We gratefully acknowledge support from
the Simons Foundation and member institutions.


New submissions

[ total of 31 entries: 1-31 ]
[ showing up to 1000 entries per page: fewer | more ]

New submissions for Thu, 25 Nov 21

[1]  arXiv:2111.12169 [pdf, ps, other]
Title: Hanson-Wright Inequality for Random Tensors under Einstein Product
Authors: Shih Yu Chang
Comments: arXiv admin note: text overlap with arXiv:2012.15428
Subjects: Probability (math.PR)

The Hanson-Wright inequality is an upper bound for tails of real quadratic forms in independent subgaussian random variables. In this work, we extend the Hanson-Wright inequality for the maximum eigenvalue of the quadratic sum of random Hermitian tensors under Einstein product. We first prove Weyl inequality for tensors under Einstein product and apply this fact to separate the quadratic form of random Hermitian tensors into diagonal sum and coupling (non-diagonal) sum parts. For the diagonal part, we can apply Bernstein inequality to bound the tail probability of the maximum eigenvalue of the sum of independent random Hermitian tensors directly. For coupling sum part, we have to apply decoupling method first, i.e., decoupling inequality to bound expressions with dependent random Hermitian tensors with independent random Hermitian tensors, before applying Bernstein inequality again to bound the tail probability of the maximum eigenvalue of the coupling sum of independent random Hermitian tensors. Finally, the Hanson-Wright inequality for the maximum eigenvalue of the quadratic sum of random Hermitian tensors under Einstein product can be obtained by the combination of the bound from the diagonal sum part and the bound from the coupling (non-diagonal) sum part.

[2]  arXiv:2111.12312 [pdf, ps, other]
Title: Lossy Compression of General Random Variables
Subjects: Probability (math.PR); Information Theory (cs.IT)

This paper is concerned with the lossy compression of general random variables, specifically with rate-distortion theory and quantization of random variables taking values in general measurable spaces such as, e.g., manifolds and fractal sets. Manifold structures are prevalent in data science, e.g., in compressed sensing, machine learning, image processing, and handwritten digit recognition. Fractal sets find application in image compression and in the modeling of Ethernet traffic. Our main contributions are bounds on the rate-distortion function and the quantization error. These bounds are very general and essentially only require the existence of reference measures satisfying certain regularity conditions in terms of small ball probabilities. To illustrate the wide applicability of our results, we particularize them to random variables taking values in i) manifolds, namely, hyperspheres and Grassmannians, and ii) self-similar sets characterized by iterated function systems satisfying the weak separation property.

[3]  arXiv:2111.12354 [pdf, ps, other]
Title: Dynamical noise sensitivity for the voter model
Comments: 7 pages
Subjects: Probability (math.PR)

We study noise sensitivity of the consensus opinion of the voter model on finite graphs, with respect to noise affecting the initial opinions and noise affecting the dynamics.
We prove that the final opinion is stable with respect to small perturbations of the initial configuration, and is sensitive to perturbations of the dynamics governing the evolution of the process.
Our proofs rely on the duality relationship between the voter model and coalescing random walks, and on a precise description of this evolution when we have coupled dynamics.

[4]  arXiv:2111.12383 [pdf, ps, other]
Title: Besov-Orlicz path regularity of non-Gaussian processes
Comments: 26 pages, 3 figures
Subjects: Probability (math.PR)

In the article, Besov-Orlicz regularity of sample paths of stochastic processes that are represented by multiple integrals of order $n\in\mathbb{N}$ is treated. We give sufficient conditions for the considered processes to have paths in the exponential Besov-Orlicz space $$B_{\varPhi_{2/n},\infty}^\alpha(0,T)\qquad \mbox{with}\qquad \varPhi_{2/n}(x)=\mathrm{e}^{x^{2/n}}-1.$$ These results provide an extension of what is known for scalar Gaussian stochastic processes to stochastic processes in an arbitrary finite Wiener chaos. As an application, the Besov-Orlicz path regularity of fractionally filtered Hermite processes is studied. But while the main focus is on the non-Gaussian case, some new path properties are obtained even for fractional Brownian motions.

