arXiv subjects

Rami Atar

Publications and source records attributed to Rami Atar.

At least 19 recordsLinked to original sources

arXiv2026

Mean field limits of large Jackson networks in heavy traffic

We consider an open Jackson network with $n$ exchangeable single-server stations and weak all-to-all interaction through routing: upon service completion at station $i$, a job is routed to station $j$ with probability $p/n$, where $p\in(0,1)$, or leaves the system with probability $q=1-p$. We study a joint asymptotic regime in which the number of stations tends to infinity while the system approaches heavy traffic. Under the critical-load condition and diffusive scaling of time and queue length, we prove propagation of chaos for the queue-length and cumulative-idleness processes. The limiting McKean--Vlasov dynamics are described by the nonlinear reflected Brownian motion \[ \bar X(t)=\bar X_0+\bar W(t)+\hat\gamma t+\bar L(t)-p\,\mathbb{E} \bar L(t), \] where $\bar W$ is a Brownian motion with variance parameter $2$ and $\bar L$ is the reflection term at the origin. The proof proceeds by coupling the rescaled queueing network to a Brownian particle system interacting through boundary local times. A key step is a decoupling construction that replaces the correlated routing noise in the prelimit by asymptotically equivalent independent driving noises.

math.PR↗

arXiv2026

High-dimensional limits for reflected Brownian motion in the orthant

We study interacting Brownian particles on the half-line whose interaction occurs through boundary local times at the origin. The particle system is given by \[ X_i^n(t)=X^n_{0,i}+W_i^n(t)+L_i^n(t) +\frac{1}{n-1}\sum_{j\ne i}\rho^n_{ij}L_j^n(t), \qquad i\in[n],\ t\ge0, \] where the initial conditions are exchangeable, the driving Brownian motions $W_i^n$ are i.i.d., and $L_i^n$ denotes the boundary local time of $X_i^n$ at zero. For each fixed coefficient array $\{\rho^n_{ij}\}$, the system can be viewed as a semimartingale reflected Brownian motion in the orthant. We first consider the homogeneous case $\rho^n_{ij}=a$. In this case, global well-posedness holds under the completely-$\mathcal S$ condition $a>-1$. We prove propagation of chaos under this condition; the subregime $a\in(-1,0]$, in the homogeneous setting, was previously covered as part of the results of \cite{baker2025particle}. The limiting process is the nonlinear reflected Brownian motion \[ \bar X(t)=\bar X_0+\bar W(t)+\bar L(t)+a\mathbb E[\bar L(t)], \qquad t\ge0. \] We also treat heterogeneous random coefficients $\rho^n_{ij}$, assumed to have mean $a$, support in a compact subset of $(-1,1)$, and to be independent across $j$ for each $i$. In both the quenched and annealed settings, the particle system converges to the same McKean--Vlasov limit as in the homogeneous case. The model is motivated by large Jackson networks in heavy traffic.

math.PR↗

arXiv2026

Trimmed branching random walk and a free obstacle problem

Consider $N$ particles performing random walks on the $\epsilon$-grid $(\epsilon Z)^d$, $\epsilon>0$ with branching and density-dependent selection: When one of the particles branches, a particle is removed from the most populated site. The walks are assumed to be asymptotic, as $\epsilon\to0$, to diffusion processes of the form \[ dX_i(t)=b(X_i(t))dt+\sqrt{2}dW_i(t), \] for $b$ a given vector field. Denoting $L^*=\Delta-\nabla\cdot(b\,\cdot)$, the hydrodynamic limit, as $N\to\infty$ followed by $\epsilon\to0$, is characterized in terms of a parabolic free obstacle problem \[ \partial_t u=L^*u+u-\beta \] where $\beta$ is a measure on $R^d\times[0,\infty)$ supported on $\{(x,t):u(x,t)=|u(\cdot,t)|_\infty\}$. Here, the unknowns are $u$, the mass density, and $\beta$, the removal measure, for which $t\mapsto\beta(R^d\times[0,t])$ is prescribed. This is analogous to the well-understood relation between particle systems with spatial selection and free boundary problems, but the techniques require quite different ideas. The key ingredients of the proof include PDE uniqueness for continuous densities and a uniform-in-$\epsilon$ estimate on modulus of continuity of prelimit densities. The work gives rise to open problems such as ``flat top'' versus ``sharp top'' solutions, which are discussed based on concrete examples.

