arXiv++ Combinatorics

In a stable matching problem there are two groups of agents, with agents on one side having their individual preferences for agents on another side as a potential match. It is assumed silently that agents can freely and costlessly ``switch" partners. A matching is called stable if no two unmatched agents prefer each other to their matches. Half a century ago, for equinumerous sides, Knuth demonstrated existence of preferences for which there are exponentially many stable matchings, and he posed a problem of evaluating an expected number of stable matchings when the preferences are uniformly random and independent. It was shown later by Pittel that this expectation is quite moderate, asymptotic to $e^{-1}n\log n$, $n$ being the number of agents on each side. The proof used Knuth's integral formula for the expectation based on a classic inclusion-exclusion counting. In later papers by Pittel this and other integral formulas were obtained by generating preference lists via a pair of random matrices, with independently random, $[0,1]$-uniform entries, rather than from a progressively problematic counting approach. The novelty of this paper is that we view the matrix entries as a basis for cardinal utilities that each agent ascribes to the agents on the other side. Relaxing the notion of stability, we declare matching stable if no unmatched pair of agents {\it strongly\/} prefer each other to their partners, with ``strength'' measured by a parameter $\eps>0$, $\eps=0$ corresponding to classic stability. We show that for $\eps$ of order $n^{-1}\log n$ the expected number of $\eps$-stable matchings is polynomially large, but for $\eps\gg n^{-1}\log n$ {\it however slightly\/} the expectation is suddenly super-polynomially large. This ``explosion'' phenomenon for a very small strength parameter $\eps$ continues to hold for imbalanced matchings, regardless of the imbalance size.

We study a mathematical model of voting contest with $m$ voters and $n$ candidates, with each voter ranking the candidates in order of preference, without ties. A Condorcet winner is a candidate who gets more than $m/2$ votes in pairwise contest with every other candidate. An ``impartial culture'' setting is the case when each voter chooses his/her candidate preference list uniformly at random from all $n!$ preferences, and does it independently of all other voters. For impartial culture case, Robert May and Lisa Sauermann showed that when $m=2k-1$ is fixed ($k=2$ and $k>2$ respectively), and $n$ grows indefinitely, the probability of a Condorcet winner is small, of order $n^{-(k-1)/k}$. We show if $m, n\to\infty$ and $m\gg n^4$, then for each fixed $\ell$ the probability of a Condercet winner is at most of order $n^{-\ell} + n^2/m^{1/2}$, thus converges to zero.

Consider $S$, a set of $n$ points chosen uniformly at random and independently from the unit hypercube of dimension $t>2$. Order $S$ by using the Cartesian product of the $t$ standard orders of $[0,1]$. We determine a constant $\bar x(t)<e$ such that, with probability $\ge 1-\exp(-Θ(\eps)n^{1/t})$, cardinality of a largest subset of comparable points is at most $(\bar x(t)+\eps)n^{1/t}$. The bound $\bar x(t)$ complements an explicit lower bound obtained by Bollobás and Winkler in 1982. Furthermore, we use Dilworth's theorem on partitions of a set into chains to prove that the cardinality of a largest antichain, i. e. a largest subset of incomparable points, is at least $(1-\eps) (n/e)^{1-1/t}$ with probability exponentially close to $1$.

Counting permutations of $[n]$ by the number of records, i.e. left-to-right maxima, is a classic problem in combinatorial enumeration. In the first volume of ``The Art of Computer Programming", Donald Knuth demonstrated its relevance for analysis of average case complexity of a basic algorithm for determining a maximum in a linear list of numbers. It is well known that the expected, and likely, number of those records in a {\it uniformly\/} random permutation is asymptotic to $\log n$. Cyril Banderier, Rene Beier, and Kurt Mehlhorn studied the case of a non-uniform random permutation, which is obtained from a generic permutation of $[n]$ by selecting its elements one after another independently with probability $p$, and permuting the selected elements uniformly at random. They proved that $E_n(p)$, the largest expected number of the maxima, is between $\text{const}\sqrt{n/p}$ and $O\bigl(\sqrt{(n/p)\log n}\bigr)$ if $p$ is fixed. For $p\gg 1/n$ and simultaneously $1-p\ge \text{const }n^{-1/2}\log n$, we prove that $E_n(p)$ is exactly of order $(1-p)\sqrt{n/p}$.

