Capabilities

Computing Optical Fiber Modes Using fibermode


This code repository contains `fibermode`, a Python package for computing guided and leaky modes of optical fibers. It provides both semi-analytical closed-form solutions and numerical solutions based on finite element discretizations of the scalar Helmholtz and vector Maxwell curl-curl eigenproblems. The numerical facilities are based on finite elements and are built atop NGSolve (https://ngsolve.org). The eigensolver is the FEAST contour integral eigensolver. 		fibermode logo
Logo made by Gemini

Capabilities#

Semi-analytical methods

  • LP modes of step-index fibers via characteristic equations and Bessel functions — StepIndexExact

  • Scalar and vector leaky modes of Bragg fibers via the transfer matrix method — BraggExactScalar, BraggExactVector

Numerical methods (FEM + FEAST)

  • Guided and leaky modes (scalar and vector) of arbitrary fiber geometries, including

    • step-index fibers — StepIndex,

    • Bragg fibers — Bragg,

    • antiresonant fibers — ARF, NANF,

    • photonic bandgap fibers — PBG.

Key algorithmic features include the FEAST polynomial eigensolver with contour integration in the complex plane, a frequency-dependent PML for accurate leaky mode eigenvalues, nondimensional Z-plane formulation to avoid numerical round-off in large eigenvalue problems, confinement loss (dB/m) computed directly from complex propagation constants, and mode visualization tools.

Installation#

git clone git@github.com:jayggg/fibermode.git
cd fibermode
pip install -e .

pip install -e . installs fibermode in editable mode and pulls in its PyPI dependencies (numpy, scipy, matplotlib, sympy, cxroots) automatically. Two dependencies are deliberately left out of install_requires and must be managed yourself:

  • NGSolve, a high-performance finite element library. Most users already have NGSolve installed (conda, a source build, or an MPI/MKL-enabled build) — pip install ngsolve would fetch an unrelated plain-PyPI copy that can shadow or conflict with it, so it’s intentionally not listed. If you don’t have NGSolve yet, pip install ngsolve is fine.

  • pyeigfeast, which implements the FEAST polynomial eigensolver for NGSolve. It has no PyPI release and no setup.py, so pip cannot install it regardless — clone it and add it to your PYTHONPATH:

    git clone git@bitbucket.org:jayggg/pyeigfeast.git
export PYTHONPATH="$PWD/pyeigfeast:$PYTHONPATH"

(If you don’t want an editable install, pip install . works too, but since this is an actively developed research codebase, -e is recommended so local edits take effect without reinstalling.)

Quick start#

from fibermode import StepIndexExact, StepIndex, Bragg, BraggExactScalar

# Semi-analytical LP modes of a named fiber
f = StepIndexExact('Nufern_Yb')
betas = f.XtoBeta(f.propagation_constants(ell=0))

# Numerical guided modes (FEM + FEAST)
fiber = StepIndex(fibername='Nufern_Yb')
betas, zsqrs, modes = fiber.guidedmodes(p=3)

Further simple demos are available at the demos folder.

Documentation & Tutorial Notebooks#

Browse tutorial notebooks online. The notebooks can also be found in the docs folder. Additionally, the demos folder contain simple python scripts.

Cite#

If you use fibermode in your research or teaching, please cite it:

@misc{fibermode,
  author       = {Gopalakrishnan, Jay},
  title        = {Computing Optical Fiber Modes Using \texttt{fibermode}},
  howpublished = {\url{https://github.com/jayggg/fibermode}},
  year         = {2026}
}

License#

Write to Jay Gopalakrishnan for consultations, bug reports and suggestions. Pull requests are welcome at jayggg/fibermode. Code and documentation are released under the MIT License.

Contents