Hamiltonian Cycles With All Small Even Chords

Size: px
Start display at page:

Download "Hamiltonian Cycles With All Small Even Chords"

Transcription

1 Hamiltonian Cycles With All Small Even Chords Guantao Chen, Katsuhiro Ota, Akira Saito, and Yi Zhao Abstract. Let G be a graph of order n 3. An even squared hamiltonian cycle (ESHC) of G is a hamiltonian cycle C = v 1 v... v nv 1 of G with chords v i v i+3 for all 1 i n (where v n+j = v j for j 1). When n is even, an ESHC contains all bipartite -regular graphs of order n. We prove that there is a positive integer N such that for every graph G of even order n N, if the minimum degree δ(g) n +9, then G contains an ESHC. We show that the condition of n being even cannot be dropped and the constant 9 can not be replaced by 1. Our results can be easily extended to even kth powered hamiltonian cycles for all k. 1. Introduction In this paper, we will only consider simple graphs finite graphs without loops or multiple edges. The notations and definitions not defined here can be found in Diestel [7]. Let G = (V, E) be a graph with vertex set V and edge set E. For a vertex v V and a subset S V, let Γ(v, S) denote the set of neighbors of v in S, and deg(v, S) = Γ(v, S). Given another set U V, define Γ(U, S) = u U Γ(u, H) as the set of common neighbors in S of all the vertices in U, and let deg(u, S) = Γ(U, S). When U = {v 1,...,v k }, we simply write Γ(U, S) and deg(v, S) as Γ(v 1...v k, S) and deg(v 1...v k, S), respectively. When S = V, we only write Γ(U) and deg(u). A graph G is called hamiltonian if it contains a spanning cycle. The Hamiltonian Problem, determining whether a graph has a hamiltonian cycle, has long been one of few fundamental problems in graph theory. In this paper we fix G to be a graph of order n 3. Dirac [8] proved that if minimum degree δ(g) n/ then G is hamiltonian. Ore [3] extended Dirac s result by replacing the minimum degree condition with that deg(u)+deg(v) n for all nonadjacent vertices u and v. Many results have been obtained on generalizing these two classic results (see Gould [13] for a recent survey in this area). A -regular subgraph (-factor) of G consists of disjoint cycles of G. Aigner and Brandt [] proved that if minimum degree δ(g) n 1 3 then G contains all -factors as subgraphs (Alon and Fischer [3] proved this for sufficiently large n). If to-be-embedded - factors have at most k odd components, then by a conjecture of El-Zahar [9], the minimum degree condition can be reduced to δ(g) (n + k)/ (Abbasi [1] announced a proof of El-Zahar s conjecture for large n). Another generalization to Aigner and Brandt s result is to find one specific subgraph of G which contains all -factors. A squared hamiltoinian cycle of G is a hamiltonian cycle v 1 v... v n v 1 together with edges v i v i+ for all 1 i n. Note that we always assume that v n+i = v i for i 1. It is easy to see that a squared hamiltonian A. Saito s research is partially supported by Japan Society for the Promotion of Science, Grant-in-Aid for Scientific Research (C), , 004, and The Research Grant of Nihon University, College of Humanities and Sciences. Y. Zhao s research is partially supported by NSA H and H

2 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO cycle contains all -factors of G. Pósa (see [10]) conjectured that every graph G of order n 3 with δ(g) 3n contains a squared hamiltonian cycle. Fan and Kiestead [1] proved this conjecture approximately; Komlés, Sárközy, and Szemerédi [16] proved the conjecture for sufficiently large n. More generally, the k-th powered hamiltonian cycle is a hamiltonian cycle v 1 v...v n v 1 with chords v i v i+j for all 1 i n and 1 j k. Komlés, Sárközy, and Szemerédi [18, 19] proved a conjecture of Seymour for sufficiently large n: every n-vertex graph G with δ(g) (k 1)n/k contains a kth powered hamiltonian cycle. Böttcher, Schacht and Taraz [5] recently proved a conjecture of Bollobás and Komlós [15], which asymptotically includes all the results mentioned above. Given an integer b, a graph H is said to have bandwidth at most b, if there exists a labeling of the vertices by v 1, v,..., v n, such that j i b whenever v i v j E(H). It is shown in [5] that for any ε > 0 and integers r,, there exists β > 0 with the following property. Let G and H be n-vertex graphs for sufficiently large n. If δ(g) ((r 1)/r + ε)n and H is r-chromatic 1 with maximum degree and bandwidth at most βn, then G contains a copy of H. We are interested in the situation when the error term εn in the conjecture of Bollobás and Komlós can be reduced to a constant. According to the El-Zahar Conjecture, every n-vertex graph G with minimum degree δ(g) n/ contain all -factors with even components. Given a graph G, we define an Even Squared Hamiltonian Cycle (ESHC) as a hamiltonian cycle C = v 1 v...v n v 1 of G with chords v i v i+3 for all 1 i n. When n 7, an ESHC is 4-regular with chromatic number χ = for even n and χ = 3 for odd n. It is not hard to check that an n-vertex ESHC contains all bipartite graphs of order n with maximum degree (e.g., by using the fact that every ESHC of even order contains a ladder graph defined below). Below is our main result. Theorem 1.1. There exists N > 0 such that for all even integers n N, if G is a graph of order n with δ(g) n + 9, then G contains an ESHC. We show that the constant 9 in Theorem 1.1 can not be replaced by 1. Proposition 1.. Suppose that n 10. Let G be the union of two copies of K n sharing 4 vertices. Then δ(g) = n + 1 but G contain no ESHC. We also show that the condition of n being even is necessary for Theorem 1.1 even if we replace 9 by n/8 1/. Proposition 1.3. There are infinitely many odd n and graphs G of order n such that δ(g) n + n/8 1/ and G contains no ESHC. More generally, an Even kth powered Hamiltonian Cycle (EkHC) of a graph G is a hamiltonian cycle v 1 v...v n v 1 with edges v i v i+j 1 for all 1 i n and 1 j k. Then an E1HC is simply a hamiltonian cycle while an EHC is an ESHC. Using the same proof techniques for Theorem 1.1, we can derive the following result, whose proof is omitted. Theorem 1.4. For any positive integer k, there exist a constant s = s(k) and a positive integer N such that if G is a graph of even order n N and δ(g) n+s then G contains an EkHC. + 1 The kth powered hamiltonian cycle of order n has chromatic number k + 1 or k + depending on the value of n. The authors of [5] make their result applicable even when H is r + 1-chromatic but one of its color classes is fairly small, e.g., the kth powered hamiltonian cycle.

3 HAMILTONIAN CYCLES WITH ALL SMALL EVEN CHORDS 3 One may view the even kth cycle of order n = N as the following bipartite graph. Let B k (N) be the bipartite graph (X Y, E) with X = {x 1,...,x N } and Y = {y 1,...,y N } such that x i y j E if and only if i j (mod N) { k + 1,..., 1, 0, 1,...,k}. In particular, B (N), or ESHC, contains the ladder graph defined by Czygrinow and Kierstead [6], which has the same vertex sets X and Y but x i is adjacent to y j if and only if i j (mod N) { 1, 0, 1}. Note that the the ladder graph contains all -factors with even components. The structure of the paper is as follows. We prove two (easy) Propositions 1. and 1.3 in the next section. Following the approach of [16] on squared hamiltonian cycles, we prove Theorem 1.1 by the regularity method. In Section 3 we state the Regularity Lemma and the Blow-up Lemma. In Section 4 we prove Theorem 1.1 by proving the non-extremal case and two extremal cases separately. It seems harder to handle the extremal cases here than in [16]; this is also the reason why we need a large constant 9 in Theorem 1.1. The last section gathers open problems with a remark.. Proofs of Propositions 1. and 1.3 Proof of Proposition 1.. Given a graph G, a pair (A, B) of vertex subsets is called a separator of G if V (G) = A B, both A B and B A are non-empty and E(A B, B A) =. It is easy to see that Proposition 1. follows from the following more general statement. Suppose that G is a graph with an EHSC. If (A, B) is a separator of G with A B 3 and B A 3, then A B 6. Let H be an ESHC of G with supporting cycle C. Assign C an orientation. An segment P 1 [x, z] of C is called an A B-path if x A B and z B A. We claim that if P 1 [x 1, z 1 ] is an A B-path such that (A B) V (P 1 ) 1 and (B A) V (P 1 ) 1, then C V (P 1 ) contains another A B-path P. In fact, C[z 1, x 1 ] = C E(P 1 ) must contain two internal vertices x A B and z B A. Then P := C[x, z ] is an A B-path. We can further assume that V (P ) {x, z } A B by letting P be a minimal A B-path of C V (P 1 ). Now let P 1 [x 1, z 1 ] be an A B-path such that V (P 1 ) {x 1, z 1 } A B. Since A B 3, B A 3, by the claim above, C V (P 1 ) contains another A B-path P [x, z ] such that V (P ) {x, z } A B. If e(p i ) 4 for i = 1,, then A B e(p 1 )+e(p ) 6 and we are done. On the other hand, we know that e(p i ) {1, 3} for i = 1, because x i z i E(G), which follows from e(a B, B A) =. Without loss of generality, assume that e(p 1 ) =, or P 1 = x 1 y 1 z 1. By following the orientation of C, let x 1 be the predecessor of x 1 and z 1 + be the successor of z 1. Since x 1 z 1, x 1 z 1 + E(G), we have x 1, z+ 1 A B. Since P 1 and P are vertex disjoint A B-paths, V (P ) {x 1, y 1, z 1 + } =. If P contains at least three internal vertices, then A B 6 and we are done. So we may assume that P = x y z, and consequently {x, y, z + } A B. If z+ 1 x and z+ x 1, then all x 1, y 1, z 1 +, x, y, z + are distinct, and consequently A B 6. Otherwise, without loss of generality, assume that z 1 + = x. Then P 1 = P 1z 1 + P is an A B-path with two vertices in each of A B and B A. Since A B 3 and B A 3, by the earlier claim, there is an A B-path P 3 [x 3, z 3 ] vertex-disjoint from P 1 such that V (P 3) {x 3, z 3 } A B. If e(p 3 ) 4, then A B 6 because P 1, P 3 are disjoint and P 1 contains three vertices y 1, z 1 +, y from A B. Otherwise P 3 = x 3 y 3 z 3 for some y 3 A B, then {x 3, y 3, z 3 + } A B. Since six vertices y 1, z 1 +, y, x 3, y 3, z 3 + are contained in A B, we have A B 6.

