Thinking out loud about a variant of the actor model (chapter 3 here), which uses graphic lambda calculus or the chemical concrete machine. The goal is to arrive to a good definition of an Artificial Chemical Connectome.
Have remarks? Happy to read them!
A chemical actor is the following structure:
- a graph (or a molecule )
- with a unique ID name
- with a numbering (tagging) of a (possibly empty) part of it’s free arrows
- with a behaviour, to be specified further.
A communication is:
- a graph (or a molecule)
- with a source ID and a target ID
- with a part of free arrows tagged with tags compatible (i.e. the same) with the ones from the graph from the source ID
- with another part of free arrows tags with tags compatible with the ones from the graph from the target ID
The actor target ID receives a communication from the actor source ID and it becomes:
At this point the actor which has target ID exhibit the following behaviour:
- performs one, several, or in a given order, etc of graph rewrites (only the + unidirectional moves) which involve at least an arrow between A and B
- following a given algorithm, splits into a new actor and a communication B’ which has as target arrows the one from the lower part of the previous figure (but with another target ID)
- or creates a new actor by using only (-) moves
Remark: the numbers could be uniformly bounded to 2, or 4, or 6, according to user’s wishes. Take a look at the Ackermann machine, for inspiration.