Finite-Particle Rates in Interacting Systems

• Finite-particle rates are defined as the precise transition, collision, decay, and mixing rates in finite systems that include all N-dependent corrections.
• They capture critical microscopic interactions and finite-size effects that influence quantum decay, algorithmic convergence, and macroscopic transport laws.
• Recent studies leverage these rates to derive non-equilibrium dynamics, equilibrium mixing, and large deviation principles across quantum, classical, and computational domains.

Finite-particle rates describe quantitatively the transition, collision, decay, and mixing rates in systems comprising a finite number of interacting particles. The precise nature and scaling of these rates is critical in diverse contexts: quantum decay, kinetic theory, interacting Markov systems, variational inference, and statistical mechanics. Unlike limiting or "mean-field" rates, finite-particle rates retain all finite-size (N-dependent) corrections, structuring not only the short-time, non-equilibrium, and fluctuation behavior but also informing algorithmic convergence and macroscopic laws in finite systems.

1. Microscopic Models and Definition of Finite-Particle Rates

At the fundamental level, finite-particle rates arise as the transition kernels or generators governing the stochastic or deterministic evolution of systems with discrete, finite N. Common scenarios include interacting jump processes, exclusion models, systems with synchronized or coupled transitions, and quantum decay with finite spatial or temporal support.

For jump Markov processes, the rate for each elementary event can depend on the system configuration, particle identities, and sometimes global observables such as the empirical measure or center of mass. For example, in mean-field finite-state systems, the empirical measure evolves via

$$\mathcal{L}^N f(x) = N \sum_{v \in \mathcal{V}} \lambda_v^N(x)[f(x + v/N) - f(x)]$$

where $\lambda_v^N(x)$ encodes the finite-N jump rates that often converge to a deterministic function in the large-N limit (Dupuis et al., 2016).

In spatially extended or lattice systems, finite-particle rates can be configuration-dependent and modified by site occupancy rules, intrinsic heterogeneity (particle-dependent jump rates), or additional interactions (energy exchange, collisions, etc.) (Malyshev et al., 2022, Grigo et al., 2011).

2. Quantum Mechanical Finite-Size and Finite-Time Corrections

In quantum decay and scattering, the standard infinite-time Fermi Golden Rule is an emergent result from the strictly infinite measurement (T→∞) limit. At finite T, wave-packet localization and overlap lead to new correction terms. Explicitly, the decay probability includes a broad weighting function: $$D(\Delta E; T) = 4 \left[ \frac{\sin(\Delta E\, T/2)}{\Delta E} \right]^2$$ and the corrected rate at finite duration becomes: $$\Gamma(T, \sigma) = \Gamma_0 + N \frac{\sigma}{T} \frac{E}{2m^2} |F(-\Delta m^2)|^2 + \mathcal{O}(1/T^2)$$ where the $1/T$ "finite-size" term results from the overlap region of initial and final wave packets, breaking strict kinetic-energy conservation in the "wave-zone" (Ishikawa et al., 2013). These corrections are universal in relativistically-invariant systems and can dominate when the daughter is very light, for large detector packet size, or at short observation times. They are crucial in the analysis of neutrino experiments, high-precision atomic transitions, and rare decay searches.

3. Particle Systems: Mixing, Convergence, and Fluctuation Rates

Particle systems with energy, mass, or position exchange exhibit finite-particle mixing rates that determine relaxation to equilibrium and scaling laws for fluctuations. A prototypical result is the spectral gap for a nearest-neighbor energy-exchange process on a chain of N sites: $\lambda_N \geq C N^{-2}$ for explicit $C$ depending on local exchange rates and kernel properties. This $N^{-2}$ scaling matches the microscopic time to global equilibrium required for deriving macroscopic transport laws such as Fourier's law in the hydrodynamic limit (Grigo et al., 2011).

In exclusion models with particle-specific rates, the system decomposes into dynamically-aligned "stable clouds," with each cloud's collective motion and stationary distribution governed explicitly by its finite-particle rates. The dynamical partition and long-time speed of each block are computable via combinatorial algorithms or potential functions, providing explicit links from rates to transport and diffusive properties (Malyshev et al., 2022, Menshikov et al., 10 Sep 2025).

4. Propagation of Chaos, Sampling, and Finite-Particle Algorithmic Rates

When interacting particle algorithms (SVGD, R-SVGD, other particle approximations to PDEs or variational flows) are used for sampling or optimization, the finite-particle convergence rates to the target law are of fundamental importance. Standard metrics include the Kernelized Stein Discrepancy (KSD) and Wasserstein-$p$ distances.

Recent advances establish:

• For time-averaged empirical measures under SVGD,

$$\mathbb{E}\!\left[\mathrm{KSD}(\mu_{\text{av}}^N \| \pi)\right] = O\left(N^{-1/2}\right)$$

with only polynomial dimension dependence, matching i.i.d. sampling (Banerjee et al., 2024).

