Time-Discrete Particle Equations

• Time-discrete particle equations are models that update states at fixed intervals, forming the backbone of structure-preserving simulation frameworks.
• They are applied to nonlinear PDEs, kinetic theory, and quantum walks using methods like discrete gradient integrators and Markov chain formulations.
• These techniques ensure the preservation of key invariants such as mass, energy, and momentum, providing robust convergence and accuracy in multi-scale applications.

Time-discrete particle equations form the mathematical, algorithmic, and physical backbone of numerous structure-preserving, probabilistically sound, and computationally robust models in the simulation and analysis of complex systems. At their core, these equations define particle dynamics in which state updates occur at fixed, discrete time intervals, leading to difference equations or Markov chains rather than ordinary or stochastic differential equations. The resulting frameworks are indispensable across the analysis of nonlinear PDEs, kinetic theory, statistical mechanics, quantum walks, and algorithmic particle methods. The following sections present a rigorous survey of the theory, representative models, and applications of time-discrete particle equations, encompassing deterministic, stochastic, interacting, and structure-preserving settings.

1. Structure-Preserving Discrete Gradient Particle Methods

A comprehensive structure-preserving framework for discretizing nonlinear continuity equations via particle methods is established through discrete gradient integrators. Consider the continuity equation

$\partial_t f + \nabla \cdot (f U[f]) = 0,$

with $U[f]$ defined by variational derivatives of an energy functional $E[f]$. To regularize singularities arising in point-particle approximations, energies $H(f)$ are mollified: $H^\varepsilon(f) = H(f * \varphi_\varepsilon)$, with $\varphi_\varepsilon$ a Gaussian mollifier of width $\varepsilon$. Regularized energies $E_\varepsilon[f]$ admit smooth variational derivatives.

The particle ansatz

$f^N(t, x) = \sum_{p=1}^N w_p\, \delta(x - x_p(t)),$

with fixed positive weights, leads to a generalized ODE system,

$$\frac{d}{dt} x_p = -\frac{1}{w_p} \nabla_{x_p} E_\varepsilon(x_1, \ldots, x_N),$$

with explicit closed forms for aggregation-diffusion and kinetic (Landau) cases. These flows admit a discrete gradient time integrator: $$\frac{X^{n+1} - X^n}{\tau} = \tilde{B}(X^{n+1}, X^n) \, \overline{\nabla E_\varepsilon}(X^{n+1}, X^n),$$ where $\tilde{B}$ is a symmetric negative semidefinite approximation and $\overline{\nabla E_\varepsilon}$ satisfies the discrete energy identity.

This framework rigorously preserves key invariants:

• Energy dissipation: $E_\varepsilon(X^{n+1}) - E_\varepsilon(X^n) \leq 0$.
• Mass conservation: $m^{n+1} = m^n$.
• Momentum and kinetic energy conservation (Landau case): preserved to machine precision; orthogonality relations hold at the discrete level.

Numerical examples confirm monotonic energy decay, invariance of mass/momentum/kinetic energy, second-order accuracy, and the absence of restrictive CFL-type stability conditions (Hu et al., 2024).

2. Discrete-Time Particle Systems and Propagation of Chaos

Discrete-time interacting particle systems rigorously approximate nonlinear, often non-conservative McKean-Vlasov SDEs and associated nonlinear PDEs. A generic time-discretized interacting particle scheme for $N$ particles employs Euler-Maruyama updates with empirical measures,

$$\tilde{\xi}_{k+1}^{i,N} = \tilde{\xi}_k^{i,N} + \Phi(t_k, \tilde{\xi}_k^{i,N}, \tilde{u}_k(\tilde{\xi}_k^{i,N}))\, \Delta W_k^i + g(t_k, \tilde{\xi}_k^{i,N}, \tilde{u}_k(\tilde{\xi}_k^{i,N})) \, \Delta t,$$

with running weights $G^i_k$ evolving exponentially, and empirical densities $\tilde{u}_k(y)$ defined via kernel estimation.

