arXiv subjects

Yevgeniy Kovchegov

Publications and source records attributed to Yevgeniy Kovchegov.

At least 19 recordsLinked to original sources

arXiv2025

Gelation in Vector Multiplicative Coalescence and Extinction in Multi-Type Poisson Branching Processes

In this note, we present a novel connection between a multi-type (vector) multiplicative coalescent process and a multi-type branching process with Poisson offspring distributions. More specifically, we show that the equations that govern the phenomenon of gelation in the vector multiplicative coalescent process are equivalent to the equations that yield the extinction probabilities of the corresponding multi-type Poisson branching process. We then leverage this connection with two applications, one in each direction. The first is a new quick proof of gelation in the vector multiplicative coalescent process, and the second is a new series expression for the extinction probabilities of the multi-type Poisson branching process. We also use random graphs to give a new derivation of the solution to the modified Smoluchowski coagulation equations, which describe the vector multiplicative coalescent process.

math.PR↗

arXiv2022

Extensions of true skewness for unimodal distributions

A 2022 paper arXiv:2009.10305v4 introduced the notion of true positive and negative skewness for continuous random variables via Fréchet $p$-means. In this work, we find novel criteria for true skewness, establish true skewness for the Weibull, Lévy, skew-normal, and chi-squared distributions, and discuss the extension of true skewness to discrete and multivariate settings. Furthermore, some relevant properties of the $p$-means of random variables are established.

math.PR↗

arXiv2022

Invariant Galton-Watson trees: metric properties and attraction with respect to generalized dynamical pruning

Invariant Galton-Watson (IGW) tree measures is a one-parameter family of critical Galton-Watson measures invariant with respect to a large class of tree reduction operations. Such operations include the generalized dynamical pruning (also known as hereditary reduction in a real tree setting) that eliminates descendant subtrees according to the value of an arbitrary subtree function that is monotone nondecreasing with respect to an isometry-induced partial tree order. We show that, under a mild regularity condition, the IGW measures are the only attractors of critical Galton-Watson measures with respect to the generalized dynamical pruning. We also derive the distributions of height, length, and size of the IGW trees.

math.PR↗

arXiv2022

A new life of Pearson's skewness

In this work we show how coupling and stochastic dominance methods can be successfully applied to a classical problem of rigorizing Pearson's skewness. Here, we use Fréchet means to define generalized notions of positive and negative skewness that we call truly positive and truly negative. Then, we apply stochastic dominance approach in establishing criteria for determining whether a continuous random variable is truly positively skewed. Intuitively, this means that scaled right tail of the probability density function exhibits strict stochastic dominance over equivalently scaled left tail. Finally, we use the stochastic dominance criteria and establish some basic examples of true positive skewness, thus demonstrating how the approach works in general.

math.PR↗

arXiv2021

The C-SHIFT algorithm for normalizing covariances

Omics technologies are powerful tools for analyzing patterns in gene expression data for thousands of genes. Due to a number of systematic variations in experiments, the raw gene expression data is often obfuscated by undesirable technical noises. Various normalization techniques were designed in an attempt to remove these non-biological errors prior to any statistical analysis. One of the reasons for normalizing data is the need for recovering the covariance matrix used in gene network analysis. In this paper, we introduce a novel normalization technique, called the covariance shift (C-SHIFT) method. This normalization algorithm uses optimization techniques together with the blessing of dimensionality philosophy and energy minimization hypothesis for covariance matrix recovery under additive noise (in biology, known as the bias). Thus, it is perfectly suited for the analysis of logarithmic gene expression data. Numerical experiments on synthetic data demonstrate the method's advantage over the classical normalization techniques. Namely, the comparison is made with Rank, Quantile, cyclic LOESS (locally estimated scatterplot smoothing), and MAD (median absolute deviation) normalization methods. We also evaluate the performance of C-SHIFT algorithm on real biological data.

q-bio.GN↗

arXiv2021

Multidimensional Lambert-Euler inversion and vector-multiplicative coalescent processes

