Skip to content

MINT-Access/stem

Repository files navigation

stem

Version Wolfram Language License: MIT

A monorepo of Wolfram Language physics simulations, data projects, and signal processing demonstrations — each runnable from the terminal via wolframscript and producing CSV data, an animated GIF, and audio output.


Repository layout

stem/
  stem-core/        Shared library: config, sonification pipeline, PCM synthesis, file export
  pendulum/         Simple and double pendulum ODE simulation
  lorenz/           Lorenz and Rössler strange attractor simulation
  dynamical/        Logistic map and period-doubling route to chaos
  asteroids/        NASA near-Earth asteroid tracker (live API data)
  cellular/         Conway's Game of Life and Wolfram Rule 110
  signal/           Fourier analysis demonstration (chord, sweep, AM)
  quantum/          Quantum mechanics (coherent state QHO, particle-in-a-box)
  primes/           Prime number patterns (Ulam spiral, prime gap rhythm)
  images/           2D image sonification via Hilbert curve traversal (brightness, colour, HSB modes)
  relativity/       General relativity (chirp: PN binary inspiral; geodesic: Schwarzschild orbits)
  cosmology/        CMB angular power spectrum and sky map sonification
  waves/            2D wave propagation and interference (FEM simulation)
  lagrange/         Circular restricted three-body problem, Lagrange point orbits
  thermo/           Maxwell-Boltzmann distribution, ideal gas ensemble, thermal cooling, equipartition theorem
  montecarlo/       2D Ising model, Metropolis MCMC, ferromagnetic phase transition
  magnetic/         Charged particle motion: cyclotron orbits, E×B drift, magnetic mirror, multi-particle chord
  hydrogen/         Hydrogen atom: wave functions, emission spectrum, quantum cascade transitions
  bayes/            Bayesian inference: coin bias, Gaussian mean estimation, Bayes factor comparison
  scattering/       Rutherford scattering: hyperbolic trajectories, cross-section, Thomson vs Rutherford
  resonance/        Orbital resonances: Galilean 4:2:1 Laplace resonance, Kirkwood gaps, Cassini Division
  fluid/            Kármán vortex street: vortex shedding, Strouhal frequency, Reynolds sweep, flag flutter
  config/           Global config defaults (config.json)
  docs/             Workflow guides

Prerequisites

Wolfram Engine

Download the free Wolfram Engine from https://www.wolfram.com/engine/ and follow the installer for your platform. wolframscript is included in the installer on all three platforms — no separate download needed.

Activate the free licence when prompted (requires a Wolfram account).

Verify your installation:

# macOS / Linux
wolframscript -version

# Windows (Command Prompt or PowerShell)
wolframscript -version

Platform-specific audio tools

Platform Tool Install
macOS afplay Built into macOS — no install needed
Linux aplay sudo apt install alsa-utils (Debian/Ubuntu) · sudo dnf install alsa-utils (Fedora)
Linux (alt) paplay sudo apt install pulseaudio-utils
Windows Windows Media Player Built into Windows 10/11 — no install needed

Platform-specific text-to-speech (optional — STEM_SPEAK=1 only)

Platform Tool Install
macOS say Built into macOS
Linux espeak-ng sudo apt install espeak-ng · sudo dnf install espeak-ng
Windows System.Speech Built into Windows 10/11 (PowerShell 5.1+)

TTS is disabled by default. If the tool is missing the app prints a warning and continues.

NASA API key (asteroids app only — optional)

Register for a free key at https://api.nasa.gov/. The built-in DEMO_KEY works at 30 requests/hour; a personal key removes that limit.

