The Ghost Loop

A Brownian loop-soup fills space with random loops of all sizes, each appearing with an intensity proportional to a parameter. Small loops are local and numerous. Large loops are rare and extended. The two populations seem independent.

Lupu and Werner show they’re not. In high dimensions, chains of small Brownian loops spontaneously assemble into large-scale cycles that are statistically indistinguishable from actual large loops. A “ghost” loop-soup emerges — present in the correlations but absent from the individual loops — with exactly doubled critical intensity.

The mechanism is concatenation. Small loops that happen to share nearby points form chains. In high enough dimensions, these chains close into macroscopic cycles with probability sufficient to create an entire population of phantom large loops. The ghost loops carry the same statistical signature as real loops of the same scale.

The critical intensity doubles because the ghost loops contribute an additional copy of the large-loop statistics. Below the original critical point, only real large loops matter. Between the original and doubled critical points, the ghost population dominates. The phase transition occurs at twice the intensity anyone expected.

Microscopic randomness self-organizes into macroscopic structure. The large loops were never placed in the soup — they assembled themselves from the ambient small-scale fluctuations. Emergence, made mathematically precise: the whole exceeds the inventory of its parts.