4 4 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO x 1 A x 1 y 1 z 1 z + 1 B Figure 1. A segment connecting A and B Proof of Proposition 1.3. Let q be an odd prime power, by using projective planes, one can construct (e.g. [11]) C 4 -free graphs H of order h = q + q + 1 with δ(h) q. Let G := H + K h q, i.e. a graph obtained from H by adding h q vertices such that each new vertex is adjacent to all vertices of H. Let X := V (G) V (H) and n := V (G) = h q. Then n is odd and δ(g) h = n + q n n because n = q + q + < (q + 1). To see that G does not have any ESHC, consider an arbitrary hamiltonian cycle C of G (if it exists). Since X is an independent vertex set and n is odd, we conclude that C X is the union of vertex-disjoint paths such that e(c X) is odd. In particular, one path P[x, y] of C X has odd length. If V (P) 4, then xx +++ / E(G), where x +++ = ((x + ) + ) +, since H contains no C 4. Otherwise V (P) =, which in turn shows that x, y + X. Since X is independent, x y + / E(G). In all cases G does not have an ESHC based on C. 3. The Regularity Lemma and Blow-up Lemma As in [16, 19], the Regularity Lemma of Szemerédi [4] and Blow-up Lemma of Komlós, Sárközy and Szemerédi [17] are main tools in our proof of Theorem 1.1. For any two disjoint vertex-sets A and B of a graph G, the density of A and B is the ratio d(a, B) := e(a, B)/( A B ), where e(a, B) is the number of edges with one endvertex in A and the other in B. Let ε and δ be two positive real numbers. The pair (A, B) is called ε-regular if for every X A and Y B satisfying X > ε A, Y > ε B, we have d(x, Y ) d(a, B) < ε. Moreover, the pair (A, B) is called (ε, δ)-super-regular if (A, B) is ε-regular and deg B (a) > δ B for all a A and deg A (b) > δ A for all b B. Lemma 3.1 (Regularity Lemma Degree Form). For every ε > 0 there is an M = M(ε) such that, for any graph G = (V, E) and any real number d [0, 1], there is a partition of the vertex set V into l + 1 clusters V 0, V 1,...,V l, and there is a subgraph G of G with the following properties: l M, V i ε V for 0 i l, V 1 = V = = V l, deg G (v) > deg G (v) (d + ε) V for all v V, G [V i ] = (i.e. V i is an independent set in G ), for all i, each pair (V i, V j ), 1 i < j l, is ε-regular with d(v i, V j ) = 0 or d(v i, V j ) d in G.

5 HAMILTONIAN CYCLES WITH ALL SMALL EVEN CHORDS 5 The Blow-up Lemma allows us to regard a super-regular pair as a complete bipartite graph when embedding a graph with bounded degree. We need a bipartite version of this lemma which also restricts the mappings of a small number of vertices. Lemma 3. (Blow-up Lemma - bipartite version). For every δ,, c > 0, there exists an ε = ε(δ,, c) > 0 and α = α(δ,, c) > 0 such that the following holds. Let (X, Y ) be an (ε, δ)-super-regular pair with X = Y = N. If a bipartite graph H with (H) can be embedded in K N,N by a function φ, then H can be embedded in (X, Y ). Moreover, in each φ 1 (X) and φ 1 (Y ), fix at most αn special vertices z, each of which is equipped with a subset S z of X or Y of size at least cn. The embedding of H into (X, Y ) exists even if we restrict the image of z to be S z for all special vertices z. 4. Proof of Theorem 1.1 A partition V 1...V k of a set V is called balanced if V i V j 1 for all i j. In this paper we often consider a balanced bipartition of V of order n for an even n, in which both partition sets are of size n/. Let us first define two extremal cases. Extremal Case 1: There exists a balanced partition of V into V 1 and V such that the density d(v 1, V ) 1 α for some small α (to be specified later). Extremal Case : There exists a balanced partition of V into V 1 and V such that the density d(v 1, V ) α for some small α (to be specified later). The following three theorems deal with the non-extremal case and two extremal cases separately. Theorem 4.1. For every α > 0, there exist β > 0 and a positive integer n 0 such that the following holds for every even integer n n 0. For every graph G of order n with δ(g) ( 1 β)n, either G contains an ESHC or G is in one of the extremal cases with parameter α. Theorem 4.. Suppose that 0 < α 1 and n is a sufficiently large even integer. Let G be a graph on n vertices with δ(g) n + 3. If G is in Extremal Case 1 with parameter α, then G contains an ESHC. Theorem 4.3. Suppose that 0 < α 1 and n is a sufficiently large even integer. Let G be a graph on n vertices with δ(g) n + 9. If G is in Extremal Case with parameter α, then G contains an ESHC. It is easy to see that Theorem 1.1 follows from Theorems 4.1, 4. and 4.3. For this purpose we can even use a weaker version of Theorem 4.1 with β = 0, but the current Theorem 4.1 may have other applications. If G is in Extremal Case with parameter α, then there exists x V such that deg(x) < (1 + α)n/ and in turn δ(g) < (1 + α)n/. Theorems 4.1 and 4. together implies the following remark, which is a special case of the theorem of Böttcher, Schacht and Taraz [5]. Remark 4.4. For any α > 0, there exists a positive integer n 0 such that every graph G of even order n n 0 with δ(g) ( 1 + α)n contains an ESHC. This remark will be used in the proof of Theorem Non-extremal Case. In this section we prove Theorem 1.1.

6 6 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO Proof. We fix the following sequence of parameters (1) ε d β α and specify their dependence as the proof proceeds. Actually we let α be the minimum of the two parameters defined in the extremal cases. Then we choose d β such that they are much small than α. Finally we choose ε = 1 d ε(d, 4, 4 ) following the definition of ε in the Blow-up lemma. Choose n to be sufficiently large. In the proof we omit ceiling and floor functions if they are not crucial. Let G be a graph of order n such that δ(g) ( 1 β)n and G is not in either of the extremal cases. Applying the Regularity Lemma (Lemma 3.1) to G with parameters ε and d, we obtain a partition of V (G) into l + 1 clusters V 0, V 1,...,V l for some l M = M(ε), and a subgraph G of G with all described properties in Lemma 3.1. In particular, for all v V, deg G (v) > deg G (v) (d + ε)n ( 1 β ε d)n (1 β)n, provided that ε+d β. On the other hand, e(g ) e(g) (d+ε) n e(g) dn by using ε < d. We further assume that l = k is even; otherwise we eliminate the last cluster V l by removing all the vertices in this cluster to V 0. As a result, V 0 εn and () (1 ε)n ln = kn n. For each pair i and j with 1 i j l, we write V i V j if d(v i, V j ) d. As in other applications of the Regularity Lemma, we consider the reduced graph G r, whose vertex set is {1,...,l}, and two vertices i and j are adjacent if and only if V i V j. From δ(g ) > ( 1 β)n, we easily derive that δ(g r) ( 1 β)l as follows. Fix a cluster V i and a vertex v V i. Let m be the number of clusters that adjacent to V i. Since in G the vertex v has no neighbor in any cluster that is not adjacent to V i, we have ( 1 β)n deg G (v) mn. With (), this gives that m (1 β)n/n (1 β)l. The rest of the proof consists of the following five steps. Step 1: Show that G r contains a hamiltonian cycle X 1 Y 1...X k Y k. Step : For each i, initiate a connecting ESP (even squared path) P i between Y i 1 and X i (where Y 0 = Y k ) with two vertices from each Y i 1 and X i. Step 3: For each i, move at most εn vertices from X i Y i to V 0 such that the graph on the remaining vertices of X i Y i has the minimum degree at least (d ε)n. Step 4: Extend P 1,...,P k to include all the vertices in V 0 such that X i j V (P j ) = Y i j V (P j ) (1 d /)N for all i. Step 5: Apply the Blow-up Lemma to each (X i, Y i ) and obtain an ESP consisting of all the remaining vertices of X i Y i. Concatenating these ESP s with P 1,...,P k, we obtain the desired ESHC of G. We now give details to each step. The assumption that G is not in either of the extremal cases leads to the following claim, which will be used in Step 1 and Step 4. Claim 4.5. (1) G r contains no independent set U 1 of size at least ( 1 8β)l. () G r contains no two disjoint subsets U 1, U of size at least ( 1 6β)l such that e Gr (U 1, U ) = 0.

7 HAMILTONIAN CYCLES WITH ALL SMALL EVEN CHORDS 7 Proof. (1). Suppose instead, that G r contains an independent set U 1 of size ( 1 8β)l. We will show that G is in the Extremal Case 1 with parameter α. Let A = i U 1 V i and B = V (G) A. By (), ( 1 9β)n (1 8β)Nl = U 1 N = A < ( 1 β)n. For each x A, since deg G (x, A) deg G (x, A) + (d + ε)n < βn, then deg G (x, B) > ( 1 β)n βn (1 β)n. Hence e G(A, B) ( 1 9β)n(1 β)n > ( β)n. Now move at most 9βn vertices from B to A such that A and B are of size n/. We still have ( 1 e G (A, B) > 4 11 ) β n 9βn n ( ) 14 ( n ) = 10β n = (1 40β). By specializing 40β α in (1), we see that G is in the Extremal Case 1 with parameter α. (). Suppose instead, that G r contains two disjoint subsets U 1, U of size ( 1 6β)l such that e Gr (U 1, U ) = 0. We will show that G is in the Extremal Case with parameter α. Let A = i U 1 V i and B = i U V i. Since e Gr (U 1, U ) = 0, we have e G (A, B) = 0. Since e(g) e(g ) + dn, we have e G (A, B) e G (A, B) + dn = dn. Note that A = U 1 N = ( 1 6β)lN > (1 7β)n. Similarly, B > (1 7β)n. By adding at most 7βn vertices to each of A and B, we obtain two subsets of size n/ and still name them as A and B, respectively. Then, e(a, B) dn + (7βn)(n/) = 8βn, which in turn shows the density d(a, B) = e(a, B)/( n ) 3β. Since α > 3β, we obtain that G is in the Extremal Case with parameter α. Step 1. To show that G r is hamiltonian, we need the following theorem of Nash-Williams. Theorem 4.6 (Nash-Williams[]). Let G be a -connected graph of order n. If minimum degree δ(g) max{(n + )/3, α(g)}, then G contains a hamiltonian cycle. We first show that G r is βl-connected. Suppose, to the contrary, let S be a cut of G r such that S < βl and let U 1 and U be two components of G r S. Since δ(g r ) ( 1 β)l, we have U i ( 1 3β)l for i = 1,. Since e(u 1, U ) = 0, we obtain a contradiction to Claim 4.5 (). Since n = Nl + V i (l + )εn, we have l 1/ε 3/β, provided that β 5ε. Then βl 3, and G r is 3-connected. By Claim 4.5 (1), the independence number of G r is at most ( 1 8β)l, which is less than ( 1 β)l, or the minimum degree of G r. By Theorem 4.6, G r is hamiltonian. Following the order of a hamiltonian cycle of G r, we denote all the clusters of G except for V 0 by X 1, Y 1,...,X k, Y k (recall that l = k is even). We call X i the partner of Y i and vice versa. Step. For each i = 1,...k, we initiate an ESP P i of G connecting X i and Y i 1 (with Y 0 = Y k ) as follows. Given an ε-regular pair (X, Y ) of clusters and a subset Y Y, we call a vertex x X typical to Y if deg(x, Y ) (d ε) Y. By the regularity of (X, Y ), at most εn vertices of X are not typical to Y whenever Y > εn. Fix 1 i k. First let a i X i be a vertex typical to both Y i 1 and Y i and let b i X i be a vertex typical to both Γ(a i, Y i 1 ) and Γ(a i, Y i ). Since both pairs (Y i 1, X i ) and (X i, Y i ) are ε-regularity with density at least d, all but εn vertices of X can be chosen as a i. Since Γ(a i, Y i 1 ), Γ(a i, Y i ) (d ε)n > εn, all but at most εn + 1 vertices of X can be chosen as b i (note that b i a i ). Recall that Γ(a i b i, Y i 1 ) = Γ(a i, Y i 1 ) Γ(b i, Y i 1 ). The way we select a i and b i guarantee that Γ(a i b i, Y i 1 ) (d ε) N εn +.