[5]  arXiv:2111.12517 [pdf, other]
Title: Explicit formulas concerning eigenvectors of weakly non-unitary matrices
Authors: Guillaume Dubach
Comments: 12 pages, 1 figure
Subjects: Probability (math.PR); Mathematical Physics (math-ph)

We investigate eigenvector statistics of the Truncated Unitary ensemble $\mathrm{TUE}(N,M)$ in the weakly non-unitary case $M=1$, that is when only one row and column are removed. We provide an explicit description of generalized overlaps as deterministic functions of the eigenvalues, as well as a method to derive an exact formula for the expectation of diagonal overlaps (or squared eigenvalue condition numbers), conditionally on one eigenvalue. This complements recent results obtained in the opposite regime when $M \geq N$, suggesting possible extensions to $\mathrm{TUE}(N,M)$ for all values of $M$.

[6]  arXiv:2111.12571 [pdf, ps, other]
Title: Fluctuations in Salem--Zygmund almost sure central limit theorem
Comments: 42 pages
Subjects: Probability (math.PR)

Let us consider i.i.d. random variables $\{a_k,b_k\}_{k \geq 1}$ defined on a common probability space $(\Omega, \mathcal F, \mathbb P)$, following a symmetric Rademacher distribution and the associated random trigonometric polynomials $S_n(\theta)= \frac{1}{\sqrt{n}} \sum_{k=1}^n a_k \cos(k\theta)+b_k \sin(k\theta)$. A seminal result by Salem and Zygmund ensures that $\mathbb{P}-$almost surely, $\forall t\in\mathbb{R}$ \[ \lim_{n \to +\infty} \frac{1}{2\pi}\int_0^{2\pi} e^{i t S_n(\theta)}d\theta=e^{-t^2/2}. \] This result was then further generalized in various directions regarding the coefficients distribution, their dependency structure or else the dimension and the nature of the ambient manifold. To the best of our knowledge, the natural question of the fluctuations in the above limit has not been tackled so far and is precisely the object of this article. Namely, for general i.i.d. symmetric random coefficients having a finite sixth-moment and for a large class of continuous test functions $\phi$ we prove that \[ \sqrt{n}\left(\frac{1}{2\pi}\int_0^{2\pi} \phi(S_n(\theta))d\theta-\int_{\mathbb{R}}\phi(t)\frac{e^{-\frac{t^2}{2}}dt}{\sqrt{2\pi}}\right)\xrightarrow[n\to\infty]{\text{Law}}~\mathcal{N}\left(0,\sigma_{\phi}^2+\frac{c_2(\phi)^2}{2}\left(\mathbb{E}(a_1^4)-3\right)\right). \] Here, the constant $\sigma_{\phi}^2$ is explicit and corresponds to the limit variance in the case of Gaussian coefficients and $c_2(\phi)$ is the coefficient of order $2$ in the decomposition of $\phi$ in the Hermite polynomial basis. Surprisingly, it thus turns out that the fluctuations are not universal since they both involve the kurtosis of the coefficients and the second coefficient of $\phi$ in the Hermite basis.

[7]  arXiv:2111.12584 [pdf, other]
Title: Coagulation dynamics under random field: turbulence effects on rain
Authors: Andrea Papini
Comments: 25 pages, 13 figures
Subjects: Probability (math.PR)

Turbulence in growth of rain droplets and rain formation is studied under an approximating particle system representing aggregation at the level of individuals, depending on their volume and distance in space, of the Smoluchowski Coagulation equation. A random field is introduced to model the air flow interaction with the particles and it is proved that the empirical density of the individual converges to solutions of limiting equation under different setting for the random field of interaction. A brief numerical study for the continuous density is proposed using the particles approach, to analyze how noise can arise in such system and the effectiveness on rain formation.

Cross-lists for Thu, 25 Nov 21

[8]  arXiv:2111.12129 (cross-list from math.AP) [pdf, ps, other]
Title: Existence of Fractional Nonlocal Neutral Stochastic Differential Equation of Order 1 < q < 2 with Non-instantaneous Impulses and State-Dependent Delay
Comments: 18pages
Subjects: Analysis of PDEs (math.AP); Probability (math.PR)

This article addresses a new class of fractional nonlocal neutral stochastic differential system of order 1<q<2 including non-instantaneous impulses(NIIs) and state-dependent delay(SDD) with the Poisson jumps and the Wiener process in Hilbert spaces. We examine the existence of a mild solution without assuming the compactness of the resolvent operators. The Monch fixed point theorem is used along with the Hausdorff measure of noncompactness. An example is constructed to verify the theoretical developments.

