A Generalization of Orthogonal Factorizations in Graphs

Let G be a graph with vertex set V(G) and edge set E(G) and let g and f be two integer-valuated functions defined on V(G) such that g(x) ≤f(x) for all x∈V(G). Then a (g, f)-factor of G is a spanning subgraph H of G such that g(x) ≤d _H (x) ≤f(x) for all x∈V(G). A (g, f)-factorization of G is a partition of E(G) into edge-disjoint (g, f)-factors. Let = {F ₁, F ₂, ..., F _m } be a factorization of G and H be a subgraph of G with mr edges. If F _i , 1 ≤i≤m, has exactly r edges in common with H, then is said to be r-orthogonal to H. In this paper it is proved that every (mg + kr, mf−kr)-graph, where m, k and r are positive integers with k < m and g≥r, contains a subgraph R such that R has a (g, f)-factorization which is r-orthogonal to a given subgraph H with kr edges. This research is supported by the National Natural Science Foundation of China (19831080) and RSDP of China     

A note on 1-factors in graphs

Conditions on a graphG are presented which are sufficient to guarantee thatG–Z contains a 1-factor, whereZ is a set of edges ofG of restricted cardinality. These conditions provide generalizations of several known results and, further, establish the result that ifG is anr-regular, (r–2)-edge-connected graph (r2) of even order andz is an integer with 0zr–1 such thatG contains fewer thanr–z edge cut sets of cardinalityr–2, thenG–Z has a 1-factor for each setZ ofz edges ofG.     

On the maximal number of certain subgraphs inK r -free graphs

Given two graphsH andG, letH(G) denote the number of subgraphs ofG isomorphic toH. We prove that ifH is a bipartite graph with a one-factor, then for every triangle-free graphG withn verticesH(G) H(T ₂(n)), whereT ₂(n) denotes the complete bipartite graph ofn vertices whose colour classes are as equal as possible. We also prove that ifK is a completet-partite graph ofm vertices,r > t, n max(m, r – 1), then there exists a complete (r – 1)-partite graphG* withn vertices such thatK(G) K(G*) holds for everyK _r -free graphG withn vertices. In particular, in the class of allK _r -free graphs withn vertices the complete balanced (r – 1)-partite graphT _r–1(n) has the largest number of subgraphs isomorphic toK _t (t < r),C ₄,K _2,3. These generalize some theorems of Turán, Erdös and Sauer.Dedicated to Paul Turán on his 80th Birthday     

Hamilton Cycle Rich 2-factorizations of Complete Multipartite Graphs

For any two 2-regular spanning subgraphs G and H of the complete multipartite graph K, necessary and sufficient conditions are found for the existence of a 2-factorization of K in which

Hamilton Cycle Rich 2-Factorization of Complete Equipartite Graphs-II

For any given two 2-factors G and H of a complete p-partite graph K(m, p), with m vertices in each partite set, we prove the existence of a 2-factorization of K(m, p), in which one 2-factor is isomorphic to G, another 2-factor is isomorphic to H and the remaining 2-factors are hamilton cycles. Further, we prove the corresponding result for K(m, p) ? I, where I is a 1-factor of K(m, p), when K(m, p) is an odd regular graph. In fact our results together with the results of McCauley and Rodger settled the problem of 2-factorization of K(m, p), when two of the 2-factors are isomorphic to the given two 2-factors and the remaining 2-factors are hamilton cycles except for (m, p)?=?(m, 2).     

Cohomologie galoisienne des groupes quasi-déployés sur des corps de dimension cohomologique ≤2; Galois cohomology of quasi-split groups over fields of cohomological dimension ≤ 2

Let k be a perfect field with cohomological dimension 2. Serre's conjecture II claims that the Galois cohomology set H ¹(k,G) is trivial for any simply connected semi-simple algebraic G/k and this conjecture is known for groups of type ¹ A _n after Merkurjev–Suslin and for classical groups and groups of type F ₄ and G ₂ after Bayer–Parimala. For any maximal torus T of G/k, we study the map H ¹(k, T) H ¹(k, G) using an induction process on the type of the groups, and it yields conjecture II for all quasi-split simply connected absolutely almost k-simple groups with type distinct from E ₈. We also have partial results for E ₈ and for some twisted forms of simply connected quasi-split groups. In particular, this method gives a new proof of Hasse principle for quasi-split groups over number fields including the E ₈-case, which is based on the Galois cohomology of maximal tori of such groups.     