Set the key as an environment variable before running:

# macOS / Linux — add to ~/.zshrc or ~/.bashrc to persist
export NASA_API_KEY=your_key_here

# Windows PowerShell — set for the current session
$env:NASA_API_KEY = "your_key_here"

# Windows — set permanently via System Properties → Environment Variables
# Variable name: NASA_API_KEY   Value: your_key_here

Quick start

Run any project from the stem/ root:

# Physics simulations
wolframscript -file pendulum/main.wl
wolframscript -file lorenz/main.wl

# Logistic map — period-doubling route to chaos
wolframscript -file dynamical/main.wl                                      # sweep mode, r 2.5→4.0
wolframscript -file dynamical/main.wl -- --simulation.mode=iterate         # iterate at r=3.8
wolframscript -file dynamical/main.wl -- --simulation.dynamical.preset=period3_window

# Live NASA asteroid data
wolframscript -file asteroids/main.wl                                    # last 7 days
wolframscript -file asteroids/main.wl -- 2026-01-01 2026-12-31           # full year
wolframscript -file asteroids/main.wl -- 2026-01-01 2026-06-25 Phrygian  # date range + scale
wolframscript -file asteroids/main.wl -- 2026-06-20 2026-06-26 --no-orbital-elements

# Cellular automata
wolframscript -file cellular/main.wl                                     # Game of Life, R-pentomino
wolframscript -file cellular/main.wl -- --simulation.mode=rule110

# Signal processing
wolframscript -file signal/main.wl                                       # chord (default)
wolframscript -file signal/main.wl -- --simulation.mode=sweep
wolframscript -file signal/main.wl -- --simulation.mode=am

# Quantum mechanics
wolframscript -file quantum/main.wl                                      # QHO coherent state
wolframscript -file quantum/main.wl -- --simulation.mode=box             # particle-in-a-box
wolframscript -file quantum/main.wl -- --simulation.qho.alpha=3.0        # larger coherent amplitude

# Prime number patterns
wolframscript -file primes/main.wl                                       # Ulam spiral (default)
wolframscript -file primes/main.wl -- --simulation.mode=gaps             # prime gap rhythm
wolframscript -file primes/main.wl -- --simulation.ulam.size=201         # larger spiral
wolframscript -file primes/main.wl -- --simulation.gaps.count=10000      # more primes

# Gravitational waves — chirp mode
wolframscript -file relativity/main.wl                                   # GW150914 (36+29 M☉)
wolframscript -file relativity/main.wl -- --simulation.chirp.preset gw170817
wolframscript -file relativity/main.wl -- --simulation.chirp.mass1_solar 50 --simulation.chirp.mass2_solar 50
wolframscript -file relativity/main.wl -- --sonification.chirp.time_stretch 8

# Gravitational waves — geodesic mode (Schwarzschild black hole)
wolframscript -file relativity/main.wl -- --simulation.mode geodesic    # bound orbit (default)
wolframscript -file relativity/main.wl -- --simulation.mode geodesic --simulation.geodesic.orbit_type plunging
wolframscript -file relativity/main.wl -- --simulation.mode geodesic --simulation.geodesic.orbit_type photon

# CMB power spectrum sonification
wolframscript -file cosmology/main.wl                                    # spectrum mode, simulated ΛCDM
wolframscript -file cosmology/main.wl -- --simulation.mode=sky           # sky map via Hilbert traversal
wolframscript -file cosmology/main.wl -- --simulation.cosmology.source=planck  # real Planck 2018 data

# 2D wave propagation (FEM)
wolframscript -file waves/main.wl                                        # ripple mode (circular membrane)
wolframscript -file waves/main.wl -- --simulation.mode=interference      # two-source interference pattern
wolframscript -file waves/main.wl -- --simulation.waves.wave_speed=1.5   # faster propagation
wolframscript -file waves/main.wl -- --simulation.waves.source_frequency=3.0

# Lagrange point orbits (CR3BP)
wolframscript -file lagrange/main.wl                                     # L4 libration, Sun-Jupiter
wolframscript -file lagrange/main.wl -- --simulation.mode=l5             # L5 libration
wolframscript -file lagrange/main.wl -- --simulation.mode=l1             # L1 saddle-point escape
wolframscript -file lagrange/main.wl -- --simulation.lagrange.preset=earth_moon

# 2D image sonification
wolframscript -file images/main.wl                                       # brightness mode, Gaussian test image
wolframscript -file images/main.wl -- --simulation.mode=scan_horizontal  # pedagogical scan
wolframscript -file images/main.wl -- --simulation.mode=colour           # colour mode
wolframscript -file images/main.wl -- --simulation.mode=hsb              # full HSB stereo mode
wolframscript -file images/main.wl -- --simulation.images.test_image=temperature
wolframscript -file images/main.wl -- --simulation.images.brightness_scale=log