[9]  arXiv:2111.12403 (cross-list from math.CO) [pdf, other]
Title: Combinatorics of the Quantum Symmetric Simple Exclusion Process, associahedra and free cumulants
Authors: Philippe Biane
Subjects: Combinatorics (math.CO); Probability (math.PR)

The Quantum Symmetric Simple Exclusion Process (QSSEP) is a model of quantum particles hopping on a finite interval and satisfying the exclusion principle. Recently Bernard and Jin have studied the fluctuations of the invariant measure for this process, when the number of sites goes to infinity. These fluctuations are encoded into polynomials, for which they have given equations and proved that these equations determine the polynomials completely. In this paper, I give an explicit combinatorial formula for these polynomials, in terms of Schr\"oder trees. I also show that, quite surprisingly, these polynomials can be interpreted as free cumulants of a family of commuting random variables.

[10]  arXiv:2111.12426 (cross-list from math.RT) [pdf, other]
Title: Skew Howe duality and limit shapes of Young diagrams
Comments: 55 pages, 12 figures, 2 tables
Subjects: Representation Theory (math.RT); Combinatorics (math.CO); Probability (math.PR)

We consider the skew Howe duality for the action of certain dual pairs of Lie groups $(G_1, G_2)$ on the exterior algebra $\bigwedge(\mathbb{C}^{n} \otimes \mathbb{C}^{k})$ as a probability measure on Young diagrams by the decomposition into the sum of irreducible representations. We prove a combinatorial version of this skew Howe for the pairs $(\mathrm{GL}_{n}, \mathrm{GL}_{k})$, $(\mathrm{SO}_{2n+1}, \mathrm{Pin}_{2k})$, $(\mathrm{Sp}_{2n}, \mathrm{Sp}_{2k})$, and $(\mathrm{Or}_{2n}, \mathrm{SO}_{k})$ using crystal bases, which allows us to interpret the skew Howe duality as a natural consequence of lattice paths on lozenge tilings of certain partial hexagonal domains. The $G_1$-representation multiplicity is given as a determinant formula using the Lindstr\"om-Gessel-Viennot lemma and as a product formula using Dodgson condensation. These admit natural $q$-analogs that we show equals the $q$-dimension of a $G_2$-representation (up to an overall factor of $q$), giving a refined version of the combinatorial skew Howe duality. Using these product formulas (at $q =1$), we take the infinite rank limit and prove the diagrams converge uniformly to the limit shape.

[11]  arXiv:2111.12431 (cross-list from math.OC) [pdf, other]
Title: A Large-Scale Multi-Rider Matching Problem with Reneging Passengers
Comments: 48 pages, 25 figures. Part of the work has been included in a conference paper submitted to the 24th International Congress on Modelling and Simulation (MODSIM2021)
Subjects: Optimization and Control (math.OC); Probability (math.PR)

This paper studies a large-scale ride-matching problem in a transport center with a large number of passengers who are different in terms of destinations and travel preferences. Passengers with similar itineraries can match each other and share the same vehicle for their potentially different destinations; and reneging passengers, who become impatient and leave the service system after waiting long time for shared rides, are considered in our model. We aim to maximize the long-run average revenue of the ride service vendor, which is defined as the difference between the long-run average reward earned by providing ride services and the long-run average penalty incurred by reneging passengers. The problem is complicated by its scale, the heterogeneity of passengers, and the reneging behavior. To this end, we formate the ride-matching problem as a specific Markov decision process and propose a scalable ride-matching policy, referred to as Bivariate Index (BI) policy. We demonstrate that the ride-matching problem reduces to a conventional restless multi-armed bandit problem in a special case, for which the BI policy reduces to the well-known Whittle index policy that prioritizes passengers with the highest Whittle indices. For the general case, following the Whittle relaxation technique, BI prioritizes passengers according to their bivariate indices that are curves in a real-valued coordinate and are assigned to each passengers. We prove that the ranking of the bivariate indices can be derived by the ranking of corresponding Whittle indices in a surrogate system. Through extensive numerical simulations for systems with real-world travel demands, it is demonstrated that BI significantly outperforms baseline policies.

