2025-11-17 | | Total: 13
We consider the $N$-model queueing system with a waiting time dependent threshold on the diagonal: the service discipline is First--Come--First--Served, but type-1 jobs can only be served by server 2 if their waiting time exceeds a deterministic threshold. We prove the necessary and sufficient stability conditions for this model -- an intuitive result that has not been established in literature up to this point. Our proof relies on coupling the queue length process to a carefully constructed upper (and lower) bound system, and establishing stochastic dominance for the queue length process.
We consider a class of functions for which the multiple Stratonovich stochastic integral or equivalent iterated Stratonovich stochastic integral with square integrable weights is defined by the orthogonal expansion. The equality of the trace of expansion coefficients matrix for these functions and the corresponding integral trace is established.
Actions in the Airy line ensemble represent distances from an infinitely far object. We characterize the Airy sheet by S(x,.)=T^x(.,1), where T^x is the unique action in the Airy line ensemble satisfying a growth condition depending on x. This provides a new simple framework for establishing convergence to the Airy sheet. We present simple conceptual proofs of such results in the case of Brownian last passage, the O'Connell-Yor semidiscrete polymer, the log-gamma polymer and the KPZ equation.
We consider the typical Poisson-Voronoi cell in the Euclidean space R d and in particular the maximal distance D from a vertex of that cell to its nucleus. We provide a sharp asymptotics for the tail distribution of D. As a byproduct, we prove that the extremal index related to the sequence of such distances for all Voronoi cells included in a large box is equal to (2d) -1 . This confirms a conjecture formulated by Chenavier and Robert. The explicit constant appearing in the estimate of the tail probability of D is proved to be the mean volume of a random simplex formed by uniformly distributed points on the unit sphere conditioned on satisfying some spatial condition.
Fix a bounded $3$-polygon $(Ω; x_1, x_2, x_3)$ with three marked boundary points $x_1, x_2, x_3\in\partialΩ$ and suppose $(Ω^δ; x_1^δ, x_2^δ, x_3^δ)$ is an approximation of $(Ω; x_1, x_2, x_3)$ on $δ$-scaled hexagonal lattice. We consider uniform spanning tree (UST) in $Ω^δ$ with wired boundary conditions. Conditional on the event that both branches from $x_1^δ$ and $x_2^δ$ hit the boundary through $x_3^δ$, the two branches meet at a point $\mathfrak{t}^δ$ which we call trifurcation, and the union of the three branches from $x_j^δ$ to $\mathfrak{t}^δ$ form a tripod in the UST. We compute the scaling limit of the tripod: the distribution of trifurcation is absolutely continuous with respect to Lebesgue measure with explicit density; given the trifurcation, the conditional law of the tripod is three-sided radial SLE$_2$. Interestingly, the scaling limit of the observable for trifurcation coincides with the partition function for three-sided radial SLE$_2$. The proof for the distribution of the trifurcation relies on Fomin's formula [Fom01] and tools from [CS11, CW21]. The proof of the convergence to three-sided radial SLE$_2$ relies on tools developped recently from [HPW25]. We believe the conclusion is true for a large family of discrete lattice approximations, however, our proof uses the geometry of the hexagonal lattice in an essential way.
In this article, we characterize continuous stationary fields via generalized Langevin dynamics. This gives natural connections between stationary fields, stationary increment fields, self-similar fields, and generalized Langevin dynamics. Our contribution extends some recently proved similar results for stochastic processes to the case of continuous random fields. As a by-product, we introduce some new results on multiple Riemann-Stieltjes integrals.
We provide necessary and sufficient conditions for explosion and implosion of birth-and-death (non-Markov) continuous-time random walks. In other words, we obtain conditions for $\infty$ to be accessible and for it to be an entrance point. We derive the analytical regularity criteria in terms of the appropriate scale function and the speed measure, which involve transition probabilities and the Laplace transform of the waiting times. We show that these criteria closely resemble classical ones for diffusions and Markov birth-and-death processes. We then calculate explicit conditions of regularity for semi-Markov processes with waiting times that have (a) finite first moments; (b) regularly varying tails (in particular, $α$-stable distribution).
Autocatalytic chemical reaction networks are dynamical systems whose linearization around zero, dX/dt = AX, is represented by a Perron-Frobenius matrix A with positive Lyapunov exponent; this exponent gives the growth rate of the species concentration vector X in the diluted regime, i.e. in a vicinity of zero. We introduce here a new, general recursive procedure providing precise quantitative information about Lyapunov data, namely, the Lyapunov eigenvalue, and left and right eigenvectors. Our estimates are based on a multi-scale algorithm inspired from Wilson's renormalization group method in quantum field theory, and Markov chain arguments introduced in (Nghe & Unterberger). They are compatible with the very scarce knowledge of kinetic rates (coefficients of A) generally available in chemistry, and take on the form of simple rational functions of the latter.
