First-Passage Times for the Space Fractional Fokker-Planck Equation

Christopher N. Angstmann, Daniel S. Han, Bruce I. Henry, Boris Z. Huang · 2025

Read First-Passage Times for the Space Fractional Fokker-Planck Equation by Christopher N. Angstmann, Daniel S. Han, Bruce I. Henry, Boris Z. Huang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-18256.txt.

Introduction

Christopher N. Angstmann, Daniel S. Han, Bruce I. Henry, and Boris Z. Huang work out how long a wandering particle takes to first hit a wall. Their particle moves in jumps that can chain sub steps, a new class of spread tied to a space fractional equation. Forces that change with place act all through each jump, not just at its ends.

The team derive the first hit time spread for varied walls and force fields, with worked cases. For a single wall and no force, the tail of the spread falls off with a new power law. That law fits a known check but differs from the law for a close cousin, Lévy flights. One best spread rate cuts the mean hit time to its least value.

Catalog record

CREATORChristopher N. Angstmann, Daniel S. Han, Bruce I. Henry, Boris Z. Huang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-18256

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

First-Passage Times for the Space Fractional Fokker-Planck Equation Christopher N. Angstmann, Daniel S. Han, Bruce I. Henry, Boris Z. Huang Abstract We extend the random walk framework to include compounded steps, providing first-passage time (FPT) properties for a new class of superdiffusive processes, which are governed by the space-fractional spectral Fokker-Planck equation. This first-passage process leads to novel FPT properties, different from Lévy flights and walks, that account for space dependent forces and hitting boundaries throughout the path of a jump. The FPT distribution can be derived for different types of barriers and potentials, for which we also provide specific examples. For the one-sided absorbing boundary on the semi-infinite line, we find that the FPT density scales as t − 1 / ( 2 α ) − 1 t^{-1/(2\alpha)-1} , in agreement with the method of images but in violation of the Sparre-Andersen scaling. In this case, there exists an optimal space-fractional exponent α \alpha to minimize the mean FPT. From molecular diffusion to financial markets, random walks are ubiquitously important in modeling many phenomena. In such models, pivotal events may be triggered when a random walk first crosses a boundary or hits a target, conventionally called the first-passage time (FPT) or the first-hitting time 32 . Problems involving FPTs are especially important in science, where collisions between particles determine chemical reaction rates for diffusing particles 38 , confined particles 29 , particles following biased pathways 28 , time-dependent non-linear reaction networks 30 , cellular biology 9 , protein binding 33 and in animal foraging strategies 10 ; 12 , where the foraging pattern of animals dictates whether they find food and shelter 2 ; 3 ; 31 . The FPT remains a topic of increasing significance within numerous other recent studies on machine learning algorithms 37 ; 42 , trading strategies 11 and quantum tunneling 44 ; 20 ; 18 . The FPT of random walks exhibiting superdiffusive characteristics, where the mean square displacement (MSD) scales as \langle x^{2}(t)\rangle\sim t^{\nu} for \nu>1 , has been of recent interest in theoretical studies. Examples include: the optimization of searching for sparse targets 27 , searches with external biases 26 ; 21 , searches with short-lived targets 4 , asymmetric searches 22 ; 23 , extreme statistics 13 and searches with restarts 24 . In addition, there has also been vast applications within experimental settings such as in electronic relaxation in solutions 19 , interfacial diffusion 41 and animal foraging strategies 40 ; 31 ; 39 . The efficiency of superdiffusive models in the context of the animal foraging hypothesis has been the subject of sustained debate. 3 ; 27 ; 14 . These models, referred to as Lévy flights and walks, include jumps with divergent variance, causing the paths to be almost surely discontinuous. This leads to the neglect of path-dependent effects such as barriers, boundaries and potential fields. A recent study has shown that superdiffusion and the space fractional Fokker-Planck equation can arise from compounded random walks 1 with continuous trajectories. These compounded random walks share the same governing equation as the Lévy flight on \mathbb{R} : \dfrac{\partial\rho(x,t)}{\partial t}=-D_{\alpha}(-\nabla^{2})^{\alpha}\rho(x,t), (1) where \rho(x,t) is the probability density of finding the random walker at x , \alpha\in(0,1] is the fractional exponent, D_{\alpha} is the anomalous diffusion coefficient