The main results include:

• Propagation of chaos: For fixed time grids, convergence of finite-dimensional distributions to i.i.d. copies of the nonlinear process as $N \to \infty$, uniformly in $k$.
• Error bounds: Uniform-in-time $L^\infty$ and $L^2$ bounds, with overall accuracy $O(\Delta t + 1/N)$ under suitable Lipschitz assumptions and mollifier regularity.
• Stability: All hidden constants depend solely on model regularity and the finite time horizon (Cavil et al., 2016).

Table: Error bounds for time-discrete interacting particle algorithms

Quantity	Rate	Prerequisites
$$\mathbb{E}\sup_{0\leq s\leq t_k}|\tilde\xi_s^{i,N}-Y_s^i|^2$$	$O(\Delta t + 1/N)$	Lipschitz coefficients, mollifier $K$
$\mathbb{E}\|\tilde u_k-u^{m^0}_{t_k}\|_\infty^2$	$O(\Delta t + 1/N)$	As above

This theory underpins convergence and reliability guarantees for particle-based numerical schemes for nonlinear PDEs.

3. Stochastic and Markovian Time-Discrete Particle Dynamics

Time-discrete equations arise naturally in the analysis of stochastic particle systems, including Markov chains with (bi-)state or multi-state velocity structure, interacting agents, and exclusion processes.

For two-state velocity Markov chains (e.g., the discrete telegraph process):

• Particle updates are $X_{n+1} = X_n + V_n h$, with $V_n \in \{\pm c\}$ and a velocity switching chain with transition probabilities dependent on site and time.
• The Chapman-Kolmogorov system for joint up/down probabilities yields a coupled difference system whose continuum limit is the telegraph (hyperbolic) PDE, which reduces to the Klein-Gordon equation under constant parameters.
• For general multi-velocity discreteness, rate-matrix evolution produces an advection-reaction system, and ensemble-averaged dynamics recover Newton's law under a "potential-gradient" condition on the rates (Beumee et al., 2014).

In integrable stochastic interacting systems, such as the discrete time $q$-TASEPs:

• Particle dynamics are governed by explicit, exactly solvable stochastic update rules (e.g., geometric or Bernoulli $q$-deformed jumping distributions).
• Observable expectations satisfy closed systems of difference equations (with two-body boundary conditions), solved via nested contour integrals and Fredholm determinant formulas (Borodin et al., 2013).

4. Discrete-Time Particle Models for Kinetic, Reactive, and Transport Processes

Time-discrete schemes are instrumental in modeling kinetic transport, reaction-diffusion, and advective-dispersive phenomena at the mesoscopic scale.

Reactive transport example: A two-state Markov chain for adsorption/desorption models solute transport, with particle position after $n$ steps a random sum weighted by the (bimodal) Markov binomial distribution (MBD). Discrete-time dynamics exhibit double-peak concentration profiles, converging in the diffusive scaling limit to classical reactive-transport PDEs,

$$\partial_t(C_F + C_A) = D \partial_{xx} C_F - v \partial_x C_F, \qquad \partial_t C_A = -\mu C_A + \lambda C_F,$$

with Gaussian mixture densities for the discrete state (Dekking et al., 2011).

Master equation for first-passage phenomena: Discrete-time first-visit equations yield the temporal Fokker-Planck (propagation-dispersion) equation under scaling. Functional generalizations result in nonlinear or fractional temporal dispersion equations, structurally mirroring nonlinear and fractional spatial diffusion models. Discrete moments control the macroscopic advection and dispersion coefficients (Boon et al., 2016).

Driven Lorentz model: A tracer on a lattice with obstacles, driven by a constant force, is formulated by a time-discrete master equation, with generating-function analysis providing full moment hierarchies. Notable results include the identification of superdiffusive scaling ($N^3$ variance growth) and a detailed account of the transient approach to stationary velocity, including the breakdown of Einstein's relation and exact subordination relations for arbitrary waiting-time distributions (Shafir et al., 2024).

