
ALT We consider a system of branching Brownian particles on the real line with bounded, possibly discontinuous drift b, and with a selection rule that removes one particle from the most crowded cells of a uniform partition of the line at each branching event so that the number of particles is preserved. As the number of particles tends to infinity, with the cell size tending to zero (sufficiently slowly), we prove convergence in probability of the empirical measure process to a deterministic limit characterized by the parabolic free obstacle problem ∂ₜu=tfrac12∂ₓₓu-∂ₓ(bu)+u-β. Here the density u has unit mass. Disintegrating the removal measure β as βₜ(d x)d t, the probability measure βₜ is supported on the spatial maximizers of u(t,⋅) for a.e. t. We establish existence and uniqueness of weak solutions in a suitable class of continuous densities, extending earlier results about this PDE to discontinuous drifts. We then consider an inverse problem: Given a target removal measure, find a dri

ALT The completely packed O(1) loop model can be viewed as an interface representation of critical bond percolation on the square lattice. We consider submatching events, which prescribe how these interfaces pair finitely many consecutive boundary points. Romik's rationality conjectures predict that the probabilities of these events on the semi-infinite cylinder are given by specific rational functions, which even become dyadic rationals in the half-plane limit. We derive explicit Pfaffian formulae for the probability of k nested arches on the semi-infinite cylinder of circumference 2N, valid for k<N. We deduce that these probabilities are rational functions of N and, by taking the limit N→∞, that the corresponding half-plane probabilities remain rational. This establishes key aspects of Romik's conjectures. Our proof uses identities of Fonseca and Zinn-Justin, derived through the qKZ formalism, which express these probabilities as ratios of enumeration numbers for totally symmetric self-c
![For every γ∈[3/4,1], we prove pathwise uniqueness and strong existence for the stochastic heat equation on ℝ with space-time white noise, Lipschitz drift, and a diffusion coefficient locally γ-Hölder under suitable growth bounds. The result settles the critical case γ=3/4 left open by Mueller, Mytnik and Perkins. The proof is based on a newly developed short-time linear comparison method. On each short interval, we compare the difference of two solutions with an auxiliary linear process. An expected weighted L¹ bound independent of the noise multiplier, together with a vanishing sum of approximation errors, yields pathwise uniqueness. The method may also apply to other SPDEs admitting analogous linear comparison and stochastic convolution estimates.](https://pbs.twimg.com/media/HXlex44WkAEsUvC.png?name=small&format=webp)
ALT For every γ∈[3/4,1], we prove pathwise uniqueness and strong existence for the stochastic heat equation on ℝ with space-time white noise, Lipschitz drift, and a diffusion coefficient locally γ-Hölder under suitable growth bounds. The result settles the critical case γ=3/4 left open by Mueller, Mytnik and Perkins. The proof is based on a newly developed short-time linear comparison method. On each short interval, we compare the difference of two solutions with an auxiliary linear process. An expected weighted L¹ bound independent of the noise multiplier, together with a vanishing sum of approximation errors, yields pathwise uniqueness. The method may also apply to other SPDEs admitting analogous linear comparison and stochastic convolution estimates.

ALT We study empirical convergence rates for (p,q)-Gromov-Wasserstein (GW) distances D_p,q for arbitrary real exponents p>0 and q≥1. Let μ and ν be probability measures on compact metric spaces X and Y, and let widehatμ_n and widehatν_m be their empirical measures, constructed from independent i.i.d. samples of sizes n and m. If the spaces X, Y are compact subsets of the Euclidean spaces mathbbR^d_x and mathbbR^dy, respectively, we obtain an upper bound of order (nwedge m)^-(pwedge2)/((d_xwedgedy)\vee2(pwedge2)) on the plug-in error of D_p,q^p(widehatμ_n,widehatν_m) in expectation, up to a logarithmic factor at the critical dimension d_xwedge dy=2(pwedge2). On Euclidean balls, this rate is sharp up to the logarithmic factor. More generally, we show that the same upper bound holds when one space is a compact subset of a complete, connected d-dimensional Riemannian manifold and the other has polynomial covering numbers, with d_xwedge dy replaced by d. Earlier works on (2,2)-GW, (2,1)-GW and

ALT Arm separation is an important technical tool in the study of critical and near-critical planar percolation systems. When combined with RSW estimates, it proves quasi-multiplicativity of arm-event probabilities, links the derivatives of different events and allows the identification of universal arm exponents. We provide a very generic statement of arm separation that applies to arbitrary planar percolation models. Our argument is based on exploring configurations conditioned on the existence of arms in such a way that "exposed" arms are revealed with positive probability. The exploration algorithm is independent of the percolation model and applies to all polychromatic arm events. As an application, we prove a strong arm-separation result for critical planar FK-percolation.

