eps-mixture-index-inversely-correlated-irreducibility

IN premise — summaries/2026/08/24/engels-2024-not-all-features-linear-s4-s-parse-autoencoders-find-multi-d-imensional-features.md

Created 2026-08-25T02:58:01+00:00

A lower ε-mixture index M(f) value indicates higher feature irreducibility, meaning the feature is less 'mixture'-like and more genuinely multi-dimensional.

Summary

This gives the system a practical scoring rule: the lower a feature's ε-mixture index, the more it counts as a genuine, fundamental dimension of the problem rather than a blend of simpler parts. In other words, it flags which features are solid building blocks you can reason about independently and which are just composites that could be broken down further.