SearcharxivSearch

arXiv subjects

Zan-Bo Zhang

Publications and source records attributed to Zan-Bo Zhang.

10 recordsLinked to original sources

MERBIT: A GPU-Based SpMV Method for Iterative Workloads

Sparse Matrix-Vector Multiplication (SpMV) is the cornerstone in many iterative workloads, including large-scale graph analytics and sparse iterative solvers. Accelerating SpMV on real-world graphs remains challenging due to highly irregular sparsity patterns. In this paper, we propose MERBIT, a GPU SpMV method designed for repeated SpMV on irregular, graph-like sparse matrices, with PageRank as a representative motivating workload. MERBIT combines two key ideas from existing GPU SpMV methods. At the global level, it uses merge-path partitioning to balance work over nonzeros and row boundaries. At the local level, it encodes each merge-path segment using a compact bit-field descriptor. MERBIT improves workload balance and promotes coalesced memory access for both matrix loading and output writes; moreover, three optimization strategies are incorporated to further enhance performance. Experiments on 50 large irregular datasets demonstrate that MERBIT outperforms competitive baselines, including cuSPARSE, Ginkgo, and academic approaches, achieving average speedups of 1.27 and 1.25 over cuSPARSE COO in single and double precision, respectively.

cs.DC

Construction, Extension and Paths of Near-Homogeneous Tournaments

A homogeneous tournament is a tournament with $4t+3$ vertices such that every arc is contained in exactly $t+1$ cycles of length $3$. Homogeneous tournaments are the first class of tournaments that are proved to be path extendable, which means that every nonhamiltonian path $P$ in such a tournament $T$ can be extended to a path $P'$ with the same initial and terminal vertex and $V(P')=V(P)\cup \{u\}$ for a certain vertex $u\in V(T)\backslash V(P)$. In order to find more path extendable tournaments we study the generalization of homogeneous tournaments called near-homogeneous tournaments, in which every arc is contained in $t$ or $t+1$ cycles of length $3$. Near-homogeneity has been defined in tournaments with $4t+1$ vertices. In this paper, we raise a new method to construct near-homogeneous tournaments with $4t+1$ vertices. We then show that the definition of near-homogeneous tournament can be extended to tournaments with an even number of vertices. Finally we verify path extendability of near-homogeneous tournaments, thus expand the class of path extendable tournaments.

math.CO

Cycles of lengths 3 and n-1 in digraphs under a Bang-Jensen-Gutin-Li type conditon

Bang-Jensen-Gutin-Li type conditions are the conditions for hamiltonicity of digraphs which impose degree restrictions on nonadjacent vertices which have a common in-neighbor or a common out-neighbor. They can be viewed as an extension of Fan type conditions in undirected graphs, as well as generalization of locally (in-, out-)semicomplete digraphs. Since their first appearance in 1996, various Bang-Jensen-Gutin-Li type conditions for hamitonicity have come forth. In this paper we establish a condition of Bang-Jensen-Gutin-Li type which implies not only a hamiltonian cycle but also a 3-cycle and an (n-1)-cycle, with well-characterized exceptional graphs. We conjecture that this condition implies the existence of cycle of every length.

math.CO

Path Extendable Tournaments

A digraph $D$ is called \emph{path extendable} if for every nonhamiltonian (directed) path $P$ in $D$, there exists another path $P^\prime$ with the same initial and terminal vertices as $P$, and $V(P^\prime) = V (P)\cup \{w\}$ for a vertex $w \in V(D)\setminus V(P)$. Hence, path extendability implies paths of continuous lengths between every vertex pair. In earlier works of C. Thomassen and K. Zhang, it was shown that the condition of small $i(T)$ or positive $π_2(T)$ implies paths of continuous lengths between every vertex pair in a tournament $T$, where $i(T)$ is the irregularity of $T$ and $π_2(T)$ denotes for the minimum number of paths of length $2$ from $u$ to $v$ among all vertex pairs $\{u,v\}$. Motivated by these results, we study sufficient conditions in terms of $i(T)$ and $π_2(T)$ that guarantee a tournament $T$ is path extendable. We prove that (1) a tournament $T$ is path extendable if $i(T)< 2π_2(T)-(|T|+8)/6$, and (2) a tournament $T$ is path extendable if $π_2(T) > (7|T|-10)/36$. As an application, we deduce that almost all random tournaments are path extendable.