[12]  arXiv:2111.12439 (cross-list from math-ph) [pdf, ps, other]
Title: Large deviations for a binary collision model: energy evaporation
Subjects: Mathematical Physics (math-ph); Statistical Mechanics (cond-mat.stat-mech); Probability (math.PR)

We analyze the large deviations for a discrete energy Kac-like walk. In particular, we exhibit a path, with probability exponentially small in the number of particles, that looses energy.

[13]  arXiv:2111.12506 (cross-list from cs.LG) [pdf, ps, other]
Title: A Unified Approach to Variational Autoencoders and Stochastic Normalizing Flows via Markov Chains
Subjects: Machine Learning (cs.LG); Probability (math.PR)

Normalizing flows, diffusion normalizing flows and variational autoencoders are powerful generative models. In this paper, we provide a unified framework to handle these approaches via Markov chains. Indeed, we consider stochastic normalizing flows as pair of Markov chains fulfilling some properties and show that many state-of-the-art models for data generation fit into this framework. The Markov chains point of view enables us to couple both deterministic layers as invertible neural networks and stochastic layers as Metropolis-Hasting layers, Langevin layers and variational autoencoders in a mathematically sound way. Besides layers with densities as Langevin layers, diffusion layers or variational autoencoders, also layers having no densities as deterministic layers or Metropolis-Hasting layers can be handled. Hence our framework establishes a useful mathematical tool to combine the various approaches.

[14]  arXiv:2111.12533 (cross-list from math.CO) [pdf, other]
Title: Tight bounds on the expected number of holes in random point sets
Subjects: Combinatorics (math.CO); Computational Geometry (cs.CG); Discrete Mathematics (cs.DM); Probability (math.PR)

For integers $d \geq 2$ and $k \geq d+1$, a $k$-hole in a set $S$ of points in general position in $\mathbb{R}^d$ is a $k$-tuple of points from $S$ in convex position such that the interior of their convex hull does not contain any point from $S$. For a convex body $K \subseteq \mathbb{R}^d$ of unit $d$-dimensional volume, we study the expected number $EH^K_{d,k}(n)$ of $k$-holes in a set of $n$ points drawn uniformly and independently at random from $K$.
We prove an asymptotically tight lower bound on $EH^K_{d,k}(n)$ by showing that, for all fixed integers $d \geq 2$ and $k\geq d+1$, the number $EH_{d,k}^K(n)$ is at least $\Omega(n^d)$. For some small holes, we even determine the leading constant $\lim_{n \to \infty}n^{-d}EH^K_{d,k}(n)$ exactly. We improve the currently best known lower bound on $\lim_{n \to \infty}n^{-d}EH^K_{d,d+1}(n)$ by Reitzner and Temesvari (2019). In the plane, we show that the constant $\lim_{n \to \infty}n^{-2}EH^K_{2,k}(n)$ is independent of $K$ for every fixed $k \geq 3$ and we compute it exactly for $k=4$, improving earlier estimates by Fabila-Monroy, Huemer, and Mitsche (2015) and by the authors (2020).

[15]  arXiv:2111.12551 (cross-list from math.HO) [pdf, ps, other]
Title: The life, work and legacy of P. L. Chebyshev
Authors: N. H. Bingham
Subjects: History and Overview (math.HO); Probability (math.PR)

We survey briefly the life and work of P. L. Chebyshev, and his ongoing influence. We discuss his contributions to probability, number theory and mechanics, his pupils and mathematical descendants, and his role as the founding father of Russian mathematics in general and of the Russian school of probability in particular.

