define-analysis-dimensions

Purpose

Define approximately independent, measurable axes that span a problem, design, or validity space.

Input contract

required: [problem_or_artifact, target_outcome, dimension_ontology]
optional: [candidate_axes, domain_constraints, measurement_plan]
constraints: [each axis has semantics, type, direction, and measurement rule; correlated axes are flagged]

Procedure

1. Extract factors and conditions relevant to the target outcome.
2. Group candidates into dimensions and test approximate independence.
3. Define each dimension's domain, units, direction, and observability.
4. Return the dimension set with exclusions and unresolved dependencies.

If each analysis dimension has a clear meaning and admissible type, consider enumerate-dimension-values as the next tactic.

Output contract

produces: [dimension_set, axis_definitions, independence_notes, coverage_scope]
delta_fields: [findings, evidence_updates, uncertainties, open_questions]

Quality gates

• Every retained axis is independent enough for the caller's analysis and is measurable under the supplied plan.
• Parameter-space mapping covers all dimensions relevant to assertion validity; omitted axes require a reason.
• Do not confuse a value/level with the dimension that contains it.

Parameterization

The caller must provide the target problem, outcome, axis ontology, candidate factors, domain bounds, measurement units, and independence criterion.

Failure and counterexamples

Reject duplicate axes, dimensions with no observable values, or an independence claim unsupported by a comparison.

Provenance map

• resolved: deep-insight/variation-axis-definition
• resolved: stress-test/parameter-space-mapping
• resolved: creative-ideation/parameter-identification
• intermediate: Pass3/map-parameter-space
• intermediate: Pass3/identify-dimensions

Preserved source criteria ledger

source	physical line	kind	source criterion
deep-insight/variation-axis-definition	14	structural	Identify orthogonal axes, ensuring they are independent, measurable, and span the parameter space.