# Statistical mechanics — Maxwell-Boltzmann distribution
wolframscript -file thermo/main.wl                                              # distribution, helium, 100K→1000K
wolframscript -file thermo/main.wl -- --simulation.mode=ensemble               # particle ensemble at 300K
wolframscript -file thermo/main.wl -- --simulation.mode=cooling                # thermal cooling 1000K→50K
wolframscript -file thermo/main.wl -- --simulation.mode=equipartition          # monatomic vs diatomic

# Monte Carlo — 2D Ising model
wolframscript -file montecarlo/main.wl                                          # sweep, T 4.0→0.5
wolframscript -file montecarlo/main.wl -- --simulation.mode=critical            # at T_c
wolframscript -file montecarlo/main.wl -- --simulation.mode=quench              # instantaneous quench

# Charged particles in a magnetic field
wolframscript -file magnetic/main.wl                                            # cyclotron orbit
wolframscript -file magnetic/main.wl -- --simulation.mode=drift                 # E×B drift
wolframscript -file magnetic/main.wl -- --simulation.mode=mirror                # magnetic mirror
wolframscript -file magnetic/main.wl -- --simulation.mode=multi                 # proton+alpha+electron chord

# Hydrogen atom — wave functions and emission spectrum
wolframscript -file hydrogen/main.wl                                            # orbitals, 2p (210)
wolframscript -file hydrogen/main.wl -- --simulation.mode=spectrum              # emission spectrum
wolframscript -file hydrogen/main.wl -- --simulation.mode=transitions           # quantum cascade
wolframscript -file hydrogen/main.wl -- --simulation.hydrogen.orbital=320       # 3d orbital

# Bayesian inference
wolframscript -file bayes/main.wl                                               # coin, θ=0.7, 100 flips
wolframscript -file bayes/main.wl -- --simulation.mode=gaussian                 # Gaussian mean
wolframscript -file bayes/main.wl -- --simulation.mode=model                    # Bayes factor

# Rutherford scattering
wolframscript -file scattering/main.wl                                          # single trajectory b=1
wolframscript -file scattering/main.wl -- --simulation.scattering.preset=backscatter
wolframscript -file scattering/main.wl -- --simulation.mode=discovery           # Thomson vs Rutherford

# Orbital resonances
wolframscript -file resonance/main.wl                                           # Galilean 4:2:1
wolframscript -file resonance/main.wl -- --simulation.mode=kirkwood             # Kirkwood gaps
wolframscript -file resonance/main.wl -- --simulation.mode=saturn               # Cassini Division

# Kármán vortex street
wolframscript -file fluid/main.wl                                               # karman, Re=150
wolframscript -file fluid/main.wl -- --simulation.mode=strouhal                 # Reynolds sweep
wolframscript -file fluid/main.wl -- --simulation.mode=flag                     # flag flutter

Each project writes outputs into its own directory:

Project Output dir File types
all twenty-one apps output/ CSV, GIF, WAV (+ PNG for signal, quantum, primes, relativity, images, cosmology, waves, lagrange, hydrogen)

Play audio:

# macOS
afplay signal/output/chord_narrative_full.wav
afplay pendulum/output/double_audio.wav
afplay lorenz/output/lorenz_audio.wav
afplay dynamical/output/sweep_audio.wav
afplay dynamical/output/iterate_audio.wav
afplay asteroids/output/asteroids_*.wav
afplay cellular/output/life_rpentomino_audio.wav
afplay quantum/output/qho_audio.wav
afplay primes/output/ulam_audio.wav
afplay primes/output/gaps_audio.wav
afplay relativity/output/chirp.wav
afplay relativity/output/gw170817.wav
afplay relativity/output/geodesic.wav
afplay cosmology/output/cmb_spectrum_audio.wav
afplay waves/output/ripple_audio.wav
afplay waves/output/interference_audio.wav
afplay lagrange/output/l4_audio.wav
afplay lagrange/output/l1_audio.wav
afplay images/output/images_brightness_audio.wav
afplay thermo/output/distribution_audio.wav
afplay thermo/output/cooling_audio.wav
afplay montecarlo/output/sweep_audio.wav
afplay montecarlo/output/critical_audio.wav
afplay magnetic/output/mirror_audio.wav
afplay magnetic/output/cyclotron_audio.wav
afplay hydrogen/output/spectrum_audio.wav
afplay hydrogen/output/transitions_audio.wav
afplay bayes/output/coin_audio.wav
afplay bayes/output/model_audio.wav
afplay scattering/output/discovery_audio.wav
afplay resonance/output/galilean_audio.wav
afplay resonance/output/kirkwood_audio.wav
afplay fluid/output/karman_audio.wav
afplay fluid/output/strouhal_audio.wav

