Boolean Cumulants and Exact Reduced Descriptions of Renewal-Driven Systems

Abstract

Reduced descriptions of unresolved fluctuations are commonly based on Gaussian processes, despite the fact that many realistic forcings possess finite correlation times and a renewal structure. For memoryless step (Kubo-Anderson) noise, we show that the reduced dynamics admit two exact and complementary descriptions: a kernel representation governed by the Boolean cumulants of the jump distribution, and a frozen-noise representation for stationary probability densities. For exponentially distributed waiting times, the totally time-ordered $G$-cumulants coincide exactly with the Boolean cumulants of the jump law, leading to an exact resummation of the memory kernel. The resulting kernel is expressed through the Boolean generator $η$ evaluated on a resolvent operator. Within this hierarchy, second-order closure is exact if and only if the jump distribution is symmetric Bernoulli. For general jump laws, the leading closure error is controlled by the universal combination $(b4/b2)λ^2$. Unlike classical cumulant expansions, however, the Boolean hierarchy acts on the reduced kernel rather than directly on the stationary measure. A complementary exact description is provided by the frozen-noise representation. For the linear system with linear multiplicative interaction (LIMI/CAM) model, it reduces the dynamics to a random affine recursion and yields exact expressions for stationary densities and Kesten tail indices. The analysis reveals that rates, moments, and closure errors are governed by the Boolean hierarchy, whereas stationary-density properties are determined by the geometry of the jump distribution through the frozen dynamics. Finite-correlation reductions are therefore controlled not only by the variance and correlation time of the forcing, but also by the underlying combinatorial structure of the renewal process. For memoryless renewal noise, that structure is Boolean.