arXiv subjects

Elchanan Mossel

Publications and source records attributed to Elchanan Mossel.

At least 217 records · Page 12Linked to original sources

arXiv2007

Incomplete Lineage Sorting: Consistent Phylogeny Estimation From Multiple Loci

We introduce a simple algorithm for reconstructing phylogenies from multiple gene trees in the presence of incomplete lineage sorting, that is, when the topology of the gene trees may differ from that of the species tree. We show that our technique is statistically consistent under standard stochastic assumptions, that is, it returns the correct tree given sufficiently many unlinked loci. We also show that it can tolerate moderate estimation errors.

q-bio.PE↗

arXiv2007

Sorting and Selection in Posets

Classical problems of sorting and searching assume an underlying linear ordering of the objects being compared. In this paper, we study a more general setting, in which some pairs of objects are incomparable. This generalization is relevant in applications related to rankings in sports, college admissions, or conference submissions. It also has potential applications in biology, such as comparing the evolutionary fitness of different strains of bacteria, or understanding input-output relations among a set of metabolic reactions or the causal influences among a set of interacting genes or proteins. Our results improve and extend results from two decades ago of Faigle and Turán. A measure of complexity of a partially ordered set (poset) is its width. Our algorithms obtain information about a poset by queries that compare two elements. We present an algorithm that sorts, i.e. completely identifies, a width w poset of size n and has query complexity O(wn + nlog(n)), which is within a constant factor of the information-theoretic lower bound. We also show that a variant of Mergesort has query complexity O(wn(log(n/w))) and total complexity O((w^2)nlog(n/w)). Faigle and Turán have shown that the sorting problem has query complexity O(wn(log(n/w))) but did not address its total complexity. For the related problem of determining the minimal elements of a poset, we give efficient deterministic and randomized algorithms with O(wn) query and total complexity, along with matching lower bounds for the query complexity up to a factor of 2. We generalize these results to the k-selection problem of determining the elements of height at most k. We also derive upper bounds on the total complexity of some other problems of a similar flavor.

cs.DS↗

arXiv2007

Noisy Sorting Without Resampling

In this paper we study noisy sorting without re-sampling. In this problem there is an unknown order $a_{π(1)} < ... < a_{π(n)}$ where $π$ is a permutation on $n$ elements. The input is the status of $n \choose 2$ queries of the form $q(a_i,x_j)$, where $q(a_i,a_j) = +$ with probability at least $1/2+\ga$ if $π(i) > π(j)$ for all pairs $i \neq j$, where $\ga > 0$ is a constant and $q(a_i,a_j) = -q(a_j,a_i)$ for all $i$ and $j$. It is assumed that the errors are independent. Given the status of the queries the goal is to find the maximum likelihood order. In other words, the goal is find a permutation $σ$ that minimizes the number of pairs $σ(i) > σ(j)$ where $q(σ(i),σ(j)) = -$. The problem so defined is the feedback arc set problem on distributions of inputs, each of which is a tournament obtained as a noisy perturbations of a linear order. Note that when $\ga < 1/2$ and $n$ is large, it is impossible to recover the original order $π$. It is known that the weighted feedback are set problem on tournaments is NP-hard in general. Here we present an algorithm of running time $n^{O(γ^{-4})}$ and sampling complexity $O_γ(n \log n)$ that with high probability solves the noisy sorting without re-sampling problem. We also show that if $a_{σ(1)},a_{σ(2)},...,a_{σ(n)}$ is an optimal solution of the problem then it is ``close'' to the original order. More formally, with high probability it holds that $\sum_i |σ(i) - π(i)| = Θ(n)$ and $\max_i |σ(i) - π(i)| = Θ(\log n)$. Our results are of interest in applications to ranking, such as ranking in sports, or ranking of search items based on comparisons by experts.

cs.DS↗

arXiv2007

Limitations of Markov chain Monte Carlo algorithms for Bayesian Inference of phylogeny

Markov chain Monte Carlo algorithms play a key role in the Bayesian approach to phylogenetic inference. In this paper, we present the first theoretical work analyzing the rate of convergence of several Markov chains widely used in phylogenetic inference. We analyze simple, realistic examples where these Markov chains fail to converge quickly. In particular, the data studied are generated from a pair of trees, under a standard evolutionary model. We prove that many of the popular Markov chains take exponentially long to reach their stationary distribution. Our construction is pertinent since it is well known that phylogenetic trees for genes may differ within a single organism. Our results shed a cautionary light on phylogenetic analysis using Bayesian inference and highlight future directions for potential theoretical work.

