How it works
Five stages, deterministic end to end. Everything the platform writes is reproducible from the same inputs, and everything it claims is traceable back to a cited source.
- 1 · Assemble
The operator composes a scenario from validated building blocks — airframe, environment, terrain, threats, guidance profile, and engagement. Each block carries its own provenance (source, confidence, citation) that follows the block wherever it is used.
- 2 · Validate
A shared validation gate reads the assembled scenario and emits blocking errors and warnings before any run is allowed to start. Rules span every capability: a misfuzed warhead, a non-monotonic Pk curve, a visibility mismatch against the chosen seeker, or a directed-energy warhead without terrain all fail fast.
- 3 · Run
A deterministic Monte Carlo engine drives either the JSBSim-backed 6-DoF dynamics or the kinematic backend. Seeds are explicit; same inputs produce byte-equal outputs. At run start, an immutable snapshot of every parameter used is captured and persisted.
- 4 · Analyse
The run detail page shows miss statistics, detection and lock-loss events, a guidance comparison chart, and — when an engagement is attached — an overlaid p(kill) curve with a Best-Profile summary. A full report can be exported as PDF or DOCX.
- 5 · Review
Analysts submit a successful run for review. A reviewer can approve, request changes with notes (the run returns to the analyst's queue), or reject outright. Only approved runs are eligible to be promoted to golden references for later comparisons.