In this paper we show the existence of the minimal solution to the multidimensional Lambert-Euler inversion, a multidimensional generalization of $[-e^{-1} ,0)$ branch of Lambert W function $W_0(x)$. Specifically, for a given nonnegative irreducible symmetric matrix $V \in \mathbb{R}^{k \times k}$, we show that for ${\bf u}\in(0,\infty)^k$, if equation $$y_j \exp\{-{\bf e}_j^T V {\bf y} \} = u_j ~~~~~~\forall j=1,...,k,$$ has at least one solution, it must have a minimal solution ${\bf y}^*$, where the minimum is achieved in all coordinates $y_j$ simultaneously. Moreover, such ${\bf y}^*$ is the unique solution satisfying $ρ\left(V D[y^*_j] \right) \leq 1$, where $D[y^*_j]={\sf diag}(y_j^*)$ is the diagonal matrix with entries $y^*_j$ and $ρ$ denotes the spectral radius. Our main application is in the vector-multiplicative coalescent process. It is a coalescent process with $k$ types of particles and vector-valued weights that begins with $α_1n+...+α_k n$ particles partitioned into types of respective sizes, and in which two clusters of weights ${\bf x}$ and ${\bf y}$ would merge with rate $({\bf x}^{\sf T} V {\bf y})/n$. We use combinatorics to solve the corresponding modified Smoluchowski equations, obtained as a hydrodynamic limit of vector-multiplicative coalescent as $n \to \infty$, and use multidimensional Lambert-Euler inversion to establish gelation and find a closed form expression for the gelation time. We also find the asymptotic length of the minimal spanning tree for a broad range of graphs equipped with random edge lengths.

math-ph↗

arXiv2021

Critical Tokunaga model for river networks

The hierarchical organization and self-similarity in river basins have been topics of extensive research in hydrology and geomorphology starting with the pioneering work of Horton in 1945. Despite significant theoretical and applied advances however, the mathematical origin of and relation among Horton laws for different stream attributes remain unsettled. Here we capitalize on a recently developed theory of random self-similar trees to introduce a one-parametric family of self-similar critical Tokunaga trees that elucidates the origin of Horton laws, Hack's laws, basin fractal dimension, power-law distributions of link attributes, and power-law relations between distinct attributes. The proposed family includes the celebrated Shreve's random topology model and extends to trees that approximate the observed river networks with realistic exponents. The results offer tools to increase our understanding of landscape organization under different hydroclimatic forcings, and to extend scaling relationships useful for hydrologic prediction to resolutions higher that those observed.

physics.geo-ph↗

arXiv2020

Invariance and attraction properties of Galton-Watson trees

We give a description of invariants and attractors of the critical and subcritical Galton-Watson tree measures under the operation of Horton pruning (cutting tree leaves with subsequent series reduction). Under a regularity condition, the class of invariant measures consists of the critical binary Galton-Watson tree and a one-parameter family of critical Galton-Watson trees with offspring distribution $\{q_k\}$ that has a power tail $q_k\sim Ck^{-(1+1/q_0)}$, where $q_0\in(1/2,1)$. Each invariant measure has a non-empty domain of attraction under consecutive Horton pruning, specified by the tail behavior of the initial Galton-Watson offspring distribution. The invariant measures satisfy the Toeplitz property for the Tokunaga coefficients and obey the Horton law with exponent $R = (1-q_0)^{-1/q_0}$.

math.PR↗

arXiv2019

Cross-Multiplicative Coalescent Processes and Applications

We introduce and analyze a novel type of coalescent processes called cross-multiplicative coalescent that models a system with two types of particles, $A$ and $B$. The bonds are formed only between the pairs of particles of opposite types with the same rate for each bond, producing connected components made of particles of both types. We analyze and solve the Smoluchowski coagulation system of equations obtained as a hydrodynamic limit of the corresponding Marcus-Lushnikov process. We establish that the cross-multiplicative kernel is a gelling kernel, and find the gelation time. As an application, we derive the limiting mean length of a minimal spanning tree on a complete bipartite graph $K_{α[n], β[n]}$ with partitions of sizes $α[n]=αn +o(\sqrt{n})$ and $β[n]=βn +o(\sqrt{n})$ and independent edge weights, distributed uniformly over $[0, 1]$.

