arXiv subjects

David Gamarnik

Publications and source records attributed to David Gamarnik.

At least 73 records · Page 4Linked to original sources

arXiv2018

High Dimensional Linear Regression using Lattice Basis Reduction

We consider a high dimensional linear regression problem where the goal is to efficiently recover an unknown vector $β^*$ from $n$ noisy linear observations $Y=Xβ^*+W \in \mathbb{R}^n$, for known $X \in \mathbb{R}^{n \times p}$ and unknown $W \in \mathbb{R}^n$. Unlike most of the literature on this model we make no sparsity assumption on $β^*$. Instead we adopt a regularization based on assuming that the underlying vectors $β^*$ have rational entries with the same denominator $Q \in \mathbb{Z}_{>0}$. We call this $Q$-rationality assumption. We propose a new polynomial-time algorithm for this task which is based on the seminal Lenstra-Lenstra-Lovasz (LLL) lattice basis reduction algorithm. We establish that under the $Q$-rationality assumption, our algorithm recovers exactly the vector $β^*$ for a large class of distributions for the iid entries of $X$ and non-zero noise $W$. We prove that it is successful under small noise, even when the learner has access to only one observation ($n=1$). Furthermore, we prove that in the case of the Gaussian white noise for $W$, $n=o\left(p/\log p\right)$ and $Q$ sufficiently large, our algorithm tolerates a nearly optimal information-theoretic level of the noise.

math.ST↗

arXiv2018

Finding cliques using few probes

Consider algorithms with unbounded computation time that probe the entries of the adjacency matrix of an $n$ vertex graph, and need to output a clique. We show that if the input graph is drawn at random from $G_{n,\frac{1}{2}}$ (and hence is likely to have a clique of size roughly $2\log n$), then for every $δ< 2$ and constant $\ell$, there is an $α< 2$ (that may depend on $δ$ and $\ell$) such that no algorithm that makes $n^δ$ probes in $\ell$ rounds is likely (over the choice of the random graph) to output a clique of size larger than $α\log n$.

math.CO↗

arXiv2018

Suboptimality of local algorithms for a class of max-cut problems

We show that in random $K$-uniform hypergraphs of constant average degree, for even $K \geq 4$, local algorithms defined as factors of i.i.d. can not find nearly maximal cuts, when the average degree is sufficiently large. These algorithms have been used frequently to obtain lower bounds for the max-cut problem on random graphs, but it was not known whether they could be successful in finding nearly maximal cuts. This result follows from the fact that the overlap of any two nearly maximal cuts in such hypergraphs does not take values in a certain non-trivial interval - a phenomenon referred to as the overlap gap property - which is proved by comparing diluted models with large average degree with appropriate fully connected spin glass models and showing the overlap gap property in the latter setting.

math.PR↗

arXiv2018

A lower bound on the queueing delay in resource constrained load balancing

We consider the following distributed service model: jobs with unit mean, general distribution, and independent processing times arrive as a renewal process of rate $λn$, with $0<λ<1$, and are immediately dispatched to one of several queues associated with $n$ identical servers with unit processing rate. We assume that the dispatching decisions are made by a central dispatcher endowed with a finite memory, and with the ability to exchange messages with the servers. We study the fundamental resource requirements (memory bits and message exchange rate), in order to drive the expected queueing delay in steady-state of a typical job to zero, as $n$ increases. We develop a novel approach to show that, within a certain broad class of "symmetric" policies, every dispatching policy with a message rate of the order of $n$, and with a memory of the order of $\log n$ bits, results in an expected queueing delay which is bounded away from zero, uniformly as $n\to\infty$.