q-bio.PE↗

arXiv2007

On the hardness of sampling independent sets beyond the tree threshold

We consider local Markov chain Monte-Carlo algorithms for sampling from the weighted distribution of independent sets with activity $ł$, where the weight of an independent set $I$ is $ł^{|I|}$. A recent result has established that Gibbs sampling is rapidly mixing in sampling the distribution for graphs of maximum degree $d$ and $ł<ł_c(d)$, where $ł_c(d)$ is the critical activity for uniqueness of the Gibbs measure (i.e., for decay of correlations with distance in the weighted distribution over independent sets) on the $d$-regular infinite tree. We show that for $d \geq 3$, $ł$ just above $ł_c(d)$ with high probability over $d$-regular bipartite graphs, any local Markov chain Monte-Carlo algorithm takes exponential time before getting close to the stationary distribution. Our results provide a rigorous justification for ``replica'' method heuristics. These heuristics were invented in theoretical physics and are used in order to derive predictions on Gibbs measures on random graphs in terms of Gibbs measures on trees. We conjecture that $ł_c$ is in fact the exact threshold for this computational problem, i.e., that for $ł>ł_c$ it is NP-hard to approximate the above weighted sum overindependent sets to within a factor polynomial in the size of the graph.

math.PR↗

arXiv2006

Learning nonsingular phylogenies and hidden Markov models

In this paper we study the problem of learning phylogenies and hidden Markov models. We call a Markov model nonsingular if all transition matrices have determinants bounded away from 0 (and 1). We highlight the role of the nonsingularity condition for the learning problem. Learning hidden Markov models without the nonsingularity condition is at least as hard as learning parity with noise, a well-known learning problem conjectured to be computationally hard. On the other hand, we give a polynomial-time algorithm for learning nonsingular phylogenies and hidden Markov models.

cs.LG↗

arXiv2006

The Kesten-Stigum Reconstruction Bound Is Tight for Roughly Symmetric Binary Channels

We establish the exact threshold for the reconstruction problem for a binary asymmetric channel on the b-ary tree, provided that the asymmetry is sufficiently small. This is the first exact reconstruction threshold obtained in roughly a decade. We discuss the implications of our result for Glauber dynamics, phylogenetic reconstruction, and so-called ``replica symmetry breaking'' in spin glasses and random satisfiability problems.

math.PR↗

arXiv2005

A New Look at Survey Propagation and its Generalizations

This paper provides a new conceptual perspective on survey propagation, which is an iterative algorithm recently introduced by the statistical physics community that is very effective in solving random k-SAT problems even with densities close to the satisfiability threshold. We first describe how any SAT formula can be associated with a novel family of Markov random fields (MRFs), parameterized by a real number ρ\in [0,1]. We then show that applying belief propagation--a well-known ``message-passing'' technique for estimating marginal probabilities--to this family of MRFs recovers a known family of algorithms, ranging from pure survey propagation at one extreme (ρ= 1) to standard belief propagation on the uniform distribution over SAT assignments at the other extreme (ρ= 0). Configurations in these MRFs have a natural interpretation as partial satisfiability assignments, on which a partial order can be defined. We isolate cores as minimal elements in this partial ordering, which are also fixed points of survey propagation and the only assignments with positive probability in the MRF for ρ=1. Our experimental results for k=3 suggest that solutions of random formulas typically do not possess non-trivial cores. This makes it necessary to study the structure of the space of partial assignments for ρ<1 and investigate the role of assignments that are very close to being cores. To that end, we investigate the associated lattice structure, and prove a weight-preserving identity that shows how any MRF with ρ>0 can be viewed as a ``smoothed'' version of the uniform distribution over satisfying assignments (ρ=0). Finally, we isolate properties of Gibbs sampling and message-passing algorithms that are typical for an ensemble of k-SAT problems.

cs.CC↗

arXiv2005

Slow Emergence of Cooperation for Win-Stay Lose-Shift on Trees

We consider a group of agents on a graph who repeatedly play the prisoner's dilemma game against their neighbors. The players adapt their actions to the past behavior of their opponents by applying the win-stay lose-shift strategy. On a finite connected graph, it is easy to see that the system learns to cooperate by converging to the all-cooperate state in a finite time. We analyze the rate of convergence in terms of the size and structure of the graph. [Dyer et al., 2002] showed that the system converges rapidly on the cycle, but that it takes a time exponential in the size of the graph to converge to cooperation on the complete graph. We show that the emergence of cooperation is exponentially slow in some expander graphs. More surprisingly, we show that it is also exponentially slow in bounded-degree trees, where many other dynamics are known to converge rapidly.