We find exact and asymptotic formulas for the number of pairs $(p,q)$ of $N$-cycles such that the all cycles of the product $p\cdot q$ have lengths from a given integer set. We then apply these results to prove a surprisingly high lower bound for the number of permutations whose block transposition distance from the identity is at least $(n+1)/2$.

Colloquially, there are two groups, $n$ men and $n$ women, each man (woman) ranking women (men) as potential marriage partners. A complete matching is called stable if no unmatched pair prefer each other to their partners in the matching. If some pairs are not admissible, then such a matching may not exist, but a properly defined partial stable matching exists always, and all such matchings involve the same, equi-numerous, groups of men and women. Earlier we proved that, for the complete, random, preference lists, with high probability (whp) the total number of complete stable matchings is, roughly, of order $n^{1/2}$, at least. Here we consider the case that the preference lists are still complete, but a generic pair (man,woman) is admissible with probability $p$, independently of all other $n^2-1$ pairs. It is shown that the expected number of complete stable matchings tends to $0$ if, roughly, $p<\tfrac{\log^2 n}{n}$ and to infinity if $p>\tfrac{\log^2 n}{n}$. We show that whp: (a) there exists a complete stable matching if $p>(9/4)\tfrac{\log^2 n}{n}$, (b) the number of unmatched men and women is bounded if $p> \tfrac{\log^2n}{n}$, and (c) this number grows as a fractional power of $n$ for $p<\tfrac{\log^2 n}{n}$.

We call a pair of vertex-disjoint, induced subtrees of a rooted trees twins if they have the same counts of vertices by out-degrees. The likely maximum size of twins in a uniformly random, rooted Cayley tree of size $n\to\infty$ is studied. It is shown that the expected number of twins of size $(2+δ)\sqrt{\log n\cdot\log\log n}$ approaches zero, while the expected number of twins of size $(2-δ)\sqrt{\log n\cdot\log\log n}$ approaches infinity.

Let $X_1,\dots, X_n$ be independent integers distributed uniformly on $\{1,\dots, M\}$, $M=M(n)\to\infty$ however slow. A partition $S$ of $[n]$ into $ν$ non-empty subsets $S_1,\dots, S_ν$ is called perfect, if all $ν$ values $\sum_{j\in S_{\a}}X_j$ are equal. For a perfect partition to exist, $\sum_j X_j$ has to be divisible by $ν$. For $ν=2$, Borgs et al. proved, among other results, that, conditioned on $\sum_j X_j$ being even, with high probability a perfect partition exists if $κ:=\lim \tfrac{n}{\log M}>\tfrac{1}{\log 2}$, and that w.h.p. no perfect partition exists if $κ<\tfrac{1}{\log 2}$. We prove that w.h.p. no perfect partition exists if $ν\ge 3$ and $κ<\tfrac{2}{\log ν}$. We identify the range of $κ$ in which the expected number of perfect partitions is exponentially high. We show that for $κ> \tfrac{2(ν-1)}{\log[(1-2ν^{-2})^{-1}]}$ the total number of perfect partitions is exponentially high with probability $\gtrsim (1+ν^2)^{-1}$.

