Divisible and indivisible Stochastic-Quantum dynamics

TL;DR

This paper provides a geometric framework for understanding divisible and indivisible stochastic-quantum dynamics, revealing bounds and structures that distinguish classical from quantum-like evolutions in probabilistic systems.

Contribution

It introduces a novel geometric construction in stochastic matrix space that characterizes divisibility and indivisibility, extending to systems with any number of configurations.

Findings

Identifies conical bounds separating divisible and indivisible dynamics.

Shows indivisible dynamics include quantum dynamics with time-flow against information erasure.

Connects continuity of dynamics with divisibility properties across dimensions.

Abstract

This work presents a complete geometrical characterisation of divisible and indivisible time-evolution at the level of probabilities for systems with two configurations, open or closed. Our new geometrical construction in the space of stochastic matrices shows the existence of conical bounds separating divisible and indivisible dynamics, bearing analogy with the relativistic causal structure, with an emerging time pointing towards information erasure when the dynamics are divisible. Indivisible dynamics, which include quantum dynamics, are characterised by a time-flow against the information-erasure time coordinate or by being tachyonic with respect to the cones in the stochastic matrix space. This provides a geometric counterpart of other results in the literature, such as the equivalence between information-decreasing and divisible processes. The results apply under minimal assumptions: (i) the system has two configurations, (ii) one can freely ascribe initial probabilities to both and (iii) probabilities at other times are linearly related to the initial ones through conditional probabilities. The optional assumption of (iv) continuity places further constraints on the system, removing one of the past cones. Discontinuous stochastic dynamics in continuous time include cases with divisible blocks of evolution which are not themselves divisible. We show that the connection between continuity and multiplicity of divisors holds for any dimension. We extend methods of coarse graining and dilations by incorporating dynamics and uncertainty, connecting them with divisibility criteria. This is a first step towards a full geometric characterisation of indivisible stochastic dynamics for any number of configurations which, as they cannot at the level of probabilities be reduced to a composition of evolution operators, constitute fundamental elements of probabilistic time-evolution.