In a percolation model on a large lattice, you observe the mean finite cluster size growing rapidly as you increase p, but no spanning cluster has appeared yet. What does this most likely indicate?
AYou are well above p_c; the infinite cluster exists but is too diffuse to visually span the lattice
BYou are approaching p_c from below; finite cluster sizes diverge as a power law as p → p_c
CThe lattice is too small to support a spanning cluster regardless of p
DYour model has an error because mean cluster size should decrease as you add more occupied sites
The mean finite cluster size diverges as |p − p_c|^{−γ} as you approach p_c from either side. Observing rapid growth of cluster size while still below the spanning transition is precisely the signature of approaching p_c from below. This diverging correlation length — clusters of all scales appearing — is the hallmark of the critical point.
Question 2 Multiple Choice
What is the deepest reason that percolation shares critical exponents with equilibrium phase transitions like the Ising model, despite having no Hamiltonian or temperature?
ABoth models are defined on square lattices, and the lattice geometry determines the critical exponents
BPercolation and Ising models both use binary variables (occupied/empty vs. spin up/down), creating a direct mathematical equivalence
CBoth exhibit scale invariance at the critical point — no characteristic length scale — and critical exponents are determined by spatial dimension and symmetry, not microscopic details
DTemperature and occupation probability play identical mathematical roles in both models, making them formally equivalent
Universality means critical exponents depend only on spatial dimension and the symmetry of the order parameter — not on whether the system has a Hamiltonian, what the microscopic interactions are, or whether the transition is geometric or thermodynamic. Both Ising and percolation exhibit a diverging correlation length at criticality and become scale-invariant. It is this shared geometry of the critical point, not microscopic similarity, that produces identical exponents.
Question 3 True / False
Below the critical probability p_c, no clusters of any size exist in a percolation model.
TTrue
FFalse
Answer: False
Below p_c, finite clusters of all sizes exist — isolated occupied sites, pairs, small connected groups. What is absent below p_c is a *spanning* (infinite-system-size) cluster that connects opposite edges of the lattice. The phase transition is specifically about the appearance of this spanning cluster, not about whether any clusters exist at all. Even at p = 0.01, small isolated clusters form.
Question 4 True / False
At exactly p = p_c, the percolation system is scale-invariant: clusters of all sizes coexist and there is no characteristic length scale.
TTrue
FFalse
Answer: True
This is the defining feature of a continuous phase transition at criticality. At p_c, the cluster size distribution follows a pure power law n(s) ~ s^{−τ}, with equal (log-scale) weight at all sizes. The correlation length, which describes the typical cluster size, diverges to infinity. There is no characteristic scale — which is why the system at p_c has fractal geometry and why critical phenomena are described by scale-invariant (renormalization group) methods.
Question 5 Short Answer
Why does the concept of universality imply that studying percolation gives quantitative predictions applicable to very different physical systems like polymer gelation and fluid flow through porous media?
Think about your answer, then reveal below.
Model answer: Universality means that systems in the same universality class share the same critical exponents, determined only by spatial dimension and the symmetry of the order parameter — not by microscopic details. Percolation, polymer gelation, and flow through porous media all involve the same geometric question (does a connected path span the system?) in the same spatial dimension, so they share a universality class. Their critical exponents are identical. This means that exact exponents calculated for the simple percolation model transfer directly to predict the behavior of these physically very different systems near their respective transitions.
This is the scientific payoff of universality: instead of solving every physical system from scratch, you identify which universality class it belongs to and import the exact results from whatever model in that class is easiest to analyze. Percolation is simple enough to solve rigorously on many lattices, making it a reference model for an entire class of connectivity transitions.