# Linux — replace afplay with aplay
aplay signal/output/chord_narrative_full.wav
aplay pendulum/output/double_audio.wav

# Windows PowerShell — replace afplay with Start-Process wmplayer
Start-Process wmplayer signal\output\chord_narrative_full.wav
Start-Process wmplayer pendulum\output\double_audio.wav

See RELEASE_NOTES_v1.4.0.md for full app descriptions, physics notes, and listening guides.


Demo

Run all sixteen apps with their most interesting presets and collect outputs into demo/:

wolframscript -file demo.wl

With spoken announcements:

# macOS / Linux
STEM_SPEAK=1 wolframscript -file demo.wl

# Windows PowerShell
$env:STEM_SPEAK = "1"; wolframscript -file demo.wl

With NASA asteroid data (requires API key):

# macOS / Linux
NASA_API_KEY=$NASA_API_KEY wolframscript -file demo.wl

# Windows PowerShell
$env:NASA_API_KEY = "your_key_here"; wolframscript -file demo.wl

Check whether a previous demo run completed successfully:

wolframscript -file demo.wl -- --check-only

Outputs are collected in demo/ with a written report at demo/demo-report.md. A full run takes approximately four to five minutes (measured: 17/17 runs — 16 unique apps, dynamical runs twice — 4 min 13 s total, including live asteroid data).


Projects

pendulum

Solves the nonlinear pendulum ODE with NDSolve. Two modes: simple (one rod) and double (two rods, chaotic). The simple pendulum maps swing angle to an A minor pentatonic scale — each half-swing becomes one note, volume set by angular velocity. The double pendulum sonifies both rods independently in binaural stereo. See pendulum/README.md.

lorenz

Simulates the Lorenz and Rössler strange attractors. Apex events on the trajectory trigger pitched notes; spatial position controls the stereo pan. The GIF renders the growing trajectory in x-z projection with a blue→cyan→orange→red colour gradient. See lorenz/README.md.

dynamical

Sonifies the logistic map, x_{n+1} = r·x_n·(1−x_n), and its period-doubling route to chaos. sweep mode traverses r from 2.5 to 4.0, sonifying the long-term attractor at each step — the rhythm audibly doubles at r≈3.0, doubles again, and dissolves into chaos, with a distinct island of order (a clean three-note rhythm) at the period-3 window near r≈3.83, guaranteed to exist by the Li-Yorke theorem ("period three implies chaos"). iterate mode fixes r (or a named preset — fixed_point, period2, period4, period3_window, chaos) and sonifies the map's actual time evolution, transient included, as a directly countable rhythm: period-4 genuinely sounds like a four-note cycle. Four correctness checks run on every invocation, including locating the first three period-doubling bifurcation points numerically and verifying their ratio against the universal Feigenbaum constant (δ≈4.669) — a number that governs the route to chaos in any smooth one-dimensional map, not just this one. The sweep mode GIF animates the classic bifurcation diagram being drawn progressively, with a moving cursor and dashed lines marking the three named events. See dynamical/README.md.

asteroids

Fetches live close-approach data from NASA's NeoWs API. Each asteroid becomes one note — pitch reflects miss distance, timbre distinguishes hazardous from safe. The GIF shows a top-down solar system view with asteroids revealed farthest-to-closest, coloured cyan (safe) or red (hazardous). Asteroid directions are computed from Keplerian orbital elements fetched from the JPL Small Body Database. Accepts arbitrary date ranges and a musical scale argument. See asteroids/README.md.

cellular

Two cellular automata: Conway's Game of Life (2D, toroidal, B3/S23 rule) and Wolfram's Rule 110 (1D, Turing-complete). Population dynamics are mapped to pitch, pan, and volume; extinction and explosion events trigger short tone bursts. See cellular/README.md.

