Separable discrete functions: Recognition and sufficient conditions

A discrete function of $n$ variables is a mapping $g:X_1\times \dots \times X_n\to A$, where $X_1,\dots ,X_n$, and $A$ are arbitrary finite sets. Function $g$ is called separable if there exist $n$ functions $g_i:X_i\to A$ for $i=1,\dots ,n$, such that for every input $x_1,\dots ,x_n$ the function $g(x_1,\dots ,x_n)$ takes one of the values $g_1\left(x_1\right),\dots ,g_n\left(x_n\right)$. Given a discrete function $g$, it is an interesting problem to ask whether $g$ is separable or not. Although this seems to be a very basic problem concerning discrete functions, the complexity of recognition of separable discrete functions of $n$ variables is known only for $n=2$. In this paper we will show that a slightly more general recognition problem, when $g$ is not fully but only partially defined, is NP-complete for $n\ge 3$. We will then use this result to show that the recognition of fully defined separable discrete functions is NP-complete for $n\ge 4$.The general recognition problem contains the above mentioned special case for $n=2$. This case is well-studied in the context of game theory, where (separable) discrete functions of $n$ variables are referred to as (assignable) $n$-person game forms. There is a known sufficient condition for assignability (separability) of two-person game forms (discrete functions of two variables) called (weak) total tightness of a game form. This property can be tested in polynomial time, and can be easily generalized both to higher dimension and to partially defined functions. We will prove in this paper that weak total tightness implies separability for (partially defined) discrete functions of $n$ variables for any $n$, thus generalizing the above result known for $n=2$. Our proof is constructive. Using a graph-based discrete algorithm we show how for a given weakly totally tight (partially defined) discrete function $g$ of $n$ variables one can construct separating functions $g_1,\dots ,g_n$ in polynomial time with respect to the size of the input function.     

Biased graphs. VI. synthetic geometry

A biased graph is a graph with a class of selected circles (“cycles”, “circuits”), called balanced, such that no theta subgraph contains exactly two balanced circles. A biased graph $\Omega$ has two natural matroids, the frame matroid $G\left(\Omega \right)$ and the lift matroid $L\left(\Omega \right)$, and their extensions the full frame matroid $G^{•}\left(\Omega \right)$ and the extended (or complete) lift matroid $L_0\left(\Omega \right)$. In Part IV we used algebra to study the representations of these matroids by vectors over a skew field and the corresponding embeddings in Desarguesian projective spaces. Here we redevelop those representations, independently of Part IV and in greater generality, by using synthetic geometry.     

Relation between the H-rank of a mixed graph and the rank of its underlying graph

Given a simple graph $G=(V_G,E_G)$ with vertex set $V_G$ and edge set $E_G$, the mixed graph $\stackrel{?}{G}$ is obtained from $G$ by orienting some of its edges. Let $H\left(\stackrel{?}{G}\right)$ denote the Hermitian adjacency matrix of $\stackrel{?}{G}$ and $A\left(G\right)$ be the adjacency matrix of $G$. The $H$-rank (resp. rank) of $\stackrel{?}{G}$ (resp. $G$), written as $rk\left(\stackrel{?}{G}\right)$ (resp. $r\left(G\right)$), is the rank of $H\left(\stackrel{?}{G}\right)$ (resp. $A\left(G\right)$). Denote by $d\left(G\right)$ the dimension of cycle space of $G$, that is $d\left(G\right)=\left|E_G\right|?\left|V_G\right|+\omega \left(G\right)$, where $\omega \left(G\right)$ denotes the number of connected components of $G$. In this paper, we concentrate on the relation between the $H$-rank of $\stackrel{?}{G}$ and the rank of $G$. We first show that $?2d\left(G\right)?rk\left(\stackrel{?}{G}\right)?r\left(G\right)?2d\left(G\right)$ for every mixed graph $\stackrel{?}{G}$. Then we characterize all the mixed graphs that attain the above lower (resp. upper) bound. By these obtained results in the current paper, all the main results obtained in Luo et al. (2018); Wong et al. (2016) may be deduced consequently.     

Fluctuation theory for level-dependent Lévy risk processes

A level-dependent Lévy process solves the stochastic differential equation $dU\left(t\right)=dX\left(t\right)??\left(U\left(t\right)\right)dt$, where $X$ is a spectrally negative Lévy process. A special case is a multi-refracted Lévy process with ${?}_k\left(x\right)={\sum }_{j=1}^k{\delta }_j{\mathbf{1}}_{\{x\ge b_j\}}$. A general rate function $?$ that is non-decreasing and locally Lipschitz continuous is also considered. We discuss solutions of the above stochastic differential equation and investigate the so-called scale functions, which are counterparts of the scale functions from the theory of Lévy processes. We show how fluctuation identities for $U$ can be expressed via these scale functions. We demonstrate that the derivatives of the scale functions are solutions of Volterra integral equations.     