8 8 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO Now let c i 1, d i 1 Γ(a i, b i, Y i 1 ) be two (distinct) vertices of Y i 1 such that c i 1 is typical to both X i 1 and X i, and d i 1 is typical to both Γ(c i 1, X i 1 ) and Γ(c i 1, X i ). All but at most εn vertices of Γ(a i b i, Y i 1 ) can be chosen as c i 1 and d i 1. In summary P i = c i 1 a i d i 1 b i is an ESP with c i 1, d i 1 Y i 1, a i, b i X i such that (3) Step 3. For each i 1, let deg(c i 1 d i 1, X i 1 ) (d ε) N, deg(a i, Y i 1 ) (d ε)n, deg(a i b i, Y i ) (d ε) N, deg(d i 1, X i ) (d ε)n. X i := {x X i, deg(x, Y i ) (d ε)n} and Y i := {y Y i, deg(y, X i ) (d ε)n}. Since (X i, Y i ) is ε-regular, we have X i, Y i (1 ε)n. If X i Y i, say X i > Y i, then we pick an arbitrary subset of X i of size Y i and still name it X i. As a result, we have X i = Y i. Let V 0 := V k 0 i=1 (X i X i ) (Y i Y i ). From X i X i + Y i Y i εn, we derive that V 0 εn + (εn)k = 3εn by using Nk n from (). In addition, the minimum degree δ(g[x i, Y i ]) (d ε)n εn. Step 4. Consider a vertex x V (G) and an original cluster A (X i or Y i for some i), we say that x is adjacent to A, denoted by x A, if deg(x, A) (d ε)n. Given two vertices u, w, we define a u, w-chain of length t as distinct clusters A 1, B 1,...,A t, B t such that u A 1 B 1...A t B t w and each A j and B j are partners, in other words, {A j, B j } = {X ij, Y ij } for some i j. Claim 4.7. Let L be the list of at most εn pairs of vertices of G. For each {u, w} L, we can find u, w-chains of length at most four such that every cluster is used in at most d N/0 chains. Proof. Suppose that we have found chains of length at most four for the first m < εn pairs such that no cluster is contained in more than d N/0 chains. Let Ω be the set of all clusters that are used in exactly d N/0 chains. Since each chain use at most four clusters, we have d kn N Ω 4m 8εn 8ε 0 1 ε, where the last inequality follows from (). Therefore Ω 30ε k 30ε l βl provided (1 ε)d d that 1 ε 1 and 30ε d β. Now consider a pair {u, w} L. Our goal is to find a u, w-chain of length at most four by using clusters not in Ω. Let U be the set of all clusters adjacent to u but not in Ω, and W be the set of all clusters adjacent to w but not in Ω. Let P(U) and P(W) be the set of the partners of clusters in U and W, respectively. The definition of chains implies that a cluster A Ω if and only if its partner P(A) Ω. Therefore (P(U) P(W)) Ω =. We claim that P(U) = U ( 1 3β)l. To see it, we first observe that any vertex v V is adjacent to at least ( 1 β)l clusters. For instead, ( ) ( ) ( β n deg G (v) β ln + dnl + 3εn < 3 ) β n, a contradiction, provided that β d + 3ε. Since Ω βl, we thus have U (1 3β)l. Similarly P(W) = W ( 1 3β)l. If E Gr (P(U), P(W)), then there exist two adjacent clusters B 1 P(U), A P(W). If A 1, B are partners of each other, then u A B 1 w gives a u, w-chain of length two.

9 HAMILTONIAN CYCLES WITH ALL SMALL EVEN CHORDS 9 Otherwise assume that A 1 = P(B 1 ) and B = P(A ). Then u A 1 B 1 A B w gives a u, w-chain of length four. Note that all A i, B i Ω. Otherwise assume that (4) E Gr (P(U), P(W)) =. If P(U) P(W) =, then (4) contradicts with Claim 4.5 () because P(U) ( 1 3β)l and P(W) ( 1 3β)l,. Otherwise assume that A P(U) P(W). Then by (4), A is not adjacent to any cluster in P(U) P(W). Since deg Gr (A) ( 1 β)l, we derive that P(U) P(W) ( 1 + β)l. Since P(U) (1 3β)l and P(W) (1 3β)l, then P(U) P(W) ( 1 8β)l. By (4), P(U) P(W) is an independent set in G r, which contradicts with Claim 4.5 (1). We arbitrarily partition V 0 into at most εn pairs (note that V 0 is even because X i = Y i for all i). Applying Claim 4.7, we find chains of length at most four for each pair such that every cluster is used in at most d N/0 chains. For each i let m i denote the number of chains that contains (X i, Y i ). Claim 4.8. We can extend connecting ESP s to include all the vertices in V 0 such that the following holds for all i. The resulting ESP s P i = u 1 v 1...u t v t satisfies u 1, u Y i 1, v t 1, v t X i and deg(u (5) 1 u, X i 1 ) (d ε) N, deg(v 1, Y i 1 ) (d ε)n, deg(v t v t 1, Y i ) (d ε) N, deg(u t, X i ) (d ε)n. The sets Xi = X i jv (P j ) and Yi = Y i jv (P j ) satisfy (6) Xi = Yi (1 ε)n 7m i. Proof. We prove by induction on m := V 0 /. When m = 0, by (3), the initial P i = c i 1 a i d i 1 b i satisfies (5). The initial Xi = X i {a i, b i } and Yi := Y i {c i, d i }) satisfy (6). Suppose that m 1 and we have extended the connecting ESP s to include m 1 pairs from V 0 such that (5) and (6) hold for all i. Let {x, y} be the last pair from V 0. We first consider the case when the x, y-chain has length two. Without loss of generality, assume that x Y i X i y for some i. Let P i = u 1 v 1...u t v t be the current connecting ESP between Y i 1 and X i and Yi := Y i jv (P j ) and Xi := X i jv (P j ). To include x, we extend P i to P i = P iy 1 x 1 y x y 3 x 3 y 4 x with four vertices y 1, y, y 3, y 4 Yi and three vertices x 1, x, x 3 Xi such that in addition y 1 Γ(v t 1 v t ), y Γ(v t ), y 3, y 4 Γ(x), x 1 Γ(u t ), (7) y 4 is typical to X i, and x 3 is typical to Γ(x, Y i ). This is possible by using Lemma 3. (actually we only need the regularity between X i and Y i ; but applying the Blow-up Lemma makes our proof shorter). To see it, first note that (6) implies that Xi, Y i > (1 d /)N because m i d N/0 by Claim 4.7. Then, by (5), Γ(v t 1 v t, Yi ) deg(v t 1 v t, Y i ) d N (d ε) N d N > d 4 N. Similarly we can show that Γ(v t, Yi ), Γ(u t, Xi ) (d d )N. The definition of x Y i also guarantees that Γ(x, Yi ) (d d )N. Finally (7) only forbids additional εn vertices when choosing y 4 and x 3. Therefore we can apply Lemma 3. to find such an P i. By (7), we have deg(y 4, X i ) (d ε)n and deg(x 3 x, Y i ) (d ε) N. Consequently P i satisfies (5).

10 10 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO Since x behaves like a vertex of X i in P i, we call such a procedure inserting x into X i (by extending P i ). Since y X i, we can similarly insert y to Y i by extending P i+1 to P i+1 = yx 4y 5 x 5 y 6 x 6 y 7 x 7 P i+1 with x j Xi and y j Yi. Since each X i and Yi lose seven vertices totally, (6) holds. Now consider the case when the x, y-chain has length four. Assume that x A 1 B 1 A B y. We first insert x to B 1, then pick any (available) vertex in B 1 that is typical to A and insert it to B, and finally insert y to A. As a result, A 1, B 1 each loses four vertices to some connecting paths while A, B each loses seven vertices to some connecting paths. Thus (6) holds. Step 5. Fix 1 i k. Suppose that at present P i = u 1 v 1...u t v t and P i+1 = w 1 z 1...w s z s for some integers s, t, and Xi = X i jv (P j ) and Yi = Y i jv (P j ). By Claim 4.8, P i and P i+1 satisfy (5) and Xi, Y i (1 d /)N. Since (X i, Y i ) is regular, the Slicing Lemma of [14] says that (Xi, Y i ) is (ε, d/)-super-regular (note that (d ε)n d N/ > dn/). We now apply the Blow-up Lemma to each G [Xi, Y i ] to obtain an spanning ESP y 1 x 1 y... x Ni 1, y Ni, x Ni, where N i = Xi = Y i such that y 1 Γ(v t v t 1, Yi ), x 1 Γ(u t, Xi ), y Γ(v t, Yi ), x Ni 1 Γ(w 1, Xi ), y Ni Γ(z 1, Yi ), x N i Γ(w 1 w, Xi ). u t 1 w 1 w u t x Ni y Ni x Ni 1 y x 1 y 1 v t 1 v t z 1 z Y i 1 X i Y i X i+1 Figure. An ESP covering X i and Y i The restrictive mapping of y 1, x 1, y, x Ni 1, y Ni, x Ni is possible because by (5), all the targeting sets are of size at least (d ε) N d N/ > d N/4. We now complete the proof of the non-extremal case. 4.. Extremal Case 1. In this subsection we prove Theorem 4.. We start with a lemma which gives a balanced spanning bipartite subgraph that we will use throughout the section. Lemma 4.9. Suppose that 0 α (1/9) 3. Let G = (V, E) be a graph on n vertices with δ(g) n + 3 and a balanced partition V 1 V such that d(v 1, V ) 1 α. Then G contains a balanced spanning bipartite subgraph G with parts U 1, U such that

11 HAMILTONIAN CYCLES WITH ALL SMALL EVEN CHORDS 11 There is a set W of at most α /3 n vertices such that we can find vertex disjoint 4-stars (stars with four edges) in G with the vertices of W as their centers. deg G (x) (1 α 1/3 α /3 )n/ for all x W. Proof. For simplicity, let α 1 = α 1/3 and α = α /3. For each i = 1,, we define { V i = x V : deg(x, V 3 i ) (1 α 1 ) n }. Since d(v 1, V ) 1 α, we have V i V i α n/ and consequently V i (1 α )n/ for i = 1,. For any x V i, (8) deg(x, V 3 i) > (1 α 1 ) n α n. Let V 0 = V V 1 V. Then V 0 α n. For each v V 0 and i = 1,, we have deg(v, V i ) (1 α 1 ) n, which implies that deg(v, V n i) α 1 and (9) deg(v, V i ) (α 1 α ) n. We now separate cases based on V 1 and V. Case 1: V 1, V n/. In this case we partition V 0 into W 1 W such that W i = n/ V i for i = 1,. For each vertex w W i, we greedily find four neighbors from V 3 i such that the neighbors for all the vertices of W i are distinct. This is possible for i = 1, because of (9) and (α 1 α ) n 4α n 4 V 0 provided that α 1 9α or α 1 1/9. Define U i = V i W i for i = 1,. Then U 1 = U = n/. With W = V 0, the second assertion of Lemma 4.9 follows from (8). Case : one of V 1, V, say, V 1 is greater than n/. Let V 0 1 be the set of vertices v V 1 such that deg(v, V 1 ) α 1n/. First assume that V1 0 V 1 n/. Then we form a set W with V 1 n/ vertices of V 0 1 and all the vertices of V 0. Let U 1 = V 1 W and U = V W. Then U 1 = U = n/. Since (1 α )n/ V n/, we have W α n/. Then V 1 U 1 α n/, by (9) and the definition of V1 0, we have (10) deg(v, U 1 ) deg(v, V 1) α n/ (α 1 α )n/ α n/. for all v W. With α 1 6α, we have (α 1 α )n/ 4α n/ 4 W. Therefore we can greedily find four neighbors for each vertex v W such that the neighbors for all the vertices of W are distinct. The second assertion of Lemma 4.9 follows from (8) and V 1 U 1 α n/. V 0 Otherwise assume that V 0 1 < V 1 n/. In this case let U 1 = V 1 V 0 1 and U = V 1 V 0. Then U 1 = n +t 1 for some positive integer t 1 α n/. Since deg(v, V 1 ) < α 1n/ for every v U 1, the induced graph G[U 1 ] has the maximum degree α 1 n/. By the minimum degree condition δ(g) n/ + 3, G[U 1 ] has the minimum degree at least ( n ) ( n ) δ(g) U + 3 t 1 t We now need the following simple fact. Fact In a graph G 1 of order n 1 with the maximum degree (G 1 ) and the minimum degree δ(g 1 ) t, the number of disjoint 4-stars is at least (t 3)n 1 5( +t 3).

