Wave 5 — approachability & front door: CLI — discover --preset (physics/ecology/epidemiology/finance/general) + presets catalog; explicit flags override presets. library becomes a provenance-aware workspace: world/data content hashes, columns, config, tags captured on add; library search + compare over the workspace (backward-compatible index). SDK — lawsynth.recipes (parity with CLI presets) + Study.discover(recipe=..., **overrides); three runnable domain demos. Van der Pol shows the payoff: general recipe misses the cubic term, mechanics recipe recovers -x+v-x^2*v (R2 0.99995). Docs — a real product README (hero + Lorenz GIF, CLI+Python quickstart, feature matrix by product loop, surfaces + HTTP endpoints, honest tone); refreshed getting-started pages and a CHANGELOG Unreleased section. All flags verified against the shipped CLI usage. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Getting started
LawSynth turns time-series observations into executable mathematical worlds: interpretable law systems you can read, simulate, forecast, compare, and share. The whole product is one loop, and every step is deterministic and offline:
observe (CSV) → discover (laws) → understand (explain) → use (simulate /
forecast / intervene) → compare → share (report / .lsworld bundle) → organize
A discovery is a portable .lsworld bundle; everything downstream — CLI, Python
SDK, Studio, and the HTTP services — operates on that same artifact.
Read next
- installation — build the
lawsynthCLI and the Python SDK. - quickstart — run the core loop end to end.
- concepts — datasets, worlds, laws, and bundles.
- your first world — discover, inspect, and reuse a bundle.
- Pick a surface: CLI, Python, or Studio.
The checked-in examples and the workspace tests are executable and are the source of truth for supported inputs. Discovery finds a sparse fit from the implemented feature library — it is not evidence that an inferred relation is causal or that extrapolation beyond the observed window is valid.