ALT We prove that the pair correlation function of Sineᵦ extends holomorphically in β to \Reβ>0 at every fixed nonzero separation. Our proof combines a random walk representation of the stochastic sine phase moments, stretched-exponential Fourier decay obtained by coupling, and weighted energy estimates for complex β.

ALT We consider N interacting particles taking values in the real numbers and evolving in discrete time with time step 1/n. Each particle follows an individual dynamics, driven by its own noise, which depends on the total configuration via a polynomial interaction. Moreover, all particles receive the same collective perturbation. This perturbation is given by a multilinear homogeneous sum of order m ≥ 2 of independent centered random fields which are evaluated at the current positions of the particles. We show that, as N and n tend to infinity, the system converges to the solution of a conditional McKean-Vlasov equation which is driven by an additional Brownian motion which emerges in the limit from the collective perturbations. Moreover we give explicit bounds for the Wasserstein-2-distance. Our proof combines the invariance principle of Mossel, O'Donnell and Oleszkiewicz (2010), the CLT for Wiener chaoses in the Wasserstein-2-form of Ledoux, Nourdin and Peccati (2015) and a conditional l

ALT Csiszár indices Sf(X,Y), defined as the f-divergence between the joint distribution of a general random couple (X,Y) and the product of its marginal distributions, are widely used to quantify dependence between the input variable X and the output variable Y. We show that the ranking of two input variables X and X' in terms of their importance w.r.t. the output variable, given by the comparison of the Csiszár indices Sf(X,Y) and Sf(X',Y) depends on the choice of f. We introduce the Csiszár order of input variables as the uniform ordering of Csiszár indices over all proper convex functions f. We give a sufficient condition to prove Csiszár order of two input multivariate random variables. We fully characterize it in two cases: when (X,X',Y) is a Gaussian vector (in the one-dimensional case this amount to compare the square of their correlation); and when X, X' and Y are Bernoulli random variables. Finally, motivated by sensitivity analysis, we show that, when (X,X',Y) is a three-dimensio

ALT We analyse the distribution of the first-passage time of Brownian motion to a time-dependent boundary as a function of the boundary. We prove that the boundary-to-distribution map is infinitely many times differentiable along the Cameron-Martin space ℋ, and obtain a with a bound on the truncation error Taylor expansion with a bound on the truncation error. We obtain identities connecting (i) the ℋ-derivatives of the first-passage time distribution, (ii) the Wiener-Itô chaos kernels of the indicator that the Brownian motion stays below the boundary, and (iii) Krylov-Veretennikov-Zvonkin-type operators. These identities yield relatively tractable representations for the ℋ-derivatives of the first-passage time distribution when the taboo transition density is available, notably for linear boundaries. As an application of the above results we propose Taylor approximations up to second order of the first-passage time distribution. In the examples considered, the first-order approximation is

ALT We estimate the probability that a vector of U-statistics lies within C/n of a target in the range of typical fluctuations. For sufficiently large C, we prove a lower bound of the typical Gaussian order under finite variance, and a matching upper bound for bounded kernels. We show that under finite variance alone, stronger concentration may occur, and find the best possible asymptotic upper bound. Our proof is algorithmic, estimating probabilities by counting successful modifications produced by a random algorithm. We apply this result to settle a question of Beniamini, Lavee and Linial (2025), who asked if there exist permutations with k-order sub-pattern counts at most O(nᵏ⁻¹) away from one another for k>4. We prove not only existence, but enumerate these permutations up to a constant factor, showing that there are Θ(n!n⁻⁽ᵏ⁻¹²/2)) such permutations of size n.

ALT We establish quantitative approximations for the empirical measures of finite-state mean-field particle systems in a general framework. The scales of the empirical drift and covariance distinguish the ODE and critical diffusion regimes and determine the corresponding time rescaling. In the ODE regime, we obtain quantitative central limit approximations in the 1-Wasserstein distance and a Bernstein-type concentration inequality. The concentration bound has no dimensional prefactor. Its explicit time dependence also identifies growing intervals on which the ODE approximation remains accurate. In the critical regime, we prove a quantitative nonlinear diffusion approximation in the 1-Wasserstein distance. We apply our quantitative approximation results to noisy voter systems, Moran models with mutations and selections, heat-bath Glauber dynamics for the p-spin Curie–Weiss model, and autocatalytic reactions with mutations.