Maximal sets of hamilton cycles in complete multipartite graphs

A set S of edge‐disjoint hamilton cycles in a graph G is said to be maximal if the edges in the hamilton cycles in S induce a subgraph H of G such that G ? E(H) contains no hamilton cycles. In this context, the spectrum S(G) of a graph G is the set of integers m such that G contains a maximal set of m edge‐disjoint hamilton cycles. This spectrum has previously been determined for all complete graphs and for all complete bipartite graphs. In this paper, we extend these results to the complete multipartite graphs. © 2003 Wiley Periodicals, Inc. J Graph Theory 43: 49–66, 2003     

Incomplete orthogonal arrays and idempotent orthogonal arrays

First, we shall define idempotent orthogonal arrays and notice that idempotent orthogonal array of strength two are idempotent mutually orthogonal quasi-groups. Then, we shall state some properties of idempotent orthogonal arrays.Next, we shall prove that, starting from an incomplete orthogonal arrayT _EF based onE andF E, from an orthogonal arrayT _G based onG = E – F and from an idempotent orthogonal arrayT _H based onH, we are able to construct an incomplete orthogonal arrayT _(F(G×H))F based onF(G × H) andF. Finally, we shall show the relationship between the construction which lead us to this result and the singular direct product of mutually orthogonal quasi-groups given by Sade [5].     

Free Resolution Invariants for Finite Groups

Given a finite group G and a G-free resolution F _* of Z, then d _G (Im(F _m+1F _m ))–(–1) ^m–i d _G (F _i ) is almost always an invariant of G.     

The order of the differentials in the Atiyah-Hirzebruch spectral sequence

LetF _* be the homology theory corresponding to a spectrumF and consider the Atiyah-Hirzebruch spectral sequenceE _s,t ² H _s (X; _t F) F _s+t (X) for a bounded below spectrum (or CW-complex)X. This paper shows that the images of the differentials d _s,t ^r :E _s,t ^r E _s ^{–r,t+r–1r} in this spectral sequence are always torsion groupsof finite exponent and that this exponent isbounded in a very universal way: we prove the existence of integersR _r forr2 such thatR _r d _s,t ^r =0 for any spectrumF, for any bounded below spectrumX and for all integersr2,s andt. The interesting point is that these upper boundsR _r for the additive order of the differentials d _s,t ^r dependonly onr, and that the result holds without any hypothesis on the spectrumF. In certain special cases, this implies that the spectral sequence collapses and even that the extension problems given by itsE -term are trivial.     

Splitting Off Edges within a Specified Subset Preserving the Edge-Connectivity of the Graph

Splitting off a pair su, sv of edges in a graph G means the operation that deletes su and sv and adds a new edge uv. Given a graph G = (V + s, E) which is k-edge-connected (k ≥ 2) between vertices of V and a specified subset R  V, first we consider the problem of finding a longest possible sequence of disjoint pairs of edges sx, sy, (x ,y  R) which can be split off preserving k-edge-connectivity in V. If R = V and d(s) is even then a well-known theorem of Lovász asserts that a complete R-splitting exists, that is, all the edges connecting s to R can be split off in pairs. This is not the case in general. We characterize the graphs possessing a complete R-splitting and give a formula for the length of a longest R-splitting sequence. Motivated by the connection between splitting off results and connectivity augmentation problems we also investigate the following problem that we call the split completion problem: given G and R as above, find a smallest set F of new edges incident to s such that G′ = (V + s, E + F) has a complete R-splitting. We give a min-max formula for F as well as a polynomial algorithm to find a smallest F. As a corollary we show a polynomial algorithm which finds a solution of size at most k/2 + 1 more than the optimum for the following augmentation problem, raised in [[2]]: given a graph H = (V, E), an integer k ≥ 2, and a set R  V, find a smallest set F′ of new edges for which H′ = (V, E + F′) is k-edge-connected and no edge of F′ crosses R.     

Galois Stability, Integrality and Realization Fields for Representations of Finite Abelian Groups

For a given field F of characteristic 0 we consider a normal extension E/F of finite degree d and finite Abelian subgroups GGL _n (E) of a given exponent t. We assume that G is stable under the natural action of the Galois group of E/F and consider the fields E=F(G) that are obtained via adjoining all matrix coefficients of all matrices gG to F. It is proved that under some reasonable restrictions for n, any E can be realized as F(G), while if all coefficients of matrices in G are algebraic integers, there are only finitely many fields E=F(G) for prescribed integers n and t or prescribed n and d.     

The rainbow number of matchings in regular bipartite graphs

Given a graph G and a subgraph H of G, let rb(G,H) be the minimum number r for which any edge-coloring of G with r colors has a rainbow subgraph H. The number rb(G,H) is called the rainbow number of H with respect to G. Denote as mK₂ a matching of size m and as B_n,k the set of all the k-regular bipartite graphs with bipartition (X,Y) such that X=Y=n and k≤n. Let k,m,n be given positive integers, where k≥3, m≥2 and n>3(m−1). We show that for every GB_n,k, rb(G,mK₂)=k(m−2)+2. We also determine the rainbow numbers of matchings in paths and cycles.     