Minimum supports of functions on the Hamming graphs with spectral constraints

We study functions defined on the vertices of the Hamming graphs $H(n,q)$. The adjacency matrix of $H(n,q)$ has $n+1$ distinct eigenvalues $n(q-1)-q\cdot i$ with corresponding eigenspaces $U_i(n,q)$ for $0\le i\le n$. In this work, we consider the problem of finding the minimum possible support (the number of nonzeros) of functions belonging to a direct sum $U_i(n,q)\oplus U_{i+1}(n,q)\oplus \cdots \oplus U_j(n,q)$ for $0\le i\le j\le n$. For the case $i+j\le n$ and $q\ge 3$ we find the minimum cardinality of the support of such functions and obtain a characterization of functions with the minimum cardinality of the support. In the case $i+j>n$ and $q\ge 4$ we also find the minimum cardinality of the support of functions, and obtain a characterization of functions with the minimum cardinality of the support for $i=j$, $i>\frac{n}{2}$ and $q\ge 5$. In particular, we characterize eigenfunctions from the eigenspace $U_i(n,q)$ with the minimum cardinality of the support for cases $i\le \frac{n}{2}$, $q\ge 3$ and $i>\frac{n}{2}$, $q\ge 5$.     

Efficient counting of degree sequences

We present formulas to count the set $D_0\left(n\right)$ of degree sequences of simple graphs of order $n$, the set $D\left(n\right)$ of degree sequences of those graphs with no isolated vertices, and the set $D_{k\text{\_}con}\left(n\right)$ of degree sequences of those graphs that are $k$-connected for fixed $k$. The formulas all use a function of Barnes and Savage and the associated dynamic programming algorithms all run in time polynomial in $n$ and are asymptotically faster than previous known algorithms for these problems. We also show that $\left|D\left(n\right)\right|$ and $\left|D_{1\text{\_}con}\left(n\right)\right|$ are asymptotically equivalent but $\left|D\left(n\right)\right|$ and $\left|D_0\left(n\right)\right|$ as well as $\left|D\left(n\right)\right|$ and $\left|D_{k\text{\_}con}\left(n\right)\right|$ with fixed $k\ge 2$ are asymptotically inequivalent. Finally, we explain why we are unable to obtain the absolute asymptotic orders of these functions from their formulas.     

Nonseparating trees in 2-connected graphs and oriented trees in strongly connected digraphs

Mader (2010) conjectured that for every positive integer $k$ and every finite tree $T$ with order $m$, every $k$-connected, finite graph $G$ with $\delta \left(G\right)\ge \lfloor \frac{3}{2}k\rfloor +m-1$ contains a subtree $T^{\prime }$ isomorphic to $T$ such that $G-V\left(T^{\prime }\right)$ is $k$-connected. The conjecture has been verified for paths, trees when $k=1$, and stars or double-stars when $k=2$. In this paper we verify the conjecture for two classes of trees when $k=2$.For digraphs, Mader (2012) conjectured that every $k$-connected digraph $D$ with minimum semi-degree $\delta \left(D\right)=min\{{\delta }^{+}\left(D\right),{\delta }^{-}\left(D\right)\}\ge 2k+m-1$ for a positive integer $m$ has a dipath $P$ of order $m$ with $\kappa (D-V\left(P\right))\ge k$. The conjecture has only been verified for the dipath with $m=1$, and the dipath with $m=2$ and $k=1$. In this paper, we prove that every strongly connected digraph with minimum semi-degree $\delta \left(D\right)=min\{{\delta }^{+}\left(D\right),{\delta }^{-}\left(D\right)\}\ge m+1$ contains an oriented tree $T$ isomorphic to some given oriented stars or double-stars with order $m$ such that $D-V\left(T\right)$ is still strongly connected.     

