Transport Equation Framework Overview

• Transport Equation Framework is a set of mathematical formalisms that uses linear and nonlinear PDEs to describe the propagation and interaction of physical quantities.
• It integrates measure-valued techniques, nonlocal interactions, and integral operators to provide multiscale modeling for applications like traffic networks, polymeric flows, and neutron transport.
• The framework employs characteristics, time-dependent transition matrices, and well-posed discretization methods to ensure accurate simulation of complex dynamic systems.

The transport equation framework encompasses a spectrum of mathematical and computational formalisms developed to describe the propagation, interaction, and evolution of physical quantities (mass, energy, particles, momentum, etc.) through continua, networks, and phase spaces. Central to this framework are both linear and nonlinear partial differential equations—often posed on high-dimensional domains, networks, or function spaces—incorporating local and nonlocal interactions, boundary and interface conditions, and multiscale structure. The precise design, well-posedness analysis, and discretization of these transport equations underpin diverse applications from traffic on networks, polymeric flows, neutron/photon transport, to optimal mass relocation.

1. Measure-Valued Linear Transport Equations on Networks

A comprehensive framework for measure-valued linear transport on networks has been introduced, covering both the continuum on single arcs and the multiscale construction on complex networks (Camilli et al., 2016). On an arc parameterized by $x \in [0,1]$ with $v(x)>0$, one studies positive Borel measures $\mu_t$ such that

$$\partial_t \mu(t,x) + \partial_x (v(x)\mu(t,x)) = 0, \quad \mu_{t=0} = \mu_0, \quad \mu_{x=0} = \nu_0,$$

with inflow and initial data represented as measures. The unique solution is represented via characteristics: $$\mu_t = \Phi_t\#(\mu_0)|_{[0,1]} + \int_{[0,t]} \delta_{\Phi_t(0,s)}\,d\nu_0(s),$$ with pushforward denoted by $\Phi_t\#$.

On a network $\Gamma=(V,E)$, local solutions on arcs are "glued" at vertices using time-dependent transition matrices $p^i_{kj}(t)$ prescribing mass distribution across incoming/outgoing arcs, under global conservation constraints. This allows recursive construction of the global solution measure on the network, with weak balance laws incorporating both arc and vertex contributions.

The measure-valued setting enables seamless integration of microscopic (Dirac/molecular) and macroscopic (continuous/fluid) components via the Lebesgue decomposition in time and space: $\mu = \mu_\mathrm{ac} + \mu_\mathrm{s}$, transporting both via characteristics and pushforwards. This approach underpins multiscale modeling for traffic, crowds, and data networks.

2. Nonlocal and Nonlinear Extensions

Measure-valued solutions to nonlinear and nonlocal transport equations on networks have been formulated for cases where the velocity $v[\mu]$ depends on the current state, typically by subtracting weighted integrals of $v(x)>0$0 against a nonlocal interaction kernel $v(x)>0$1 (Camilli et al., 2017): $v(x)>0$2 Under boundedness and Lipschitz continuity in $v(x)>0$3 and $v(x)>0$4, unique measure-valued solutions exist, and a superposition principle applies—solutions are constructed by branching along all admissible paths through the network, with mass split at vertices as dictated by stochastic matrices on the network's edges. This is particularly relevant for interacting agents, traffic with nonlocal interactions, or crowd models with anticipation effects.

3. Transport Equations with Integral Operators

The extension of the transport equation to include inhomogeneities via integral operators is crucial for kinetic models with complex interactions, e.g., in polymeric flows (Lellis et al., 2016). Here, a scalar unknown $v(x)>0$5 evolves by [ \partial_t u + b(t, x, r)\cdot\nabla_{x, r}u = \int_{\mathbb{R}^j} K(t, x, r, \tilde{r}) u(t, x, \til