Skip to content
Back to search results

Beam Propagation Method

Introduced by
Michael Feit and John Fleck
Year
1978
Type
Paraxial propagation algorithm
Family
Computational
Font size

What does it do?

The Beam Propagation Method is a computational technique for simulating the propagation of optical beams through slowly varying, weakly guiding structures.

Developed by Feit and Fleck in 1978, BPM exploits the paraxial approximation to reduce the full vector wave equation to a scalar or vector envelope equation, enabling efficient simulation of waveguides, integrated optics, and photonic devices.

The intuition behind the method

BPM is based on the paraxial approximation: the optical field varies slowly in the transverse directions and rapidly along the propagation axis, so the backward-propagating component is negligible. This allows the wave equation to be reduced to a first-order differential equation in the propagation coordinate z. The field is stepped forward through small slices of the structure, with each slice treating scattering or mode coupling. This approach is much faster than full-wave FDTD for long, slowly varying structures but less accurate for abrupt transitions or highly multimodal systems.

History

Feit and Fleck developed BPM in 1978 for graded-index optical fibers. The method was extended to waveguides by Huang and others in the 1990s. Key improvements include wide-angle operators (Padé approximants by Hadley), vectorial formulations for polarization, and semivectorial schemes for computational efficiency. BPM is now standard in integrated-optics design and photonic-device simulation, with commercial tools such as OptiBPM and Lumerical widely used.

When to use it

BPM is ideal for long, slowly varying optical structures such as graded-index fibers, tapered waveguides, photonic crystals with gentle bends, and integrated-optic devices. Avoid BPM for structures with sharp discontinuities, backward reflections, or strong multimodal coupling. Use BPM when propagation distance is much larger than transverse wavelength, and when full-wave methods (FDTD) are computationally prohibitive.

Limitations and considerations

  • Paraxial approximation breaks down for strong refraction, large-angle scattering, or highly multimodal systems
  • Cannot model backward-propagating waves or strong reflections
  • Accuracy degrades for abrupt index steps or discontinuous geometries
  • Requires fine transverse resolution in regions with rapid field variation

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Much faster than FDTD for long propagation distances due to larger step sizes
  • Lower memory and computational cost, enabling efficient design optimization
  • Straightforward treatment of inhomogeneous media and weak guidance
  • Natural handling of mode coupling in weakly guiding structures
  • Flexible operator schemes (split-step, implicit) allow trade-off between accuracy and speed

Applications

  • Optical fiber design, including single-mode, multimode, and specialty fiber characterization
  • Photonic integrated circuit and waveguide simulation, mode mapping, and coupling analysis
  • Tapered and adiabatic optical devices, mode converters, and directional couplers
  • Photonic crystal waveguide propagation and band-edge effects
  • Nonlinear optical device simulation with gain or loss sections in semiconductor amplifiers

How it works

How it works

1

Define the optical field at z=0, typically the fundamental mode profile, source distribution, or input beam.

2

Apply the paraxial approximation, factoring out the rapid oscillation e^{ikz} and solving for the slowly varying envelope.

3

Discretize the transverse field on a grid and the propagation direction into small steps Δz, using finite-difference, finite-element, or spectral methods.

4

Update the field envelope at each propagation step using an operator such as split-step Fourier, implicit finite-difference, or Crank-Nicolson scheme.

5

Extract output field, transmission, loss, mode coupling, or far-field patterns at the exit z=L.

Common pitfalls

  • Choosing propagation step Δz too large leads to numerical dispersion and amplitude errors
  • Improper boundary conditions on the transverse grid can trap energy or cause wraparound
  • Neglecting the full vector nature of the field in scalar BPM misses polarization effects
  • Assuming the paraxial approximation is valid for structures with strong guidance or wide-angle scattering

Frequently asked questions

What is the paraxial approximation, and when does it fail?

The paraxial approximation assumes that the optical field propagates primarily in one direction (the z-axis) and changes slowly in the transverse directions. Mathematically, it requires that the transverse wavenumber k_t << k_z where k_z is the axial wavenumber. It fails when the field has significant backward components (strong reflections), propagates at large angles (>20°), or interacts with very small features (<0.5λ).

How does BPM differ from FDTD, and which should I use?

FDTD solves Maxwell's equations directly in 3D without approximation but is slow for long structures. BPM uses the paraxial approximation to reduce computational cost by stepping through the structure, making it ideal for long, slowly varying devices. Use BPM for waveguides and fibers, FDTD for cavities, antennas, or structures with strong backward scattering.

What is the split-step Fourier method, and why is it popular in BPM?

The split-step Fourier method alternates between linear propagation in Fourier space and nonlinear or diffractive effects in real space. It is accurate, efficient, handles chromatic dispersion elegantly, and works well with spectral methods. It is the de facto standard for BPM implementations.

How fine does the transverse mesh need to be?

The transverse mesh must resolve the smallest feature in the field profile, typically at least 4–10 points per wavelength in the medium. For features smaller than λ/10, use finer grids. Use non-uniform grids to concentrate points near abrupt transitions or field maxima.

Related methods

Sources

  1. Feit, M. D., & Fleck, J. A. (1978). Light propagation in graded-index optical fibers. Applied Optics, 17(24), 3990-3998.

    Open DOI reference
    Copied.
  2. Huang, W. P. (1992). The finite-difference vector beam propagation method: an analysis. Journal of Lightwave Technology, 10(3), 295-305.

    Open publication reference
    Copied.
  3. Hadley, G. R. (1992). Wide-angle beam propagation using Padé approximant operators. Optics Letters, 17(20), 1426-1428.

    Open DOI reference
    Copied.