math.PR↗

arXiv2019

Random Self-Similar Trees: A mathematical theory of Horton laws

The Horton laws originated in hydrology with a 1945 paper by Robert E. Horton, and for a long time remained a purely empirical finding. Ubiquitous in hierarchical branching systems, the Horton laws have been rediscovered in many disciplines ranging from geomorphology to genetics to computer science. Attempts to build a mathematical foundation behind the Horton laws during the 1990s revealed their close connection to the operation of pruning -- erasing a tree from the leaves down to the root. This survey synthesizes recent results on invariances and self-similarities of tree measures under various forms of pruning. We argue that pruning is an indispensable instrument for describing branching structures and representing a variety of coalescent and annihilation dynamics. The Horton laws appear as a characteristic imprint of self-similarity, which settles some questions prompted by geophysical data.

math.PR↗

arXiv2018

Random self-similar trees and a hierarchical branching process

We study self-similarity in random binary rooted trees. In a well-understood case of Galton-Watson trees, a distribution on a space of trees is said to be self-similar if it is invariant with respect to the operation of pruning, which cuts the tree leaves. This only happens for the critical Galton-Watson tree (a constant process progeny), which also exhibits other special symmetries. We extend the prune-invariance setup to arbitrary binary trees with edge lengths. In this general case the class of self-similar processes becomes much richer and covers a variety of practically important situations. The main result is construction of the hierarchical branching processes that satisfy various self-similarity definitions (including mean self-similarity and self-similarity in edge-lengths) depending on the process parameters. Taking the limit of averaged stochastic dynamics, as the number of trajectories increases, we obtain a deterministic system of differential equations that describes the process evolution. This system is used to establish a phase transition that separates fading and explosive behavior of the average process progeny. We describe a class of critical Tokunaga processes that happen at the phase transition boundary. They enjoy multiple additional symmetries and include the celebrated critical binary Galton-Watson tree with independent exponential edge length as a special case. Finally, we discuss a duality between trees and continuous functions, and introduce a class of extreme-invariant processes, constructed as the Harris paths of a self-similar hierarchical branching process, whose local minima has the same (linearly scaled) distribution as the original process.

math.PR↗

arXiv2018

Dynamical pruning of rooted trees with applications to 1D ballistic annihilation

We introduce generalized dynamical pruning on rooted binary trees with edge lengths. The pruning removes parts of a tree $T$, starting from the leaves, according to a pruning function defined on subtrees within $T$. The generalized pruning encompasses a number of discrete and continuous pruning operations, including the tree erasure and Horton pruning. The main result is invariance of a finite critical binary Galton-Watson tree with exponential edge lengths with respect to the generalized dynamical pruning for an arbitrary admissible pruning function. The second part of the paper examines the continuum 1-D ballistic annihilation model $A+A \rightarrow \emptyset$ for a constant particle density and initial velocity that alternates between the values of $\pm$1. The model evolution is equivalent to a generalized dynamical pruning of the shock tree that represents dynamics of sinks (points of particle annihilation), with the pruning function equal to the total tree length. The shock tree is isometric to the level set tree of the model potential (integral of velocity). This equivalence allows us to construct a complete probabilistic description of the annihilation dynamics for the initial velocity that alternates between the values of $\pm$1 at the epochs of a stationary Poisson process. Finally, we discuss several real tree representations of the ballistic annihilation model, closely connected to the shock wave tree.

math.PR↗

arXiv2018

Tokunaga self-similarity arises naturally from time invariance

The Tokunaga condition is an algebraic rule that provides a detailed description of the branching structure in a self-similar tree. Despite a solid empirical validation and practical convenience, the Tokunaga condition lacks a theoretical justification. Such a justification is suggested in this work. We define a geometric branching processes $\mathcal{G}(s)$ that generates self-similar rooted trees. The main result establishes the equivalence between the invariance of $\mathcal{G}(s)$ with respect to a time shift and a one-parametric version of the Tokunaga condition. In the parameter region where the process satisfies the Tokunaga condition (and hence is time invariant), $\mathcal{G}(s)$ enjoys many of the symmetries observed in a critical binary Galton-Watson branching process and reproduce the latter for a particular parameter value.

