Benchmarking wall models for confined active Brownian particles near planar and channel walls

Different microscopic wall models are frequently used in simulations of confined active Brownian particles (ABPs), but their practical equivalence for standard steady-state observables is rarely assessed against an analytical benchmark with explicit and reproducible criteria. In this work, we present a reproducible benchmark for three common wall models in confined ABPs simulations-stochastic hard, specular hard, and exponential soft-using the analytical hard-wall theory of Duzgun and Selinger as reference. After validating the soft wall against the Boltzmann limit at $\text{Pe}=0$ , we compare density and polarization profiles across $\text{Pe}\in \{1,\dots ,10\}$ using local and integrated error metrics. The stochastic wall consistently matches the analytical reference best, the specular rule fails sharply in polarization, and the exponential wall's fidelity depends strongly on both activity and observable; reducing its range improves local but not integrated agreement. A compact two-wall channel test confirms this ranking survives finite confinement. The results provide an explicit, observable-dependent criterion for selecting wall models in ABP simulations.