An introduction to random rule-based chemical networks

TL;DR

This paper introduces a probabilistic model of chemical reactivity using random rule generation, analyzing its statistical properties and phase behavior through mathematical and simulation methods.

Contribution

It presents a novel random rule-based model of chemical networks, including two different probabilistic frameworks and their phase transition behaviors.

Findings

Model II exhibits a non-trivial phase diagram.

Simulation results agree with theoretical predictions.

The model provides insights into the complexity and structure of reaction networks.

Abstract

We introduce in this article a random model of reactivity in which a primitive rule, if accepted, generates an infinite number of rules by context derivation. The model may be thought of as a toy model of chemical reactivity, where reactions are accepted if their randomly distributed activation energy is below a certain threshold. It may be simulated by induction on the level (length of the word). We describe some statistical features of the model, regarding the number and complexity of the rules, and the shape of the reaction network. The complexity index of a rule is defined as the number of covalent bonds involved in the rearrangement. The Bernoulli parameter (acceptation probability) of the rules is chosen as fixed in a first model (Model I), and exponentially decreasing in the complexity index in a second one (Model II). The two models have very different behaviors, Model II exhibiting a non-trivial phase diagram. The main tool for mathematical analysis is an approximate mapping to a Galton-Watson tree with generation-dependent progeny distributions. Detailed simulations, demonstrating a general agreement with theoretical estimates, are provided at the end. Extensions to realistic bond formation/breaking rules for molecules will be presented elsewhere.