode, tapered), directional couplers, ring/disk resonators, MZI modulators, plasma-dispersion modulators, photodetectors (PIN, avalanche), AWG, coupled-resonator chains
• Fiber optics — step-index, graded-index, photonic-crystal fibers; NLSE split-step solver; Bragg gratings; soliton propagation; supercontinuum generation; distributed sensing
• Photonic crystals — 2D/3D band structure, defect modes, DOS, slab PHC, topological PHC
• Quantum photonics — Fock states, boson sampling, Hong-Ou-Mandel effect, linear optical circuits, Jaynes-Cummings model, photon statistics, entanglement measures, quantum memory
• Nonlinear optics — nonlinear crystals (SHG, DFG, OPA), nonlinear microscopy (SHG/CARS imaging), ultrafast pulse characterization (FROG, SPIDER, autocorrelation, pulse shaping)
• Dispersive materials — Sellmeier, Drude, Drude-Lorentz, Cauchy, Brendel-Bormann, critical-point models; built-in database for Si, SiO2, Si3N4, TiO2, Au, Ag, Al, GaAs, InP, graphene, and more
• Specialized domains — plasmonics, metamaterials, metasurfaces/metalenses, topological photonics, polarimetry (Jones/Mueller), thin films, adaptive optics, structured light, frequency combs, laser rate equations, nano-lasers, polaritonics, THz photonics, X-ray optics
• Applied photonics — solar cell absorption (AM1.5), WDM interconnects, optical networks, free-space optical comms, optical CDMA, PIC design and simulation
• Sensing and measurement — optical coherence tomography (OCT), metrology, biophotonics, photoacoustics, photonic sensors, beam quality (M² analysis)
• I/O — Touchstone (.sNp), GDS layout, VTK, HDF5, Lumerical, OxiRS, STL
• Inverse design — adjoint gradient method, fabrication constraints, parametric/shape/topology optimization
• Performance — optional Rayon parallelism (parallel feature), SIMD acceleration (simd feature), GPU compute via wgpu (gpu-wgpu feature)

Module Overview

Feature Flags

Installation

Add to your Cargo.toml with default features:

[dependencies]
oxiphoton = "0.1"

To enable specific optional features:

[dependencies]
oxiphoton = { version = "0.1", features = ["parallel", "simd"] }

To use only a minimal subset:

[dependencies]
oxiphoton = { version = "0.1", default-features = false, features = ["smatrix", "materials-builtin"] }

Quick Start

Unit conversions and physical constants

use oxiphoton::prelude::*;

// Construct wavelength and convert to frequency and wave number
let wl = Wavelength::from_nm(1550.0);          // telecom C-band
let freq = wl.to_frequency();                   // Frequency in Hz
let k0 = wl.to_wavenumber();                    // vacuum wave number in rad/m

println!("λ = {:.1} nm", wl.as_nm());
println!("f = {:.4e} Hz", freq.0);
println!("k₀ = {:.4e} rad/m", k0.0);
println!("c = {:.6e} m/s", SPEED_OF_LIGHT);

Refractive index and material properties

use oxiphoton::prelude::*;

// Complex refractive index of a lossy material
let gold = RefractiveIndex::new(0.55, 11.5);    // Au at ~1550 nm
let eps = gold.to_permittivity_scalar();
let alpha = gold.absorption_coefficient(1550e-9);

println!("ε = {:.3} + {:.3}i", eps.re, eps.im);
println!("α = {:.4e} m⁻¹", alpha);

// Lossless dielectric
let sio2 = RefractiveIndex::real(1.444);
println!("n(SiO2) = {}", sio2.n);

Transfer Matrix Method — antireflection coating

use oxiphoton::prelude::*;

// Quarter-wave SiO2 AR coating on Si at 1550 nm
let layers = vec![
    Layer::from_boxed(Box::new(Sellmeier::sio2()), 268e-9),  // λ/4 in SiO2
];

let wavelengths: Vec<Wavelength> = (1400..=1700)
    .step_by(10)
    .map(|nm| Wavelength::from_nm(nm as f64))
    .collect();

let results = TransferMatrix::spectrum(
    &layers,
    RefractiveIndex::real(1.0),    // air
    RefractiveIndex::real(3.48),   // Si substrate
    &wavelengths,
    Angle(0.0),
    Polarization::TE,
);

for (wl, r) in wavelengths.iter().zip(&results) {
    println!("{:.0} nm  R={:.4}  T={:.4}", wl.as_nm(), r.reflectance, r.transmittance);
}

Examples

The examples/ directory contains runnable demonstrations:

Run any example with:

cargo run --example ring_resonator

Architecture

oxiphoton is built on the COOLJAPAN pure-Rust ecosystem:

oxiblas 0.1.2  (pure-Rust BLAS/LAPACK)
oxifft  0.1.3  (pure-Rust FFT)
    └── oxiphoton 0.1.0

Core dependencies:

There are zero C or Fortran dependencies in the default feature set. All numerical kernels — matrix factorization, FFT, eigensolvers — are implemented in pure Rust via the COOLJAPAN ecosystem crates.

Performance

• Parallelism: Enable the parallel feature to activate Rayon-based threading across FDTD time steps, mode solver iterations, and spectrum sweeps. Scales linearly with core count on large grids.
• SIMD: The simd feature enables explicit SIMD intrinsics in Yee-cell update loops (x86-64 AVX2 and ARM NEON).
• GPU: The gpu-wgpu feature offloads FDTD field updates to the GPU via wgpu compute shaders, targeting both Vulkan and Metal backends.
• Benchmarks: Three Criterion benchmark suites cover 2D FDTD throughput, TMM spectral sweep, and mode solver convergence. Run with cargo bench.

License

Licensed under the Apache License, Version 2.0.

Copyright 2026 COOLJAPAN OU (Team Kitasan). All rights reserved.

Repository: https://github.com/cool-japan/oxiphoton Documentation: https://docs.rs/oxiphoton

Contributing

Contributions are welcome. Please open an issue before submitting large pull requests, and ensure all tests pass (cargo test --all-features) and code is formatted (cargo fmt) before review.