A Cellular-Automaton Model of Computation
February 14, 2019

This early model-of-computation note tried to make cellular automata programmable directly rather than using fragile glider constructions as an indirect way to simulate logic.

The source ideas were actor-style objects, object capabilities, membrane computing, functional reactive programming, interaction nets, finite-state automata, and a lattice intended for local spatial computation.

Computation-model noteOriginally posted 2019-02-14; expanded here around the actual model ingredients.

Related: Experiments with Cellular Automata

Article focus: programmable cellular automata, local state, object capabilities, membranes, interaction nets, lattice geometry, and claytronics.

The Design Question

Conway-style glider constructions show universality, but they are not a practical programming interface. The question in this note was whether a cellular automaton could be designed so code maps directly onto local cells, ports, state, and interactions.

That makes the problem closer to a programming model than an animation. A useful cellular machine has to expose ownership, boundaries, local rules, composition, and update behavior.

Ingredients

Actor model
Independent entities with local state and message-style interaction.
Object capabilities
Explicit authority and access boundaries between parts of the system.
Membranes
Spatial boundaries that can contain state and regulate exchange.
Reactive programming
Local updates that respond to changing state over time.
Interaction nets
Computation expressed through local graph-like interactions.
Finite automata
Cell behavior grounded in explicit state transitions.

Lattice And Goal

The experimental lattice mentioned in the original post was hexagonal, with a later 3D target tied to cuboctahedral symmetry and kissing-number geometry.

The longer-term claytronics question was whether programmable local cells could support locomotion, damage recovery, and composition from repeated parts. The practical claim is bounded: the model sought a direct programming surface for local cellular computation.

Links From the Original Post