12 1 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO To see it, suppose G 1 has a largest family of disjoint 4-stars on M of size m. Then (t 3)(n 1 5m) e(m, V (G) M) 5m and the fact follows. Applying Fact 4.10, there are at least n (t ) U 1 5( + t ) t 1 U 1 n 5(α 1 + t 1) t 1 5(α 1 + α ) n t 1 vertex disjoint 4-starts in G[U 1 ]. Pick t 1 such 4-stars and move their centers to U. As a result, U 1 = U = n/. Let W 0 = V1 0 V 0 and W be the union of W 0 with the new vertices of U. Since V 1 U 1 = V 1 n/ α n/, we have (10) for all v W 0. Since W = n/ V α n/, we can find disjoint 4-stars in G[U 1, W 0 ] with all the vertices of W 0 as centers such that these 4-stars are also disjoint from the existing 4-stars. In addition, the second assertion of Lemma 4.9 holds as before. Proposition 4.11 below shows that if G is a graph or a bipartite graph with large minimum degree and it contains not many vertex disjoint 4-stars, then we can find an ESC or ESP containing all the vertices in these stars. We need its part () for this subsection, and part (1) for Extremal Case. Proposition Fix 0 < ε 1 1/5. (1) Let G be a graph of order N with a subset W of size t ε 1 N/8. Suppose that G contains t vertex-disjoint 4-stars with the vertices of W as centers. For every vertex x / W, we have deg(x) (1 ε 1 )N. Then G contains an ESC C of length 8t which contains all the vertices of W such that any two nearest vertices of W are separated by exactly seven vertices not in W. () Let G be a bipartite graph on two parts U 1, U of size N. Let W be a vertex subset of size t ε 1 N/5. Suppose that G contains t vertex-disjoint 4-stars with the vertices of W as centers. For every vertex x / W, we have deg(x) (1 ε 1 )N. Then G contains an ESP of length 8t + 4 which contains all the vertices of W and whose first and last three vertices are not from W. Proof. (1). Suppose that W = {x 1,...,x t }, and denote the four leaves of x i by a i, b i, c i, d i. For each x i, we greedily choose three new vertices u i, v i, w i (that are not contained in any existing star) such that u i Γ(c i 1 d i 1 a i b i ), v i Γ(d i 1 a i b i c i ), w i Γ(c i d i a i+1 b i+1 ), in which the indices are modulo t. This is possible because each a i, b i, c i, d i, 1 i t, has at least (1 ε 1 )N neighbors and any four of them have at least (1 4ε 1 )N ε 1 N 8t common neighbors (where 8t is the total number of vertices used at the end of this greedy algorithm). We thus obtain an ESC u i a i v i b i x i c i w i d i : i = 1,...,t, in which the vertices of W are distributed evenly. (). Partition W = W 1 W with W 1 = U 1 W and W = U W. For each W i, we follow the procedure in (1) to find two vertex disjoint ESC s C 1 and C of length 8 W 1 and 8 W in which the vertices of W are distributed evenly. The calculation is similar except that any four vertices in U 1 W (or U W) have at least (1 4ε 1 )N ε 1 N > 4t common neighbors in U (or U 1 ), where 4t is the total number of the vertices used in one partition set. We next break C 1 into P 1 = x 1 x x 3...u 3 u u 1 and break C into P 1 = v 1 v v 3...y 3 y y 1 such that x i, u i, v i, y i / W for i = 1,, 3 and u 1, u 3, v U 1, u, v 1, v 3 U. Assume that

13 HAMILTONIAN CYCLES WITH ALL SMALL EVEN CHORDS 13 t otherwise we are done. Choose four new vertices not in W (in this order) z 1 Γ(u 1 u 3 ), z 3 Γ(u 1 v ) and z Γ(u z 1 z 3 v 1 ), z 4 Γ(z 1 z 3 v 1 v 3 ). This is possible because the number of common neighbors of any four vertices not in W is at least (1 4ε 1 )N 5t 4t +, where 4t + is the total number of the vertices used in one partition set. As a result, P 1 z 1 z z 3 z 4 P is an ESP which contains all the vertices of W and whose first and last three vertices are not from W. We finally observe that a bipartite graph with very large minimum degree is superregular. Proposition 4.1. Given 0 < ρ < 1, let G be a bipartite graph on X Y such that (11) δ(x, Y ) (1 ρ) Y, δ(y, X) (1 ρ) X. Then G is ( ρ, 1 ρ)-super-regular. Proof. It suffices to show that G is ρ-regular. Consider subsets A X, B Y with A = ε 1 X and B = ε Y for some ε 1, ε > ρ. By (11), we have δ(a, Y ) Y ρ Y and consequently δ(a, B) B ρ Y = (ε ρ) Y. The density between A and B satisfies d(a, B) δ(a, B) A A B (ε ρ) Y B = ε ρ ε > 1 ρ ρ = 1 ρ. Since 1 ρ < d(a, B) 1 and in particular, 1 ρ < d(x, Y ) 1, we have d(a, B) d(x, Y ) < ρ. We are ready to prove Theorem 4. now. Proof of Theorem 4.. Let 0 α 1, in particular α (1/9) 3. Write α 1 = α 1/3 and α = α /3. Let G = (V, E) be a graph oh n vertices with δ(g) n + 3. Suppose G is in the extremal case 1 with parameter α. We first apply Lemma 4.9 to G and obtain a bipartite subgraph G with two partition sets U 1, U of size n/ which contains at most α /3 n vertex disjoint 4-stars. Denote by W the set of the centers of the 4-stars. We also have (1) deg G (x) (1 α 1 α )n/ for all x W. Since α 1 + α 1/5 and α n α 1+α n 5, we may apply Proposition 4.11 () to G with ε 1 = α 1 + α and N = n/. We thus obtain an ESP P 0 = x 1 x x 3...y 3 y y 1 of length 8 W + 4 which contains all the vertices of W such that x i, y i V 1 V for 1 i 3. In order to find an ESHC of G, it suffices to find an ESP P = u 1 u u 3...v 3 v v 1 on V (G) V (P 0 ) such that x 3 x x 1 P y 1 y y 3 = x 3 x x 1 u 1 u u 3...v 3 v v 1 y 1 y y 3 is also an ESP. Let U i = U i V (P 0 ) for i = 1,, and n = U 1 = U. Then n = n/ (4 V 0 + ) n/ (4α n + ). By (1), the bipartite subgraph G[U 1, U ] has its minimum degree at least (1 α 1 α ) n (4α n + ) (1 3α 1 ) n (1 3α 1)n. by using α 1 9α. Similarly the degree from x i or y i, i = 1,, 3, to U 1 or U is at least (1 3α 1 )n. By Proposition 4.1, (U 1, U ) is ( 3α 1, 3 )-super-regular (using α 1 1/9 again).

14 14 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO Since 3α 1 1, we can apply the Blow-up Lemma to obtain an ESHP u 1 u u 3... v 3 v v 1 of G[U 1, U ] such that u 1 Γ(x 1, x 3 ), u Γ(x ), u 3 Γ(x 1 ), v 1 Γ(y 1, y 3 ), v Γ(y ), v 3 Γ(y 1 ). Since Γ(x i ), Γ(y i ) (1 3α 1 )n for i = 1,, 3 and Γ(x 1, x 3 ), Γ(y 1, y 3 ) (1 6α 1 )n, the restrictive mapping of u 1, u, u 3 and v 1, v, v 3 is possible Extremal Case. In this subsection we prove Theorem 4.3. In Extremal Case 1, we used the Blow-up Lemma to find an ESHP with certain properties in a bipartite graph with very large minimum degree. In this subsection we first prove such a lemma for arbitrary graphs. Lemma Let k 3 and n 1 be sufficiently large. Let G be a graph of order k+n Suppose that G contains an ESP P 0 = u 1...u k. Let X = V (G) V (P 0 ). Suppose that x 1 x x 3 and y 1 y y 3 are two paths of X and let X be the set of the remaining vertices of X (then X = n 1 ). If deg(x, X ) 7 8 n for all vertices x X and deg(u j, X) 7 8 X + 1 for j = 1,, 3, k, k 1, k, then G contains an ESHP that starts with x 3 x x 1, finishes with y 1 y y 3 and contains P 0 as an internal path. Proof. Our proof consists of three steps. Step 1: we find an ESC on X. Let G 1 = G[X ]. Since δ(g 1 ) 7 8 n 1 and n 1 is sufficiently large, by Remark 4.4, G 1 contains an ESHC. Step : we find an ESP on X such that it starts with x 3 x x 1 and finishes with y 1 y y 3. Let v 1,...,v n1 be the ESC given by Step 1. We will form an ESP x 3 x x 1 v i...v 1 v n1 v n1 1...v i+1 y 1 y y 3. for some 1 i n 1. It suffices to have the following adjacency (13) x 3 v i, x v i 1, x 1 v i, x 1 v i, y 1 v i+1, y 1 v i+3, y v i+, y 3 v i+1, in which we assume that v j = v j+n1 for all integers j. Since deg(x, X ) 7n 1 /8 + 1 for any vertex x x 1, x, x 3, y 1, y, y 3, the number of 1 i n 1 satisfying (13) is at least n 1 8(n 1 /8 1) = 8. Thus (13) hods for some 1 i n 1. Step 3: we find an ESHP of G that starts with x 3 x x 1, finishes with y 1 y y 3 and contains P 0 as an internal path. Let n = X = n Denote the ESP found in Step by v 1,...v n, where v 1 = x 3, v = x, v 3 = x 1, v n = y 1, v n 1 = y, v n = y 3. Our goal is to find an index 3 i n 3 such that v 1...v i P 0 v i+1...v n is an ESP. Since P 0 = u 1...u k, it suffices to have the following adjacency (14) u 1 v i, u v i 1, u 3 v i, u 1 v i, u k v i+1, u k v i+3, u k 1 v i+, u k v i+1, for some 3 i n 3. Since deg(u j, X) 7n /8 + 1 for j = 1,, 3, k, k 1, k, the number of 1 i n satisfying (13) is at least n 8(n /8 1) = 8. Thus (13) hods for some 3 i n 3.