This paper analyzes a class of recursive distributional equations (RDE's) proposed by Gurel-Gurevich [17] and involving a bias parameter $p$, which includes the logarithm of the resistance of the series-parallel graph. A discrete-time evolution equation resembling a nonlinear, fractional Fisher-KPP equation is derived to describe the CDF's of solutions. When the bias parameter $p = \frac{1}{2}$, this equation is shown to have a PDE scaling limit, from which distributional limit theorems for the RDE are derived. Applied to the series-parallel graph, the results imply that $N^{-1/3} \log R^{(N)}$ has a nondegenerate limit when $p = \frac{1}{2}$, as conjectured by Addario-Berry, Cairns, Devroye, Kerriou, and Mitchell [1].
Nonlinear filtering problems are encountered in many applications, and one solution approach is the extended Kalman filter, which is not always convergent. Therefore, it is crucial to identify conditions under which the extended Kalman filter provides accurate approximations. This paper generalizes two significant results from Picard (1991) on the efficiency of the continuous-time extended Kalman filter to a more general setting where the observation noise may be state-dependent but does not allow signal reconstruction from the quadratic variation of the observation process as in epidemic models. Firstly, we show that when the observation's drift coefficient is strongly injective and the signal's and observation's drift become nearly linear for the diffusion scaling coefficient $ε\to 0$, the estimation error is of order $\sqrtε$. Subsequently, we establish conditions under which the impact of the initial filtering error decays exponentially fast.
Motivated by the estimation of covariance matrices by importance sampling arising in the cross-entropy (CE) algorithm, we study a random matrix model $\hat Σ= {\bf X} L {\bf X}^\top$ with two distinct features: $\bf X$ and $L$ are dependent, and $L$ is heavy-tailed. In the high-dimensional regime $d \to \infty$, we prove under suitable assumptions that a phase transition occurs in the polynomial regime $n = d^κ$, with $n$ the sample size. Namely, we prove that $\lVert \hat Σ- E \hat Σ\rVert \Rightarrow 0$ if and only if $κ> κ_*$ for some threshold $κ_*$ determined by the behavior of the maximum likelihood ratios. Moreover, we identify general situations where $κ_* = 1/λ_1$, with $λ_1$ the smallest eigenvalue of the covariance matrix of the auxiliary distribution used to estimate $\hat Σ$ by importance sampling. This suggests that importance sampling will work better with covariance matrices having a large smallest eigenvalue. We carry this insight into recent CE schemes proposed to estimate the probability of high-dimensional rare events. Through numerical simulations, we demonstrate that better CE schemes are also the ones with larger smallest eigenvalue, even though these algorithms were not designed to smooth the spectrum. This new spectral interpretation raises stimulating questions and opens research directions for the design of efficient high-dimensional algorithms.
We generalize the construction of Compactified Imaginary Liouville Theory (CILT), a non-unitary logarithmic Conformal Field Theory (CFT) defined on closed surfaces, to surfaces with boundary. Starting from a compactified Gaussian Free Field (GFF) with Neumann boundary condition, we perturb it by adding in curvature terms and exponential potentials on both the bulk and the boundary. In physics, this theory is conjectured to describe the scaling limit of loop models such as the Potts and $O(n)$ models. To define it mathematically, the curvature terms require a detailed analysis of the topology, and the potential terms are defined using the imaginary Guassian Multiplicative Chaos (GMC). We prove that the resulting probabilistic path integral satisfies the axioms of CFT, including Segal's gluing axioms. This work provides the foundation for future studies of boundary CILT and will also help with the understanding of CILT.
We study here the social network generated by the asynchronous visits, to a fixed set of sites, of mobile agents modelled as independent random walks on the plane lattice. The social network is constructed by assuming that a group of agents are associated if they have visited the same set of sites within a finite time interval. This construction is an instance of a random intersection graph, and has been used in the literature to study association networks in a number of animal species. We characterize the mathematical structure of these networks, which we view as one-mode projections of suitable bipartite graphs or, equivalently, as 2-sections of the corresponding hypergraphs. We determine analytically the probability distribution of the random bipartite graphs and hypergraphs associated to this construction, and suggest that association networks generated by the use of common resources are better described by hypergraphs rather than simple projected graphs, that miss important information regarding the actual associations among the agents.