Skip to content

Ticket 037: Monte Carlo Uncertainty Modeling

Goal

Add an optional uncertainty-analysis layer around deterministic mission estimation without changing deterministic default behavior.

Current Gap

The estimator produces single deterministic outputs. It does not model input uncertainty, run repeated samples, or report confidence intervals for mission time, energy reserve, wind margins, or feasibility outcomes.

Scope

  • Define versioned uncertainty inputs for selected mission, vehicle, wind, resource, link, and energy parameters.
  • Add seeded Monte Carlo execution that wraps the deterministic estimator.
  • Report distributions and summary statistics for key outputs.
  • Preserve deterministic reproducibility for a given seed and sample count.
  • Add CLI/API output contracts for uncertainty reports.
  • Add tests for reproducibility, validation, and failure aggregation.

Integration Requirements

  • Add uncertainty configuration through YAML so users can pair it with existing mission, vehicle, terrain, wind-grid, geofence, landing-zone, and scenario files.
  • Include resource and link feasibility abstractions from Ticket 034 so sampled runs can vary communication availability, external power availability, backup battery policy, or link failover state without bypassing deterministic defaults.
  • Keep the deterministic estimate command as the baseline path and add uncertainty execution as an explicit opt-in mode or command option.
  • Ensure scenario execution can reference uncertainty outputs only through a documented contract, without making ordinary scenarios nondeterministic.
  • Add YAML examples for uncertainty runs that reuse existing mission, vehicle, terrain, and wind examples.
  • Include uncertainty report output in JSON/Markdown rendering and fixture coverage if it becomes a public output surface.

Acceptance Criteria

  • A Monte Carlo run is reproducible with the same inputs and seed.
  • The default estimator path remains deterministic and unchanged.
  • Uncertainty outputs clearly distinguish sampled results from baseline deterministic estimates.
  • Uncertainty runs compose with all implemented deterministic inputs instead of bypassing terrain, wind, geofence, landing-zone, resource, link, or energy behavior.

Out of Scope

  • Real-time risk scoring.
  • Regulatory approval calculations.
  • Replacing deterministic feasibility checks.