A model of irreversible jam formation in dense traffic

TL;DR

This paper analyzes a one-dimensional stochastic traffic model exhibiting phase transitions and jam formation, deriving exact expressions for local density and jam probability, and explaining phase behavior through finite-size effects and random walk theory.

Contribution

It extends previous work by providing exact stationary properties and analyzing phase transition mechanisms in a stochastic traffic model.

Findings

Exact expressions for local density at the first site and jam probability P(1)

Identification of three regimes near the phase boundary using random walk theory

Support of theoretical results with extensive Monte Carlo simulations

Abstract

We study an one-dimensional stochastic model of vehicular traffic on open segments of a single-lane road of finite size $L$. The vehicles obey a stochastic discrete-time dynamics which is a limiting case of the generalized Totally Asymmetric Simple Exclusion Process. This dynamics has been previously used by Bunzarova and Pesheva [Phys. Rev. E 95, 052105 (2017)] for an one-dimensional model of irreversible aggregation. The model was shown to have three stationary phases: a many-particle one, MP, a phase with completely filled configuration, CF, and a boundary perturbed MP+CF phase, depending on the values of the particle injection ($\alpha$), ejection ($\beta$) and hopping ($p$) probabilities. Here we extend the results for the stationary properties of the MP+CF phase, by deriving exact expressions for the local density at the first site of the chain and the probability P(1) of a completely jammed configuration. The unusual phase transition, characterized by jumps in both the bulk density and the current (in the thermodynamic limit), as $\alpha$ crosses the boundary $\alpha =p$ from the MP to the CF phase, is explained by the finite-size behavior of P(1). By using a random walk theory, we find that, when $\alpha$ approaches from below the boundary $\alpha =p$, three different regimes appear, as the size $L\rightarrow \infty$: (i) the lifetime of the gap between the rightmost clusters is of the order $O(L)$ in the MP phase; (ii) small jams, separated by gaps with lifetime $O(1)$, exist in the MP+CF phase close to the left chain boundary; and (iii) when $\beta =p$, the jams are divided by gaps with lifetime of the order $O(L^{1/2})$. These results are supported by extensive Monte Carlo calculations.