math.PR↗

arXiv2025

On free boundary problems for the Atlas model

For $n\in\mathbb{N}$, let $\{X^n_i\}$ be an infinite collection of Brownian particles on the real line where the leftmost particle $\min_iX^n_i(t)$ is given a drift $n$, and let $\mu^n_t=n^{-1}\sum_i\delta_{X^n_i(t)}$, $t\ge0$ denote the normalized configuration measure. The case where the initial particle positions follow a Poisson point process on $[0,\infty)$ of intensity $n\lambda$, $\lambda>0$ was studied where it was shown that $\mu^n_t$ converge, as $n\to\infty$, to a limit characterized by a Stefan problem of melting solid (respectively, freezing supercooled liquid) type when $\lambda\ge 2$ (respectively, $0<\lambda<2$). In this paper it is assumed that $\mu^n_0\to\mu_0$ in probability, where $\mu_0$ is supported on $[0,\infty)$ and satisfies a polynomial growth condition. Because $(y-x)^{-1}\mu_0((x,y])$, $0 0$, the free boundary exists as a continuous trajectory, and the process determined by the leftmost particle converges to it.

math.PR↗

arXiv2025

A Durrett-Remenik particle system in $\mathbb{R}^d$

This paper studies a branching-selection model of motionless particles in $\mathbb{R}^d$, with nonlocal branching, introduced by Durrett and Remenik in dimension $1$. The assumptions on the fitness function, $F$, and on the inhomogeneous branching distribution, are mild. The evolution equation for the macroscopic density is given by an integro-differential free boundary problem in $\mathbb{R}^d$, in which the free boundary represents the least $F$-value in the population. The main result is the characterization of the limit in probability of the empirical measure process in terms of the unique solution to this free boundary problem.

math.PR↗

arXiv2025

Heavy traffic limit with discontinuous coefficients via a non-standard semimartingale decomposition

This paper studies a single server queue in heavy traffic, with general inter-arrival and service time distributions, where arrival and service rates vary discontinuously as a function of the (diffusively scaled) queue length. It is proved that the weak limit is given by the unique-in-law solution to a stochastic differential equation in $[0,\infty)$ with discontinuous drift and diffusion coefficients. The main tool is a semimartingale decomposition for point processes introduced in \cite{dal-miy}, which is distinct from the Doob-Meyer decomposition of a counting process. Whereas the use of this tool is demonstrated here for a particular model, we believe it may be useful for investigating the scaling limits of queueing models very broadly.

math.PR↗

arXiv2024

Rank-based stochastic differential inclusions and diffusion limits for a load balancing model

In an earlier paper, a randomized load balancing model was studied in a heavy traffic asymptotic regime where the load balancing stream is thin compared to the total arrival stream. It was shown that the limit is given by a system of rank-based Brownian particles on the half-line. This paper extends these results from the case of exponential service time to an invariance principle, where service times have finite second moment. The main tool is a new notion of rank-based stochastic differential inclusion, which may be of interest in its own right.

math.PR↗

arXiv2024

Diffusion limits in the quarter plane and non-semimartingale reflected Brownian motion