• For strongly regularized variants (R-SVGD), the Fisher information and $W_1$ error obey

$$\mathbb{E}\!\left[W_1(\mu_{av}^N, \pi)\right]^2 = \mathcal{O}(N^{-2/3}) \vee \mathcal{O}(N^{-1})$$

under matching conditions, with explicit dependencies on tuning parameters (He et al., 5 Feb 2026).

These rates are optimal or near-optimal in particle number, and their sharpness relies on a detailed entropy-splitting analysis and careful control of finite-N algorithmic noise. The balance between bias (kernelization) and variance (particle count) is explicitly quantified.

5. Collision, Aggregation, and Reaction Rates in Finite Particle Systems

In collisional particle dynamics—relevant in atmospheric physics, cloud microphysics, and aggregation—finite-size and finite-charge corrections to collision rates lead to nontrivial dependence on particle geometry, hydrodynamics, and electrostatics.

For two dielectric spheres in flow, the collision rate is given by: $$R_{\text{coll}} = n_1 n_2 E_{\text{coll}} \pi (a_1 + a_2)^2 \Delta U$$ where $E_{\text{coll}}$ (collision efficiency) is determined by integrating the equations of motion (including hydrodynamic and electrostatic forces) for the pair-trajectory up to the minimal approach. Non-monotonic dependence on size and charge ratios arises: certain charge and size combinations yield enhanced collisional growth due to attractive near-field electrostatics, counter to naive point-charge intuition (Warrier et al., 30 Dec 2025).

Similarly, spatial birth-death processes with finite configurations admit explicit generators and martingale structures for rates, with extinction probabilities and growth rates determined by linear or sublinear growth conditions on total rate functions (Bezborodov et al., 2015).

6. Large Deviations and Macroscopic Limits from Finite-Particle Rates

The development of large deviation principles (LDPs) for empirical measures in finite interacting particle systems fundamentally depends on the explicit finite-N transition rates. For mean-field (finite-state) systems, the empirical measure process is a jump Markov process whose generator and associated rate function are determined by the finite-particle rates: $$L(x, \beta) = \inf_{q_v \geq 0, \sum_v q_v v = \beta} \sum_{v \in \mathcal{V}} \lambda_v(x) \left[ \frac{q_v}{\lambda_v(x)} \log \frac{q_v}{\lambda_v(x)} - \frac{q_v}{\lambda_v(x)} + 1 \right ]$$ and the macroscopic LDP for trajectories is constructed using these local cost structures (Dupuis et al., 2016). Close connection to ergodicity and boundary behavior is governed by the finite-particle rates—e.g., polynomial decay to zero rates at the simplex boundary, or irreducibility conditions for pathwise estimates.

7. Finite-Particle Rates in Cosmology, Finite-Time Observables, and Precision Physics

Finite-particle and finite-time rates are essential in cosmological contexts where the S-matrix framework fails due to an expanding background and a finite particle horizon. Here, spectral and adiabatic expansions yield time-dependent rates: $$\Gamma(\eta) = \frac{\lambda^2}{4V \Omega_{k_1}(\eta) \Omega_{k_2}(\eta)} \int_{-\infty}^\infty dK_0 \rho(K_0, K) \frac{\sin[(K_0 - \Omega_T(\eta)) (\eta - \eta_i)]}{K_0 - \Omega_T(\eta)}$$ These rates explicitly account for non-instantaneous transitions, sub-threshold production (the anti-Zeno effect), and enhanced dark matter yields in the early universe—a direct phenomenological consequence of finite-particle and finite-duration corrections (Rai et al., 2020, Ho et al., 2015). Universal properties, freeze-out at finite times, and explicit violation of local Lorentz invariance occur in these settings.

Summary Table: Representative Finite-Particle Rate Phenomena

Context	Characteristic Finite-Particle Rate/Scaling	Reference
Quantum decay (finite-time, wavepacket overlap)	$\Gamma(T,\sigma) = \Gamma_0 + \mathcal{O}(1/T)$	(Ishikawa et al., 2013)
SVGD (sampling, empirical measure)	$E[\mathrm{KSD}] = \mathcal{O}(N^{-1/2})$	(Banerjee et al., 2024)
SVGD (Wasserstein-2, bilinear kernels)	$E[W_2] = \mathcal{O}(N^{-\Omega(1/d)})$	(Banerjee et al., 2024)
Mixing in energy exchange models	Spectral gap $\lambda_N = O(N^{-2})$	(Grigo et al., 2011)
Collision/aggregation (finite hydrodynamics)	$R_{\text{coll}} \sim E_{\text{coll}} (a_1 + a_2)^2$	(Warrier et al., 30 Dec 2025)
Large deviations (mean-field systems)	Rate function $L(x,\beta)$ via $\lambda^N_v(x)$	(Dupuis et al., 2016)
Cosmological reaction rates	$\Gamma(\eta),\ \sigma(\eta)$ spectral integrals, finite-horizon	(Rai et al., 2020)

In each case, the explicit form and scaling of finite-particle rates serve as the bridge between microscopic rules, emergent macroscopic laws, precise statistical control, and algorithmic effectiveness across quantum, classical, and computational domains.