15 HAMILTONIAN CYCLES WITH ALL SMALL EVEN CHORDS 15 Proof of Theorem 4.3. We start with defining two new sets, which are variants of V 1 and V. Let α 1 = α 1/3 and α = α /3. We define { V i n } = x V : deg(x, V 3 i ) < α 1. for i = 1,. Since δ(g) > n/, we have deg(x, V i ) > (1 α 1 )n/ for every x V i. Since d(v 1, V ) α, we have V i V i α n/ and V i (1 α )n/ for i = 1,. Consequently, (15) deg(x, V i n ) > deg(x, V i ) α 1 (1 α 1 α ) n for all x V i Let V 0 = V V 1 V. Then V 0 α n and deg(x, V i ) (α 1 α )n/ for all x V 0. Our proof consists of the following two steps which together provide an ESHC of G. Step 1. Find two disjoint ESP s x 1...x p and y 1...y p of length 6 p 14 such that x 1, x, x 3, y 1, y, y 3 V 1 and x p, x p 1, x p, y p, y p 1, y p V. Step. Find two ESP s P 1 and P consisting of all the remaining vertices in V 1 and V V 0, respectively, such that x 3 x x 1 P 1 y 1 y y 3, and x p x p 1 x p P y p y p 1 y p are also ESP s. While Step follows from Proposition 4.11 and Lemma 4.13 easily, Step 1 is much harder (at least from the point of our view) it is where we need the large constant 9 in the min-degree condition. Below we present Step first Step : Find two ESP s covering the remaining vertices. Let P 1 = x 1...x p and P = y 1...y p be the two ESP s of length 6 p 14 provided by Step 1. Let S = V (P 1 ) V (P ). Let U 1 = V 1 S and U 1 = U 1 {x 1, x, x 3, y 1, y, y 3 }. Fix x U 1, by (15), we have (16) deg(x, U 1) (1 α 1 α )n/ p (1 α 1 α )n/ + 1, where the second inequality follows from α n/ 9 p + 1. Using U 1 (1 + α )n/ and α 1 3α, we derive that deg(x, U 1 ) (1 α 1) U Using α , we have deg(x, U 1 ) 7 8 U We then apply Lemma 4.13 to G[U 1] with P 0 = and obtain an ESP x 3 x x 1...y 1 y y 3 on U 1. On the other hand, let U = V V 0 S and U = U {x p, x p 1, x p, y p, y p 1, y p }. Partition U into V = V S and V 0 = V 0 S. We have V 0 α n and for each x V 0, deg(x, V ) (α 1 α ) n p α 1 n 4 V 0, by using α n/ p and α 1 16α. We then greedily find V 0 disjoint 4-stars with the vertices of V 0 as centers and vertices in V as leaves. Let N = (1 + α )n/. Then U V V 0 = n V 1 N. For any vertex x V, the arguments above for G[U 1] give that (17) deg(x, V ) (1 α 1 )N + 1 We then apply Proposition 4.11 (1) to G[U ] and obtain an ESC C 0 of length k = 8 V 0 8α n such that it contains all the vertices of V in a way that every two nearest vertices of V 0 0 are separated by exactly seven vertices of V. We then break C 0 into an ESP P 0 = u 1...u k such that u 1, u, u 3, u k, u k 1, u k V. Let X = U V (P 0) and X = X {x p, x p 1, x p, y p, y p 1, y p }. For any vertex x V, by (17) and using 8α n α 1 n α 1N, deg(x, X ) (1 α 1 )N + 1 8α n (1 3α 1 )N + 1.

16 16 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO Since X V, u 1, u, u 3, u k, u k 1, u k V and X < X N, by letting α 1 1 4, the degree conditions in Lemma 4.13 hold. We then apply Lemma 4.13 to G[U ] and obtain the desired ESHP x p x p 1 x p...p 0...y p y p 1 y p Step 1: Connect V 1 and V. Given two disjoint sets A and B, an ESP on A B is called an (A, B)-connector if its first three vertices are from A, and the last three vertices are from B. Our goal is to find two disjoint (V 1, V )-connectors of length at most 14. The simplest connector is an ESP x 1...x 6 with x 1, x, x 3 V 1 and x 4, x 5, x 6 V. Unfortunately such a simple connector may not exist if e(v 1, V ) is very small but e(v i, V 0) is relatively large. Let us sketch our proof. We first separate the vertices of V 0 with large degree to both V 1 and V : V 0 = {x V 0 : deg(x, V 1), deg(x, V ) n/6 α n} The reason why we choose n/6 can be seen from (0), in which we use n/ = 3(n/6). If V 0 > 165, then we can find two disjoint copies of T,3, from the 3-partite subgraph G[V 1, V 0, V ], where T,3, is the union of two copies K,3 sharing the three vertices in one partition set. Each copy of T,3, can be easily extended to an (V 1, V )-connector. If V 0 165, then V 0 will not be used any more. We add the vertices of V 0 V 0 into V 1 or V forming two new (disjoint) sets U 1, U such that any three vertices in U i, i = 1,, have many common neighbors. What remains is to find two disjoint (U 1, U )-connectors by using the minimum degree condition δ(g) (n+ V 0 )/+9. One way to construct such a connector is to find two adjacent vertices x U 1, y U such that there is 4-vertex path between Γ(x, U 1 ) and Γ(y, U ). If this can not be done, then we find six vertices x 1, x, x 3 U 1 and x 4, x 5, x 6 U such that x 1 x 4 and x 3 x 6 are edges and x, x 3, x 4, x 5 form a copy of K, in G[U 1, U ]. After finding one connector, we remove all or some of its vertices and repeat the procedure above. Note that ignoring V 0 of size at most 165 is the major reason for the large constant 9 in δ(g); for example, when V 0 =, then δ(g) n/ + 9 suffices. We need the following propositions on the existence of T,3, and K,, whose proofs are standard counting arguments. Given two disjoint vertex sets A, B in a graph H, we write K s,t (A, B) if H[A, B] contains a copy of K s,t with s vertices from A and t vertices from B. Similarly T,3, (A, B, C) means that there are subsets X A, Y B and Z C such that H[X, Y ] = H[Z, Y ] = K,3. We denote by δ(a, B) the minimum degree deg(a, B) over all a A. Proposition (1) Give any integer t > 0, there exist ε 0 > 0 and n 0 such that the following holds for any ε ε 0 and n n 0. Let A, B be two disjoint vertex set in a graph such that δ(b, A) n/6 3εn, A (1 + ε)n/ and B > 9(t 1). Then K,t (A, B). () There exist ε 0 > 0 and n 0 such that the following holds for any ε ε 0 and n n 0. Let A, B, C be three disjoint vertex set in a graph such that δ(b, A), δ(b, C) n/6 3εn, A, C (1 + ε)n/ and B > 16. Then T,3, (A, B, C). Proof. (1). If the graph contains no K,t with vertices in A and t vertices in B, then ( ) ( ) deg(x, A) A (18) (t 1). x B

17 HAMILTONIAN CYCLES WITH ALL SMALL EVEN CHORDS 17 Since δ(b, A) n/6 3εn and A (1 + ε)n/, ( ) ( ) n/6 3εn (1 + ε)n/ B (t 1). As n and ε 0, we obtain B 9(t 1), contradiction. (). Since B > 9(19 1), we can apply (1) with t = 19 to (A, B) and obtain a copy of K,19 on X A of size and Y B of size 19. Then since Y = 19 > 9(3 1), we can apply (1) again with t = 3 to (C, Y ) and obtain a copy of K,3 on Z C of size and Y Y of size 3. The next proposition easily follows from a classical result of Kővári, Sós, and Turán [0]. For completeness, we include its proof below. Proposition Let H = (A B, E) be a bipartite graph such that A = n, B = m. Then H contains a copy of K, if either of the following holds. (1) deg(x) m for all x A and n > m + m, () e := E max{3n, m /}. Proof. Suppose instead, H contains no copy of K,. Then (18) hold with t =. (1). We have ( ) m n ( ) ( ) deg(x) m, x A m(m 1) which implies that n m( m 1) = m( m + 1), a contradiction. (). By convexity, we have n ( ) ( e/n m ), which implies e(e/n 1) < m. Since e m /, we have (e/n 1) < or e < 3n, a contradiction. Now we start our proof. First assume that V 0 > 165. Since δ(v 0, V i ) n/6 α n and V i (1 + α )n/, we can apply Proposition 4.14 () to the 3-partite subgraph on V 1 V 0 V and find a copy of T,3, on X V 1, Y V 0 and Z V such that X = Z =, Y = 3. Let V = V 1 X, V 0 = V 0 Y, and V = V Z. Then V 0 > 16 and δ(v 0, V i ) n/6 α n. We apply Proposition 4.14 () again to 3-partite subgraph on V 1 V 0 V and find another copy of T,3,. We next extend each copy of T,3, to a (V 1, V )-connector of length 11 as follows. Assume X = {x 3, x 5 }, Y = {x 4, x 6, x 8 }, and Z = {x 7, x 9 }. Then x 3, x 4,...,x 8, x 9 is an ESP but it is not a (V 1, V )-connector because x 4 V 1 and x 8 V. We extend this ESP by adding two vertices from V 1 in the beginning and two vertices from V at the end. Since x 3, x 5 V 1, by (15), we can find a vertex x Γ(x 3 x 5, V 1 ). Since deg(x, V 1 ) > V 1 α 1n/ and deg(x 4, V 1 ) > n/6 α n, we can find a vertex x 1 Γ(x x 4, V 1 ), which is different from x 3, x 5. Therefore x 1 x... x 9 is an ESP. Similarly we find x 10, x 11 V such that x 1x...,x 10 x 11 is an ESP, which is a (V 1, V )-connector. Now assume that c 0 := V We will not use the vertices of V 0 any more. Since V 0 α n, all vertices x V 0 satisfy deg(x, V 1 V ) > n/ α n. If x V 0 V 0, then exactly one of deg(x, V 1 ) and deg(x, V ) is less than n/6 α n. We thus partition V 0 V 0 into W 1 and W such that W i = {x V 0 V 0 : deg(x, V 3 i ) < n/6 α n}. For i = 1,, we have δ(w i, V i ) δ(w i, V 1 V ) n 6 + α n n α n n 6 + α n = n 3 + α n.

18 18 GUANTAO CHEN, KATSUHIRO OTA, AKIRA SAITO, AND YI ZHAO Let U i = V i W i for i = 1,. The above bound for δ(w i, V i ) and (15) together imply that (19) δ(u i, V i ) n 3 + α n. Since V i (1 + α )n/, for any three vertices x 1, x, x 3 U i, we have (0) deg(x 1 x x 3, V i ) 3( n 3 + α n) V i α n. Without loss of generality, assume that U 1 U. Since U 1 (1 α )n/, we have U (1 + α )n/. It suffices to find two disjoint (U 1, U )-connectors u 1...u q and v 1...v q for some q 8. In fact, if any of u 1, u, u 3 is not from V 1 (note that u 1, u, u 3 U 1 by the definition of (U 1, U )-connectors), then we find at most three new vertices x 1, x, x 3 V 1 such that x 1 x x 3 u 1 u u 3 is an ESP. For example, assume that u 1 V 1. Then we first find x 3 Γ(u 1 u 3, V 1 ), then x Γ(x 3 u, V 1 ), and finally x 1 Γ(x u 1, V 1 ) by applying (0) three times. Similar we add at most three new vertices x q+1, x q+, x q+3 from V such that u q u q 1 u q x q+1 x q+ x q+3 is an ESP. The resulting ESP x 1 x x 3 u 1...u q x q+1 x q+ x q+3 is a (V 1, V )-connector of length at most 14. We obtain the other connector similarly. The following technical lemma is the main step in our proof, we postpone its proof to the end. Lemma Let ε 1 and n be sufficiently large. Suppose that G is a graph of order n with a vertex partition U 0 U 1 U such that U 0 = c 0 n; U 1 U (1 + ε)n/; U 1 contains a subset V 1 such that δ(v 1, V 1 ) (1 ε)n/; δ(u, U ) n/3. If δ(g) n+c 0 + 6, then G[U 1, U ] contain either of the following 6-vertex subgraphs. H 1 : Two vertices x U 1, x 5 U are adjacent; two vertices x 1, x 3 Γ(x, V 1 ) and two vertices x 4, x 6 Γ(x 5, U ) form a path x 1 x 4 x 3 x 6. H : Four vertices x, x 3 U 1 and x 4, x 5 U form a copy of K, ; a vertex x 1 Γ(x 4, U 1 ), and a vertex x 6 Γ(x 3, U ). We observe that subgraphs H 1 and H given by Lemma 4.16 can be easily converted to (U 1, U )-connectors. In the case of H 1, x 1 x x 3 x 4 x 5 x 6 is an ESP and thus a (U 1, U )- connector. In the case of H, by (0), x 1, x, x 3 have a common neighbor x 7 V 1, and x 4, x 5, x 6 have a common neighbor x 8 V. The x 1x 7 x x 4 x 3 x 5 x 8 x 6 is an ESP and thus a (U 1, U )-connector. x x 5 x 1 x 4 x 1 x 4 x x 5 x 3 x 6 x 3 x 6 H 1 H Figure 3. Convert H 1 and H to ESP s

