arXiv subjects
Publications and source records attributed to Satya N. Majumdar.
Motivated by a putative model of black holes as quantum objects we consider what types of operators would have a corresponding spectrum. We find that there are indeed such operators, but of a rather unusual types, and for which the wave functions are only barely localized. We point out a tension between such almost delocalized states and black holes as compact objects.
In this paper, we introduce a new stochastic process of $N$ interacting particles on the line that evolve via Dyson Brownian motion (DBM) with Dyson's index $β> 0$ and undergo simultaneous resetting to their initial positions at a constant rate $r$. We call this process the resetting Dyson Brownian motion (RDBM) -- in short the $β$-RDBM. For $β= 1,2,4$, the positions of the particles in the RDBM can be interpreted as the eigenvalues of a random matrix ensemble where the entries of an $N x N$ Gaussian matrix evolve as simultaneously resetting Brownian motions (with rate $r$) in the presence or absence of a harmonic trap. For $r=0$ and in the presence of a harmonic trap, this system reaches an equilibrium Gibbs-Boltzmann state of the so called Dyson log-gas. However, the stochastic resetting drives the system at long time to a nonequilibrium stationary state (NESS). We compute exactly the joint distribution of the positions of the particles in this NESS for all $β>0$ and calculate several observables for large $N$: the average density profile of the gas, the extreme value statistics, the spacing between two consecutive particles and the full counting statistics. We show that a nonzero resetting rate $r>0$ drastically changes the nature of the fluctuations in the stationary state: while the log-gas is rather rigid, the $β$-RDBM in its NESS becomes fluffy, i.e., the fluctuations of different observables are of the same order as their mean. In the absence of a harmonic trap, our results for the $β= 2$-RDBM can be related to nonintersecting Brownian motions in the presence of resetting. Our model demonstrates interesting effects arising from the interplay between the eigenvalue repulsion and the all-to-all attraction (generated by stochastic resetting) in an interacting particle system. Numerical simulations are in excellent agreement with our analytical results.
We study the nonequilibrium stationary state (NESS) induced by quantum resetting of a system of $N$ noninteracting bosons in a harmonic trap. Our protocol consists of preparing initially the system in the ground state of a harmonic oscillator centered at $+a$, followed by a rapid quench where the center is shifted to $-a$ and the system is allowed to evolve unitarily up to a random Poissonian time $τ$ distributed via $r\, e^{-r τ}$. Then the trap center is reset to $+a$ again and the system is assumed to cool instantaneously to the initial ground state. The system is again allowed to evolve unitarily in the trap centered at $-a$ up to a random time, and the procedure is repeated. Under repeated resetting, the system reaches a NESS where the positions of bosons get strongly correlated due to simultaneous resetting induced by the trap. We fully characterize the steady state by analytically computing several physical observables such as the average density, extreme value statistics, order and gap statistics, and also the distribution of the number of particles in a region $[-L,L]$, known as the full counting statistics (FCS). In particular, we show that in the large $N$ limit, the scaling function describing the FCS exhibits a striking feature: it is supported over a nontrivial finite interval, and moreover is discontinuous at an interior point of the support. Our results are supported by numerical simulations. This is a rare example of a strongly correlated quantum many-body NESS where various observables can be exactly computed.
We study the relaxation of a diffusive particle confined in an arbitrary external potential and subject to a non-Markovian resetting protocol. With a constant rate $r$, a previous time $τ$ between the initial time and the present time $t$ is chosen from a given probability distribution $K(τ,t)$, and the particle is reset to the position that it occupied at time $τ$. Depending on the shape of $K(τ,t)$, the particle either relaxes toward the Gibbs-Boltzmann distribution or toward a non-trivial stationary distribution that breaks ergodicity and depends on the initial position and the resetting protocol. From a general asymptotic theory, we find that if the kernel $K(τ,t)$ is sufficiently localized near $τ=0$, i.e., mostly the initial part of the trajectory is remembered and revisited, the steady state is non-Gibbs-Boltzmann. Conversely, if $K(τ,t)$ decays slowly enough or increases with $τ$, i.e., recent positions are more likely to be revisited, the probability distribution of the particle tends toward the Gibbs-Boltzmann state at large times. In the latter case, however, the temporal approach to the stationary state is generally anomalously slow, following for instance an inverse power law or a stretched exponential, if $K(τ,t)$ is not too strongly peaked at the current time $t$. These findings are verified by the analysis of several exactly solvable cases and by numerical simulations.