signal

Demonstrates the discrete Fourier transform. Three modes — chord (sum of sinusoids), sweep (linear chirp), and am (amplitude modulation) — each generate a signal, corrupt it with Gaussian noise, recover it via frequency-domain filtering, and export the three stages as WAV files plus a spoken narrative. Unlike all other apps, the WAV output is the phenomenon rather than a sonification of something else. See signal/README.md.

quantum

Simulates quantum mechanical wave-packet evolution in two exactly-solvable systems. qho mode evolves a coherent state |α⟩ in the quantum harmonic oscillator using a truncated Hermite-Gauss basis (ħ=m=1). box mode evolves an equal superposition of the ground state and first excited state in a particle-in-a-box. Both modes export an animated probability-density GIF, a 3×3 snapshot PNG, and a time-series CSV of ⟨x⟩, Var(x), and speed. Stereo pan tracks mean position, pitch encodes position variance, and volume follows |d⟨x⟩/dt|. See quantum/README.md.

primes

Visualises prime number structure in two modes. ulam mode generates a size×size grid winding the integers outward in a spiral — prime cells appear white, composites black — revealing the diagonal stripes that emerge from polynomial prime-rich progressions. The app also exports a 31×31 centre zoom with cell borders. gaps mode maps the sequence of gaps between consecutive primes to percussive audio: each prime triggers a short sine burst at a time proportional to its distance from p₁, so twin primes (gap=2) produce near-simultaneous attacks and large gaps leave audible rests. A second WAV at quarter tempo stretches the rhythm so individual gap lengths become easier to count by ear. See primes/README.md.

relativity

Two modes of general relativity simulation.

chirp — gravitational wave strain from a binary inspiral using the post-Newtonian (PN) approximation, the same analytic model behind LIGO's matched filters. The strain h(t) is literally an audio waveform: a chirp sweeping upward in frequency and amplitude, ending in an abrupt merger followed by an exponentially damped ringdown. Three preset comparison WAVs are produced automatically (gw150914, gw170817, stellar). Four physical correctness checks verify the PN formulas on each run and abort if they fail.

geodesic — test-particle and photon orbits around a Schwarzschild black hole, integrated numerically by NDSolve. Three orbit types: bound (elliptical orbit; GR periapsis precession traces a rosette), plunging (particle spirals past the event horizon in finite proper time), photon (light deflected by gravity — impact parameter controls deflection vs. capture). Each orbit is visualised as a polar plot showing the event horizon, photon sphere, and ISCO, and sonified with pitch mapped to the orbital angular frequency or gravitational blueshift depending on orbit type. See relativity/README.md.

cosmology

Sonifies the Cosmic Microwave Background (CMB) angular power spectrum — the oldest light in the universe. Two modes: spectrum traverses the angular power spectrum from ℓ = 2 to ℓ = 2000, mapping each multipole to a note so the Sachs-Wolfe plateau, the first acoustic peak near ℓ ≈ 220, and the Silk damping tail are directly audible. sky generates a flat-sky Gaussian random field from the spectrum and traverses it pixel by pixel in Hilbert curve order, making the spatial temperature pattern of the CMB sound as a correlated noise texture. Both modes accept a --simulation.cosmology.source=planck flag to fetch the actual Planck 2018 best-fit spectrum from the Planck Legacy Archive (internet required). The PNG output shows the power spectrum with peak markers. See cosmology/README.md.

waves

Solves the 2D wave equation using the finite element method (NDSolveValue on a spatial Region). Two modes: ripple places a Gaussian impulse at the centre of a circular membrane and records the wavefront as it reaches a set of listening points at increasing distances — the stereo sweep of arrivals from left to right makes the propagation speed directly audible. interference drives two coherent point sources in a rectangular tank and sweeps a listening point across the resulting fringe pattern — constructive and destructive fringes produce swells and silences in the audio. Four sanity checks (numerical stability, arrival time, causality, Dirichlet boundary) run on every execution. See waves/README.md.

