Numerical integration of non-local master equations for renewal shot noise with drift and state-dependent amplitude

The probability density of a system driven by a drift and a renewal train of kicks with state-dependent amplitude obeys a master equation that is non-local in space, because the kick operator $\hat{p}(i\partial_x I(x))$ is of infinite order, and, for non-exponential waiting times, non-local in time, with a memory kernel containing the propagator of the drift. Its solutions contain quasi-singular components, the never-kicked mass, which collapse onto the attractors of the drift. We present an accurate numerical method for such equations. The kick operator is a push-forward along the Marcus flow; in the Lamperti variable it becomes a translation, whose exact cell-to-cell weights follow from the cumulative distribution of the kicks, without quadrature. The drift is integrated by a conservative semi-Lagrangian step exact in time. The memory is represented by a sum of exponentials whose auxiliary fields are advected by the drift, at a cost linear in the number of steps. The never-kicked mass is transported exactly. For each ingredient we quantify the errors of simpler alternatives. The method is verified against exact solutions and event-driven Monte Carlo simulations with exact flow maps, from which it is statistically indistinguishable with $10^8$ trajectories. Parameters tuned for Gaussian kicks carry over unchanged to other kick distributions, and near a zero of the coupling, where the density vanishes faster than any power, the solution follows the known asymptotic law down to densities of order $10^{-37}$, far beyond Monte Carlo reach. As an application we compare the exact equation with its local and Poissonian approximations: in the range explored, the local equation deviates by up to 8% of the density width, also for waiting times with finite mean, consistently with an error estimate predicting larger deviations for slow drifts at long times. A Fortran implementation is available.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-18
DOI
https://doi.org/10.48550/arxiv.2611.24390
Primary Topic
stochastic dynamics and bifurcation
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Numerical integration of non-local master equations for renewal shot noise with drift and state-dependent amplitude

R. Mannella, Marco Bianucci
arXiv (Cornell University)
stochastic dynamics and bifurcation
preprint

Numerical integration of non-local master equations for renewal shot noise with drift and state-dependent amplitude

R. Mannella, Marco Bianucci
preprint en

Abstract

The probability density of a system driven by a drift and a renewal train of kicks with state-dependent amplitude obeys a master equation that is non-local in space, because the kick operator $\hat{p}(i\partial_x I(x))$ is of infinite order, and, for non-exponential waiting times, non-local in time, with a memory kernel containing the propagator of the drift. Its solutions contain quasi-singular components, the never-kicked mass, which collapse onto the attractors of the drift. We present an accurate numerical method for such equations. The kick operator is a push-forward along the Marcus flow; in the Lamperti variable it becomes a translation, whose exact cell-to-cell weights follow from the cumulative distribution of the kicks, without quadrature. The drift is integrated by a conservative semi-Lagrangian step exact in time. The memory is represented by a sum of exponentials whose auxiliary fields are advected by the drift, at a cost linear in the number of steps. The never-kicked mass is transported exactly. For each ingredient we quantify the errors of simpler alternatives. The method is verified against exact solutions and event-driven Monte Carlo simulations with exact flow maps, from which it is statistically indistinguishable with $10^8$ trajectories. Parameters tuned for Gaussian kicks carry over unchanged to other kick distributions, and near a zero of the coupling, where the density vanishes faster than any power, the solution follows the known asymptotic law down to densities of order $10^{-37}$, far beyond Monte Carlo reach. As an application we compare the exact equation with its local and Poissonian approximations: in the range explored, the local equation deviates by up to 8% of the density width, also for waiting times with finite mean, consistently with an error estimate predicting larger deviations for slow drifts at long times. A Fortran implementation is available.

arXiv (Cornell University)
stochastic dynamics and bifurcation
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon