Finite-size scaling

Finite-size scaling analyzes how observables in finite systems depend on system size, especially near a critical point, to infer behavior in the infinite-system limit.

Finite-size scaling ansatz: A scaling form that expresses an observable as a power of system size multiplied by a function of scaled variables. It organizes size dependence and distance from criticality into one testable relation.
Phase transition: A qualitative change in a system's macroscopic state as an external condition varies. Finite-size scaling is most often used to study transitions rounded by finite dimensions.
Binder cumulant: A dimensionless ratio of moments of an order parameter used to characterize fluctuations and criticality. Curves for different sizes often intersect near a critical point.
Finite-time scaling: A method that analyzes critical behavior using the duration of a process as a limiting scale. It replaces or supplements system size when dynamics are observed for limited time.
Critical exponent: A number describing how a physical quantity changes near a continuous phase transition. Its value determines the powers of system size used in scaling predictions.
Critical phenomenon: Collective behavior near a continuous phase transition, often marked by scale-free fluctuations. Its divergent scales make system-size dependence especially informative.
Percolation threshold: The occupation probability at which a system first forms a spanning connected cluster in the infinite-size limit. Cluster statistics across finite lattices can locate this threshold.
Corrections to scaling: Subleading terms that modify leading scaling predictions at finite sizes. Ignoring them can bias exponent estimates from small systems.
Correlation length: A length scale over which fluctuations in a system remain correlated. Finite-size effects become pronounced when this length approaches the system's dimensions.
Order parameter: A quantity that distinguishes phases by taking different values in different macroscopic states. Its size-dependent distribution often tracks the onset of order.

Linked from 10 pages