and (-\nabla^{2})^{\alpha} is the spectral fractional Laplacian, which can be defined on bounded or unbounded domains 15 . In this Letter, we explore the FPT of the compounded random walk 1 and show that the FPT properties differ from Lévy flights and walks. Physically, the differences arise due to the compounded random walk having connected paths in the continuum limit. Unlike previous formulations, this compounding allows the superdiffusive particle to interact with boundaries and sample local forces across its entire path. The link between the compounded random walk and the spectral fractional Laplacian 1 allows us to formulate FPT problems in the spectral sense. This is possible due to the homogeneous boundary conditions which arise from the formulation, as opposed to the exterior conditions necessary in Lévy flight models. In addition, the compounded random walk can be easily extended to include space-dependent biases, which leads to the space-fractional spectral Fokker-Planck equation 36 ; 5 ; 1 \dfrac{\partial\rho(x,t)}{\partial t}=-D_{\alpha}(-\mathcal{L})^{\alpha}\rho(x,t), (2) where (-\mathcal{L})^{\alpha} is the space-fractional spectral Fokker-Planck operator and D_{\alpha} is the anomalous diffusion coefficient. This operator is spectrally defined by (-\mathcal{L})^{\alpha}\rho(x,t)=\sum_{n=1}^{\infty}\langle\rho,\phi_{n}\rangle\lambda_{n}^{\alpha}\phi_{n}(x). (3) Here, \phi_{n} is the n th normalized eigenfunction, with eigenvalue \lambda_{n} , and \langle\cdot,\cdot\rangle is the weighted inner product associated with the Fokker-Planck operator \mathcal{L} defined by \mathcal{L}\rho(x,t)=\dfrac{\partial}{\partial x}\left(\dfrac{\partial\rho(x,t)}{\partial x}+\beta V^{\prime}(x)\rho(x,t)\right), (4) with appropriate boundary conditions, where \beta is a constant and V(x) is the potential function. If V(x) is constant, then ( 2 ) reduces to ( 1 ) and, if \alpha=1 , we recover the standard Fokker-Planck equation. In what follows, we derive the FPT distribution from the compounded random walk, provide examples of FPT problems for different boundary conditions and outline a Monte Carlo scheme for stochastic simulations. The examples on a finite interval and the semi-infintie line highlight how the FPT distribution, density and mean governed by ( 2 ) differ from the standard Fokker-Planck equation and Lévy flight models. Model Framework. Consider a discrete-time random walk on the set \Omega=\{x_{i}=i\Delta x|0\leq i\leq M,\Delta x=L/M\} , where a particle makes a step after every time increment. We denote the random variable Y_{n}\in\Omega as the location of the particle after the n time increments, given some initial position Y_{0}\in\Omega . Let \Omega^{\prime}\subseteq\{0,L\}\subset\Omega be a set of points such that the particle is absorbed once it enters \Omega^{\prime} . We are only concerned with the FPT into \Omega^{\prime} where it is absorbed, so the particle is considered to have stopped once it enters \Omega^{\prime} . We will denote the probability of stepping from x_{i} to x_{j} via the probability transition function \Lambda_{1}(x_{i},x_{j})=P(Y_{n}=x_{j}|Y_{n-1}=x_{i}), (5) for all n . Thus, the probability to transition from x_{i} to x_{j} in exactly n steps, without entering \Omega^{\prime} during the intermittent steps, is defined recursively by \Lambda_{n}(x_{i},x_{j})=\sum_{x_{k}\notin\Omega^{\prime}}\Lambda_{n-1}(x_{i},x_{k})\Lambda_{1}(x_{k},x_{j}). (6) Note that the sum does not include the probability of stepping into the absorbing set so the sum over all x_{j} may not be equal to 1 , which is to say \sum_{x_{j}\notin\Omega^{\prime}}\Lambda_{n}(x_{i},x_{j})\leq\sum_{x_{j}\in\Omega}\Lambda_{n}(x_{i},x_{j})\leq 1. (7) The FPT into \Omega^{\prime} is the first time step for which Y_{n}\in\Omega^{\prime} and is written as n^{*}=\inf\{n|Y_{n}\in\Omega^{\prime}\}. (8) For a random walk initially at x_{i}\notin\Omega^{\prime} , the probability of making n steps without being absorbed is called the survival function given by P(n^{*}>n)=\sum_{x_{j}\notin\Omega^{\prime}}\Lambda_{n}(x_{i},x_{j}). (9) To obtain the FPT for the compounded random walk, we embed the steps from the discrete-time random walk above onto real values of time t>0 . Suppose that for some \Delta t>0 , the particle undergoes a compounded stepping process, where in the time interval ((m-1)\Delta t,m\Delta t] , the particle steps K_{m} times, where K_{m}\geq 1 is an i.i.d. random variable, or it has been absorbed into \Omega^{\prime} . The probability of making k steps in the interval is denoted as p(k)=P(K_{m}=k) . For convenience, we will assume that each of these K_{m} steps are uniformly spaced in time over the interval with the last step occurring at time m\Delta t . The total number of steps taken

Read the complete text