math.PR↗

arXiv2017

Delay, memory, and messaging tradeoffs in distributed service systems

We consider the following distributed service model: jobs with unit mean, exponentially distributed, and independent processing times arrive as a Poisson process of rate $λn$, with $0<λ<1$, and are immediately dispatched by a centralized dispatcher to one of $n$ First-In-First-Out queues associated with $n$ identical servers. The dispatcher is endowed with a finite memory, and with the ability to exchange messages with the servers. We propose and study a resource-constrained "pull-based" dispatching policy that involves two parameters: (i) the number of memory bits available at the dispatcher, and (ii) the average rate at which servers communicate with the dispatcher. We establish (using a fluid limit approach) that the asymptotic, as $n\to\infty$, expected queueing delay is zero when either (i) the number of memory bits grows logarithmically with $n$ and the message rate grows superlinearly with $n$, or (ii) the number of memory bits grows superlogarithmically with $n$ and the message rate is at least $λn$. Furthermore, when the number of memory bits grows only logarithmically with $n$ and the message rate is proportional to $n$, we obtain a closed-form expression for the (now positive) asymptotic delay. Finally, we demonstrate an interesting phase transition in the resource-constrained regime where the asymptotic delay is non-zero. In particular, we show that for any given $α>0$ (no matter how small), if our policy only uses a linear message rate $αn$, the resulting asymptotic delay is upper bounded, uniformly over all $λ<1$; this is in sharp contrast to the delay obtained when no messages are used ($α= 0$), which grows as $1/(1-λ)$ when $λ\uparrow 1$, or when the popular power-of-$d$-choices is used, in which the delay grows as $\log(1/(1-λ))$.

math.PR↗

arXiv2017

Matrix Completion from $O(n)$ Samples in Linear Time

We consider the problem of reconstructing a rank-$k$ $n \times n$ matrix $M$ from a sampling of its entries. Under a certain incoherence assumption on $M$ and for the case when both the rank and the condition number of $M$ are bounded, it was shown in \cite{CandesRecht2009, CandesTao2010, keshavan2010, Recht2011, Jain2012, Hardt2014} that $M$ can be recovered exactly or approximately (depending on some trade-off between accuracy and computational complexity) using $O(n \, \text{poly}(\log n))$ samples in super-linear time $O(n^{a} \, \text{poly}(\log n))$ for some constant $a \geq 1$. In this paper, we propose a new matrix completion algorithm using a novel sampling scheme based on a union of independent sparse random regular bipartite graphs. We show that under the same conditions w.h.p. our algorithm recovers an $ε$-approximation of $M$ in terms of the Frobenius norm using $O(n \log^2(1/ε))$ samples and in linear time $O(n \log^2(1/ε))$. This provides the best known bounds both on the sample complexity and computational complexity for reconstructing (approximately) an unknown low-rank matrix. The novelty of our algorithm is two new steps of thresholding singular values and rescaling singular vectors in the application of the "vanilla" alternating minimization algorithm. The structure of sparse random regular graphs is used heavily for controlling the impact of these regularization steps.

stat.ML↗

arXiv2017

Uniqueness of Gibbs Measures for Continuous Hardcore Models

We formulate a continuous version of the well known discrete hardcore (or independent set) model on a locally finite graph, parameterized by the so-called activity parameter $λ> 0$. In this version, the state or "spin value" $x_u$ of any node $u$ of the graph lies in the interval $[0,1]$, the hardcore constraint $x_u + x_v \leq 1$ is satisfied for every edge $(u,v)$ of the graph, and the space of feasible configurations is given by a convex polytope. When the graph is a regular tree, we show that there is a unique Gibbs measure associated to each activity parameter $λ>0$. Our result shows that, in contrast to the standard discrete hardcore model, the continuous hardcore model does not exhibit a phase transition on the infinite regular tree. We also consider a family of continuous models that interpolate between the discrete and continuous hardcore models on a regular tree when $λ= 1$ and show that each member of the family has a unique Gibbs measure, even when the discrete model does not. In each case, the proof entails the analysis of an associated Hamiltonian dynamical system that describes a certain limit of the marginal distribution at a node. Furthermore, given any sequence of regular graphs with fixed degree and girth diverging to infinity, we apply our results to compute the asymptotic limit of suitably normalized volumes of the corresponding sequence of convex polytopes of feasible configurations. In particular, this yields an approximation for the partition function of the continuous hard core model on a regular graph with large girth in the case $λ= 1$.