We investigate the overdamped dynamics of a `passive' particle driven by nonreciprocal interaction with a `driver' Brownian particle. When the interaction between them is short-ranged, the long-time behavior of the driven particle is remarkably universal -- the mean-squared displacement (MSD) and the typical position of the driven particle exhibits the same qualitative behaviors independent of the specific form of the potential. In particular, the MSD grows as $t^{1/2}$ in one dimension and $\log t$ in two spatial dimensions. We compute the exact scaling functions for the position distribution in $d=1$ and $d=2$. These functions are universal when the interaction is short-ranged. For long-ranged interactions, the MSD of the driven particle grows as $t^ϕ$ with exponent $ϕ$ depending on the tail of the potential.
We study a simple one dimensional sluggish random walk model with subdiffusive growth. In the continuum hydrodynamic limit, the model corresponds to a particle diffusing on a line with a space dependent diffusion constant D(x)= |x|^{-α} and a drift potential U(x)=|x|^{-α}, where α\geq 0 parametrizes the model. For α=0 it reduces to the standard diffusion, while for α>0 it leads to a slow subdiffusive dynamics with the distance scaling as x\sim t^μ at late times with μ= 1/(α+2)\leq 1/2. In this paper, we compute exactly, for all α\ge 0, the full probability distributions of three observables for a sluggish walker of duration T starting at the origin: (i) the occupation time t_+ denoting the time spent on the positive side of the origin, (ii) the last passage time t_{\rm l} through the origin before T, and (iii) the time t_M at which the walker is maximally displaced on the positive side of the origin. We show that while for α=0 all three distributions are identical and exhibit the celebrated arcsine laws of Lévy, they become different from each other for any α>0 and have nontrivial shapes dependent on α. This generalizes the Lévy's three arcsine laws for normal diffusion (α=0) to the subdiffusive sluggish walker model with a general α\geq 0. Numerical simulations are in excellent agreement with our analytical predictions.
We study a system of $N$ noninteracting particles on a line in the presence of a harmonic trap $U(x)=μ\bigl[x-z(t)\bigr]^2/2$, where the trap center $z(t)$ undergoes a bounded stochastic modulation. We show that this stochastic modulation drives the system into a nonequilibrium stationary state, where the joint distribution of the positions of the particles is not factorizable. This indicates strong correlations between the positions of the particles that are not inbuilt, but rather get generated by the dynamics itself. Moreover, we show that the stationary joint distribution can be fully characterized and has a special conditionally independent and identically distributed (CIID) structure. This special structure allows us to compute several observables analytically even in such a strongly correlated system, for an arbitrary bounded drive $z(t)$. These observables include the average density profile, the correlations between particle positions, the order and gap statistics, as well as the full counting statistics. We then apply our general results to two specific examples where (i) $z(t)$ represents a dichotomous telegraphic noise, and (ii) $z(t)$ represents an Ornstein-Uhlenbeck process. Our analytical predictions are verified in numerical simulations, finding excellent agreement.
We introduce the profligacy of a search process as a competition between its expected cost and the probability of finding the target. The arbiter of the competition is a parameter $λ$ that represents how much a searcher invests into increasing the chance of success. Minimizing the profligacy with respect to the search strategy specifies the optimal search. We show that in the case of diffusion with stochastic resetting, the amount of resetting in the optimal strategy has a highly nontrivial dependence on model parameters resulting in classical continuous transitions, discontinuous transitions and tricritical points as well as non-standard discontinuous transitions exhibiting re-entrant behavior and overhangs.
We study a one-dimensional run-and-tumble particle (RTP), which is a prototypical model for active system, moving within an arbitrary external potential. Using backward Fokker-Planck equations, we derive the differential equation satisfied by its mean first-passage time (MFPT) to an absorbing target, which, without any loss of generality, is placed at the origin. Depending on the shape of the potential, we identify four distinct ``phases'', with a corresponding expression for the MFPT in every case, which we derive explicitly. To illustrate these general expressions, we derive explicit formulae for two specific cases which we study in detail: a double-well potential and a logarithmic potential. We then present different applications of these general formulae to (i) the generalization of the Kramer's escape law for an RTP in the presence of a potential barrier, (ii) the ``trapping'' time of an RTP moving in a harmonic well and (iii) characterizing the efficiency of the optimal search strategy of an RTP subjected to stochastic resetting. Our results reveal that the MFPT of an RTP in an external potential exhibits a far more complex and, at times, counter-intuitive behavior compared to that of a passive particle (e.g., Brownian) in the same potential.