Bipartite Graph Tiling

Bipartite Graph Tiling Bipartite Graph Tiling Yi Zhao Department of Mathematics and Statistics Georgia State University Atlanta, GA 30303 December 4, 008 Abstract For each s, there exists m 0 such that the following holds for

More information

arxiv: v1 [math.co] 1 Oct 2013

arxiv: v1 [math.co] 1 Oct 2013 Tiling in bipartite graphs with asymmetric minimum degrees Andrzej Czygrinow and Louis DeBiasio November 9, 018 arxiv:1310.0481v1 [math.co] 1 Oct 013 Abstract The problem of determining the optimal minimum

More information

Proof of the (n/2 n/2 n/2) Conjecture for large n

Proof of the (n/2 n/2 n/2) Conjecture for large n Proof of the (n/2 n/2 n/2) Conjecture for large n Yi Zhao Department of Mathematics and Statistics Georgia State University Atlanta, GA 30303 September 9, 2009 Abstract A conjecture of Loebl, also known

More information

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution

More information

Compatible Hamilton cycles in Dirac graphs

Compatible Hamilton cycles in Dirac graphs Compatible Hamilton cycles in Dirac graphs Michael Krivelevich Choongbum Lee Benny Sudakov Abstract A graph is Hamiltonian if it contains a cycle passing through every vertex exactly once. A celebrated

More information

Partitioning 2-edge-colored Ore-type graphs by monochromatic cycles

Partitioning 2-edge-colored Ore-type graphs by monochromatic cycles Partitioning 2-edge-colored Ore-type graphs by monochromatic cycles János Barát MTA-ELTE Geometric and Algebraic Combinatorics Research Group barat@cs.elte.hu and Gábor N. Sárközy Alfréd Rényi Institute

More information

Improved degree conditions for 2-factors with k cycles in hamiltonian graphs

Improved degree conditions for 2-factors with k cycles in hamiltonian graphs Improved degree conditions for -factors with k cycles in hamiltonian graphs Louis DeBiasio 1,4, Michael Ferrara,, Timothy Morris, December 4, 01 Abstract In this paper, we consider conditions that ensure

More information

On the Turán number of forests

On the Turán number of forests On the Turán number of forests Bernard Lidický Hong Liu Cory Palmer April 13, 01 Abstract The Turán number of a graph H, ex(n, H, is the maximum number of edges in a graph on n vertices which does not

More information

On the Pósa-Seymour Conjecture

On the Pósa-Seymour Conjecture On the Pósa-Seymour Conjecture János Komlós, 1 Gábor N. Sárközy, 2 and Endre Szemerédi 3 1 DEPT. OF MATHEMATICS, RUTGERS UNIVERSITY, NEW BRUNSWICK, NJ 08903 2 DEPT. OF COMPUTER SCIENCE, WORCESTER POLYTECHNIC

More information

Graphs with large maximum degree containing no odd cycles of a given length

Graphs with large maximum degree containing no odd cycles of a given length Graphs with large maximum degree containing no odd cycles of a given length Paul Balister Béla Bollobás Oliver Riordan Richard H. Schelp October 7, 2002 Abstract Let us write f(n, ; C 2k+1 ) for the maximal

More information

arxiv: v1 [math.co] 1 Aug 2013

arxiv: v1 [math.co] 1 Aug 2013 Semi-degree threshold for anti-directed Hamiltonian cycles Louis DeBiasio and Theodore Molla May 11, 014 arxiv:1308.069v1 [math.co] 1 Aug 013 Abstract In 1960 Ghouila-Houri extended Dirac s theorem to

More information

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and This article appeared in a journal published by Elsevier The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution

More information

Chromatic number, clique subdivisions, and the conjectures of Hajós and Erdős-Fajtlowicz

Chromatic number, clique subdivisions, and the conjectures of Hajós and Erdős-Fajtlowicz Chromatic number, clique subdivisions, and the conjectures of Hajós and Erdős-Fajtlowicz Jacob Fox Choongbum Lee Benny Sudakov Abstract For a graph G, let χ(g) denote its chromatic number and σ(g) denote

More information

arxiv: v2 [math.co] 20 Jun 2018

arxiv: v2 [math.co] 20 Jun 2018 ON ORDERED RAMSEY NUMBERS OF BOUNDED-DEGREE GRAPHS MARTIN BALKO, VÍT JELÍNEK, AND PAVEL VALTR arxiv:1606.0568v [math.co] 0 Jun 018 Abstract. An ordered graph is a pair G = G, ) where G is a graph and is

More information

1. Introduction Given k 2, a k-uniform hypergraph (in short, k-graph) consists of a vertex set V and an edge set E ( V

1. Introduction Given k 2, a k-uniform hypergraph (in short, k-graph) consists of a vertex set V and an edge set E ( V MINIMUM VERTEX DEGREE THRESHOLD FOR C 4-TILING JIE HAN AND YI ZHAO Abstract. We prove that the vertex degree threshold for tiling C4 (the - uniform hypergraph with four vertices and two triples in a -uniform

More information

Packing and decomposition of graphs with trees

Packing and decomposition of graphs with trees Packing and decomposition of graphs with trees Raphael Yuster Department of Mathematics University of Haifa-ORANIM Tivon 36006, Israel. e-mail: raphy@math.tau.ac.il Abstract Let H be a tree on h 2 vertices.

More information

Disjoint Hamiltonian Cycles in Bipartite Graphs

Disjoint Hamiltonian Cycles in Bipartite Graphs Disjoint Hamiltonian Cycles in Bipartite Graphs Michael Ferrara 1, Ronald Gould 1, Gerard Tansey 1 Thor Whalen Abstract Let G = (X, Y ) be a bipartite graph and define σ (G) = min{d(x) + d(y) : xy / E(G),

More information

THE STRUCTURE OF ALMOST ALL GRAPHS IN A HEREDITARY PROPERTY

THE STRUCTURE OF ALMOST ALL GRAPHS IN A HEREDITARY PROPERTY THE STRUCTURE OF ALMOST ALL GRAPHS IN A HEREDITARY PROPERTY NOGA ALON, JÓZSEF BALOGH, BÉLA BOLLOBÁS, AND ROBERT MORRIS Abstract. A hereditary property of graphs is a collection of graphs which is closed

More information

2-factors of bipartite graphs with asymmetric minimum degrees

2-factors of bipartite graphs with asymmetric minimum degrees 2-factors of bipartite graphs with asymmetric minimum degrees Andrzej Czygrinow, Louis DeBiasio, H. A. Kierstead Department of Mathematics and Statistics Arizona State University Tempe, AZ 85287 Submitted

More information

Proof of a Tiling Conjecture of Komlós

Proof of a Tiling Conjecture of Komlós Proof of a Tiling Conjecture of Komlós Ali Shooufandeh, 1 Yi Zhao 2 1 Department of Computer Science, Drexel University, Philadelphia, Pennsylvania 19104 2 Department of Mathematics, Statistics, and Computer

More information

ON THE STRUCTURE OF ORIENTED GRAPHS AND DIGRAPHS WITH FORBIDDEN TOURNAMENTS OR CYCLES

ON THE STRUCTURE OF ORIENTED GRAPHS AND DIGRAPHS WITH FORBIDDEN TOURNAMENTS OR CYCLES ON THE STRUCTURE OF ORIENTED GRAPHS AND DIGRAPHS WITH FORBIDDEN TOURNAMENTS OR CYCLES DANIELA KÜHN, DERYK OSTHUS, TIMOTHY TOWNSEND, YI ZHAO Abstract. Motivated by his work on the classification of countable

More information

Rao s degree sequence conjecture

Rao s degree sequence conjecture Rao s degree sequence conjecture Maria Chudnovsky 1 Columbia University, New York, NY 10027 Paul Seymour 2 Princeton University, Princeton, NJ 08544 July 31, 2009; revised December 10, 2013 1 Supported

More information

EXACT MINIMUM CODEGREE THRESHOLD FOR K 4 -FACTORS. MSC2000: 5C35, 5C65, 5C70. Keywords: Tiling, Hypergraphs, Absorbing method.

EXACT MINIMUM CODEGREE THRESHOLD FOR K 4 -FACTORS. MSC2000: 5C35, 5C65, 5C70. Keywords: Tiling, Hypergraphs, Absorbing method. EXACT MINIMUM CODEGREE THRESHOLD FOR K 4 -FACTORS JIE HAN, ALLAN LO, ANDREW TREGLOWN AND YI ZHAO Abstract. Given hypergraphs F and H, an F -factor in H is a set of vertex-disjoint copies of F which cover

More information

Discrete Mathematics

Discrete Mathematics Discrete Mathematics 309 (009) 3811 380 Contents lists available at ScienceDirect Discrete Mathematics journal homepage: www.elsevier.com/locate/disc Disjoint hamiltonian cycles in bipartite graphs Michael

More information

Stability for vertex cycle covers

Stability for vertex cycle covers József Balogh, Frank Mousset, Jozef Skokan Stability for vertex cycle covers Article (Published version) (Refereed) Original citation: Balogh, József and Mousset, Frank and Skokan, Jozef (2017) Stability

More information

MATCHINGS IN k-partite k-uniform HYPERGRAPHS

MATCHINGS IN k-partite k-uniform HYPERGRAPHS MATCHINGS IN k-partite k-uniform HYPERGRAPHS JIE HAN, CHUANYUN ZANG, AND YI ZHAO Abstract. For k 3 and ɛ > 0, let H be a k-partite k-graph with parts V 1,..., V k each of size n, where n is sufficiently

More information

On the Regularity Method

On the Regularity Method On the Regularity Method Gábor N. Sárközy 1 Worcester Polytechnic Institute USA 2 Computer and Automation Research Institute of the Hungarian Academy of Sciences Budapest, Hungary Co-authors: P. Dorbec,

More information

Monochromatic subgraphs of 2-edge-colored graphs

Monochromatic subgraphs of 2-edge-colored graphs Monochromatic subgraphs of 2-edge-colored graphs Luke Nelsen, Miami University June 10, 2014 Abstract Lehel conjectured that for all n, any 2-edge-coloring of K n admits a partition of the vertex set into