math.PR↗

arXiv2017

On the Max-Cut of Sparse Random Graphs

We consider the problem of estimating the size of a maximum cut (Max-Cut problem) in a random Erdős-Rényi graph on $n$ nodes and $\lfloor cn \rfloor$ edges. It is shown in Coppersmith et al. ~\cite{Coppersmith2004} that the size of the maximum cut in this graph normalized by the number of nodes belongs to the asymptotic region $[c/2+0.37613\sqrt{c},c/2+0.58870\sqrt{c}]$ with high probability (w.h.p.) as $n$ increases, for all sufficiently large $c$. In this paper we improve both upper and lower bounds by introducing a novel bounding technique. Specifically, we establish that the size of the maximum cut normalized by the number of nodes belongs to the interval $[c/2+0.47523\sqrt{c},c/2+0.55909\sqrt{c}]$ w.h.p. as $n$ increases, for all sufficiently large $c$. Instead of considering the expected number of cuts achieving a particular value as is done in the application of the first moment method, we observe that every maximum size cut satisfies a certain local optimality property, and we compute the expected number of cuts with a given value satisfying this local optimality property. Estimating this expectation amounts to solving a rather involved two dimensional large deviations problem. We solve this underlying large deviation problem asymptotically as $c$ increases and use it to obtain an improved upper bound on the Max-Cut value. The lower bound is obtained by application of the second moment method, coupled with the same local optimality constraint, and is shown to work up to the stated lower bound value $c/2+0.47523\sqrt{c}$. It is worth noting that both bounds are stronger than the ones obtained by standard first and second moment methods. Finally, we also obtain an improved lower bound of $1.36000n$ on the Max-Cut for the random cubic graph or any cubic graph with large girth, improving the previous best bound of $1.33773n$.

math.PR↗

arXiv2017

Supermarket Queueing System in the Heavy Traffic Regime. Short Queue Dynamics

We consider a queueing system with $n$ parallel queues operating according to the so-called "supermarket model" in which arriving customers join the shortest of $d$ randomly selected queues. Assuming rate $nλ_{n}$ Poisson arrivals and rate $1$ exponentially distributed service times, we consider this model in the heavy traffic regime, described by $λ_{n}\uparrow 1$ as $n\to\infty$. We give a simple expectation argument establishing that majority of queues have steady state length at least $\log_d(1-λ_{n})^{-1} - O(1)$ with probability approaching one as $n\rightarrow\infty$, implying the same for the steady state delay of a typical customer. Our main result concerns the detailed behavior of queues with length smaller than $\log_d(1-λ_{n})^{-1}-O(1)$. Assuming $λ_{n}$ converges to $1$ at rate at most $\sqrt{n}$, we show that the dynamics of such queues does not follow a diffusion process, as is typical for queueing systems in heavy traffic, but is described instead by a deterministic infinite system of linear differential equations, after an appropriate rescaling. The unique fixed point solution of this system is shown explicitly to be of the form $π_{1}(d^{i}-1)/(d-1), i\ge 1$, which we conjecture describes the steady state behavior of the queue lengths after the same rescaling. Our result is obtained by combination of several technical ideas including establishing the existence and uniqueness of an associated infinite dimensional system of non-linear integral equations and adopting an appropriate stopped process as an intermediate step.

math.PR↗

arXiv2016

A Message Passing Algorithm for the Problem of Path Packing in Graphs

We consider the problem of packing node-disjoint directed paths in a directed graph. We consider a variant of this problem where each path starts within a fixed subset of root nodes, subject to a given bound on the length of paths. This problem is motivated by the so-called kidney exchange problem, but has potential other applications and is interesting in its own right. We propose a new algorithm for this problem based on the message passing/belief propagation technique. A priori this problem does not have an associated graphical model, so in order to apply a belief propagation algorithm we provide a novel representation of the problem as a graphical model. Standard belief propagation on this model has poor scaling behavior, so we provide an efficient implementation that significantly decreases the complexity. We provide numerical results comparing the performance of our algorithm on both artificially created graphs and real world networks to several alternative algorithms, including algorithms based on integer programming (IP) techniques. These comparisons show that our algorithm scales better to large instances than IP-based algorithms and often finds better solutions than a simple algorithm that greedily selects the longest path from each root node. In some cases it also finds better solutions than the ones found by IP-based algorithms even when the latter are allowed to run significantly longer than our algorithm.

