The critical value of an occupation parameter at which a system of randomly occupied sites or bonds first develops a macroscopic connected component that spans the domain.

Principle

A competition between local occupancy and global connectivity produces a sharp change in connectivity structure at a parameter value where cluster correlation length diverges and macroscopic connectivity appears.

Demonstration

In bond percolation on the infinite square lattice, edges independently occupied above a characteristic probability yield, with nonzero probability, an unbounded connected cluster; numerically the threshold is approximately 0.5 for that lattice geometry.

Misapplication

Estimating the threshold from a single moderate‑size simulation without finite‑size scaling or ignoring correlations among occupation events leads to biased conclusions about the true critical value.

Consequence

Identifying the threshold distinguishes regimes with and without system‑spanning connectivity and informs transport, conductivity, and robustness properties of random media or networks.

Reversal

Below the threshold the network decomposes into only finite clusters and lacks long‑range connectivity; macroscopic transport is suppressed.

Boundary

Defined for random occupancy models (site or bond) on lattices and graphs with independent placement or specified correlation structure; thresholds differ for directed, correlated, or continuum variants and are not universally identical across geometries.

Semantic Tension

Closely related but distinct from epidemic or dynamical thresholds: percolation is a static geometric connectivity transition, whereas dynamical thresholds refer to sustained activity in a process evolving on the substrate.

Synthesis

The percolation threshold is the parameter value separating a disconnected regime of finite clusters from a connected regime where a macroscopic component emerges, marking a geometric connectivity transition in random occupancy models.