math.PR↗

arXiv2005

Noise stability of functions with low influences: invariance and optimality

In this paper we study functions with low influences on product probability spaces. The analysis of boolean functions with low influences has become a central problem in discrete Fourier analysis. It is motivated by fundamental questions arising from the construction of probabilistically checkable proofs in theoretical computer science and from problems in the theory of social choice in economics. We prove an invariance principle for multilinear polynomials with low influences and bounded degree; it shows that under mild conditions the distribution of such polynomials is essentially invariant for all product spaces. Ours is one of the very few known non-linear invariance principles. It has the advantage that its proof is simple and that the error bounds are explicit. We also show that the assumption of bounded degree can be eliminated if the polynomials are slightly ``smoothed''; this extension is essential for our applications to ``noise stability''-type problems. In particular, as applications of the invariance principle we prove two conjectures: the ``Majority Is Stablest'' conjecture from theoretical computer science, which was the original motivation for this work, and the ``It Ain't Over Till It's Over'' conjecture from social choice theory.

math.PR↗

arXiv2005

Conditional Hardness for Approximate Coloring

We study the coloring problem: Given a graph G, decide whether $c(G) \leq q$ or $c(G) \ge Q$, where c(G) is the chromatic number of G. We derive conditional hardness for this problem for any constant $3 \le q < Q$. For $q\ge 4$, our result is based on Khot's 2-to-1 conjecture [Khot'02]. For $q=3$, we base our hardness result on a certain `fish shaped' variant of his conjecture. We also prove that the problem almost coloring is hard for any constant $\eps>0$, assuming Khot's Unique Games conjecture. This is the problem of deciding for a given graph, between the case where one can 3-color all but a $\eps$ fraction of the vertices without monochromatic edges, and the case where the graph contains no independent set of relative size at least $\eps$. Our result is based on bounding various generalized noise-stability quantities using the invariance principle of Mossel et al [MOO'05].

cs.CC↗

arXiv2005

Robust reconstruction on trees is determined by the second eigenvalue

Consider a Markov chain on an infinite tree T=(V,E) rooted at ρ. In such a chain, once the initial root state σ(ρ) is chosen, each vertex iteratively chooses its state from the one of its parent by an application of a Markov transition rule (and all such applications are independent). Let μ_j denote the resulting measure for σ(ρ)=j. The resulting measure μ_j is defined on configurations σ=(σ(x))_{x\in V}\in A^V, where A is some finite set. Let μ_j^n denote the restriction of μto the sigma-algebra generated by the variables σ(x), where x is at distance exactly n from ρ. Letting α_n=max_{i,j\in A}d_{TV}(μ_i^n,μ_j^n), where d_{TV} denotes total variation distance, we say that the reconstruction problem is solvable if lim inf_{n\to\infty}α_n>0. Reconstruction solvability roughly means that the nth level of the tree contains a nonvanishing amount of information on the root of the tree as n\to\infty. In this paper we study the problem of robust reconstruction. Let νbe a nondegenerate distribution on A and ε>0. Let σbe chosen according to μ_j^n and σ' be obtained from σby letting for each node independently, σ(v)=σ'(v) with probability 1-εand σ'(v) be an independent sample from νotherwise. We denote by μ_j^n[ν,ε] the resulting measure on σ'. The measure μ_j^n[ν,ε] is a perturbation of the measure μ_j^n.

math.PR↗

arXiv2004

Non-interactive correlation distillation, inhomogeneous Markov chains, and the reverse Bonami-Beckner inequality

In this paper we study non-interactive correlation distillation (NICD), a generalization of the study of noise sensitivity of boolean functions. We extend the model to NICD on trees. In this model there is a fixed undirected tree with players at some of the nodes. One node is given a uniformly random string and this string is distributed throughout the network, with the edges of the tree acting as independent binary symmetric channels. The goal of the players is to agree on a shared random bit without communicating. Our new contributions include the following: 1. In the case of a k-leaf star graph, we resolve the open question of whether the success probability must go to zero as k goes to infinity. We show that this is indeed the case and provide matching upper and lower bounds on the asymptotically optimal rate (a slowly-decaying polynomial). 2. In the case of the k-vertex path graph, we show that it is always optimal for all players to use the same 1-bit function. 3. In the general case we show that all players should use monotone functions. 4. For certain trees it is better if not all players use the same function. Our techniques include the use of the reverse Bonami-Beckner inequality.