We provide a uniform framework to compute the exact distribution of the number of minima/maxima in three different random walk landscape models in one dimension. The landscape is generated by the trajectory of a discrete-time continuous space random walk with arbitrary symmetric and continuous jump distribution at each step. In model I, we consider a ``free'' random walk of $N$ steps. In model II, we consider a ``meander landscape'' where the random walk, starting at the origin, stays non-negative up to $N$ steps. In model III, we study a ``first-passage landscape'' which is generated by the trajectory of a random walk that starts at the origin and stops when it crosses the origin for the first time. We demonstrate that while the exact distribution of the number of minima is different in the three models, for each model it is universal for all $N$, in the sense that it does not depend on the jump distribution as long as it is symmetric and continuous. In the last two cases we show that this universality follows from a non trivial mapping to the Sparre Andersen theorem known for the first-passage probability of discrete-time random walks with symmetric and continuous jump distribution. Our analytical results are in excellent agreement with our numerical simulations.
In this paper, we consider the problem of numerical investigation of the counting statistics for a class of one-dimensional systems. Importance sampling, the cornerstone technique usually implemented for such problems, critically hinges on selecting an appropriate biased distribution. While exponential tilt in the observable stands as the conventional choice for various problems, its efficiency in the context of counting statistics may be significantly hindered by the genuine discreteness of the observable. To address this challenge, we propose an alternative strategy which we call importance sampling with the local tilt. We demonstrate the efficiency of the proposed approach through the analysis of three prototypical examples: a set of independent Gaussian random variables, Dyson gas, and Symmetric Simple Exclusion Process (SSEP) with a steplike initial condition.
In this paper, we study a simple model of a diffusive particle on a line, undergoing a stochastic resetting with rate $r$, via rescaling its current position by a factor $a$, which can be either positive or negative. For $|a|<1$, the position distribution becomes stationary at long times and we compute this limiting distribution exactly for all $|a|<1$. This symmetric distribution has a Gaussian shape near its peak at $x=0$, but decays exponentially for large $|x|$. We also studied the mean first-passage time (MFPT) $T(0)$ to a target located at a distance $L$ from the initial position (the origin) of the particle. As a function of the initial position $x$, the MFPT $T(x)$ satisfies a nonlocal second order differential equation and we have solved it explicitly for $0 \leq a < 1$. For $-1<a\leq 0$, we also solved it analytically but up to a constant factor $κ$ whose value can be determined independently from numerical simulations. Our results show that, for all $-1<a<1$, the MFPT $T(0)$ (starting from the origin) shows a minimum at $r=r^*(a)$. However, the optimised MFPT $T_{\rm opt}(a)$ turns out to be a monotonically increasing function of $a$ for $-1<a<1$. This demonstrates that, compared to the standard resetting to the origin ($a=0$), while the positive rescaling is not beneficial for the search of a target, the negative rescaling is. Thus resetting via rescaling followed by a reflection around the origin expedites the search of a target in one dimension.
In this Chapter, we consider a model of $N$ independent random walkers, each of duration $t$, and each starting from the origin, on a lattice in $d$ dimensions. We focus on two observables, namely $D_N(t)$ and $C_N(t)$ denoting respectively the number of distinct and common sites visited by the walkers. For large $t$, where the lattice random walkers converge to independent Brownian motions, we compute exactly the mean $\langle D_N(t) \rangle$ and $\langle C_N(t) \rangle$. Our main interest is on the $N$-dependence of these quantities. While for $\langle D_N(t) \rangle$ the $N$-dependence only appears in the prefactor of the power-law growth with time, a more interesting behavior emerges for $\langle C_N(t) \rangle$. For this latter case, we show that there is a ``phase transition'' in the $(N, d)$ plane where the two critical line $d=2$ and $d=d_c(N) = 2N/(N-1)$ separate three phases of the growth of $\langle C_N(t)\rangle$. The results are extended to the mean number of sites visited exactly by $K$ of the $N$ walkers. Furthermore in $d=1$, the full distribution of $D_N(t)$ and $C_N(t)$ are computed, exploiting a mapping to the extreme value statistics. Extensions to two other models, namely $N$ independent Brownian bridges and $N$ independent resetting Brownian motions/bridges are also discussed.
We consider a discrete-time random walk on a line starting at $x_0\geq 0$ where a cost is incurred at each jump. We obtain an exact analytical formula for the distribution of the total cost of a trajectory until the process crosses the origin for the first time. The formula is valid for arbitrary jump distribution and cost function (heavy- and light-tailed alike), provided they are symmetric and continuous. We analyze the formula in different scaling regimes, and find a high degree of universality with respect to the details of the jump distribution and the cost function. Applications are given to the motion of an active run-and-tumble particle in one dimension and extensions to multiple cost variables are considered. The analytical results are in perfect agreement with numerical simulations.