We consider a continuous-time random walk in the quarter plane for which the transition intensities are constant on each of the four faces $(0,\infty)^2$, $F_1=\{0\}\times(0,\infty)$, $F_2=(0,\infty)\times\{0\}$ and $\{(0,0)\}$. We show that when rescaled diffusively it converges in law to a Brownian motion with oblique reflection direction $d^{(i)}$ on face $F_i$, $i=1,2$, defined via the Varadhan-Williams submartingale problem. A parameter denoted by $\alpha$ was introduced in \cite{vw}, measuring the extent to which $d^{(i)}$ are inclined toward the origin. In the case of the quarter plane, $\alpha$ takes values in $(-2,2)$, and it is known that the reflected Brownian motion is a semimartingale if and only if $\alpha\in(-2,1)$. Convergence results via both the Skorohod map and the invariance principle for semimartingale reflected Brownian motion are known to hold in various settings in arbitrary dimension. In the case of the quarter plane, the invariance principle was proved for $\alpha \in (-2,1)$ whereas for tools based on the Skorohod map to be applicable it is necessary (but not sufficient) that $\alpha \in [-1,1)$. Another tool that has been used to prove convergence in general dimension is the extended Skorohod map, which in the case of the quarter plane provides convergence for $\alpha=1$. This paper focuses on the range $\alpha \in (1,2)$, where the Skorohod problem and the extended Skorohod problem do not possess a unique solution, the limit process is not a semimartingale, and convergence to reflected Brownian motion has not been shown before. The result has implications on the asymptotic analysis of two Markovian queueing models: The {\it generalized processor sharing model with parallelization slowdown}, and the {\it coupled processor model}.

math.PR↗

arXiv2024

Invariance principle and McKean-Vlasov limit for randomized load balancing in heavy traffic

We consider a load balancing model where a Poisson stream of jobs arrive at a system of many servers whose service time distribution possesses a finite second moment. A small fraction of arrivals pass through the so called power-of-choice algorithm, which assigns a job to the shortest among $\ell$, $\ell\ge 2$, randomly chosen queues, and the remaining jobs are assigned to queues chosen uniformly at random. The system is analyzed at critical load in an asymptotic regime where both the number of servers and the usual heavy traffic parameter associated with individual queue lengths grow to infinity. The first main result is a hydrodynamic limit, where the empirical measure of the diffusively normalized queue lengths is shown to converge to a path in measure space whose density is given by the unique solution of a parabolic PDE with nonlocal coefficients. Further, two forms of an invariance principle are proved, corresponding to two different assumptions on the initial distribution, where individual normalized queue lengths converge weakly to solutions of SDE. In one of these results, the limit is given by a McKean-Vlasov SDE, and propagation of chaos holds. The McKean-Vlasov limit is closely related to limit results for Brownian particles on $\mathbb{R}_+$ interacting through their rank (with a specific interaction). However, an entirely different set of tools is required, as the collection of $n$ prelimit particles does not obey a Markovian evolution on $\mathbb{R}_+^n$.

math.PR↗

arXiv2022

Asymptotic optimality of switched control policies in a simple parallel server system under an extended heavy traffic condition

This paper studies a 2-class, 2-server parallel server system under the recently introduced extended heavy traffic condition, which states that the underlying 'static allocation' linear program (LP) is critical, but does not require that it has a unique solution. The main result is the construction of policies that asymptotically achieve a lower bound, proved in [1], on an expected discounted linear combination of diffusion-scaled queue lengths, and are therefore asymptotically optimal (AO). Each extreme point solution to the LP determines a control mode, i.e., a set of activities (class--server pairs) that are operational. When there are multiple solutions, these modes can be selected dynamically. It is shown that the number of modes required for AO is either one or two. In the latter case there is a switching point in the (normalized) workload domain, characterized in terms of a free boundary problem. Our policies are defined by identifying pairs of elementary policies and switching between them at this switching point. They provide the first example in the heavy traffic literature where weak limits under an AO policy are given by a diffusion process where both the drift and diffusion coefficients are discontinuous.

math.OC↗

arXiv2022

Scheduling in the high uncertainty heavy traffic regime

We propose a model uncertainty approach to heavy traffic asymptotics that allows for a high level of uncertainty. That is, the uncertainty classes of underlying distributions accommodate disturbances that are of order 1 at the usual diffusion scale, as opposed to asymptotically vanishing disturbances studied previously in relation to heavy traffic. A main advantage of the approach is that the invariance principle underlying diffusion limits makes it possible to define uncertainty classes in terms of the first two moments only. The model we consider is a single server queue with multiple job types. The problem is formulated as a zero sum stochastic game played between the system controller, who determines scheduling and attempts to minimize an expected linear holding cost, and an adversary, who dynamically controls the service time distributions of arriving jobs, and attempts to maximize the cost. The heavy traffic asymptotics of the game are fully solved. It is shown that an asymptotically optimal policy for the system controller is to prioritize according to an index rule and for the adversary it is to select distributions based on the system's current workload. The workload-to-distribution feedback mapping is determined by an HJB equation, which also characterizes the game's limit value. Unlike in the vast majority of results in the heavy traffic theory, and as a direct consequence of the diffusive size disturbances, the limiting dynamics under asymptotically optimal play are captured by a stochastic differential equation where both the drift and the diffusion coefficients may be discontinuous.

