ginc-property-transition-matrix-construction
IN premise — summaries/2026/08/24/xie-2021-icl-bayesian-sR-references-chunk-2.md
Created 2026-08-25T02:58:56+00:00
GINC property transition matrices are generated as a convex combination of 100 random permutation matrices with softmax-distributed weights using temperature t=0.1.
Summary
The GINC property transition matrix is built by blending 100 random shuffle patterns, but a low temperature setting makes the weighting heavily concentrated on just a few of them. In practice, each transition is nearly deterministic — close to one specific rearrangement — with only a thin spread of randomness, so the system's property updates are highly predictable rather than broadly random.