Module precision 

Algorithm 3: Precision Assessment

Evaluates whether a fitted BLR+ARD model meets the user’s precision goal, using percentile-based semantics (per CLARIFY-1, option C):

Goal met when: 95th-percentile posterior std ≤ target_std

Rationale: 95th percentile is interpretable (“95% of measurements meet spec”) while allowing a small tail of higher-uncertainty points (e.g., at input-space boundaries before dense sampling).

Status thresholds:

• MetGoal: percentile_95_std ≤ target_std
• Near: percentile_95_std ≤ 1.1 × target_std (within 10% of goal)
• Unmet: percentile_95_std > 1.1 × target_std
• NoiseFloorHit: set externally by Algorithm 4 integration

Structs

PrecisionAssessment

Summary of the current predictive precision.

Enums

PrecisionStatus

Multi-level precision status.

Functions

assess_precision

Assess how close the current model is to the precision goal.

percentile

Compute the p-th percentile (0.0–1.0) of a slice.