cs.DS↗

arXiv2016

Finding a Large Submatrix of a Gaussian Random Matrix

We consider the problem of finding a $k\times k$ submatrix of an $n\times n$ matrix with i.i.d. standard Gaussian entries, which has a large average entry. It was shown earlier by Bhamidi et al. that the largest average value of such a matrix is $2\sqrt{\log n/k}$ with high probability. In the same paper an evidence was provided that a natural greedy algorithm called Largest Average Submatrix ($\LAS$) should produce a matrix with average entry approximately $\sqrt{2}$ smaller. In this paper we show that the matrix produced by the $\LAS$ algorithm is indeed $\sqrt{2\log n/k}$ w.h.p. Then by drawing an analogy with the problem of finding cliques in random graphs, we propose a simple greedy algorithm which produces a $k\times k$ matrix with asymptotically the same average value. Since the greedy algorithm is the best known algorithm for finding cliques in random graphs, it is tempting to believe that beating the factor $\sqrt{2}$ performance gap suffered by both algorithms might be very challenging. Surprisingly, we show the existence of a very simple algorithm which produces a matrix with average value $(4/3)\sqrt{2\log n/k}$. To get an insight into the algorithmic hardness of this problem, and motivated by methods originating in the theory of spin glasses, we conduct the so-called expected overlap analysis of matrices with average value asymptotically $α\sqrt{2\log n/k}$. The overlap corresponds to the number of common rows and common columns for pairs of matrices achieving this value. We discover numerically an intriguing phase transition at $α^*\approx 1.3608..$: when $α<α^*$ the space of overlaps is a continuous subset of $[0,1]^2$, whereas $α=α^*$ marks the onset of discontinuity, and the model exhibits the Overlap Gap Property when $α>α^*$. We conjecture that $α>α^*$ marks the onset of the algorithmic hardness.

math.PR↗

arXiv2016

A Note on Alternating Minimization Algorithm for the Matrix Completion Problem

We consider the problem of reconstructing a low rank matrix from a subset of its entries and analyze two variants of the so-called Alternating Minimization algorithm, which has been proposed in the past. We establish that when the underlying matrix has rank $r=1$, has positive bounded entries, and the graph $\mathcal{G}$ underlying the revealed entries has bounded degree and diameter which is at most logarithmic in the size of the matrix, both algorithms succeed in reconstructing the matrix approximately in polynomial time starting from an arbitrary initialization. We further provide simulation results which suggest that the second algorithm which is based on the message passing type updates, performs significantly better.

stat.ML↗

arXiv2015

Join the Shortest Queue with Many Servers. The Heavy Traffic Asymptotics

We consider queueing systems with n parallel queues under a Join the Shortest Queue (JSQ) policy in the Halfin-Whitt heavy traffic regime. We use the martingale method to prove that a scaled process counting the number of idle servers and queues of length exactly 2 weakly converges to a two-dimensional reflected Ornstein-Uhlenbeck process, while processes counting longer queues converge to a deterministic system decaying to zero in constant time. This limiting system is comparable to that of the traditional Halfin-Whitt model, but there are key differences in the queueing behavior of the JSQ model. In particular, only a vanishing fraction of customers will have to wait, but those who do will incur a constant order waiting time.

math.PR↗

arXiv2014

Structure learning of antiferromagnetic Ising models

In this paper we investigate the computational complexity of learning the graph structure underlying a discrete undirected graphical model from i.i.d. samples. We first observe that the notoriously difficult problem of learning parities with noise can be captured as a special case of learning graphical models. This leads to an unconditional computational lower bound of $Ω(p^{d/2})$ for learning general graphical models on $p$ nodes of maximum degree $d$, for the class of so-called statistical algorithms recently introduced by Feldman et al (2013). The lower bound suggests that the $O(p^d)$ runtime required to exhaustively search over neighborhoods cannot be significantly improved without restricting the class of models. Aside from structural assumptions on the graph such as it being a tree, hypertree, tree-like, etc., many recent papers on structure learning assume that the model has the correlation decay property. Indeed, focusing on ferromagnetic Ising models, Bento and Montanari (2009) showed that all known low-complexity algorithms fail to learn simple graphs when the interaction strength exceeds a number related to the correlation decay threshold. Our second set of results gives a class of repelling (antiferromagnetic) models that have the opposite behavior: very strong interaction allows efficient learning in time $O(p^2)$. We provide an algorithm whose performance interpolates between $O(p^2)$ and $O(p^{d+2})$ depending on the strength of the repulsion.

