nucleation

LIT-1126: Computational Mechanics: Pattern and Prediction, Structure and Simplicity

1 min read

Properties4
StatusActive — read 2026-10-29 (NOTE-1372) in full, from arXiv's own copy (v2); worth reading for the proofs that a stationary process's causal states are maximally predictive, minimal, unique up to measure zero and minimally stochastic among predictive states, and that their entropy Cµ bounds the excess entropy E from above. Read it knowing that minimality holds only for predicting the whole future (for a finite horizon a coarser state can do as well), that the theorems assume the joint distribution known, a discrete stationary time series and prediction from its own past, and that "causal" means Humean succession, with nothing on interventions or estimation from data.
Version2
Tagsinformation-theory, complex-systems
Cited as “paper” byNOTE-1372 — Shalizi and Crutchfield — Computational mechanics

Graph View

Shalizi, C. R. & Crutchfield, J. P. (2001). Journal of Statistical Physics 104(3–4):817–879. doi:10.1023/A:1010388907793.

Read in full on 2026-10-29 (NOTE-1372).

Why this is here

A10 of the exchange of 2026-10-16: the minimal history a collective must keep to predict its future, as a candidate for the slow state that survives turnover (CLAIM-327). A10: causal states of a process are not thereby the boundaries of a social individual.

Backlinks