The study of diffusion with preferential returns to places visited in the past has attracted an increased attention in recent years. In these highly non-Markov processes, a standard diffusive particle intermittently resets at a given rate to previously visited positions. At each reset, a position to be revisited is randomly chosen with a probability proportional to the accumulated amount of time spent by the particle at that position. These preferential revisits typically generate a very slow diffusion, logarithmic in time, but still with a Gaussian position distribution at late times. Here we consider an active version of this model, where between resets the particle is self-propelled with constant speed and switches direction in one dimension according to a telegraphic noise. Hence there are two sources of non-Markovianity in the problem. We exactly derive the position distribution in Fourier space, as well as the variance of the position at all times. The crossover from the short-time ballistic regime, dominated by activity, to the large-time anomalous logarithmic growth induced by memory is studied. We also analytically derive a large deviation principle for the position, which exhibits a logarithmic time-scaling instead of the usual algebraic form. Interestingly, at large distances, the large deviations become independent of time and match the non-equilibrium steady state of a particle under resetting to its starting position only.
We consider a system of non-interacting Brownian particles on the line with steplike initial condition and study the statistics of the occupation time on the positive half-line. We demonstrate that this system exhibits long-lasting memory effects of the initialization. Specifically, we calculate the mean and the variance of the occupation time, demonstrating that the memory effects in the variance are determined by a generalized compressibility (or Fano factor), associated with the initial condition. In the particular case of the uncorrelated uniform initial condition we conduct a detailed study of two probability distributions of the occupation time: annealed (averaged over all possible initial configurations) and quenched (for a typical configuration). We show that at large times both the annealed and the quenched distributions admit large deviation form and we compute analytically the associated rate functions. We verify our analytical predictions via numerical simulations using Importance Sampling Monte-Carlo strategy.
We investigate the full counting statistics (FCS) of a harmonically confined 1d short-range Riesz gas consisting of $N$ particles in equilibrium at finite temperature. The particles interact with each other through a repulsive power-law interaction with an exponent $k>1$ which includes the Calogero-Moser model for $k=2$. We examine the probability distribution of the number of particles in a finite domain $[-W, W]$ called number distribution, denoted by $\mathcal{N}(W, N)$. We analyze the probability distribution of $\mathcal{N}(W, N)$ and show that it exhibits a large deviation form for large $N$ characterised by a speed $N^{\frac{3k+2}{k+2}}$ and by a large deviation function of the fraction $c = \mathcal{N}(W, N)/N$ of the particles inside the domain and $W$. We show that the density profiles that create the large deviations display interesting shape transitions as one varies $c$ and $W$. This is manifested by a third-order phase transition exhibited by the large deviation function that has discontinuous third derivatives. Monte-Carlo (MC) simulations show good agreement with our analytical expressions for the corresponding density profiles. We find that the typical fluctuations of $\mathcal{N}(W, N)$, obtained from our field theoretic calculations are Gaussian distributed with a variance that scales as $N^{ν_k}$, with $ν_k = (2-k)/(2+k)$. We also present some numerical findings on the mean and the variance. Furthermore, we adapt our formalism to study the index distribution (where the domain is semi-infinite $(-\infty, W])$, linear statistics (the variance), thermodynamic pressure and bulk modulus.
We compute the connected two-time correlator of the maximum $M_N(t)$ of $N$ independent Gaussian stochastic processes (GSP) characterised by a common correlation coefficient $ρ$ that depends on the two times $t_1$ and $t_2$. We show analytically that this correlator, for fixed times $t_1$ and $t_2$, decays for large $N$ as a power law $N^{-γ}$ (with logarithmic corrections) with a decorrelation exponent $γ= (1-ρ)/(1+ ρ)$ that depends only on $ρ$, but otherwise is universal for any GSP. We study several examples of physical processes including the fractional Brownian motion (fBm) with Hurst exponent $H$ and the Ornstein-Uhlenbeck (OU) process. For the fBm, $ρ$ is only a function of $τ= \sqrt{t_1/t_2}$ and we find an interesting ``freezing'' transition at a critical value $τ= τ_c=(3-\sqrt{5})/2$. For $τ< τ_c$, there is an optimal $H^*(τ) > 0$ that maximises the exponent $γ$ and this maximal value freezes to $γ= 1/3$ for $τ>τ_c$. For the OU process, we show that $γ= {\rm tanh}(μ\,|t_1-t_2|/2)$ where $μ$ is the stiffness of the harmonic trap. Numerical simulations confirm our analytical predictions.