Stationary patterns and their selection mechanism of Urban crime models with heterogeneous near-repeat victimization effect

TL;DR

This paper analyzes pattern formation in an urban crime PDE model with spatial heterogeneity, identifying stability conditions, bifurcation points, and a wavemode selection mechanism that predicts stable criminal aggregation patterns.

Contribution

It introduces a generalized PDE model incorporating heterogeneity and derives a wavemode selection mechanism for stable pattern formation, supported by rigorous analysis and numerical simulations.

Findings

Stable patterns emerge as near-repeat victimization rate decreases.

The dominant stable pattern has the wavenumber maximizing the bifurcation value.

Large domains support more stable criminal aggregation patterns.

Abstract

In this paper, we study two PDEs that generalize the urban crime model proposed by Short \emph{et al}. [Math. Models Methods Appl. Sci., 18 (2008), pp. 1249-1267]. Our modifications are made under assumption of the spatial heterogeneity of both the near-repeat victimization effect and the dispersal strategy of criminal agents. We investigate pattern formations in the reaction-advection-diffusion systems with nonlinear diffusion over multi-dimensional bounded domains subject to homogeneous Neumann boundary conditions. It is shown that the positive homogeneous steady state loses its stability as the intrinsic near-repeat victimization rate $\epsilon$ decreases and spatially nonconstant solutions emerge through bifurcation. Moreover, we find the wavemode selection mechanism through rigorous stability analysis of these nontrivial patterns, which shows that the only stable pattern must have wavenumber that maximizes the bifurcation value. Based on this wavemode selection mechanism, we will be able to precisely predict the formation of stable aggregates of the house attractiveness and criminal population density, at least when the diffusion rate $\epsilon$ is around the principal bifurcation value. Our theoretical results also suggest that large domains support more stable aggregates than small domains. Finally, we perform extensive numerical simulations over 1D intervals and 2D squares to illustrate and verify our theoretical findings. Our numerics also include some interesting phenomena such as the merging of two interior spikes and the emerging of new spikes, etc. These nontrivial solutions can model the well observed aggregation phenomenon in urban criminal activities.