Cross-Correlation
Cross-correlation evaluates dependence between multiple entropy sources.
Implemented in openentropy_core::analysis via
cross_correlation_matrix().
What it is: A pairwise dependence matrix across multiple sources.
Use it for: Independence screening before combining sources in a pool or extractor.
Input shape: Two or more named byte streams.
Use this when
Section titled “Use this when”- You are combining multiple sources and want an independence sanity check.
- You suspect shared hardware paths or environmental coupling.
- You need to identify source pairs that should not be treated as independent.
What It Computes
Section titled “What It Computes”- Pairwise Pearson correlation for all source pairs
- Flag count for pairs whose absolute correlation exceeds threshold
| Metric | Description |
|---|---|
pairs | Correlation coefficient for each source pair |
flagged_count | Number of pairs with ` |
Interpretation
Section titled “Interpretation”Independent sources should show low pairwise correlation. Elevated cross-correlation can indicate shared hardware pathways, scheduling coupling, or environmental coupling that reduces effective independence.
Usage Notes
Section titled “Usage Notes”- Requires two or more sources
- Enabled by
--profile deepor--cross-correlation