More information

The Turán number of sparse spanning graphs

The Turán number of sparse spanning graphs The Turán number of sparse spanning graphs Noga Alon Raphael Yuster Abstract For a graph H, the extremal number ex(n, H) is the maximum number of edges in a graph of order n not containing a subgraph isomorphic

More information

Subdivisions of a large clique in C 6 -free graphs

Subdivisions of a large clique in C 6 -free graphs Subdivisions of a large clique in C 6 -free graphs József Balogh Hong Liu Maryam Sharifzadeh October 8, 2014 Abstract Mader conjectured that every C 4 -free graph has a subdivision of a clique of order

More information

On the number of cycles in a graph with restricted cycle lengths

On the number of cycles in a graph with restricted cycle lengths On the number of cycles in a graph with restricted cycle lengths Dániel Gerbner, Balázs Keszegh, Cory Palmer, Balázs Patkós arxiv:1610.03476v1 [math.co] 11 Oct 2016 October 12, 2016 Abstract Let L be a

More information

MINIMALLY NON-PFAFFIAN GRAPHS

MINIMALLY NON-PFAFFIAN GRAPHS MINIMALLY NON-PFAFFIAN GRAPHS SERGUEI NORINE AND ROBIN THOMAS Abstract. We consider the question of characterizing Pfaffian graphs. We exhibit an infinite family of non-pfaffian graphs minimal with respect

More information

VERTEX DEGREE SUMS FOR PERFECT MATCHINGS IN 3-UNIFORM HYPERGRAPHS

VERTEX DEGREE SUMS FOR PERFECT MATCHINGS IN 3-UNIFORM HYPERGRAPHS VERTEX DEGREE SUMS FOR PERFECT MATCHINGS IN 3-UNIFORM HYPERGRAPHS YI ZHANG, YI ZHAO, AND MEI LU Abstract. We determine the minimum degree sum of two adjacent vertices that ensures a perfect matching in

More information

The regularity method and blow-up lemmas for sparse graphs

The regularity method and blow-up lemmas for sparse graphs The regularity method and blow-up lemmas for sparse graphs Y. Kohayakawa (São Paulo) SPSAS Algorithms, Combinatorics and Optimization University of São Paulo July 2016 Partially supported CAPES/DAAD (430/15),

More information

Spanning Paths in Infinite Planar Graphs

Spanning Paths in Infinite Planar Graphs Spanning Paths in Infinite Planar Graphs Nathaniel Dean AT&T, ROOM 2C-415 600 MOUNTAIN AVENUE MURRAY HILL, NEW JERSEY 07974-0636, USA Robin Thomas* Xingxing Yu SCHOOL OF MATHEMATICS GEORGIA INSTITUTE OF

More information

On the bandwidth conjecture for 3-colourable graphs

On the bandwidth conjecture for 3-colourable graphs On the bandwidth conjecture for 3-colourable graphs Julia Böttcher Technische Universität München Symposium on Discrete Algorithms, January 2007, New Orleans (joint work with Mathias Schacht & Anusch Taraz)

More information

New lower bounds for hypergraph Ramsey numbers

New lower bounds for hypergraph Ramsey numbers New lower bounds for hypergraph Ramsey numbers Dhruv Mubayi Andrew Suk Abstract The Ramsey number r k (s, n) is the minimum N such that for every red-blue coloring of the k-tuples of {1,..., N}, there

More information

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution

More information

Graph Packing - Conjectures and Results

Graph Packing - Conjectures and Results Graph Packing p.1/23 Graph Packing - Conjectures and Results Hemanshu Kaul kaul@math.iit.edu www.math.iit.edu/ kaul. Illinois Institute of Technology Graph Packing p.2/23 Introduction Let G 1 = (V 1,E

More information

Cycle lengths in sparse graphs

Cycle lengths in sparse graphs Cycle lengths in sparse graphs Benny Sudakov Jacques Verstraëte Abstract Let C(G) denote the set of lengths of cycles in a graph G. In the first part of this paper, we study the minimum possible value

More information

arxiv: v1 [math.co] 13 May 2016

arxiv: v1 [math.co] 13 May 2016 GENERALISED RAMSEY NUMBERS FOR TWO SETS OF CYCLES MIKAEL HANSSON arxiv:1605.04301v1 [math.co] 13 May 2016 Abstract. We determine several generalised Ramsey numbers for two sets Γ 1 and Γ 2 of cycles, in

More information

Corrádi and Hajnal s theorem for sparse random graphs

Corrádi and Hajnal s theorem for sparse random graphs Corrádi and Hajnal s theorem for sparse random graphs József Balogh Choongbum Lee Wojciech Samotij Abstract In this paper we extend a classical theorem of Corrádi and Hajnal into the setting of sparse

More information

arxiv: v1 [math.co] 23 Nov 2015

arxiv: v1 [math.co] 23 Nov 2015 arxiv:1511.07306v1 [math.co] 23 Nov 2015 RAMSEY NUMBERS OF TREES AND UNICYCLIC GRAPHS VERSUS FANS MATTHEW BRENNAN Abstract. The generalized Ramsey number R(H, K) is the smallest positive integer n such

More information

HAMILTONIAN CYCLES AVOIDING SETS OF EDGES IN A GRAPH

HAMILTONIAN CYCLES AVOIDING SETS OF EDGES IN A GRAPH HAMILTONIAN CYCLES AVOIDING SETS OF EDGES IN A GRAPH MICHAEL J. FERRARA, MICHAEL S. JACOBSON UNIVERSITY OF COLORADO DENVER DENVER, CO 8017 ANGELA HARRIS UNIVERSITY OF WISCONSIN-WHITEWATER WHITEWATER, WI

More information

Disjoint Subgraphs in Sparse Graphs 1

Disjoint Subgraphs in Sparse Graphs 1 Disjoint Subgraphs in Sparse Graphs 1 Jacques Verstraëte Department of Pure Mathematics and Mathematical Statistics Centre for Mathematical Sciences Wilberforce Road Cambridge CB3 OWB, UK jbav2@dpmms.cam.ac.uk

More information

Tiling on multipartite graphs

Tiling on multipartite graphs Tiling on multipartite graphs Ryan Martin Mathematics Department Iowa State University rymartin@iastate.edu SIAM Minisymposium on Graph Theory Joint Mathematics Meetings San Francisco, CA Ryan Martin (Iowa

More information

Stability of the path-path Ramsey number

Stability of the path-path Ramsey number Worcester Polytechnic Institute Digital WPI Computer Science Faculty Publications Department of Computer Science 9-12-2008 Stability of the path-path Ramsey number András Gyárfás Computer and Automation

More information

On colorability of graphs with forbidden minors along paths and circuits

On colorability of graphs with forbidden minors along paths and circuits On colorability of graphs with forbidden minors along paths and circuits Elad Horev horevel@cs.bgu.ac.il Department of Computer Science Ben-Gurion University of the Negev Beer-Sheva 84105, Israel Abstract.

More information

Perfect matchings in highly cyclically connected regular graphs

Perfect matchings in highly cyclically connected regular graphs Perfect matchings in highly cyclically connected regular graphs arxiv:1709.08891v1 [math.co] 6 Sep 017 Robert Lukot ka Comenius University, Bratislava lukotka@dcs.fmph.uniba.sk Edita Rollová University

More information

Star Versus Two Stripes Ramsey Numbers and a Conjecture of Schelp

Star Versus Two Stripes Ramsey Numbers and a Conjecture of Schelp Combinatorics, Probability and Computing (2012) 21, 179 186. c Cambridge University Press 2012 doi:10.1017/s096358311000599 Star Versus Two Stripes Ramsey Numbers and a Conjecture of Schelp ANDRÁS GYÁRFÁS1

More information

arxiv: v2 [math.co] 7 Jan 2016

arxiv: v2 [math.co] 7 Jan 2016 Global Cycle Properties in Locally Isometric Graphs arxiv:1506.03310v2 [math.co] 7 Jan 2016 Adam Borchert, Skylar Nicol, Ortrud R. Oellermann Department of Mathematics and Statistics University of Winnipeg,

More information

Independent Transversals in r-partite Graphs

Independent Transversals in r-partite Graphs Independent Transversals in r-partite Graphs Raphael Yuster Department of Mathematics Raymond and Beverly Sackler Faculty of Exact Sciences Tel Aviv University, Tel Aviv, Israel Abstract Let G(r, n) denote

More information

Out-colourings of Digraphs

Out-colourings of Digraphs Out-colourings of Digraphs N. Alon J. Bang-Jensen S. Bessy July 13, 2017 Abstract We study vertex colourings of digraphs so that no out-neighbourhood is monochromatic and call such a colouring an out-colouring.

More information

Upper Bounds of Dynamic Chromatic Number

Upper Bounds of Dynamic Chromatic Number Upper Bounds of Dynamic Chromatic Number Hong-Jian Lai, Bruce Montgomery and Hoifung Poon Department of Mathematics West Virginia University, Morgantown, WV 26506-6310 June 22, 2000 Abstract A proper vertex

More information

1.3 Vertex Degrees. Vertex Degree for Undirected Graphs: Let G be an undirected. Vertex Degree for Digraphs: Let D be a digraph and y V (D).

1.3 Vertex Degrees. Vertex Degree for Undirected Graphs: Let G be an undirected. Vertex Degree for Digraphs: Let D be a digraph and y V (D). 1.3. VERTEX DEGREES 11 1.3 Vertex Degrees Vertex Degree for Undirected Graphs: Let G be an undirected graph and x V (G). The degree d G (x) of x in G: the number of edges incident with x, each loop counting

More information

Antoni Marczyk A NOTE ON ARBITRARILY VERTEX DECOMPOSABLE GRAPHS

Antoni Marczyk A NOTE ON ARBITRARILY VERTEX DECOMPOSABLE GRAPHS Opuscula Mathematica Vol. 6 No. 1 006 Antoni Marczyk A NOTE ON ARBITRARILY VERTEX DECOMPOSABLE GRAPHS Abstract. A graph G of order n is said to be arbitrarily vertex decomposable if for each sequence (n

More information

4 Packing T-joins and T-cuts

4 Packing T-joins and T-cuts 4 Packing T-joins and T-cuts Introduction Graft: A graft consists of a connected graph G = (V, E) with a distinguished subset T V where T is even. T-cut: A T -cut of G is an edge-cut C which separates

More information

Graph Theory Notes. January 15, H. A. Kierstead

Graph Theory Notes. January 15, H. A. Kierstead Graph Theory Notes January 15, 2019 H. A. Kierstead CHAPTER 1 Introduction This is an on going project. There will be changes to the material not yet covered, especially just before each lecture. There

More information

The expansion of random regular graphs

The expansion of random regular graphs The expansion of random regular graphs David Ellis Introduction Our aim is now to show that for any d 3, almost all d-regular graphs on {1, 2,..., n} have edge-expansion ratio at least c d d (if nd is

More information

Finding Hamilton cycles in robustly expanding digraphs

Finding Hamilton cycles in robustly expanding digraphs Journal of Graph Algorithms and Applications http://jgaa.info/ vol. 16, no. 2, pp. 335 358 (2012) Finding Hamilton cycles in robustly expanding digraphs Demetres Christofides 1 Peter Keevash 1 Daniela

More information

Odd independent transversals are odd

Odd independent transversals are odd Odd independent transversals are odd Penny Haxell Tibor Szabó Dedicated to Béla Bollobás on the occasion of his 60th birthday Abstract We put the final piece into a puzzle first introduced by Bollobás,

More information

Relating minimum degree and the existence of a k-factor

Relating minimum degree and the existence of a k-factor Relating minimum degree and the existence of a k-factor Stephen G Hartke, Ryan Martin, and Tyler Seacrest October 6, 010 Abstract A k-factor in a graph G is a spanning regular subgraph in which every vertex

More information

Equitable coloring of random graphs

Equitable coloring of random graphs Michael Krivelevich 1, Balázs Patkós 2 1 Tel Aviv University, Tel Aviv, Israel 2 Central European University, Budapest, Hungary Phenomena in High Dimensions, Samos 2007 A set of vertices U V (G) is said

More information

Induced subgraphs of prescribed size

Induced subgraphs of prescribed size Induced subgraphs of prescribed size Noga Alon Michael Krivelevich Benny Sudakov Abstract A subgraph of a graph G is called trivial if it is either a clique or an independent set. Let q(g denote the maximum

More information

Graph coloring, perfect graphs

Graph coloring, perfect graphs Lecture 5 (05.04.2013) Graph coloring, perfect graphs Scribe: Tomasz Kociumaka Lecturer: Marcin Pilipczuk 1 Introduction to graph coloring Definition 1. Let G be a simple undirected graph and k a positive

More information

Advanced Topics in Discrete Math: Graph Theory Fall 2010

Advanced Topics in Discrete Math: Graph Theory Fall 2010 21-801 Advanced Topics in Discrete Math: Graph Theory Fall 2010 Prof. Andrzej Dudek notes by Brendan Sullivan October 18, 2010 Contents 0 Introduction 1 1 Matchings 1 1.1 Matchings in Bipartite Graphs...................................

