Finite-Time Finite-Size Scaling (FTFSS)

• Finite-Time Finite-Size Scaling (FTFSS) is a framework that integrates finite-time and finite-size effects to characterize critical phenomena and dynamic transitions.
• It employs universal scaling forms and critical exponents to collapse data from diverse systems, including low-dimensional maps, synchronization dynamics, and phase transitions.
• FTFSS enables accurate extraction of universal behavior in both equilibrium and nonequilibrium settings, providing insights into system limitations and crossover effects.

Finite-Time–Finite-Size Scaling (FTFSS) is an analytical and numerical framework that systematically characterizes critical phenomena and dynamic transitions in complex systems subject to both finite observation windows (finite time) and finite macroscopic extents (finite system size). FTFSS, developed by extending classical finite-size scaling (FSS) theory from equilibrium statistical physics, unifies temporal and spatial finite effects into universal scaling forms that are central for extracting critical exponents, identifying dynamical regimes, and probing universality classes across deterministic, stochastic, discrete, and continuous-time systems. Its implementations have rigorously demonstrated surface and curve collapse for observables in systems ranging from low-dimensional map bifurcations, synchronization dynamics, and first-order phase transitions, to nonequilibrium large-deviation sampling and driven condensed matter, establishing FTFSS as a foundational pillar in the study of finite-system criticality.

1. Core Principles and Theoretical Framework

The foundation of FTFSS is the recognition that close to a critical point, diverging relaxation timescales and spatial correlation lengths render both temporal and spatial cutoffs essential, breaking down the asymptotic assumption of infinite equilibration and system volume. Observables $A(t,L,\epsilon)$ (with $t$ time, $L$ linear system size, $\epsilon$ the control parameter) are hypothesized to obey scaling forms:

$$A(t,L,\epsilon) = L^{-\alpha} \; \mathcal{F}\left( t L^{-z},\, \epsilon L^{1/\nu} \right)$$

where $\alpha$ is related to the order-parameter exponent, $z$ the dynamic critical exponent, and $\nu$ the correlation-length exponent (Lee et al., 2014). Variants include:

• Discrete-time dynamical systems: The number of iterations $T$ functions as an effective system size, yielding

$$A(T,\mu) \simeq T^{-\beta/\alpha} F \left( (\mu - \mu_c) T^{1/\alpha} \right)$$

where $\mu$ is a control parameter and $F$ is a universal collapse function (Martin et al., 30 May 2025).

• Stochastic processes: Both observation time $t$ and population size $N_c$ enter joint scaling forms for estimators of trajectory-additive observables, such as

$$\overline{\Psi_s^{(N_c)}(t)} = \Psi_s^{(\infty)} + \frac{a}{t} + \frac{b}{N_c} + \cdots$$

with convergence controlled by leading $1/t$ and $1/N_c$ corrections (Hidalgo et al., 2016).

FTFSS interpolates between pure FSS (where time is infinite) and finite-time scaling (FTS) (where size is infinite), producing a full scaling surface on which all conventional singular behaviors appear as cross-sections (Lee et al., 2014).

2. Critical Exponents, Scaling Variables, and Collapse Procedures

Critical exponents in FTFSS are inherited or extended from standard critical phenomena. For continuous transitions:

• Static exponents (e.g., $\beta$, $\nu$) characterize order-parameter and correlation-length divergences.
• The dynamic exponent $z$ relates divergence of the correlation time $\tau \sim \xi^z$.

Scaling variables encode both temporal and spatial distances from criticality:

• For temporal proximity: $x = t L^{-z}$ or $T (\mu - \mu_c)$ for maps (Martin et al., 30 May 2025).
• For spatial proximity: $y = \epsilon L^{1/\nu}$, with $\epsilon$ the reduced control parameter.

FTFSS collapse methodology involves:

1. Computation of observable $A$ for a set of sizes and times.
2. Rescaling by the hypothesized exponents.
3. Plotting $A\,L^{\alpha}$ vs the scaling variables and optimizing exponents for collapse (Huang et al., 2014, Martin et al., 30 May 2025).

Extraction of exponents further utilizes the power-law decay at criticality, saturation off criticality, and logarithmic or stretched exponential corrections depending on bifurcation order and universality class (Martin et al., 30 May 2025, Choi et al., 2013).

3. Dynamical Classes and Regimes: Continuous and First-Order Transitions

FTFSS applies robustly to both continuous and first-order transitions:

Continuous transitions:

• For systems such as the Ising model under finite-rate quench, FTFSS describes crossover from adiabatic to impulse regimes, controlled by the Kibble-Zurek mechanism. The scaling variable $R L^{z + 1/\nu}$ with rate $R$ determines the governing regime (Huang et al., 2014).
• In oscillator synchronization (Kuramoto, q-state clock), the scaling surface unites equilibrium FSS and early-time dynamical relaxation, with universality class affected by quenched vs. thermal disorder (Lee et al., 2014, Choi et al., 2013).

