Sisyphus random walks in the presence of moving traps

Shahar Hod · 2025

Read Sisyphus random walks in the presence of moving traps by Shahar Hod (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-2503-08171.txt.

Introduction

A random walker that keeps being put back to the start, like Sisyphus and his boulder, faces a different fate when a trap can catch it. Shahar Hod, a solo author, proves in 2025 that a slow-moving trap changes the walker's survival odds in a sharp way. With a fixed trap, earlier work showed survival chances fall off exponentially fast. When the trap's speed fades toward zero in step with one over time, survival instead fades as a gentler power law, keeping walkers alive far longer.

Catalog record

CREATORShahar Hod
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE13 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-08171

Browse the full library · About the Tower of Babel

Full text

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

Sisyphus random walks in the presence of moving traps Shahar Hod Abstract It has recently been proved that, in the presence of a static absorbing trap, Sisyphus random walkers with a restart mechanism are characterized by exponentially decreasing asymptotic survival probability functions. Interestingly, in the present compact paper we prove analytically that, in the presence of a moving trap whose velocity approaches zero asymptotically in time as v trap ∼ 1 / t v_{\text{trap}}\sim 1/t , the survival probabilities of the Sisyphus walkers are dramatically changed into inverse power-law decaying tails. I Introduction The dynamics of random walkers in the presence of absorbing traps have attracted much attention from physicists and mathematicians during the last three decades RW1 ; RW2 ; RW3 ; RW4 ; RW5 ; RW6 . In particular, random walk models provide useful physical insights in various studies of complex adaptive systems DV2 ; Hodetl , granular segregation models DV3 , and polymer adsorption phenomena DV4 . Random walk models are also useful in the mathematical analysis of absorbing-state phase transitions DV1 . In addition, some biological toy models of epidemic spreading are based on the physical and mathematical properties of biased random walkers DV5 . One of the most important mathematical quantities that characterizes the dynamics of random walkers in the presence of absorbing traps is the survival probability function S(t) of the population. This physically interesting quantity describes the time-dependent probability that a random walker has not been absorbed by the trap after taking t steps. It is well known that, for biased random walkers who are characterized by a constant drift velocity towards the absorbing trap, the asymptotic survival probability decays exponentially in time DV2 ; Jsp , S(t\to\infty)\sim t^{-3/2}e^{-\gamma t}\ \ \ \ \text{with}\ \ \ \ \gamma(q)=2\sqrt{q(1-q)}\ , (1) where q>1/2 is the probability of the random walker to step towards the trap (and 1-q<1/2 is the complementary probability of the walker to step in the opposite direction of the trap). Interestingly, it has been proved in Jsp that if the absorbing trap moves with a constant (time-independent) velocity which equals the average drift velocity v_{\text{drift}}=2q-1>0 of the walkers, then the temporal behavior of the survival probability function changes dramatically. In particular, one finds that in this case the survival probability decays asymptotically as an inverse power of time (instead of exponentially in time) DV2 ; Jsp : S(t)\sim t^{-1/2}\ \ \ \ \text{for}\ \ \ \ v_{\text{trap}}=2q-1\ . (2) In recent years there is a growing interest of physicists and mathematicians in analyzing the properties of the Sisyphus random walk model RS1 ; RS2 ; RS3 ; RS4 ; RS5 ; Sisy ; Hodanp ; Hodbia , a stochastic process with a simple restart mechanism in which the random walker, regardless of her current position, has a finite probability 1-q to return to her initial position (which, we assume, is located at the origin). In this physically interesting model, the asymptotic survival probability of the walkers in the presence of a static trap decays exponentially in time for all physically acceptable values q\in(0,1) of the biased jumping probability Hodbia : S(t\to\infty)\sim e^{-\gamma_{\text{Sis}}t}\ \ \ \ \text{with}\ \ \ \ \gamma_{\text{Sis}}(x_{T}\gg 1)=-\ln[1-q^{x_{T}}(1-q)]\ , (3) where x_{T} is the location of the trap (the initial distance between the walkers and the absorbing trap). Similarly to the case of ordinary random walks (with no restart mechanisms) Jsp , it is natural to expect that the asymptotic late-time behavior of the survival probability function of the Sisyphus random walkers may be changed dramatically if the absorbing trap moves away from the starting point of the walkers. Based on this expectation, we here raise the following physically interesting question: How fast should the trap move in order to change the qualitative behavior of the survival probability function of the Sisyphus random walkers from an exponentially decaying function [see Eq. ( 3 )] into an inverse power-law decaying function? As we shall explicitly prove below, the answer to the above stated question is highly non-trivial. In particular, in this paper we shall reveal the somewhat surprising fact that, contrary to the familiar case of ordinary random walk models with no restart mechanisms in which the survival probability function decays as an inverse power of time if the trap moves with a constant (non-decreasing in time) velocity [see Eq. ( 2 )] Jsp , the survival probability function that characterizes the dynamics of the Sisyphus random walkers decays asymptotically as an inverse power of time in the presence of moving traps whose velocities go to zero asymptotically as v_{\text{trap}}\sim 1/t . II Description of the system We analyze the survival probability function S[t;x_{T}(t),q] which characterizes the dynamics of biased Sisyphus random walkers in the presence of a moving trap. A Sisyphus agent, who starts her dynamics at the origin x_{0}\equiv x(t=0)=0 , is characterized by a finite probability q to take a unit step towards the absorbing trap, whose location x_{T}(t)>0 is a non-decreasing function of time, and a complementary probability 1-q to jump back all the way to her initial location at the origin. The dynamics of a biased Sisyphus random walker is therefore governed by the mathematically compact jumping rule x(t+1)=\begin{cases}x(t)+1&\text{with probability}\ \ q\ ;\\ 0&\text{with probability}\ \ 1-q\ ,\end{cases} (4) supplemented by the normalized boundary condition N_{\text{tot}}(t)=1\ \ \ \ \text{for}\ \ \ \ t<x_{T}(t)\ . (5) Here N_{\text{tot}}(t)=\sum_{k=0}^{k=x_{T}(t)-1}N_{k}(t)\ (6) is the time-dependent total number of Sisyphus walkers [normalized to N_{\text{tot}}(t=0)=1 at the beginning of the dynamics] who have not been absorbed by the moving trap after taking t steps according to the jumping rule ( 4 ), and N_{k}(t) is the number of agents who are located at 0\leq x(t)=k<x_{T}(t) at time t . III The characteristic recurrence relation for the survival probability function In the present section we shall use analytical techniques in order to derive the recurrence relation [see Eq. ( 12 ) below] that determines the functional behavior of the survival probability function S(t) , the time-dependent (monotonically decreasing) fraction of Sisyphus walkers who have not been absorbed by the moving trap after taking t steps. This important physical quantity is defined by the relation [see the normalization ( 5 )] S[t;x_{T}(t),q]\equiv N_{\text{tot}}(t)\ . (7) We first note that, taking cognizance of the jumping rule ( 4 ), one finds that the number d(t)\equiv N_{\text{tot}}(t-1)-N_{\text{tot}}(t)\ (8) of walkers who, after taking exactly t steps, are absorbed by the trap at x=x_{T}(t) is given by the relation d(t)=q\cdot N_{x_{T}(t)-1}(t-1)=q^{2}\cdot N_{x_{T}(t)-2}(t-2)=\cdots=Q\cdot N_{0}[t-x_{T}(t)]\ , (9) where Q\equiv q^{x_{T}(t)}\ . (10) Here we have used the jumping rule ( 4 ), according to which an agent who is located at the origin at time t-x_{T}(t) has to go x_{T}(t) steps in a row to the right (towards the absorbing trap) in order to be absorbed at time t by the trap, which is now located at x_{T}(t) [remember that each step to the right is characterized by the finite probability q ]. The normalized number N_{0}[t-x_{T}(t)] of walkers who are located at the origin at time t-x_{T}(t) is given by N_{0}[t-x_{T}(t)]=(1-q)\cdot N_{\text{tot}}[t-x_{T}(t)-1]\ . (11) Here we have used the jumping rule ( 4 ), according to which a Sisyphus random walker has a finite probability

Read the complete text