math.PR↗

arXiv2022

Parallel server systems under an extended heavy traffic condition: A lower bound

The standard setting for studying parallel server systems (PSS) at the diffusion scale is based on the heavy traffic condition (HTC), which assumes that the underlying static allocation linear program (LP) is critical and has a unique solution. This solution determines the graph of basic activities, which identifies the set of activities (i.e., class-server pairs) that are operational. In this paper we explore the extended HTC, where the LP is merely assumed to be critical. Because multiple solutions are allowed, multiple sets of operational activities, referred to as modes, are available. Formally, the scaling limit for the control problem associated with the model is given by a so called workload control problem (WCP) in which a cost associated with a diffusion process is to be minimized by dynamically switching between these modes. Our main result is that the WCP's value constitutes an asymptotic lower bound on the cost associated with the PSS model.

math.OC↗

arXiv2021

Large-time limit of nonlinearly coupled measure-valued equations that model many-server queues with reneging

The large-time behavior of a nonlinearly coupled pair of measure-valued transport equations with discontinuous boundary conditions, parameterized by a positive real-valued parameter $\lambda$, is considered. These equations describe the hydrodynamic or fluid limit of many-server queues with reneging (with traffic intensity $\lambda$), which model phenomena in diverse disciplines, including biology and operations research. For a broad class of reneging distributions with finite mean and service distributions with finite mean and hazard rate function that is either decreasing or bounded away from zero and infinity, it is shown that if the fluid equations have a unique invariant state, then the Dirac measure at this state is the unique random fixed point of the fluid equations, which implies that the stationary distributions of scaled $N$-server systems converge to the unique invariant state of the corresponding fluid equations. Moreover, when $\lambda\ne 1$, it is shown that the solution to the fluid equation starting from any initial condition converges to this unique invariant state in the large time limit. The proof techniques are different under the two sets of assumptions on the service distribution. When the hazard rate function is decreasing, a reformulation of the dynamics in terms of a certain renewal equation is used, in conjunction with recursive asymptotic estimates. When the hazard rate function is bounded away from zero and infinity, the proof uses an extended relative entropy functional as a Lyapunov function. Analogous large-time convergence results are also established for a system of coupled measure-valued equations modeling a multiclass queue.

math.PR↗

arXiv2021

On the Persistent-Idle Load Distribution Policy Under Batch Arrivals and Random Service Capacity

The Persistent-Idle (PI) load distribution policy was recently introduced as an appealing alternative to current low-communication load balancing techniques. In PI, servers only update the dispatcher when they become idle, and the dispatcher always sends jobs to the last server that reported being idle. PI is unique in that it does not seek to push the server queue lengths towards equalization greedily. Rather, it aggressively pulls the servers away from starvation. As a result, PI's analysis requires different tools than other load balancing approaches. So far, PI was proven to achieve the stability region for Bernoulli arrivals and deterministic and constant service capacities. Our main contribution is proving that PI achieves the stability region in a model with batch arrivals and random service capacities. Proving this result requires developing tighter bounds on quantities of interest and proving the fulfillment of a generalized version of the state-dependent drift criteria previously considered. We also present PI-Split, which allows batches to be split among idle servers and prove its stability. Finally, we conduct simulations comparing PI and PI-Split to other load balancing policies. Simulation results indicate that both achieve excellent performance for the model under consideration for a wide range of parameters.

math.PR↗

arXiv2020

A weak formulation of free boundary problems and its application to hydrodynamic limits of particle systems with selection

