propagate-uncertainty

Purpose

Propagate input uncertainty through a model or reasoning chain to estimate output distributions, tails, and contributions.

Input contract

required: [model, uncertain_inputs, output_variables]
optional: [sampling_plan, dependency_structure, tail_metrics]
constraints: [input distributions, dependencies, and sampling method must be declared]

Procedure

1. Represent each uncertain input with its distribution and dependencies.
2. Sample or propagate the inputs through the model.
3. Summarize output distributions, tail behavior, and dominant uncertainty contributors.

Output contract

produces: [output_distributions, tail_summary, uncertainty_contributions]
delta_fields: [findings, evidence_updates, uncertainties]

Quality gates