First-order transitions:

• Dynamic FTFSS incorporates an exponentially large tunneling timescale, leading to universal two-state dynamics for observables such as magnetization and autocorrelation. Characteristic time scales obey $\tau(L) \sim L^\alpha \exp(\sigma L^{d-1})$ (Pelissetto et al., 2017).
• In hysteresis and driven systems (e.g., $\phi^4$ and Curie-Weiss models), FTFSS captures the emergence and scaling of plateaus with universal and model-specific exponents across driving regimes (Chen et al., 10 Jul 2025).

4. FTFSS in Discrete Dynamical Maps and Low-Dimensional Systems

For deterministic maps with bifurcations, finite-time scaling is directly analogous to FSS, with the number of iterations taking the role of system size:

• The scaling ansatz is $$A(T,\mu) \approx T^{-\beta/\alpha} F\left( (\mu-\mu_c) T^{1/\alpha}\right)$$.
• The critical exponent $\beta$ is determined by the order of the leading nonlinearity at the bifurcation: $\beta = 1/(k-1)$, where $k$ is the nonlinearity order (Martin et al., 30 May 2025).
• Finite-time susceptibility and Lyapunov exponents display scaling forms and critical divergence/saturation, with explicit forms derivable from the nonlinear normal form expansion.
• Extensions to two-dimensional maps generalize the scaling variable via the geometric mean of Jacobian eigenvalues, preserving the collapse structure (Martin et al., 30 May 2025).

5. Stochastic Systems, Rare-Event Sampling, and Large Deviation Theory

In population-based numerical approaches for sampling rare-event distributions (e.g., large deviation of activity in stochastic processes), FTFSS quantifies convergence and systematic error:

• Trajectory-additive CGF estimators exhibit finite-time and finite-population corrections $\sim 1/t$, $\sim 1/N_c$ (Hidalgo et al., 2016).
• Two-step extrapolation (first in time, then in system size) or joint regression recovers the infinite-size and infinite-time limit, enabling improved estimates of large-deviation properties.
• Application to models such as the contact process demonstrates FTFSS as essential for numerical precision and convergence beyond naïve sampling (Hidalgo et al., 2016).

6. Nonequilibrium Driving, Self-Similarity Breaking, and Anomalous Scaling

FTFSS captures anomalous scaling behaviors arising when self-similarity is broken by strong driving or when system size and finite-time cutoff are not separable:

• Under sufficiently rapid external drive or for small enough systems, new exponents (“bressy exponents,” Editor's term) appear, resulting in distinct scaling laws on either side of the transition. For example, in the 2D Ising model under linear heating/cooling, the ordered and disordered phases show different leading exponents in the scaling regime (Yuan et al., 2020).
• The affected scaling forms are, e.g.:

$$O(\tau, L, R) = L^{-x/\nu} \mathcal{F}_\pm( \tau L^{1/\nu},\, R L^r ),\qquad (bressy\ corrections\ for\ R L^r \gg 1)$$

with new exponents $\sigma$ manifesting as power-law prefactors in the scaling functions in the fast-driving limit.

• These phenomena signal the breakdown of naïve single-exponent universality and require extended FTFSS analysis to resolve crossovers and the failure of equilibrium scaling (Yuan et al., 2020).

7. Applications, Limitations, and Outlook

FTFSS has found application in:

• Nonequilibrium relaxation and transient determination of universality classes in continuous/dynamical transitions (Choi et al., 2013, Lee et al., 2014).
• Dynamic Hysteresis and irreversible processes in driven phase transitions (Chen et al., 10 Jul 2025).
• Phase transitions in expanding QCD matter with finite volume and timescale effects (Cherif et al., 2019).
• First-order phase transitions in discrete models and energy landscapes (Pelissetto et al., 2017).
• Numerical rare-event sampling and improved estimation of large-deviation statistics in stochastic processes (Hidalgo et al., 2016).

Limitations are primarily computational (the need for extensive data across two scaling variables), and care is required to distinguish between asymptotic scaling and crossover effects. Furthermore, systems with multiple diverging timescales or strong corrections to scaling may require even richer FTFSS variants to obtain accurate exponents.

A plausible implication is that FTFSS offers a universal language for analyzing critical phenomena where both spatiotemporal cutoffs are unavoidable, with direct implications for the interpretation of simulation and experiment and for uncovering new classes of universal, nonequilibrium scaling behavior (Lee et al., 2014, Martin et al., 30 May 2025, Chen et al., 10 Jul 2025, Yuan et al., 2020, Pelissetto et al., 2017).