Skip to content
Choose an application

07 / Records & prediction

Keep the distinctions
that the future needs.

Inspect the supplied examples, then enter your own finite deterministic update and find the coarsest partition that predicts every future displayed output.

Exact finite-state calculation; one supplied autonomous update

A present record can hide a future distinction

Refine the record partition until every class has one well-defined successor class. Stable agreement concerns all future outputs of the supplied update.

Inspect the hidden distinction

Begin with classes that retain only the present binary output. Each refinement separates states whose successors fall into different current classes. Stop when every class has one well-defined successor class.

Model and refinement

The default example below is complete. Interactive controls become available when the script loads.

Current partition

{A, B} | {C, D}

Round 0: two present-record classes. A and B both display 0, but their next outputs are 1 and 0. The stable partition is {A} | {B} | {C, D}: three predictive states, requiring two fixed-length bits in total.

One-step hidden memory: exact update and partition
StateOutput R(x)Successor F(x)Next outputClass nowStable class
A0C111
B0A012
C1C123
D1D123
First eight outputs from each initial state (time starts at zero)
Initial statet=0t=1t=2t=3t=4t=5t=6t=7
A01111111
B00111111
C11111111
D11111111
Complete refinement history
Complete refinement history
RoundClassesNumber of classes
0{A, B} | {C, D}2
1{A} | {B} | {C, D}3
Open this example with its default parameters

Exports are calculated examples, not experimental data. No information is uploaded.

Try this investigationSelect the delayed example. Determine the first step at which A and B become distinguishable in their output traces.

Test your own finite model

Start with the partition defined by the displayed output R. Split each class according to the pair (current class, successor class) until it stops changing.

Analyze a supplied transition table

Use 2–64 uniquely named states. Every successor must appear in the state column. Output labels may be any short letters or numbers; names can also contain underscores or hyphens. The calculation stays in this browser.

Run the table to inspect its predictive classes.

Try this investigationCreate two states with the same present output but different successor outputs. Then make their entire futures identical and check that the stable partition merges them.

What the result establishes

At stability, R*∘F=T∘R* defines an autonomous update on predictive classes. States with identical future output sequences remain in the same class.

The displayed bit count is ⌈log₂(number of classes)⌉ for a fixed-length label. It is not a thermodynamic entropy, energy requirement or average code length. Stochastic updates and external interventions require a different equivalence test and are outside this tool.

THE QUESTIONS BEHIND THE CALCULATION

Connect this laboratory to the fundamental problems

Follow inner-time generation and outer-time completion from a worked model to its physical research target.

View all related problems →

SMONAD

Share this page

Share this page with its own title, summary and preview image.

More platforms