←Research

Self-organized criticality

Systems that drive themselves to the critical point, with nobody tuning the parameters.

The avalanche distribution follows a power law: no characteristic scale.

Power laws are everywhere: earthquake magnitudes, landslide and avalanche sizes, extinctions in the fossil record, 1/f noise, city sizes. In each case there is no characteristic scale, and the question is why a system would sit exactly at a critical point, given that the critical point of an ordinary transition requires tuning an external parameter with precision.

The answer Bak, Tang and Wiesenfeld gave in 1987 is a minimal model: grains of sand dropped one at a time onto a lattice. When a site exceeds the threshold it redistributes to its neighbours, which may exceed the threshold in turn: an avalanche. After a transient the system settles into a stationary state where avalanches have no typical size.

The mechanism rests on a separation of time scales: driving is slow, avalanches are fast. Accumulation pushes up, avalanches discharge, and the critical point is the attractor of that competition rather than a value imposed from outside. No fine-tuning is needed because the dynamics produces it.

It is worth being precise about what is verified and what is not. The exponent BTW report for the three-dimensional case is about 1.35, and that one is extracted from their paper. For the two-dimensional case the original figure is unreadable in the scan I have, and the value circulating in secondary literature could not be confirmed against the source: in the vault it stays flagged as unverified rather than settled.

Sources

  • Bak, Tang & Wiesenfeld (1987)
  • Bak, How Nature Works