[16]  arXiv:2111.12602 (cross-list from cs.CV) [pdf, other]
Title: Hierarchical Graph-Convolutional Variational AutoEncoding for Generative Modelling of Human Motion
Comments: Under Review at CVPR
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Probability (math.PR)

Models of human motion commonly focus either on trajectory prediction or action classification but rarely both. The marked heterogeneity and intricate compositionality of human motion render each task vulnerable to the data degradation and distributional shift common to real-world scenarios. A sufficiently expressive generative model of action could in theory enable data conditioning and distributional resilience within a unified framework applicable to both tasks. Here we propose a novel architecture based on hierarchical variational autoencoders and deep graph convolutional neural networks for generating a holistic model of action over multiple time-scales. We show this Hierarchical Graph-convolutional Variational Autoencoder (HG-VAE) to be capable of generating coherent actions, detecting out-of-distribution data, and imputing missing data by gradient ascent on the model's posterior. Trained and evaluated on H3.6M and the largest collection of open source human motion data, AMASS, we show HG-VAE can facilitate downstream discriminative learning better than baseline models.

[17]  arXiv:2111.12603 (cross-list from math.ST) [pdf, ps, other]
Title: Strong Invariance Principles for Ergodic Markov Processes
Subjects: Statistics Theory (math.ST); Probability (math.PR); Computation (stat.CO)

Strong invariance principles describe the error term of a Brownian approximation of the partial sums of a stochastic process. While these strong approximation results have many applications, the results for continuous-time settings have been limited. In this paper, we obtain strong invariance principles for a broad class of ergodic Markov processes. The main results rely on ergodicity requirements and an application of Nummelin splitting for continuous-time processes. Strong invariance principles provide a unified framework for analysing commonly used estimators of the asymptotic variance in settings with a dependence structure. We demonstrate how this can be used to analyse the batch means method for simulation output of Piecewise Deterministic Monte Carlo samplers. We also derive a fluctuation result for additive functionals of ergodic diffusions using our strong approximation results.

[18]  arXiv:2111.12607 (cross-list from math.DG) [pdf, ps, other]
Title: Heat kernel bounds and Ricci curvature for Lipschitz manifolds
Comments: 28 pages. Comments welcome
Subjects: Differential Geometry (math.DG); Analysis of PDEs (math.AP); Probability (math.PR)

Given any $d$-dimensional Lipschitz Riemannian manifold $(M,g)$ with heat kernel $\mathsf{p}$, we establish uniform upper bounds on $\mathsf{p}$ which can always be decoupled in space and time. More precisely, we prove the existence of a constant $C>0$ and a bounded Lipschitz function $R\colon M \to (0,\infty)$ such that for every $x\in M$ and every $t>0$, \begin{align*} \sup_{y\in M} \mathsf{p}(t,x,y) \leq C\min\{t, R^2(x)\}^{-d/2}. \end{align*} This allows us to identify suitable weighted Lebesgue spaces w.r.t. the given volume measure as subsets of the Kato class induced by $(M,g)$. In the case $\partial M \neq \emptyset$, we also provide an analogous inclusion for Lebesgue spaces w.r.t. the surface measure on $\partial M$.
We use these insights to give sufficient conditions for a possibly noncomplete Lipschitz Riemannian manifold to be tamed, i.e. to admit a measure-valued lower bound on the Ricci curvature, formulated in a synthetic sense.

[19]  arXiv:2111.12633 (cross-list from math.DS) [pdf, other]
Title: Untangling the role of temporal and spatial variations in persistance of populations
Subjects: Dynamical Systems (math.DS); Probability (math.PR); Populations and Evolution (q-bio.PE)

We consider a population distributed between two habitats, in each of which it experiences a growth rate that switches periodically between two values, $1- \varepsilon > 0$ or $ - (1 + \varepsilon) < 0$. We study the specific case where the growth rate is positive in one habitat and negative in the other one for the first half of the period, and conversely for the second half of the period, that we refer as the $(\pm 1)$ model. In the absence of migration, the population goes to $0$ exponentially fast in each environment. In this paper, we show that, when the period is sufficiently large, a small dispersal between the two patches is able to produce a very high positive exponential growth rate for the whole population, a phenomena called inflation. We prove in particular that the threshold of the dispersal rate at which the inflation appears is exponentially small with the period. We show that inflation is robust to random perturbation, by considering a model where the values of the growth rate in each patch are switched at random times: we prove, using theory of Piecewise Deterministic Markov Processes (PDMP) that inflation occurs for low switching rate and small dispersal. Finally, we provide some extensions to more complicated models, especially epidemiological and density dependent models.

