Rogue — finite-time blow-up & rogue-wave predictor

Real-time physics-based extreme-wave forecast.
Ocean · aerospace · plasma · fiber-optics — NLS solver with mathematically rigorous blow-up detection.

---

What it does

Rogue predicts extreme wave events — rogue ocean waves (≥ 2.2× significant wave height), clear-air turbulence, plasma instabilities, and fiber-optic damage — from the breakthrough mathematics of finite-time blow-up in supercritical defocusing NLS (Merle–Raphael–Rodnianski–Szeftel, 2026 Breakthrough Prize in Mathematics).

Two surfaces: a browser dashboard (no install) and a Rust CLI (for servers, automation, streaming instruments). Both run the same symplectic split-step NLS solver — deterministic, seeded, and reproducible.

Why

• $2B+ / year in shipping damage from rogue waves
• ~100 cruise-ship deaths / year
• Offshore platform, submarine cable, and wind turbine damage
• Existing tools (WAVEWATCH III, SWAN) are spectral — they don't resolve individual waves
• CFD is accurate but far too slow for real-time
• Statistical methods (Rayleigh) systematically underestimate extremes

Finite-time blow-up detection gives a mathematically rigorous, real-time flag.

---

• [Docs] - Live site  nbsp;·- Live site  nbsp; Live site - Live site  nbsp;·  Education dashboard  ·  Production forecast
• Download binary  ·  crates.io  ·  Contributing

---

Quickstart

No Rust — download a binary

Pre-built binaries for Linux, macOS, and Windows are attached to every GitHub Release. Download, extract, and run:

chmod +x rogue
./rogue predict data/sample-buoy.csv
./rogue help

Web (browser, no build)

cd education && npm install && npm run dev

Open localhost:5173 — the home page links to the interactive education playground and the production forecast form.

From source (Rust)

git clone https://github.com/jesedv/rogue.git && cd rogue
./dev setup     # cargo fetch + npm install + wasm-pack build
./dev test      # cargo test --workspace
./dev ui        # browser dashboard at localhost:5173

CLI

./dev b                           # release build
./dev r predict sea.csv           # production forecast
./dev r predict sea.csv --json    # machine-readable JSON
./dev r akhmediev 1.0 1.0         # Akhmediev breather
./dev r peregrine 1.0             # Peregrine soliton
./dev r ocean 60.0                 # JONSWAP random sea
./dev r blowup 7.0 0.6            # supercritical blow-up
./dev r soliton 4.0               # soliton walk

Or install system-wide:

cargo install --path . --root ~/.local
rogue predict data/sample-buoy.csv --json

Production — live data intake

rogue predict reads a CSV/TSV of sea-state observations (buoys, AIS, bridge, streaming instruments) and outputs a physical, dimensionality-correct forecast per row.

Input format (t, hs, tp — optional gamma):

t,hs,tp,gamma
0,1.5,8.0,3.3
3600,2.8,9.5,3.3

Column aliases accepted: time, timestamp, Hs, significant_wave_height, Tp, peakperiod, gamma.

./dev p data/sample-buoy.csv    # human-readable report
./dev p data/sample-buoy.csv --json  # JSON for alarms / routing systems

The bridge derives carrier wavenumber k₀ = ω₀²/g (deep-water), group-velocity dispersion β = ω₀/(8k₀²) m²/s, and nonlinearity γ = ω₀k₀²/2 /m·s from Tp. Output is dimensional: crest factor in σ, significant wave height in metres.

Connecting instruments

Any buoy, bridge, AIS feed, or streaming source that writes t, hs, tp can feed the solver directly:

watch -n 60 'rogue predict /var/feed/current.csv --json >> alerts.log'

The forecast is deterministic per seed — wind back an AIS log and the alerts reproduce identically for audit.

Repository layout

crates/
  rogue-nls          # split-step NLS solver (1D/2D), FFT, diagnostics, scenarios
  rogue-blow-up      # virial / energy + H¹-acceleration blow-up detection, ETA
  rogue-rogue        # Akhmediev / KM / Peregrine breathers, Benjamin–Feir, crest detection
  rogue-fluid        # Madelung ↔ compressible Euler bridge
  rogue-plasma       # plasma modulational instability
  rogue-fiber        # fiber-optic soliton power / bandwidth (supercontinuum)
  rogue-turbulence   # clear-air turbulence intermittency
  rogue-wasm         # wasm-bindgen bindings for the browser dashboard
  rogue-production   # real-data intake + physical forecast bridge
education/           # interactive browser dashboard (Vite + TS + WASM)
data/                # sample sea-state CSV records
docs/math.md         # full mathematical basis
scripts/             # regression + bench suite (including production regress)

Regression suite

./dev regress          # Akhmediev / Peregrine / ocean / blowup physics suite
./dev regress-prod     # production CSV-intake regression

Benchmarks (release, 12 threads)

Constraint	Target	Measured
2D step 1024×1024	< 100 ms	59 ms
Blow-up detection	< 1 s	streaming fit
WASM bundle	≤ 8 MB	verified at build

Math

docs/math.md — governing equation i∂tψ = −(β₂/2)∇²ψ + γ|ψ|^{p−1}ψ, invariants, Strang splitting, blow-up detection (virial + H¹ ETA), Benjamin–Feir law γ(ν) = ν√(2A² − ν²), unified breather formula, Madelung / Euler bridge.

License

Free software. MIT — do whatever you want. Built for maritime, aviation, plasma, and optics operators worldwide with zero restrictions.

Contributing

Contributions are welcome. Open an issue, fork the repo, and send a PR.

• Branch from main
• Run ./dev check before pushing (cargo check + test)
• For physics changes, include a regression run (./dev regress)
• For production changes, include ./dev regress-prod

Links

• Live site: rogue.jesed.dev
• Author: jesed.dev
• GitHub: github.com/jesedv/rogue
• crates.io: rogue-nls
• Docs: docs/math.md