We prove that for any fixed $k$, the probability that a random vertex of a random increasing plane tree is of rank $k$, that is, the probability that a random vertex is at distance $k$ from the leaves, converges to a constant $c_k$ as the size $n$ of the tree goes to infinity. {\color{blue} We prove that $1-\sum_{j\le k} c_k<\tfrac{3^{k+1}}{(2k+1)!}$, so that the tail of the limiting rank distribution is super-exponentially narrow. We prove that the latter property holds uniformly for all finite $n$ as well.} More generally, we prove that the ranks of a finite uniformly random set of vertices are asymptotically independent, each with distribution $\{c_k\}$. We compute the exact value of $c_k$ for $0\leq k\leq 3$, demonstrating that the limiting expected fraction of vertices with rank $\le 3$ is $0.9997\dots$. We show that with probability $1-n^{-0.99\eps}$ the highest rank of a vertex in the tree is sandwiched between $(1-\eps)\log n /\log\log n$ and $(1.5+\eps)\log n/\log\log n$, {\color{blue} and that this rank is asymptotic to $\log n/\log\log n$ with probability $1-o(1)$.}

Consider a rooted tree $T$ with leaf-set $[n]$, and with all non-leaf vertices having out-degree $2$, at least. A rooted tree $\mathcal T$ with leaf-set $S\subset [n]$ is induced by $S$ in $T$ if $\mathcal T$ is the lowest common ancestor subtree for $S$, with all its degree-2 vertices suppressed. A "maximum agreement subtree" (MAST) for a pair of two trees $T'$ and $T"$ is a tree $\mathcal T$ with a largest leaf-set $S\subset [n]$ such that $\mathcal T$ is induced by $S$ both in $T'$ and $T"$. Bryant et al. \cite{BryMcKSte} and Bernstein et al. \cite{Ber} proved, among other results, that for $T'$ and $T"$ being two independent copies of a random binary (uniform or Yule-Harding distributed) tree $T$, the likely magnitude order of $\text{MAST}(T',T")$ is $O(n^{1/2})$. We prove this bound for a wide class of random rooted trees : $T$ is a terminal tree of a branching, Galton--Watson, process with an ordered-offspring distribution of mean $1$, conditioned on "total number of leaves is $n$".

For a two-sided ($n$ men/$n$ women) stable matching problem) Gale and Shapley studied a proposal algorithm (men propose/women select, or the other way around), that determines a matching, not blocked by any unmatched pair. Irving used this algorithm as a first phase of his algorithm for one-sided (stable roommates) matching problem with $n$ agents. We analyze a fully extended version of Irving's proposal algorithm that runs all the way until either each agent holds a proposal or an agent gets rejected by everybody on the agent's preference list. It is shown that the terminal, directed, partnerships form a stable permutation with matched pairs remaining matched in any other stable permutation. A likely behavior of the proposal algorithm is studied under assumption that all $n$ rankings are independently uniform. It is proved that with high probability (w.h.p.) every agent has a partner, and that both the number of agents in cycles of length $\ge 3$ and the total number of stable matchings are bounded in probability. W.h.p. the total number of proposals is asymptotic to $0.5 n^{3/2}$.

Consider a cyclically ordered collection of $r$ equinumerous agent sets with strict preferences of every agent over the agents from the next agent set. A weakly stable cyclic matching is a partition of the set of agents into disjoint union of $r$-long cycles, one agent from each set per cycle, such that there are no destabilizing $r$-long cycles, i.e. cycles in which every agent strictly prefers its successor to its successor in the matching. Assuming that the preferences are uniformly random and independent, we show that the expected number of stable matchings grows with $n$ (cardinality of each agent set) as $(n\log n)^{r-1}$. We also consider a bipartite stable matching problem where preference list of each agent forms a partially ordered set. Each partial order is an intersection of several, $k_i$ for side $i$, independent, uniformly random, strict orders. For $k_1+k_2>2$, the expected number of stable matchings is analyzed for three, progressively stronger, notions of stability. The expected number of weakly stable matchings is shown to grow super-exponentially fast. In contrast, for $\min(k_1,k_2)>1$, the fraction of instances with at least one strongly stable (super-stable) matching is super-exponentially small.