math.CO

Component factors in $K_{1,r}$-free graphs

A graph is said to be $K_{1,r}$-free if it does not contain an induced subgraph isomorphic to $K_{1,r}$. An $\mathcal{F}$-factor is a spanning subgraph $H$ such that each connected component of $H$ is isomorphic to some graph in $\mathcal{F}$. In particular, $H$ is called an $\{P_2,P_3\}$-factor of $G$ if $\mathcal{F}=\{P_2,P_3\}$; $H$ is called an $\mathcal{S}_n$-factor of $G$ if $\mathcal{F}=\{K_{1,1},K_{1,2},K_{1,3},...,K_{1,n}\}$, where $n\geq2$. A spanning subgraph of a graph $G$ is called a $\mathcal{P}_{\geq k}$-factor of $G$ if its each component is isomorphic to a path of order at least $k$, where $k\geq2$. A graph $G$ is called a $\mathcal{F}$-factor covered graph if there is a $\mathcal{F}$-factor of $G$ including $e$ for any $e\in E(G)$. In this paper, we give a minimum degree condition for a $K_{1,r}$-free graph to have an $\mathcal{S}_n$-factor and a $\mathcal{P}_{\geq 3}$-factor, respectively. Further, we obtain sufficient conditions for $K_{1,r}$-free graphs to be $\mathcal{P}_{\geq 2}$-factor, $\mathcal{P}_{\geq 3}$-factor or $\{P_2,P_3\}$-factor covered graphs. In addition, examples show that our results are sharp.

math.CO

Directed Hamilton cycles in digraphs and matching alternating Hamilton cycles in bipartite graphs

In 1972, Woodall raised the following Ore type condition for directed Hamilton cycles in digraphs: Let $D$ be a digraph. If for every vertex pair $u$ and $v$, where there is no arc from $u$ to $v$, we have $d^+u)+d^-(v)\geq |D|$, then $D$ has a directed Hamilton cycle. By a correspondence between bipartite graphs and digraphs, the above result is equivalent to the following result of Las Vergnas: Let $G = (B,W)$ be a balanced bipartite graph. If for any $b \in B$ and $w \in W$, where $b$ and $w$ are nonadjacent, we have $d(w)+d(b) \geq |G|/2 + 1$, then every perfect matching of $G$ is contained in a Hamilton cycle. The lower bounds in both results are tight. In this paper, we reduce both bounds by $1$, and prove that the conclusions still hold, with only a few exceptional cases that can be clearly characterized.

math.CO

M-alternating Hamilton paths and M-alternating Hamilton cycles

We study $M$-alternating Hamilton paths and $M$-alternating Hamilton cycles in a simple connected graph $G$ on $ν$ vertices with a perfect matching $M$. Let $G$ be a bipartite graph, we prove that if for any two vertices $x$ and $y$ in different parts of $G$, $d(x)+d(y)\geq ν/2+2$, then $G$ has an $M$-alternating Hamilton cycle. For general graphs, a condition for the existence of an $M$-alternating Hamilton path starting and ending with edges in $M$ is put forward. Then we prove that if $κ(G)\geqν/2$, where $κ(G)$ denotes the connectivity of $G$, then $G$ has an $M$-alternating Hamilton cycle or belongs to one class of exceptional graphs. Lou and Yu \cite{LY} have proved that every $k$-extendable graph $H$ with $k\geqν/4$ is bipartite or satisfies $κ(H)\geq 2k$. Combining this result with those we obtain we prove the existence of $M$-alternating Hamilton cycles in $H$.

math.CO

Bipartite graphs with a perfect matching and digraphs

In this paper, we introduce a corresponding between bipartite graphs with a perfect matching and digraphs, which implicates an equivalent relation between the extendibility of bipartite graphs and the strongly connectivity of digraphs. Such an equivalent relation explains the similar results on $k$-extendable bipartite graphs and $k$-strong digraphs. We also study the relation among $k$-extendable bipartite graphs, $k$-strong digraphs and combinatorial matrices. For bipartite graphs that are not 1-extendable and digraphs that are not strong, we prove that the elementary components and strong components are counterparts.

math.CO

Equivalence between Extendibility and Factor-Criticality

In this paper, we show that if $k\geq (ν+2)/4$, where $ν$ denotes the order of a graph, a non-bipartite graph $G$ is $k$-extendable if and only if it is $2k$-factor-critical. If $k\geq (ν-3)/4$, a graph $G$ is $k\ 1/2$-extendable if and only if it is $(2k+1)$-factor-critical. We also give examples to show that the two bounds are best possible. Our results are answers to a problem posted by Favaron [3] and Yu [11].

math.CO