Skip to content

Runtime (Entropy)

This page is a conceptual reference for entropy-style measures.

Entropy can be used to summarize uncertainty, dispersion, or heterogeneity.

Shannon entropy (concept)

For a discrete distribution with probabilities p_i:

H = -\sum_i p_i \log p_i

For continuous variables, differential entropy exists but is not directly comparable across unit changes.

Reference engine status

The reference engine in this repo does not currently compute entropy-based risk metrics as first-class outputs.

What it does have today:

  • A planning-time warning when a bound asset set appears heterogeneous (tags and/or criticality_index.type differ). This is used as a guardrail, not as an entropy measure.

If you want entropy-based measures in results, they should be treated as an engine-defined extension.