“…To this end, one can implement spatially discrete stochastic (Markovian) reaction–diffusion models. For reactions in 1D nanoporous systems , as described in section , one can divide the pore into cells comparable to reactant size and treat continuous diffusion by hopping and exchange between adjacent cells. − ,− For reactions on 2D crystalline surfaces , we shall assume that reactants are localized on well-defined adsorption sites possibly of multiple types, and then implement appropriate detailed and realistic single-site or more generally multisite lattice-gas (msLG) models. ,,− Extensive input is required for these models for both system thermodynamics (adsorption energies, adspecies interactions) and kinetics (adsorption and desorption dynamics, reaction pathways and barriers), which is obtained from or by comparing with experiment and increasingly from density functional theory (DFT) analysis . The behavior of these models is in principle described exactly by master equations. , However, given the difficulty of reliable analysis of the master equations, model behavior is usually instead determined precisely by kinetic Monte Carlo (KMC) simulation, − as described extensively in recent reviews. ,, More details are provided in section .…”