JEKrayPro My benches
What this models, and what it does not
In three dimensions. Every free-space part at a height of its own, tipped and rolled; rays traced exactly through real surfaces in 3D, skew and rising alike; Gaussian beams through turned optics as general astigmatic beams, their higher-order modes with them; the diffracted field through tipped, rolled and off-centre optics; periscopes that turn the polarisation and the image as real ones do; and misaligned cavities, whose light walks round to their own axis and settles there. The 3D view shows it all, and moves and turns any part along all six axes.
Modelled. Exact tracing of conic and aspheric surfaces with real glass dispersion; reflection, refraction and total internal reflection; Fresnel coefficients and thin-film coatings with phase; polarisation as complex s and p amplitudes, with waveplates and polarisers; optical path, so wavefront error, interference and cavity round trips; Gaussian beam propagation along the axis, including every branch at splitters and round trips in cavities; physical optics propagation of the beam's diffracted field through the system, with each surface a ray-traced phase screen, as a round field or, where tilts and decentres break the symmetry, a full 2D one that carries the polarisation at every point (each sample through its own Fresnel, coating or waveplate matrix), with branches of one beam adding as fields, phases and polarisation included, wherever they meet on the bench or on a screen; stops (iris, slit, double slit, slit comb, knife edge or beam block, anywhere across the beam) and mask elements with any amplitude and delay you write as a formula, acting on the field and on the rays; and, zoomed in below 50 µm, a close-up that samples the field on every pixel, so standing waves and crossing-beam fringes show, with the fields added as vectors; and at each screen the field that arrives, from Gaussian beams and from ray fans taken as uniform pupils, giving its irradiance map, point spread, Strehl ratio, MTF and enclosed energy. Also diffraction gratings, each order a branch of its own; birefringent crystals (beam displacers with walk-off, Wollaston and Glan–Taylor prisms), where s and p go their own ways; Faraday rotators and isolators, non-reciprocal as they should be; filters and dichroics; cylindrical lenses and axicons; a gain medium that saturates, so a cavity shows whether it lases and, above threshold, the power it settles at and sends out of each partly transmitting mirror as a beam; atomic vapour cells (rubidium, caesium, potassium; D1 and D2) with their hyperfine, Doppler-broadened absorption and dispersion at the vapour's temperature, saturated by the light itself slice by slice along the cell and point by point across each beam (by its real profile, so a vortex pump leaves its dark core unsaturated), and saturated absorption between beams crossing them, head on or at an angle, where they overlap; and fibres and photodiodes, which take in the arriving field as a real one would and report coupled power, photocurrent, output voltage and signal to noise. Sources can have a line of their own (a coherence length, Lorentzian or Gaussian), so fringes fade as two paths grow apart (by their group paths, so glass counts by its group index) and a cavity builds up the average over the line, and a beam quality M² above 1, carried as a sum of Hermite–Gauss modes added as powers, as from a multimode laser or an LED; or they can launch a higher-order mode, Hermite–Gauss HGmn or Laguerre–Gauss LGpl (a vortex of any charge), carried as a coherent sum of Hermite–Gauss modes each with its own Gouy phase, so a cylindrical-lens mode converter works as in the lab. Mask presets cover spatial light modulator patterns: spiral phase plates, fork holograms, blazed gratings and a π step, and a mask whose detail is finer than the field's grid is flagged rather than quietly smoothed. A source can be split from another (two fibre launches off one coupler): the two are one laser, so their light interferes, with a phase and an extra length of fibre between them; separate lasers add as powers. A project can hold several benches, each on a tab of its own, joined by link planes: a pair of them is one plane seen from two benches, so beams, fields and cavities carry across. Fibre subsystems (lasers, couplers, WDMs, circulators, isolators, attenuators, polarisation controllers and polarisers, Faraday mirrors, Bragg gratings, terminators, photodiodes and feedthroughs between tabs) are joined port to port by fibres, with fibres that keep polarisation, scramble it as laid fibre does, or maintain it (PM, with a key angle and extinction ratio), multimode fibre and tips, and fibre amplifiers that saturate and add their spontaneous emission and its noise, and solved by following each path with its own phase and delay, so rings, reflections and a laser's coherence come out exactly; fibre tips take free-space light in by mode overlap and send the network's light back out as part of the same laser's launch. Short pulses: a source or fibre laser given a pulse duration traces reference rays at 33 wavelengths across its spectrum, and their optical paths through glass, prisms, gratings, chirped mirrors and fibre are the pulse's spectral phase, so a screen or photodiode shows the pulse that arrives (its duration, group delay, GDD and TOD, spatial chirp and pulse-front tilt), and prism and grating compressors, stretchers and delay lines come out without formulas. The diffracted field (on the bench, at each screen, in a cavity and in the close-up) is worked out in a background thread, so the page stays responsive while it runs.
Not modelled. Fibre, electronic and chip parts in 3D: the 3D view leaves them out (free-space parts take all six degrees of freedom: x, y, z, facing, tip and roll). Aberrations across the field are the tangential section only. Also no scatter, no thermal effects, no coating dispersion. A laser runs on one mode with a homogeneous gain line, with no spatial hole burning, relaxation or Q-switching; a vapour cell has no buffer gas or optical pumping between levels (saturated-absorption dips are where and as wide as they should be, their depths only roughly), and beams in it cross in the bench plane only; a multimode fibre's speckle is averaged away. Nonlinear optics is second-harmonic generation in poled crystals (continuous-wave, by Boyd and Kleinman's focusing) and self-phase modulation in fibre only: no Kerr lensing, sum- or difference-frequency mixing, parametric oscillation, Raman scattering or self-steepening. The pulse is taken on the traced reference rays, at the centre of the beam; fringes between two paths of a pulsed laser on the ordinary screen views use the phase path and a Gaussian line of the pulse's bandwidth (sech² spectra taken as Gaussian), so for the pulse itself use the Pulse tab. Beat notes on a free-space photodiode leave out the radio-frequency delay of the free-space path after a modulator (a phase of 2πf L/c, 2.4° per 100 mm at 20 MHz): the cavity’s round trips and fibre carry it. Power figures ignore absorption unless you set it on a material.
How far to trust it. The validation page sets JEKray against published results and independent codes, live in your browser; the list grows as each part of the model is checked. For anything that gets built, funded or published, reproduce the number in an established package first.