math.DS↗

arXiv2017

The aggregate path coupling method for the Potts model on bipartite graph

In this paper, we derive the large deviations principle for the Potts model on the complete bipartite graph $K_{n,n}$ as $n$ increases to infinity. Next, for the Potts model on $K_{n,n}$, we provide an extension of the method of aggregate path coupling that was originally developed in Kovchegov et al 2011 for the mean-field Blume-Capel model and in Kovchegov and Otto 2015 for a general mean-field setting that included the Generalized Curie-Weiss-Potts model analyzed in Cuff et al 2012. We use the aggregate path coupling method to identify and prove the interface value $β_s$ separating the rapid and slow mixing regimes for the Glauber dynamics of the Potts model on $K_{n,n}$.

math.PR↗

arXiv2015

Horton Law in Self-Similar Trees

Self-similarity of random trees is related to the operation of pruning. Pruning $R$ cuts the leaves and their parental edges and removes the resulting chains of degree-two nodes from a finite tree. A Horton-Strahler order of a vertex $v$ and its parental edge is defined as the minimal number of prunings necessary to eliminate the subtree rooted at $v$. A branch is a group of neighboring vertices and edges of the same order. The Horton numbers $N_k[K]$ and $N_{ij}[K]$ are defined as the expected number of branches of order $k$, and the expected number of order-$i$ branches that merged order-$j$ branches, $j>i$, respectively, in a finite tree of order $K$. The Tokunaga coefficients are defined as $T_{ij}[K]=N_{ij}[K]/N_j[K]$. The pruning decreases the orders of tree vertices by unity. A rooted full binary tree is said to be mean-self-similar if its Tokunaga coefficients are invariant with respect to pruning: $T_k:=T_{i,i+k}[K]$. We show that for self-similar trees, the condition $\limsup(T_k)^{1/k}<\infty$ is necessary and sufficient for the existence of the strong Horton law: $N_k[K]/N_1[K] \rightarrow R^{1-k}$, as $K \rightarrow \infty$ for some $R>0$ and every $k\geq 1$. This work is a step toward providing rigorous foundations for the Horton law that, being omnipresent in natural branching systems, has escaped so far a formal explanation.

cs.DM↗

arXiv2015

Horton self-similarity of Kingman's coalescent tree

The paper establishes a weak version of Horton self-similarity for a tree representation of Kingman's coalescent process. The proof is based on a Smoluchowski-type system of ordinary differential equations for the number of branches of a given Horton-Strahler order in a tree that represents Kingman's N-coalescent process with a constant kernel, in a hydrodynamic limit. We also demonstrate a close connection between the combinatorial Kingman's tree and the combinatorial level set tree of a white noise, which implies Horton self-similarity for the latter.

math.PR↗

arXiv2015

Unexpected links reflect the noise in networks

Gene covariation networks are commonly used to study biological processes. The inference of gene covariation networks from observational data can be challenging, especially considering the large number of players involved and the small number of biological replicates available for analysis. We propose a new statistical method for estimating the number of erroneous edges in reconstructed networks that strongly enhances commonly used inference approaches. This method is based on a special relationship between sign of correlation (positive/negative) and directionality (up/down) of gene regulation, and allows for the identification and removal of approximately half of all erroneous edges. Using the mathematical model of Bayesian networks and positive correlation inequalities we establish a mathematical foundation for our method. Analyzing existing biological datasets, we find a strong correlation between the results of our method and false discovery rate (FDR). Furthermore, simulation analysis demonstrates that our method provides a more accurate estimate of network error than FDR.

q-bio.MN↗

arXiv2015

Rapid Mixing of Glauber Dynamics of Gibbs Ensembles via Aggregate Path Coupling and Large Deviations Methods

In this paper, we present a novel extension to the classical path coupling method to statistical mechanical models which we refer to as aggregate path coupling. In conjunction with large deviations estimates, we use this aggregate path coupling method to prove rapid mixing of Glauber dynamics for a large class of statistical mechanical models, including models that exhibit discontinuous phase transitions which have traditionally been more difficult to analyze rigorously. The parameter region for rapid mixing for the generalized Curie-Weiss-Potts model is derived as a new application of the aggregate path coupling method.

math.PR↗