5. Structure and Analysis of Time-Discrete Particle Algorithms

Time-discrete particle methods are represented in both deterministic (e.g., splitting, Lie-Trotter, and discrete gradient integrators) and stochastic (e.g., Markov chains, random walk models) contexts.

• Hamiltonian particle-in-cell (PIC) methods: A Hamiltonian splitting on the discretized Vlasov-Maxwell system ensures explicit, Poisson-preserving, and energy-conserving time evolution. The construction respects all Casimir invariants, including discrete charge and divergence constraints, provided appropriate finite element spaces are used (He et al., 2016).
• Feynman-Kac particle integration: Discrete-time FK particle schemes analyze the fluctuation properties (variance, bias) and uniform-in-time convergence of empirical measures to the FK-flow, with explicit $O(1/N)$ and, for continuous-time mesh approximations, $O(1/m)$ error bounds. First-order fluctuation analysis yields precise nonasymptotic control (Moral et al., 2012).
• Weakly interacting particle systems: In discrete time, nonlinear Markov chains and their empirical measures admit unique fixed points. Convergence to equilibrium is exponentially fast in Wasserstein distance, and propagation of chaos is uniform-in-time, with explicit concentration inequalities for the empirical process (Budhiraja et al., 2014).

Representative table of algorithmic approaches:

Method Type	Update Scheme	Conservation Properties
Discrete gradient integrator	$(X^{n+1}-X^n)/\tau$ via grad	Exact energy/mass, some cases momentum/kinetic energy (Hu et al., 2024)
Hamiltonian splitting (PIC)	Lie-Trotter/Strang composition	Poisson structure, energy, charge, Gauss's, div-B (He et al., 2016)
Euler + threshold (lubrication)	Euler with contact sticking rule	Position $O(q_s)$ error, robust near singularity (Hillairet et al., 2010)

6. Discrete Spacetime Particle Dynamics and Quantum Emergence

In discrete spacetime geometries, time-discrete particle worldlines ("world chains") are the primitive dynamical objects. The "skeleton" equations for free particles enforce discrete analogs of geodesic motion:

• Fixed-length and parallelism of links, determined by a geometrically deformed world function $\sigma_d$.
• Inherent underdeterminacy yields a stochastic discrete process, whose wavefunction is constructed by ensemble averaging over all possible chains.
• The discrete-time path-integral yields, via appropriate scaling ($\lambda_0^2 = \hbar/(mc)$), the Schrödinger equation in the continuum limit: $$i\hbar \partial_t \psi = -\frac{\hbar^2}{2m} \nabla^2 \psi.$$
• Quantum behavior thus emerges from the discrete geometry plus a Planckian length-mass relation, with no a priori postulated wave dynamics (Rylov, 2011).

7. Discrete-Time Quantum Walks and Lattice Gauge Particle Dynamics

On spacetime lattices, time-discrete particle equations underpin strictly local, unitary, and gauge-invariant lattice gauge theories for single-particle dynamics:

• The update law is realized as a unitary, strictly causal, nearest-neighbor operator (quantum walk), generated by a discretized Dirac action with minimal coupling to gauge fields via Wilson line factors.
• The equations of motion, discrete Noether theorem, and exact lattice Maxwell equations are all realized at the difference level, with amplitudes strictly vanishing outside a lattice light cone. The framework admits arbitrary spacetime dimension (Arnault et al., 2022).

---

Time-discrete particle equations constitute a foundational element in modern applied analysis, mathematical physics, probability, and computational modeling. Their rigorous structure enables the implementation of schemes that are provably invariant-preserving, amenable to large-scale computation, and capable of capturing multiscale effects, from stochastic fluctuation propagation to quantum emergent behavior. The referenced works provide a comprehensive mathematical infrastructure and empirical validation for these techniques.