Connected k-dominating graphs

For a graph $G=(V,E)$, the $k$-dominating graph $D_k\left(G\right)$ of $G$ has vertices corresponding to the dominating sets of $G$ having cardinality at most $k$, where two vertices of $D_k\left(G\right)$ are adjacent if and only if the dominating set corresponding to one of the vertices can be obtained from the dominating set corresponding to the second vertex by the addition or deletion of a single vertex. We denote the domination and upper domination numbers of $G$ by $\gamma \left(G\right)$ and $\Gamma \left(G\right)$, respectively, and the smallest integer $\epsilon$ for which $D_k\left(G\right)$ is connected for all $k\ge \epsilon$ by $d_0\left(G\right)$. It is known that $\Gamma \left(G\right)+1\le d_0\left(G\right)\le \left|V\right|$, but constructing a graph $G$ such that $d_0\left(G\right)>\Gamma \left(G\right)+1$ appears to be difficult.We present two related constructions. The first construction shows that for each integer $k\ge 3$ and each integer $r$ such that $1\le r\le k?1$, there exists a graph $G_{k,r}$ such that $\Gamma \left(G_{k,r}\right)=k$, $\gamma \left(G_{k,r}\right)=r+1$ and $d_0\left(G_{k,r}\right)=k+r=\Gamma \left(G\right)+\gamma \left(G\right)?1$. The second construction shows that for each integer $k\ge 3$ and each integer $r$ such that $1\le r\le k?1$, there exists a graph $Q_{k,r}$ such that $\Gamma \left(Q_{k,r}\right)=k$, $\gamma \left(Q_{k,r}\right)=r$ and $d_0\left(Q_{k,r}\right)=k+r=\Gamma \left(G\right)+\gamma \left(G\right)$.     

Constructing sparse Davenport–Schinzel sequences

For any sequence $u$, the extremal function $Ex(u,j,n)$ is the maximum possible length of a $j$-sparse sequence with $n$ distinct letters that avoids $u$. We prove that if $u$ is an alternating sequence $abab\dots$ of length $s$, then $Ex(u,j,n)=\Theta \left(s{n}^2\right)$ for all $j\ge 2$ and $s\ge n$, answering a question of Wellman and Pettie (2018) and extending the result of Roselle and Stanton that $Ex(u,2,n)=\Theta \left(s{n}^2\right)$ for any alternation $u$ of length $s\ge n$ (Roselle and Stanton, 1971).Wellman and Pettie also asked how large must $s\left(n\right)$ be for there to exist $n$-block $DS(n,s\left(n\right))$ sequences of length $\Omega \left(n^{2-o\left(1\right)}\right)$. We answer this question by showing that the maximum possible length of an $n$-block $DS(n,s\left(n\right))$ sequence is $\Omega \left(n^{2-o\left(1\right)}\right)$ if and only if $s\left(n\right)=\Omega \left(n^{1-o\left(1\right)}\right)$. We also show related results for extremal functions of forbidden 0–1 matrices with any constant number of rows and extremal functions of forbidden sequences with any constant number of distinct letters.     

Colouring of generalized signed triangle-free planar graphs

We view an undirected graph $G$ as a symmetric digraph, where each edge $xy$ is replaced by two opposite arcs $e=(x,y)$ and $e^{?1}=(y,x)$. Assume $S$ is an inverse closed subset of permutations of positive integers. We say $G$ is $S$-$k$-colourable if for any mapping $\sigma :E\left(G\right)\to S$ with $\sigma (x,y)={\left(\sigma (y,x)\right)}^{?1}$, there is a mapping $f:V\left(G\right)\to \left[k\right]=\{1,2,\dots ,k\}$ such that ${\sigma }_e\left(f\left(x\right)\right)\ne f\left(y\right)$ for each arc $e=(x,y)$. The concept of $S$-$k$-colourable is a common generalization of several other colouring concepts. This paper is focused on finding the sets $S$ such that every triangle-free planar graph is $S$-3-colourable. Such a set $S$ is called TFP-good. Grötzsch’s theorem is equivalent to say that $S=\left\{id\right\}$ is TFP-good. We prove that for any inverse closed subset $S$ of $S_3$ which is not isomorphic to $\{id,\left(12\right)\}$, $S$ is TFP-good if and only if either $S=\left\{id\right\}$ or there exists $a\in \left[3\right]$ such that for each $\pi \in S$, $\pi \left(a\right)\ne a$. It remains an open question to determine whether or not $S=\{id,\left(12\right)\}$ is TFP-good.     

Relating broadcast independence and independence

An independent broadcast on a connected graph $G$ is a function $f:V\left(G\right)\to {\mathbb{N}}_0$ such that, for every vertex $x$ of $G$, the value $f\left(x\right)$ is at most the eccentricity of $x$ in $G$, and $f\left(x\right)>0$ implies that $f\left(y\right)=0$ for every vertex $y$ of $G$ within distance at most $f\left(x\right)$ from $x$. The broadcast independence number ${\alpha }_b\left(G\right)$ of $G$ is the largest weight $\sum _{x\in V\left(G\right)}f\left(x\right)$ of an independent broadcast $f$ on $G$. Clearly, ${\alpha }_b\left(G\right)$ is at least the independence number $\alpha \left(G\right)$ for every connected graph $G$. Our main result implies ${\alpha }_b\left(G\right)\le 4\alpha \left(G\right)$. We prove a tight inequality and characterize all extremal graphs.     

