Paperlane

no. 16 · Tuesday 17 November 2026

Exact mean first-passage time and a Poisson-smoothing crossover in renewal shot noise

Yuxin Wang and Jiajun Zhang

math.PR · via q-bio.QM

Abstract

Threshold crossings in exponentially relaxing shot noise depend on the interplay between incoming impulses and the residual signal left by earlier arrivals. For renewal input, existing high-threshold product approximations of the mean first-passage time (MFPT) leave open when evaluation at a single index preserves the mean with vanishing relative error. Here we derive an exact finite-threshold MFPT series for exponentially marked renewal shot noise and recast it as an average over an auxiliary Poisson index. When the waiting-time density exhibits short-time power-law behavior, $w(t)\sim ct^{κ-1}$, the dominant series contributions shift from the Poisson center by an amount of order $x^{1-κ}$, whereas the averaging kernel has a fluctuation width of order $\sqrt{x}$, with $x=λb$. The competition between these scales yields a crossover at $κ=1/2$, where the deterministic-index product is asymptotically exact in ratio above this boundary, requires a finite multiplicative correction at criticality, and exhibits a growing logarithmic discrepancy below it. A moving-saddle approximation also yields an asymptotically exact MFPT ratio under general strictly positive, finite-mean waiting times. Numerical comparisons across Gamma, Weibull, and generalized-Gamma renewal laws illustrate the approximation regimes and their finite-threshold accuracy. Together, these results provide an exact finite-threshold benchmark for the mean first-passage time and reveal how short-gap renewal statistics determine the asymptotic accuracy of high-threshold approximations.

fields
Probability
pdf ↗ page arXiv:2611.18675 ↗