lagrange

Integrates test-particle trajectories in the circular restricted three-body problem (CR3BP) in the co-rotating reference frame. Three modes: l4 and l5 place a particle near the stable triangular Lagrange points and let it librate in a tadpole pattern for six orbital periods — the bounded, quasi-periodic orbit of Jupiter's Trojan asteroids. l1 places a particle near the unstable L1 saddle point, where it escapes within one to three orbital periods along the unstable manifold. In all modes, pitch maps to instantaneous angular velocity around the barycentre, pan to x-position in the co-rotating frame, and volume to inverse distance to the nearest primary. Named presets cover Sun-Jupiter, Earth-Moon, and Sun-Earth systems. See lagrange/README.md.

images

Converts 2D images into audio via Hilbert curve traversal. The Hilbert curve visits every pixel in locality-preserving order — pixels adjacent in the audio timeline are also spatially nearby — so spatial gradients become smooth pitch sweeps and sharp edges become abrupt jumps. Four sonification modes: brightness maps grayscale intensity to frequency, logarithmically by default to match how human hearing perceives pitch (simplest, best for first-time listeners); scan_horizontal is a pedagogical raster-scan counterpart to brightness, letting a listener hear the Hilbert-curve locality benefit directly by comparison; colour maps each pixel to the nearest of nine colours ordered by position in the visible light spectrum — violet the lowest pitch, red the highest — using perceptually uniform Lab colour distance (good for categorical images); hsb encodes hue as a shared pitch and brightness as timbre (pure tone → richer harmonics) on top of a saturation-controlled amplitude (most information-dense). Three built-in scientific test images are included: a 2D Gaussian, a false-colour temperature map, and a quantum probability density. Every run opens with a short spoken introduction describing the image, mode, and mapping in use. See images/README.md and images/LISTENING_GUIDE.md for the recommended listening sequence.

thermo

Sonifies classical statistical mechanics — what temperature means at the molecular level. At thermal equilibrium, molecular speeds in an ideal gas follow the Maxwell-Boltzmann distribution, a curve with three characteristic speeds (most probable, mean, and RMS, always in that increasing order) that broadens and shifts to higher speeds as the gas heats up. Four modes: distribution sweeps temperature and synthesises the distribution curve itself as a spectral envelope — many simultaneous sine partials whose amplitudes trace f(v), so the sound literally is the distribution, not a sonification of something else; ensemble samples a fixed set of particles from that distribution and lets them exchange speeds via elastic collisions, audible as a chord that continuously reshuffles while its overall character stays fixed; cooling relaxes the gas from hot to cold following Newton's law of cooling; equipartition compares a monatomic and a diatomic gas side by side, making audible the extra heat capacity a diatomic gas gets from rotational degrees of freedom. Named gas presets cover hydrogen, helium, nitrogen, oxygen, and argon. See thermo/README.md.

montecarlo

Sonifies the 2D ferromagnetic Ising model — the canonical example of a phase transition, and one of the few exactly solved in statistical physics (Onsager, 1944). A square lattice of up/down spins is updated by the Metropolis algorithm, which samples configurations in proportion to their Boltzmann probability by occasionally accepting energetically unfavourable moves — the trick that lets the system properly explore thermal fluctuations rather than getting stuck in a local minimum. Below the critical temperature T_c ≈ 2.269 J/k the spins spontaneously align (ferromagnetic order); above it, thermal noise wins (disorder). Near T_c, magnetisation and susceptibility follow universal power laws shared by every system in the same critical-phenomena universality class, from magnets to the liquid-gas transition. Three modes: sweep descends through T_c in one continuous run; critical holds at T_c for an extended listen to scale-free fluctuation; quench drops the temperature instantaneously and follows the disorder-to-order coarsening that results. Two simultaneous sonification layers: global observables (magnetisation, energy, susceptibility) as a continuous carrier, plus a quieter Hilbert-curve scan of the spin grid itself. See montecarlo/README.md.