We study two models of an age-biased graph process: the $δ$-version of the preferential attachment graph model (PAM) and the uniform attachment graph model (UAM), with $m$ attachments for each of incoming vertices. We show that almost surely the scaled size of a breadth-first (descendant) tree rooted at a fixed vertex converges, for $m=1$, to a limit whose distribution is a mixture of two beta-distributions and a single beta-distribution respectively, and that for $m>1$ the limit is $1$. We also analyze the likely performance of two greedy (online) algorithms, for a large matching set and a large independent set, and determine--for each model and each greedy algorithm--both a limiting fraction of vertices involved and an almost sure convergence rate.

Existence of a perfect matching in a random bipartite digraph with bipartition $(V_1, V_2)$, $|V_i|=n$, is studied. The graph is generated in two rounds of random selections of a potential matching partner such that the average number of selections made by each vertex overall is below $2$. More precisely, in the first round each vertex chooses a potential mate uniformly at random, and independently of all vertices. Given a fixed integer $m$, a vertex is classified as unpopular if it has been chosen by at most $m$ vertices from the other side. Each unpopular vertex makes yet another uniform/independent selection of a potential mate. The expected number of selections made by a generic vertex $v$, i.e. its out-degree, is asymptotic to $1+\Bbb P(\text{Poisson}(1)\le m)\in (1,2)$. Aided by Matlab software, we prove that for $m=1$, whence for all $m\ge 1$, the resulting bipartite graph has a perfect matching a.a.s. (asymptotically almost surely). On the other hand, for $m=0$ a.a.s. a perfect matching does not exist, and the graph consists of a single giant component of size $2n -O(n^{1/2+o(1)})$ and possibly some components of size $O(\log n)$. This is a thorough revision of the joint paper (JCT(B) 88 (2003), 1-16) by the first author and the third author.

Bollobás-Riordan random pairing model of a preferential attachment graph $G_m^n$ is studied. Let $\{W_j\}_{j\le mn+1}$ be the process of sums of independent exponentials with mean $1$. We prove that the degrees of the first $ν_m^n:=n^{\frac{m}{m+2}-ε}$ vertices are jointly, and uniformly, asymptotic to $\{2(mn)^{1/2}\bigl(W^{1/2}_{mj}-W^{1/2}_{m(j-1)}\bigr)\}_{j\in [t]}$, and that with high probability (whp) the smallest of these degrees is $n^{\frac{ε(m+2)}{2m}}$, at least. In contrast, the degrees of vertices below the top by any fraction of $n$ are whp of $O(\log n)$ order. Next we bound the probability that there exists a pair of large vertex sets with no edges joining them, and apply the bound to several special cases. We propose to measure an influence of a vertex $v$ by the size of a maximal recursive tree (max-tree) rooted at $v$. The set of the first $ν_m^n$ vertices is shown whp not to contain a max-tree of any size. Whp the largest recursive tree has size of order $n$. We prove that, for $m>1$, $\Bbb P(G_m^n\text{ is connected})\ge 1- O\bigl((\log n)^{-(m-1)/3+o(1)}\bigr)$. We show that the distribution of the scaled size of a generic max-tree in $G_1^n$ converges to a mixture of two beta distributions.

A uniform attachment graph (with parameter $k$), denoted $G_{n,k}$ in the paper, is a random graph on the vertex set $[n]$, where each vertex $v$ makes $k$ selections from $[v-1]$ uniformly and independently, and these selections determine the edge set. We study several aspects of this graph. Our motivation comes from two similarly constructed, well-studied random graphs: $k$-out graphs and preferential attachment graphs. In this paper, we find the asymptotic distribution of its minimum degree and connectivity, and study the expansion properties of $G_{n,k}$ to show that the conductance of $G_{n,k}$ is of order $(\log n)^{-1}$. We also study the bootstrap percolation on $G_{n,k}$, where, each vertex is either initially infected with probability $p$, independently of others, or gets infected later as a result of having $r$ infected neighbors at some point. We show that, for $2\le r\le k-1$, if $p\ll (\log n)^{-r/(r-1)}$, then, with probability approaching 1, the process ends before all vertices get infected. On the other hand, if $p\ge ω(\log n)^{-r/(r-1)}$, where $ω$ is a certain very slowly growing function, then all the vertices get infected with probability approaching 1.