Some Existence Theorems on All Fractional (<Emphasis Type="Italic">g</Emphasis>, <Emphasis Type="Italic">f</Emphasis>)-factors with Prescribed Properties

Let G be a graph, and g, f: V (G) → Z⁺ with g(x) ≤ f(x) for each x ∈ V (G). We say that G admits all fractional (g, f)-factors if G contains an fractional r-factor for every r: V (G) → Z⁺ with g(x) ≤ r(x) ≤ f(x) for any x ∈ V (G). Let H be a subgraph of G. We say that G has all fractional (g, f)-factors excluding H if for every r: V (G) → Z⁺ with g(x) ≤ r(x) ≤ f(x) for all x ∈ V (G), G has a fractional r-factor F _h such that E(H) ∩ E(F _h ) = θ, where h: E(G) → [0, 1] is a function. In this paper, we show a characterization for the existence of all fractional (g, f)-factors excluding H and obtain two sufficient conditions for a graph to have all fractional (g, f)-factors excluding H.     

The empirical process of a short-range dependent stationary sequence under Gaussian subordination

Summary Consider the stationary sequenceX ₁=G(Z ₁),X ₂=G(Z ₂),..., whereG(·) is an arbitrary Borel function andZ ₁,Z ₂,... is a mean-zero stationary Gaussian sequence with covariance functionr(k)=E(Z ₁ Z _k+1) satisfyingr(0)=1 and _k=1 |r(k)| ^m < , where, withI{·} denoting the indicator function andF(·) the continuous marginal distribution function of the sequence {X _n }, the integerm is the Hermite rank of the family {I{G(·) x} –F(x):xR}. LetF _n (·) be the empirical distribution function ofX ₁,...,X _n . We prove that, asn, the empirical processn ^1/2{F _n (·)-F(·)} converges in distribution to a Gaussian process in the spaceD[–,].Partially supported by NSF Grant DMS-9208067     

Riesz Exponential Families on Symmetric Cones

Let E be a simple Euclidean Jordan algebra of rank r and let be its symmetric cone. Given a Jordan frame on E, the generalized power _s (– ^–1) defined on – is the Laplace transform of some positive measure R _s on E if and only if s is in a given subset of R ^r . The aim of this paper is to study the natural exponential families (NEFs) F(R _s ) associated to the measures R _s . We give a condition on s so that R _s generates a NEF, we calculate the variance function of F(R _s ) and we show that a NEF F on E is invariant by the triangular group if and only if there exists s in such that either F=F(R _s ) or F is the image of F(R _s ) under the map x–x.     

Nowhere-Zero 5-Flows and Even (1,2)-Factors

A graph G = (V, E) admits a nowhere-zero k-flow if there exists an orientation H = (V, A) of G and an integer flow ${\varphi:A \to \mathbb{Z}}$ such that for all ${a \in A, 0 < |\varphi(a)| < k}$ . Tutte conjectured that every bridgeless graphs admits a nowhere-zero 5-flow. A (1,2)-factor of G is a set ${F \subseteq E}$ such that the degree of any vertex v in the subgraph induced by F is 1 or 2. Let us call an edge of G, F-balanced if either it belongs to F or both its ends have the same degree in F. Call a cycle of G F-even if it has an even number of F-balanced edges. A (1,2)-factor F of G is even if each cycle of G is F-even. The main result of the paper is that a cubic graph G admits a nowhere-zero 5-flow if and only if G has an even (1,2)-factor.     

A note on the signed edge domination number in graphs

Let G=(V(G),E(G)) be a graph. A function f:E(G)→{+1,−1} is called the signed edge domination function (SEDF) of G if ∑_e^′N[e]f(e^′)≥1 for every eE(G). The signed edge domination number of G is defined as is a SEDF of G}. Xu [Baogen Xu, Two classes of edge domination in graphs, Discrete Applied Mathematics 154 (2006) 1541–1546] researched on the edge domination in graphs and proved that for any graph G of order n(n≥4). In the article, he conjectured that: For any 2-connected graph G of order n(n≥2), . In this note, we present some counterexamples to the above conjecture and prove that there exists a family of k-connected graphs G_m,k with .     

Globally sparse vertex-ramsey graphs

For graphs G and F we write F → (G)¹_r if every r-coloring of the vertices of F results in a monochromatic copy of G. The global density m(F) of F is the maximum ratio of the number of edges to the number of vertices taken over all subgraphs of F. Let We show that The lower bound is achieved by complete graphs, whereas, for all r ≥ 2 and ? > 0, m_cr(S_k, r) > r - ? for sufficiently large k, where S_k is the star with k arms. In particular, we prove that