Automata dynamics modelled via monoid actions

Lulof Pirée

Tallinn University of Technology




Abstract: Many familiar models of computation, such as Boolean circuits, deterministic automata, Turing machines, etc., can be seen as a graph of memory cells that change their content according to a local update rule. The behaviour of cells over multiple steps in time depends recursively on the behaviour of connected cells, which quickly leads to complicated expressions in inductive proofs. However, when viewing time as a monoid, then coKleisli composition allows to scale local rules to arbitrary long intervals in time. This not only simplifies proofs, but also provides other fun opportunities.

The talk will (1) show several examples of how familiar automata fit in the framework, and (2) try to explain how (co)Kleisli composition recursively computes the multi-step behaviour. Familiarity with models of computation such as Boolean circuits and Turing machines is assumed, and acquaintance with (co)monads and (co)Kleisli categories is useful.