Kinetic models for traffic flow resulting in a reduced space of microscopic velocities (Q503016)

The authors of this interesting paper study a class of kinetic models for the traffic flow. The main elements taking part here are mainly vehicles (also can be used particles), and note that one may apply it in biology, population dynamics, ecology and so on. The main mathematical model in the paper is constructed on a continuous space of microscopic velocities with a Boltzmann-type approach. A kinetic distribution function \(f = f (t, v)\) is defined, where \(f : \mathbb{R}^{+}\times V \to \mathbb{R}^{+}\), \(V = [0, V_{\max }]\) is the domain of the microscopic speeds, and \(V_{\max }\) is the maximal speed, which may depend on the mechanical characteristics of the vehicles, imposed speed limits and environmental conditions. In the homogeneous case, the Boltzmann-type equation has the form \(\partial_tf(t,v)=Q[f,f](t,v)\), where \(Q[f, f ](t, v)\) is the ``collisional operator'', and \(v\) is the test speed which describes the relaxation to equilibrium due to the microscopic binary interactions among vehicles. The mass conservation follows from the equality \(\int_{V}Q[f, f ](t, v)dv=0\). Thus in the space homogeneous case, the density remains constant in time. The collisional operator is usually split into two terms: a gain term \(G[f, f ]\), and a loss term \(L[f, f ]\), that model statistically the interactions which lead to gain or to loose the test speed \(v\). Thus the present model includes an integro-differential equation \(\partial_{t}f(t,v)=G[f, f ](t,v) + L[f, f ](t,v)\). Both operators \(G\) and \(L\) have integral representation, and it is taken into account the probability that the velocity \(v \in V\) results from a microscopic interaction between ``candidate vehicles'' with some velocity \(v_{\ast }\) and ``field vehicles'' with speed \(v^{\ast }\). Here the interaction rate possibly depends on the relative speed of the ``interacting vehicle''. For this model it is noted that ``the particular structure of the collision kernel allows one to find the analytical expression of a class of steady-state distributions, which are characterized by being supported on a quantized space of microscopic speeds, and the number of the above stated velocities is determined by a physical parameter describing the typical acceleration of a vehicle and the uniqueness of this class of solutions is supported by numerical investigations''. The authors investigate two different models of special interest. The first one is ``Quantized acceleration (\(\delta \)-model)'' characterized with the property that the output velocity \(v\) is obtained by accelerating instantaneously from \(v_{\ast }\) to the velocity \(\min \{v_{\ast } + \triangle v, V_{\max }\}\). The second model is ``Uniformly distributed acceleration (\(\chi \)-model)'' characterized with the property that the new velocity \(v\) is uniformly distributed between \(v_{\ast }\) and \(\min \{v_{\ast } + \triangle v, V_{\max }\}\). The above stated models have different resulting probability distributions. For a \(\delta \)-model, steady-state case, is shown an important result concerning the existence of a particular class of stationary solutions, i.e. given set of velocities in \([0,V_{\max }]\), then the distribution function is a weak stationary solution of the \(\delta \)-model provided the velocities have the form \(v_j = v_1 + j\triangle v\) (\(j=1,2,\ldots ,N\)). Here the reader may observe that this statement ensures the existence of a class of steady-state solutions for this \(\delta \)-model which are characterized by the property that the total mass of vehicles on the road is distributed uniquely on the velocities multiplied of \(\triangle v\). Further, the authors investigate some discretization of this model. The next interesting result shows that all equilibria of the discrete model are of the quantized form described by the above statement.