math.PR↗

arXiv2004

Random autocatalytic networks

We determine conditions under which a random biochemical system is likely to contain a subsystem that is both autocatalytic and able to survive on some ambient `food' source. Such systems have previously been investigated for their relevance to origin-of-life models. In this paper we extend earlier work, by finding precisely the order of catalysation required for the emergence of such self-sustaining autocatalytic networks. This answers questions raised in earlier papers, yet also allows for a more general class of models. We also show that a recently-described polynomial time algorithm for determining whether a catalytic reaction system contains an autocatalytic, self-sustaining subsystem is unlikely to adapt to allow inhibitory catalysation - in this case we show that the associated decision problem is NP-complete.

q-bio.MN↗

arXiv2004

Coin flipping from a cosmic source: On error correction of truly random bits

We study a problem related to coin flipping, coding theory, and noise sensitivity. Consider a source of truly random bits $x \in \bits^n$, and $k$ parties, who have noisy versions of the source bits $y^i \in \bits^n$, where for all $i$ and $j$, it holds that $\Pr[y^i_j = x_j] = 1 - \eps$, independently for all $i$ and $j$. That is, each party sees each bit correctly with probability $1-ε$, and incorrectly (flipped) with probability $ε$, independently for all bits and all parties. The parties, who cannot communicate, wish to agree beforehand on {\em balanced} functions $f_i : \bits^n \to \bits$ such that $\Pr[f_1(y^1) = ... = f_k(y^k)]$ is maximized. In other words, each party wants to toss a fair coin so that the probability that all parties have the same coin is maximized. The functions $f_i$ may be thought of as an error correcting procedure for the source $x$. When $k=2,3$ no error correction is possible, as the optimal protocol is given by $f_i(x^i) = y^i_1$. On the other hand, for large values of $k$, better protocols exist. We study general properties of the optimal protocols and the asymptotic behavior of the problem with respect to $k$, $n$ and $\eps$. Our analysis uses tools from probability, discrete Fourier analysis, convexity and discrete symmetrization.

math.PR↗

arXiv2004

A Law of Large Numbers for Weighted Majority

Consider an election between two candidates in which the voters' choices are random and independent and the probability of a voter choosing the first candidate is $p>1/2$. Condorcet's Jury Theorem which he derived from the weak law of large numbers asserts that if the number of voters tends to infinity then the probability that the first candidate will be elected tends to one. The notion of influence of a voter or its voting power is relevant for extensions of the weak law of large numbers for voting rules which are more general than simple majority. In this paper we point out two different ways to extend the classical notions of voting power and influences to arbitrary probability distributions. The extension relevant to us is the ``effect'' of a voter, which is a weighted version of the correlation between the voter's vote and the election's outcomes. We prove an extension of the weak law of large numbers to weighted majority games when all individual effects are small and show that this result does not apply to any voting rule which is not based on weighted majority.

math.PR↗

arXiv2004

How much can evolved characters tell us about the tree that generated them?

In this paper we review some recent results that shed light on a fundamental question in molecular systematics: how much phylogenetic `signal' can we expect from characters that have evolved under some Markov process? There are many sides to this question and we begin by describing some explicit bounds on the probability of correctly reconstructing an ancestral state from the states observed at the tips. We show how this bound sets upper limits on the probability of tree reconstruction from aligned sequences, and we provide some new extensions that allow site-to-site rate variation or a covarion mechanism. We then explore the relationship between the number of sites required for accurate tree reconstruction and other model parameters - such as the number of species, and substitution probabilities, and we describe a phase transition that occurs when substitution probabilities exceed a critical value. In the remainder of this paper we turn to models of character evolution where the state space is assumed to be either infinite or very large. These models have some relevance to certain types of genomic data (such as gene order) and here we again investigate how many characters are required for accurate tree reconstruction.

q-bio.PE↗

arXiv2004

Survey: Information flow on trees

Consider a tree network T, where each edge acts as an independent copy of a given channel M, and information is propagated from the root. For which T and M does the configuration obtained at level n of T typically contain significant information on the root variable? This model appeared independently in biology, information theory, and statistical physics. Its analysis uses techniques from the theory of finite markov chains, statistics, statistical physics, information theory, cryptography and noisy computation. In this paper, we survey developments and challenges related to this problem.

math.PR↗