A weak formulation for a class of parabolic free boundary problems (FBP) is proposed that does not involve the notion of a free boundary but reduces to a FBP when classical solutions exist. It is aimed at hydrodynamic limits (HDL) of particle systems with selection in circumstances where the macroscopic model does not possess (or is hard to prove to possess) a regular free boundary in the classical sense. The formulation involves the macroscopic density of particles and a measure that accounts for selection. It consists of a second order parabolic equation satisfied by the density and driven by the measure, coupled with a complementarity condition satisfied by the density-measure pair. The approach is applied to an injection-branching-selection particle system of diffusion on $\R$ under arbitrarily varying injection and removal rates, for which the corresponding FBP is not in general known to be classically solvable. The HDL is characterized as the unique solution to the weak formulation. The proof of convergence is based on PDE uniqueness, which in turn relies on the barrier method.

math.PR↗

arXiv2020

Fluid limits for earliest-deadline-first networks

This paper analyzes fluid scale asymptotics of two models of generalized Jackson networks employing the earliest deadline first (EDF) policy. One applies the 'soft' EDF policy, where deadlines are used to determine priority but jobs do not renege, and the other implements 'hard' EDF, where jobs renege when deadlines expire, and deadlines are postponed with each migration to a new station. The arrival rates, deadline distribution and service capacity are allowed to fluctuate over time at the fluid scale. Earlier work on EDF network fluid limits, used as a tool to obtain stability of these networks, addressed only the soft version of the policy, and moreover did not contain a full fluid limit result. In this paper, tools that extend the notion of the measure-valued Skorokhod map are developed and used to establish for the first time fluid limits for both the soft and hard EDF network models.

math.PR↗

arXiv2020

Customer-server population dynamics in heavy traffic

We study a many-server queueing model with server vacations, where the population size dynamics of servers and customers are coupled: a server may leave for vacation only when no customers await, and the capacity available to customers is directly affected by the number of servers on vacation. We focus on scaling regimes in which server dynamics and queue dynamics fluctuate at matching time scales, so that their limiting dynamics are coupled. Specifically, we argue that interesting coupled dynamics occur in (a) the Halfin-Whitt regime, (b) the nondegenerate slowdown regime, and (c) the intermediate, near Halfin-Whitt regime; whereas the dynamics asymptotically decouple in the other heavy traffic regimes. We characterize the limiting dynamics, which are different for each scaling regime. We consider relevant respective performance measures for regimes (a) and (b) --- namely, the probability of wait and the slowdown. While closed form formulas for these performance measures have been derived for models that do not accommodate server vacations, it is difficult to obtain closed form formulas for these performance measures in the setting with server vacations. Instead, we propose formulas that approximate these performance measures, and depend on the steady-state mean number of available servers and previously derived formulas for models without server vacations. We test the accuracy of these formulas numerically.

math.PR↗

arXiv2020

Robust bounds and optimization at the large deviations scale for queueing models via R\'enyi divergence

This paper develops tools to obtain robust probabilistic estimates for queueing models at the large deviations (LD) scale. These tools are based on the recently introduced robust R\'enyi bounds, which provide LD estimates (and more generally risk-sensitive (RS) cost estimates) that hold uniformly over an uncertainty class of models, provided that the class is defined in terms of R\'enyi divergence with respect to a reference model and that estimates are available for the reference model. One very attractive quality of the approach is that the class to which the estimates apply may consist of hard models, such as highly non-Markovian models and ones for which the LD principle is not available. Our treatment provides exact expressions as well as bounds on the R\'enyi divergence rate on families of marked point processes, including as a special case renewal processes. Another contribution is a general result that translates robust RS control problems, where robustness is formulated via R\'enyi divergence, to finite dimensional convex optimization problems, when the control set is a finite dimensional convex set. The implications to queueing are vast, as they apply in great generality. This is demonstrated on two non-Markovian queueing models. One is the multiclass single-server queue considered as a RS control problem, with scheduling as the control process and exponential weighted queue length as cost. The second is the many-server queue with reneging, with the probability of atypically large reneging count as performance criterion. As far as LD analysis is concerned, no robust estimates or non-Markovian treatment were previously available for either of these models.

math.PR↗