stat.ML↗

arXiv2014

Learning graphical models from the Glauber dynamics

In this paper we consider the problem of learning undirected graphical models from data generated according to the Glauber dynamics. The Glauber dynamics is a Markov chain that sequentially updates individual nodes (variables) in a graphical model and it is frequently used to sample from the stationary distribution (to which it converges given sufficient time). Additionally, the Glauber dynamics is a natural dynamical model in a variety of settings. This work deviates from the standard formulation of graphical model learning in the literature, where one assumes access to i.i.d. samples from the distribution. Much of the research on graphical model learning has been directed towards finding algorithms with low computational cost. As the main result of this work, we establish that the problem of reconstructing binary pairwise graphical models is computationally tractable when we observe the Glauber dynamics. Specifically, we show that a binary pairwise graphical model on $p$ nodes with maximum degree $d$ can be learned in time $f(d)p^2\log p$, for a function $f(d)$, using nearly the information-theoretic minimum number of samples.

cs.LG↗

arXiv2014

Performance of the Survey Propagation-guided decimation algorithm for the random NAE-K-SAT problem

We show that the Survey Propagation-guided decimation algorithm fails to find satisfying assignments on random instances of the "Not-All-Equal-$K$-SAT" problem if the number of message passing iterations is bounded by a constant independent of the size of the instance and the clause-to-variable ratio is above $(1+o_K(1)){2^{K-1}\over K}\log^2 K$ for sufficiently large $K$. Our analysis in fact applies to a broad class of algorithms described as "sequential local algorithms". Such algorithms iteratively set variables based on some local information and then recurse on the reduced instance. Survey Propagation-guided as well as Belief Propagation-guided decimation algorithms - two widely studied message passing based algorithms, fall under this category of algorithms provided the number of message passing iterations is bounded by a constant. Another well-known algorithm falling into this category is the Unit Clause algorithm. Our work constitutes the first rigorous analysis of the performance of the SP-guided decimation algorithm. The approach underlying our paper is based on an intricate geometry of the solution space of random NAE-$K$-SAT problem. We show that above the $(1+o_K(1)){2^{K-1}\over K}\log^2 K$ threshold, the overlap structure of $m$-tuples of satisfying assignments exhibit a certain clustering behavior expressed in the form of constraints on distances between the $m$ assignments, for appropriately chosen $m$. We further show that if a sequential local algorithm succeeds in finding a satisfying assignment with probability bounded away from zero, then one can construct an $m$-tuple of solutions violating these constraints, thus leading to a contradiction. Along with (citation), this result is the first work which directly links the clustering property of random constraint satisfaction problems to the computational hardness of finding satisfying assignments.

math.PR↗

arXiv2014

Local Algorithms for Graphs

We are going to analyze local algorithms over sparse random graphs. These algorithms are based on local information where local regards to a decision made by the exploration of a small neighbourhood of a certain vertex plus a believe of the structure of the whole graph and maybe added some randomness. This kind of algorithms can be a natural response to the given problem or an efficient approximation such as the Belief Propagation Algorithm.

cond-mat.dis-nn↗

arXiv2014

Hardness of parameter estimation in graphical models

We consider the problem of learning the canonical parameters specifying an undirected graphical model (Markov random field) from the mean parameters. For graphical models representing a minimal exponential family, the canonical parameters are uniquely determined by the mean parameters, so the problem is feasible in principle. The goal of this paper is to investigate the computational feasibility of this statistical task. Our main result shows that parameter estimation is in general intractable: no algorithm can learn the canonical parameters of a generic pair-wise binary graphical model from the mean parameters in time bounded by a polynomial in the number of variables (unless RP = NP). Indeed, such a result has been believed to be true (see the monograph by Wainwright and Jordan (2008)) but no proof was known. Our proof gives a polynomial time reduction from approximating the partition function of the hard-core model, known to be hard, to learning approximate parameters. Our reduction entails showing that the marginal polytope boundary has an inherent repulsive property, which validates an optimization procedure over the polytope that does not use any knowledge of its structure (as required by the ellipsoid method and others).

cs.CC↗