On a problem of Specker about Euclidean representations of finite graphs

Say that a graph $G$ is representable in ${\mathbb{R}}^n$ if there is a map $f$ from its vertex set into the Euclidean space ${\mathbb{R}}^n$ such that $6f\left(x\right)?f\left(x^{\prime }\right)6=6f\left(y\right)?f\left(y^{\prime }\right)6$ iff $\{x,x^{\prime }\}$ and$\{y,y^{\prime }\}$ are both edges or both non-edges in $G$. The purpose of this note is to present the proof of the following result, due to Einhorn and Schoenberg in Einhorn and Schoenberg (1966): if $G$ finite is neither complete nor independent, then it is representable in ${\mathbb{R}}^{\left|G\right|?2}$. A similar result also holds in the case of finite complete edge-colored graphs.     

Group-invariant solutions for the Ricci curvature equation and the Einstein equation

We consider the pseudo-Euclidean space $({\mathbb{R}}^n,g)$, $n\ge 3$, with coordinates $x=(x_1,\dots ,x_n)$ and metric $g_{ij}={\delta }_{ij}{?}_i$, ${?}_i=\pm 1$, where at least one ${?}_i$ is positive, and also tensors of the form $A={\sum }_{i,j}{A}_{ij}d{x}_{i}d{x}_j$, such that $A_{ij}$ are differentiable functions of x. For such tensors, we use Lie point symmetries to find metrics $\stackrel{￣}{g}=\frac{1}{u^2}g$ that solve the Ricci curvature and the Einstein equations. We provide a large class of group-invariant solutions and examples of complete metrics $\stackrel{￣}{g}$ defined globally in ${\mathbb{R}}^n$. As consequences, for certain functions $\stackrel{￣}{K}$, we show complete metrics $\stackrel{￣}{g}$, conformal to the pseudo-Euclidean metric g, whose scalar curvature is $\stackrel{￣}{K}$.     

Graph homomorphisms via vector colorings

In this paper we study the existence of homomorphisms $G\to H$ using semidefinite programming. Specifically, we use the vector chromatic number of a graph, defined as the smallest real number $t\ge 2$ for which there exists an assignment of unit vectors $i\mapsto p_i$ to its vertices such that $〈p_i,p_j〉\le -1∕(t-1),$ when $i\sim j$. Our approach allows to reprove, without using the Erdős–Ko–Rado Theorem, that for $n>2r$ the Kneser graph $K_{n:r}$ and the $q$-Kneser graph $q{K}_{n:r}$ are cores, and furthermore, that for $n∕r=n^{\prime }∕r^{\prime }$ there exists a homomorphism $K_{n:r}\to K_{n^{\prime }:r^{\prime }}$ if and only if $n$ divides $n^{\prime }$. In terms of new applications, we show that the even-weight component of the distance $k$-graph of the $n$-cube $H_{n,k}$ is a core and also, that non-bipartite Taylor graphs are cores. Additionally, we give a necessary and sufficient condition for the existence of homomorphisms $H_{n,k}\to H_{n^{\prime },k^{\prime }}$ when $n∕k=n^{\prime }∕k^{\prime }$. Lastly, we show that if a 2-walk-regular graph (which is non-bipartite and not complete multipartite) has a unique optimal vector coloring, it is a core. Based on this sufficient condition we conducted a computational study on Ted Spence’s list of strongly regular graphs (http://www.maths.gla.ac.uk/ es/srgraphs.php) and found that at least 84% are cores.     

A C5-decomposition of the λ-fold line graph of the complete graph

In 1996, Cox and Rodger [Cycle systems of the line graph of the complete graph, J. Graph Theory 21 (1996) 173–182] raised the following question: For what values of $m$ and $n$ does there exist an $m$-cycle decomposition of $L\left(K_n\right)?$ In this paper, the above question is answered for $m=5.$ In fact, it is shown that $L\left(K_n\right)\left(\lambda \right),$ the $\lambda$-fold line graph of the complete graph $K_n,$ has a $C_5$-decomposition if and only if $5\mid \lambda \left(\genfrac{}{}{0ex}{}{n}{2}\right)(n?2)$ and $n\ge 4.$