[20]  arXiv:2111.12640 (cross-list from q-fin.MF) [pdf, other]
Title: Completing correlation matrices
Comments: 16 pages, 8 figures
Subjects: Mathematical Finance (q-fin.MF); Probability (math.PR)

We describe a way to complete a correlation matrix that is not fully specified. Such matrices often arise in financial applications when the number of stochastic variables becomes large or when several smaller models are combined in a larger model. We argue that the proper completion to consider is the matrix that maximizes the entropy of the distribution described by the matrix. We then give a way to construct this matrix starting from the graph associated with the incomplete matrix. If this graph is chordal our construction will result in a proper correlation matrix. We give a detailed description of the construction for a cross-currency model with six stochastic variables and describe extensions to larger models involving more currencies.

Replacements for Thu, 25 Nov 21

[21]  arXiv:1907.09876 (replaced) [pdf, other]
Title: Multi-point distribution of TASEP
Authors: Zhipeng Liu
Comments: 63 pages, minor changes
Subjects: Probability (math.PR); Mathematical Physics (math-ph)
[22]  arXiv:1912.08012 (replaced) [pdf, ps, other]
Title: Unit boundary length quantum disk: a study of two different perspectives and their equivalence
Authors: Baptiste Cerclé
Comments: Minor corrections; presentation has been improved and more detailed
Subjects: Probability (math.PR); Mathematical Physics (math-ph)
[23]  arXiv:2005.02915 (replaced) [pdf, ps, other]
Title: An almost sure invariance principle for some classes of non-stationary mixing sequences
Authors: Yeor Hafouta
Comments: The results are now obtained for alpha mixing sequences, and not only for phi mixing sequences
Subjects: Probability (math.PR)
[24]  arXiv:2103.02457 (replaced) [pdf, other]
Title: Continuous scaled phase-type distributions
Subjects: Probability (math.PR); Statistics Theory (math.ST)
[25]  arXiv:2107.00006 (replaced) [pdf, ps, other]
Title: A proof of the continuous martingale convergence theorem
Authors: Joe Ghafari
Comments: 6 pages
Subjects: Probability (math.PR)
[26]  arXiv:2108.11634 (replaced) [pdf, ps, other]
Title: Higher order fluctuations of extremal eigenvalues of sparse random matrices
Authors: Jaehun Lee
Comments: Revised. Some of the main estimates (Theorem 2.6 and Proposition 3.1) were fixed
Subjects: Probability (math.PR)
[27]  arXiv:2110.05950 (replaced) [pdf, other]
Title: An epidemic model in inhomogeneous environment
Comments: simulations and comparison with the homogeneous setting added
Subjects: Probability (math.PR)
[28]  arXiv:2111.09310 (replaced) [pdf, other]
Title: On an estimate of the rate of convergence for regenerative processes in queuing theory and related problems
Comments: 19 pages, in Russian, 5 figures
Subjects: Probability (math.PR)
[29]  arXiv:2105.00548 (replaced) [pdf, ps, other]
Title: Quenched limit theorems for expanding on average cocycles
Comments: 48 pages, comments are welcome !
Subjects: Dynamical Systems (math.DS); Probability (math.PR)
[30]  arXiv:2105.09429 (replaced) [pdf, other]
Title: Point process simulation of generalised inverse Gaussian processes and estimation of the Jaeger integral
Subjects: Methodology (stat.ME); Signal Processing (eess.SP); Probability (math.PR)
[31]  arXiv:2110.06798 (replaced) [pdf, ps, other]
Title: Quantitative Stability of Regularized Optimal Transport and Convergence of Sinkhorn's Algorithm
Subjects: Optimization and Control (math.OC); Analysis of PDEs (math.AP); Probability (math.PR)
[ total of 31 entries: 1-31 ]
[ showing up to 1000 entries per page: fewer | more ]

Disable MathJax (What is MathJax?)

Links to: arXiv, form interface, find, math, recent, 2111, contact, help  (Access key information)