Multifractality Breaking From Bounded Random Measures

Multifractal systems usually have singularity spectra defined on bounded sets of Holder exponents. As a consequence, their associated multifractal scaling exponents are expected to depend linearly on statistical moment orders at high-enough orders-a phenomenon referred to as the linearization effect. Motivated by general ideas taken from models of turbulent intermittency and focusing on the case of two-dimensional systems, we investigate the issue within the framework of Gaussian multiplicative chaos. As verified by means of Monte Carlo simulations, it turns out that the linearization effect can be accounted for by Liouville-like random measures defined in terms of upper-bounded scalar fields. The coarse-grained statistical properties of Gaussian multiplicative chaos are furthermore found to be preserved in the linear regime of the scaling exponents. As a related application, we look at the problem of turbulent circulation statistics, and obtain a remarkably accurate evaluation of circulation statistical moments, recently determined with the help of massive numerical simulations.