Documentation

Complete manual for the Fragility Discovery Engine — from first run to algorithm internals.

Sections

What the engine is, what problem it solves, the six domains, and who it fits.

Public site layout: server-hosted samples, viewers, runs, and what is not uploaded from your machine.

Platform-specific setup notes for Windows, Linux, macOS, and WSL. Git configuration and CI scripts for contributors.

Step-by-step for all six domains: replay, counterfactuals, co-evolution, benchmarks, and more.

How the code is organized, how a search run flows through it, and how to add new simulation domains.

All command-line flags, environment variables, output file formats, and a full script index.

Which algorithms are original to this project, which are standard, and where each one is cited.

Fragility Evidence Language — schedules, rollouts, interventions, Δ⁻/Δ⁺ attribution, and evidence schemas.

What scales how, when parallelism is safe, and honest complexity bounds for sweeps and search.

What the sixth simulation domain models and how it differs from the other five worlds.

In-step peg–overload coupling, mutation-chain attribution, and downloadable fork JSON.

Quick links

Install and verify — Python and a virtual environment in under five minutes.
Tutorials — first replay, search run, attacker/defender simulation, counterfactuals.
Simulation domains — aggregate, network, resource cascade, service backlog, liquidity ladder, inventory buffer.
Output file formats — all schema IDs and what produces them.
Scale and limits — complexity, parallelism, and sweep cost.
Why inventory buffer — the sixth simulation domain explained.
Coupled fork — research physics with in-step coupling (replay, attribution chain, JSON bundle).
Algorithms — what we built vs what we borrowed, with citations.
Run a scenario — start a search on this server right now.