magnetic

Simulates and sonifies a charged particle moving under the Lorentz force, F = q(E + v×B), across four field configurations. A uniform magnetic field alone can never speed a particle up or slow it down — it only turns it — producing circular motion at the cyclotron frequency, a quantity that depends only on the particle's charge-to-mass ratio and the field strength, never on its speed; the cyclotron mode's tone literally is that frequency. Adding a perpendicular electric field produces drift mode: a steady sideways drift riding underneath the circular motion, tracing a cycloid — the same physics that makes stray fields such a problem for confining plasma in a tokamak. A magnetic field that strengthens away from a midplane creates a mirror — the principle behind the Van Allen radiation belts, where solar-wind particles bounce between reflection points near Earth's poles. Finally, multi mode sounds a proton, an alpha particle, and an electron gyrating in the same field simultaneously — a three-tone chord whose frequency ratios directly encode the particles' relative masses. See magnetic/README.md.

hydrogen

Sonifies the quantum mechanics of the hydrogen atom — the only atom whose Schrödinger equation can be solved exactly in closed form, and the foundation of all atomic physics and spectroscopy. Energy levels follow E_n = -13.6057/n^2 eV, and a transition from an excited state releases a photon whose energy is the difference — the Rydberg formula. The Balmer series (transitions down to n=2) is the visible light emitted by hydrogen, empirically discovered in 1885 and theoretically explained by Bohr in 1913; these same wavelengths appear as absorption lines in the spectra of stars across the universe, including the Sun. Three modes: orbitals sonifies |psi_nlm|^2 wave functions via Hilbert curve traversal, so an orbital's lobes and nodes become audible sweeps and silences; spectrum sonifies the full n=2..n_max emission spectrum as a chord and then a UV-to-IR sweep; transitions simulates an electron cascading down to the ground state through a sequence of random, selection-rule-respecting quantum jumps — |Delta l| = 1 — audible as a different "melody" on every run. See hydrogen/README.md.

bayes

Sonifies Bayesian inference — belief updating made directly audible. A prior distribution, combined with observed data via the likelihood, produces a posterior: the updated belief. Three conjugate-update scenarios, all exact: coin mode updates a Beta prior over a coin's bias flip by flip, the clearest possible demonstration of a broad, uncertain prior narrowing into a focused posterior as evidence accumulates; gaussian mode updates a Normal prior over an unknown mean, both narrowing and shifting in pitch toward the true value; model mode computes a Bayes factor between two competing hypotheses about a coin, with stereo position drifting toward whichever hypothesis the data favours (interpreted via the commonly-cited Jeffreys scale: anecdotal, moderate, strong, very strong evidence). The coin/gaussian posteriors are sonified with the same spectral-narrowing additive-synthesis technique as thermo/'s Maxwell-Boltzmann distribution mode — but driven by accumulating information rather than temperature. See bayes/README.md.

scattering

Simulates and sonifies Rutherford alpha-particle scattering — the 1909-1911 Geiger-Marsden experiment that discovered the atomic nucleus. J.J. Thomson's "plum pudding" model, with positive charge spread evenly through the atom, could not produce large-angle scattering; Rutherford described the observed result as "almost as incredible as if you fired a 15-inch shell at a piece of tissue paper and it came back and hit you." The only explanation was a tiny, dense, positively charged nucleus. Three modes: scatter integrates a single alpha particle's hyperbolic trajectory around the nucleus, pitch and volume rising through closest approach; distribution sonifies a realistic beam of particles as a dense, quiet stream of small-angle events punctuated by rare, loud backscatter accents — the Rutherford cross-section 1/sin^4(theta/2) played directly rather than plotted; discovery recreates the historical comparison in binaural stereo, Thomson's model (left channel, confined to under one degree) against Rutherford's (right channel, with occasional dramatic backscatter events the Thomson model structurally cannot produce). See scattering/README.md.

resonance

Sonifies orbital resonance — the phenomenon where orbiting bodies' periods lock into simple integer ratios — across three settings, in the app where planetary mechanics and music theory meet most precisely: a 2:1 orbital resonance is a musical octave. galilean mode simulates Jupiter's moons Io, Europa, and Ganymede, whose periods lock into an almost exact 4:2:1 ratio (the Laplace resonance), sonified as a rhythmic canon — a two-octave chord, C3:C4:C5, in a 1:2:4 ratio of note counts, locked in place by gravity. kirkwood mode sonifies the asteroid belt's Kirkwood gaps — sharp silences at semi-major axes where Jupiter's resonant gravitational kicks have swept the belt clear. saturn mode applies the same technique to Saturn's rings, where the Cassini Division is audible as a long silence, the same 2:1 resonance mechanism at a different scale, caused by the moon Mimas. Kepler spent much of his career searching for a "harmony of the spheres" among the planets; the Galilean moons, discovered just nine years before his death, are the genuine article he never got to see. See resonance/README.md.