ALT We study the boundary behavior of nonnegative harmonic functions for a class of purely discontinuous symmetric Markov processes in C¹η open sets D⊂ℝᵈ. The jump kernels are allowed to blow up at the boundary of D and the processes may have a critical killing potential. We identify the exact boundary decay exponent p, characterized by the coefficient of the critical killing potential. We establish a boundary Harnack principle which yields the comparability of f(x)/δ_D(x)ᵖ for nonnegative harmonic functions that vanish continuously on a portion of the boundary. The key ingredient of the proof combines estimates of the exact decay rate for the survival probability of the killed process with a modified box method and uses a regularized distance function to handle the limited boundary regularity.

ALT We develop a second-moment method for the empirical spectral distribution (ESD) of non-Hermitian random matrices X_N=(√sᵢⱼξᵢⱼ) with a deterministic variance profile S_N=(sᵢⱼ). When the entries are independent and centered, the second moment of det(zI-X_N) is the permanent per(S_N+|z|²I) of a diagonal shift of S_N, and the method approximates log|{det(zI}-X_N)|by half its logarithm. For i.i.d. centered ξᵢⱼ of unit variance satisfying a Dini-type condition on their Lévy concentration function, we prove the circular law for every doubly stochastic profile whose largest entry tends to zero, however slowly. This condition holds for every law with a bounded density on ℝ or ℂ. We also allow the two entries Xᵢⱼ and Xⱼᵢ of each reciprocal pair to be correlated, with a nondegenerate joint law that has a finite mixed fourth moment. If the pairs and the diagonal entries also have bounded densities on their real linear supports, we prove an elliptic law for every symmetric doubly stochastic profile

ALT We prove dynamical edge universality for three families of interacting particle systems for every β>0: β-Dyson Brownian motion with an analytic potential, weighted β-corners processes, and a class of discrete-time Markov chains. Our results apply to general non-equilibrium initial data and show that the Airyᵦ line ensemble emerges at positive times whenever the limiting density has a regular square-root edge. More precisely, after centering at the moving deterministic edge and rescaling space and time on the n⁻²/³ and n⁻¹/³ scales, respectively, the trajectories of the top k particles converge jointly in distribution to the top k curves of the Airyᵦ line ensemble, for every fixed k, in the topology of uniform convergence on compact time intervals. The proof combines dynamical loop equations with characteristic-flow estimates to establish local laws near the moving edge and upper bounds on the extremal particle. We then identify the stochastic evolution of every subsequential limit of t

ALT Many problems in statistics, combinatorics, econometrics, and other fields are expressible in terms of the upper probability of all copulas. We formally study this capacity, which we call the upper copula capacity, as a mathematical object. In particular, we examine the capacity of given sets and integration with respect to this capacity. For two-dimensional convex sets, the upper copula capacity equals the smaller of the two coordinate projection lengths. We also give an exact formula for the capacity of lower-tail events for positively weighted sums in any dimension and characterize when the upper Choquet integral equals the supremum of expectations over all copulas. Various problems in different fields can be formulated via the upper-copula capacity, and we discuss how known and new results connect to offer new perspectives on these problems.

ALT We introduce a random Dirac operator on L²(ℝ,ℝ²) that we call 1d Stochastic Dirac Operator and which is indexed by two parameters κ₀,κ. In two companion articles, we show that this operator arises generically as the small disorder limit of one-dimensional random Schrödinger operators, thus revealing a universal behavior. In the present paper, we provide a detailed construction of this singular operator, and investigate its spectral properties. In particular, we show that its spectrum is either absolutely continuous when κ=0, or pure point when κ>0. In the latter case, we show that the shape of the eigenfunctions is given by the exponential of a two sided Brownian motion with drift, in connection with a result of Rifkind and Virág. Finally, we establish a delicate convergence result of the sequence of eigenvalues / eigenfunctions from the finite-volume operator to the infinite-volume one.

ALT We propose a modification of the bisexual branching process with random mating and overlapping generations. Unlike the model of Molina and Mota, our modification introduces probabilistic age-dependent survival that depends on the state of the population. Our modification preserves the general properties of the original process. We establish two sufficient conditions characterizing the subcritical and supercritical regimes of the process, i.e., extinction of the population with probability one and survival of the population with positive probability.The results are illustrated with numerical simulations of the process performed in Python.

ALT Limit theorems are proved for the signed Regge curvature of independent oriented hyperplanes in Euclidean space under the piecewise-flat metric induced by unit linear ramps on their positive half-spaces. Provided each hyperplane law is invariant under orientation reversal, the curvature has zero mean and zero first Hoeffding component at every finite sample size. In particular, it is uncorrelated with all square-integrable unoriented functionals. For isotropic Gaussian normals on bounded windows, the rescaled curvature is approximated in mean square by a degenerate U-statistic and converges to a second-chaos series diagonalised on odd spherical harmonics. In the plane, the global cone-angle deficit admits a closed-form limit law universal up to scale under finite fourth moments.