Consider the group of $n$ men and $n$ women, each with their own preference list for a potential marriage partner. The stable marriage is a bipartite matching such that no unmatched pair (man, woman) prefer each other to their partners in the matching. Its non-bipartite version, with an even number $n$ of members, is known as the stable roommates problem. Jose Alcalde introduced an alternative notion of exchange-stable, one-sided, matching: no two members prefer each other's partners to their own partners in the matching. Katarina Cechlárová and David Manlove showed that the e-stable matching decision problem is $NP$-complete for both types of matchings. We prove that the expected number of e-stable matchings is asymptotic to $\left(\frac{πn}{2}\right)^{1/2}$ for two-sided case, and to $e^{1/2}$ for one-sided case. However, the standard deviation of this number exceeds $1.13^n$, ($1.06^n$ resp.). As an obvious byproduct, there exist instances of preference lists with at least $1.13^n$ ($1.06^n$ resp.) e-stable matchings. The probability that there is no matching which is stable and e-stable is at least $1-e^{-n^{1/6+o(1)}}$, ($1-O(2^{-n/2})$ resp.).

The stable roommates problem does not necessarily have a solution, i.e. a stable matching. We had found that, for the uniformly random instance, the expected number of solutions converges to $e^{1/2}$ as $n$, the number of members, grows, and with Rob Irving we proved that the limiting probability of solvability is $e^{1/2}/2$, at most. Stephan Mertens's extensive numerics compelled him to conjecture that this probability is of order $n^{-1/4}$. Jimmy Tan introduced a notion of a stable cyclic partition, and proved existence of such a partition for every system of members' preferences, discovering that presence of odd cycles in a stable partition is equivalent to absence of a stable matching. In this paper we show that the expected number of stable partitions with odd cycles grows as $n^{1/4}$. However the standard deviation of that number is of order $n^{3/8}\gg n^{1/4}$, too large to conclude that the odd cycles exist with high probability (whp). Still, as a byproduct, we show that whp the fraction of members with more than one stable "predecessor" is of order $n^{-1/4}$. Furthermore, whp the average rank of a predecessor in every stable partition is of order $n^{1/2}$. The likely size of the largest stable matching is $n/2-O(n^{1/4+o(1)})$, and the likely number of pairs of unmatched members blocking the optimal complete matching is $O(n^{3/4+o(1)})$.

Following up a recent work by Ashlagi, Kanoria and Leshno, we study a stable matching problem with unequal numbers of men and women, and independent uniform preferences. The asymptotic formulas for the expected number of stable matchings, and for the probabilities of one point--concentration for the range of husbands' total ranks and for the range of wives' total ranks are obtained.

Consider the set of all digraphs on $[N]$ with $M$ edges, whose minimum in-degree and minimum out-degree are at least $k_1$ and $k_2$ respectively. For $k:=\min\{k_1,k_2\}\ge 2$ and $M/N>\max\{k_1,k_2\}$, $M=Θ(N)$, we show that, among those digraphs, the fraction of $k$-strongly connected digraphs is $1-O\bigl(N^{-(k-1)})$. Earlier with Dan Poole we identified a sharp edge-density threshold $c^*(k_1,k_2)$ for birth of a giant $(k_1,k_2)$-core in the random digraph $D(n,m=[cn])$. Combining the claims, for $c>c^*(k_1,k_2)$ with probability $1-O\bigl(N^{-(k-1)})$ the giant $(k_1,k_2)$-core exists and is $k$-strongly connected.