fluid

Simulates and sonifies vortex shedding behind a bluff body — the Kármán vortex street — making the Strouhal frequency directly audible: the same pitch a wire, flagpole, or power line sings in the wind (an Aeolian tone). karman mode fixes the Reynolds number and sonifies the resulting periodic vortex shedding as a steady tone with alternating shed-event clicks. strouhal mode sweeps the Reynolds number from onset (Re≈47 — a genuine bifurcation from steady to periodic flow) through the laminar regime toward turbulence, audible as silence giving way to a sudden tone that gradually broadens. flag mode models a flexible flag flapping in the flow as a damped, driven oscillator — a simplified cousin of the aeroelastic flutter that destroyed the original Tacoma Narrows Bridge in 1940 — with pitch and pan following the flapping tip. The wake itself is modelled with the vortex particle method: a handful of discrete point vortices advected by mutual Biot-Savart induction, an educational approximation rather than a full Navier-Stokes solve. See fluid/README.md.


Config system

Every project uses a four-layer configuration system:

$HardcodedDefaults  →  config/config.json  →  <app>/config.json  →  CLI --key=value

Inspect the fully merged config for any project:

wolframscript -file pendulum/main.wl -- --config-dump
wolframscript -file cellular/main.wl -- --config-dump | python3 -m json.tool

Override any parameter at runtime using dot-separated key paths:

wolframscript -file pendulum/main.wl -- --simulation.mode=double
wolframscript -file lorenz/main.wl -- --simulation.mode=rossler
wolframscript -file cellular/main.wl -- --simulation.life.starting_pattern=gliderlgun
wolframscript -file signal/main.wl -- --simulation.chord.noise_level=0.8
wolframscript -file asteroids/main.wl -- --simulation.days_ahead=14
wolframscript -file quantum/main.wl -- --simulation.qho.alpha=3.0
wolframscript -file relativity/main.wl -- --simulation.chirp.mass1_solar 50

See docs/APPS.md for a full listing of each app's config keys.


stem-core

All twenty-one projects load stem-core as a shared library. It provides:

  • ConfigLoadConfig, GetCfg, DeepMerge — four-layer config merging and safe key lookup
  • Sonification pipelineSonifyTrajectory, SpatialLayer, MotionLayer, EventLayer, MixLayers, RenderAudio
  • Scale and synthScaleLookup, StemSynthNote, NormalizeBuffer, ExportAudioBuffer
  • ExportExportCSV, ExportGIF
  • AccessibilitySTEMHeading, STEMSection, STEMPrintN, STEMDescribeCSV/WAV/GIF, STEMSay
  • UtilsEnsureDir, FmtN, LogError

See stem-core/README.md for the full API and stem-core/AGENTS.md for parameter details.


Accessibility

All projects run fully headlessly and write plain WAV files. Every console output line is a self-contained announcement so screen readers (VoiceOver on macOS, Orca on Linux, Narrator on Windows) read each item cleanly without splitting numbers across lines.

To enable spoken announcements alongside normal printed output, set STEM_SPEAK=1 before running:

# macOS / Linux
STEM_SPEAK=1 wolframscript -file pendulum/main.wl
STEM_SPEAK=1 wolframscript -file signal/main.wl

# Windows PowerShell
$env:STEM_SPEAK = "1"; wolframscript -file pendulum/main.wl

Platform TTS: macOS uses say, Linux uses espeak-ng (install with sudo apt install espeak-ng), Windows uses the built-in System.Speech. If TTS is unavailable the app prints a warning and continues silently.

The signal app's {mode}_narrative_full.wav is the most accessible single output — it chains spoken text with the clean, noisy, and recovered signals so the entire Fourier demonstration can be followed by listening alone.

For the complete VoiceOver + wolframscript workflow see docs/voiceover-wolframscript-guide.md and stem-core/ACCESSIBILITY.md.


About MINT Access

MINT Access is a Swiss organisation promoting accessible STEM education. Website: mintaccess.ch (German)