More information

Bulletin of the Iranian Mathematical Society

Bulletin of the Iranian Mathematical Society ISSN: 117-6X (Print) ISSN: 1735-8515 (Online) Bulletin of the Iranian Mathematical Society Vol. 4 (14), No. 6, pp. 1491 154. Title: The locating chromatic number of the join of graphs Author(s): A. Behtoei

More information

Set-orderedness as a generalization of k-orderedness and cyclability

Set-orderedness as a generalization of k-orderedness and cyclability Set-orderedness as a generalization of k-orderedness and cyclability Keishi Ishii Kenta Ozeki National Institute of Informatics, Tokyo 101-8430, Japan e-mail: ozeki@nii.ac.jp Kiyoshi Yoshimoto Department

More information

On (δ, χ)-bounded families of graphs

On (δ, χ)-bounded families of graphs On (δ, χ)-bounded families of graphs András Gyárfás Computer and Automation Research Institute Hungarian Academy of Sciences Budapest, P.O. Box 63 Budapest, Hungary, H-1518 gyarfas@sztaki.hu Manouchehr

More information

The Algorithmic Aspects of the Regularity Lemma

The Algorithmic Aspects of the Regularity Lemma The Algorithmic Aspects of the Regularity Lemma N. Alon R. A. Duke H. Lefmann V. Rödl R. Yuster Abstract The Regularity Lemma of Szemerédi is a result that asserts that every graph can be partitioned in

More information

Sparse halves in dense triangle-free graphs

Sparse halves in dense triangle-free graphs Sparse halves in dense triangle-free graphs Sergey Norin and Liana Yepremyan 2 Department of Mathematics and Statistics, McGill University 2 School of Computer Science, McGill University arxiv:3.588v2

More information

AN IMPROVED ERROR TERM FOR MINIMUM H-DECOMPOSITIONS OF GRAPHS. 1. Introduction We study edge decompositions of a graph G into disjoint copies of

AN IMPROVED ERROR TERM FOR MINIMUM H-DECOMPOSITIONS OF GRAPHS. 1. Introduction We study edge decompositions of a graph G into disjoint copies of AN IMPROVED ERROR TERM FOR MINIMUM H-DECOMPOSITIONS OF GRAPHS PETER ALLEN*, JULIA BÖTTCHER*, AND YURY PERSON Abstract. We consider partitions of the edge set of a graph G into copies of a fixed graph H

More information

Smaller subgraphs of minimum degree k

Smaller subgraphs of minimum degree k Smaller subgraphs of minimum degree k Frank Mousset Institute of Theoretical Computer Science ETH Zürich 8092 Zürich, Switzerland moussetf@inf.ethz.ch Nemanja Škorić Institute of Theoretical Computer Science

More information

Multi-coloring and Mycielski s construction

Multi-coloring and Mycielski s construction Multi-coloring and Mycielski s construction Tim Meagher Fall 2010 Abstract We consider a number of related results taken from two papers one by W. Lin [1], and the other D. C. Fisher[2]. These articles

More information

Large topological cliques in graphs without a 4-cycle

Large topological cliques in graphs without a 4-cycle Large topological cliques in graphs without a 4-cycle Daniela Kühn Deryk Osthus Abstract Mader asked whether every C 4 -free graph G contains a subdivision of a complete graph whose order is at least linear

More information

CYCLE STRUCTURES IN GRAPHS. Angela K. Harris. A thesis submitted to the. University of Colorado Denver. in partial fulfillment

CYCLE STRUCTURES IN GRAPHS. Angela K. Harris. A thesis submitted to the. University of Colorado Denver. in partial fulfillment CYCLE STRUCTURES IN GRAPHS by Angela K. Harris Master of Science, University of South Alabama, 003 A thesis submitted to the University of Colorado Denver in partial fulfillment of the requirements for

More information

THE ROBUST COMPONENT STRUCTURE OF DENSE REGULAR GRAPHS AND APPLICATIONS

THE ROBUST COMPONENT STRUCTURE OF DENSE REGULAR GRAPHS AND APPLICATIONS THE ROBUST COMPONENT STRUCTURE OF DENSE REGULAR GRAPHS AND APPLICATIONS DANIELA KÜHN, ALLAN LO, DERYK OSTHUS AND KATHERINE STADEN Abstract. In this paper, we study the large-scale structure of dense regular

More information

The number of edge colorings with no monochromatic cliques

The number of edge colorings with no monochromatic cliques The number of edge colorings with no monochromatic cliques Noga Alon József Balogh Peter Keevash Benny Sudaov Abstract Let F n, r, ) denote the maximum possible number of distinct edge-colorings of a simple

More information

A Generalization of a result of Catlin: 2-factors in line graphs

A Generalization of a result of Catlin: 2-factors in line graphs AUSTRALASIAN JOURNAL OF COMBINATORICS Volume 72(2) (2018), Pages 164 184 A Generalization of a result of Catlin: 2-factors in line graphs Ronald J. Gould Emory University Atlanta, Georgia U.S.A. rg@mathcs.emory.edu

More information

Graph Theory Notes. August 8, H. A. Kierstead

Graph Theory Notes. August 8, H. A. Kierstead Graph Theory Notes August 8, 2018 H. A. Kierstead CHAPTER 1 Introduction This is an on going project. There will be changes to the material not yet covered, especially just before each lecture. There

More information

An asymptotic multipartite Kühn-Osthus theorem

An asymptotic multipartite Kühn-Osthus theorem An asymptotic multipartite Kühn-Osthus theorem Ryan R. Martin 1 Richard Mycroft 2 Jozef Skokan 3 1 Iowa State University 2 University of Birmingham 3 London School of Economics 08 August 2017 Algebraic

More information

Ramsey-type problem for an almost monochromatic K 4

Ramsey-type problem for an almost monochromatic K 4 Ramsey-type problem for an almost monochromatic K 4 Jacob Fox Benny Sudakov Abstract In this short note we prove that there is a constant c such that every k-edge-coloring of the complete graph K n with

More information

On directed versions of the Corrádi-Hajnal Corollary

On directed versions of the Corrádi-Hajnal Corollary On directed versions of the Corrádi-Hajnal Corollary Andrzej Czygrinow H. A. Kierstead heodore Molla October 4, 01 Abstract For k N, Corrádi and Hajnal proved that every graph G on k vertices with minimum

More information

Extremal H-colorings of graphs with fixed minimum degree

Extremal H-colorings of graphs with fixed minimum degree Extremal H-colorings of graphs with fixed minimum degree John Engbers July 18, 2014 Abstract For graphs G and H, a homomorphism from G to H, or H-coloring of G, is a map from the vertices of G to the vertices

More information

Strongly chordal and chordal bipartite graphs are sandwich monotone

Strongly chordal and chordal bipartite graphs are sandwich monotone Strongly chordal and chordal bipartite graphs are sandwich monotone Pinar Heggernes Federico Mancini Charis Papadopoulos R. Sritharan Abstract A graph class is sandwich monotone if, for every pair of its

More information

Graph Theory. Thomas Bloom. February 6, 2015

Graph Theory. Thomas Bloom. February 6, 2015 Graph Theory Thomas Bloom February 6, 2015 1 Lecture 1 Introduction A graph (for the purposes of these lectures) is a finite set of vertices, some of which are connected by a single edge. Most importantly,

More information

Decompositions of graphs into cycles with chords

Decompositions of graphs into cycles with chords Decompositions of graphs into cycles with chords Paul Balister Hao Li Richard Schelp May 22, 2017 In memory of Dick Schelp, who passed away shortly after the submission of this paper Abstract We show that

More information

THE STRUCTURE AND EXISTENCE OF 2-FACTORS IN ITERATED LINE GRAPHS

THE STRUCTURE AND EXISTENCE OF 2-FACTORS IN ITERATED LINE GRAPHS Discussiones Mathematicae Graph Theory 27 (2007) 507 526 THE STRUCTURE AND EXISTENCE OF 2-FACTORS IN ITERATED LINE GRAPHS Michael Ferrara Department of Theoretical and Applied Mathematics The University

More information

Degree Sum Conditions and Vertex-Disjoint Cycles in a Graph

Degree Sum Conditions and Vertex-Disjoint Cycles in a Graph Degree Sum Conditions and Vertex-Disjoint Cycles in a Graph Shinya Fujita Hajime Matsumura Department of Mathematics Keio University Yokohama 223-8522, Japan shinyaa@comb.math.keio.ac.jp musimaru@comb.math.keio.ac.jp

More information

arxiv: v1 [math.co] 6 Apr 2018

arxiv: v1 [math.co] 6 Apr 2018 New Perspectives on Neighborhood-Prime Labelings of Graphs John Asplund Department of Technology and Mathematics, Dalton State College, Dalton, GA 30720, USA jasplund@daltonstate.edu arxiv:1804.02473v1

More information

Properly colored Hamilton cycles in edge colored complete graphs

Properly colored Hamilton cycles in edge colored complete graphs Properly colored Hamilton cycles in edge colored complete graphs N. Alon G. Gutin Dedicated to the memory of Paul Erdős Abstract It is shown that for every ɛ > 0 and n > n 0 (ɛ), any complete graph K on

More information

HAMILTONIAN PROPERTIES OF TRIANGULAR GRID GRAPHS. 1. Introduction

HAMILTONIAN PROPERTIES OF TRIANGULAR GRID GRAPHS. 1. Introduction HAMILTONIAN PROPERTIES OF TRIANGULAR GRID GRAPHS VALERY S. GORDON, YURY L. ORLOVICH, FRANK WERNER Abstract. A triangular grid graph is a finite induced subgraph of the infinite graph associated with the

More information

A necessary and sufficient condition for the existence of a spanning tree with specified vertices having large degrees

A necessary and sufficient condition for the existence of a spanning tree with specified vertices having large degrees A necessary and sufficient condition for the existence of a spanning tree with specified vertices having large degrees Yoshimi Egawa Department of Mathematical Information Science, Tokyo University of

More information

arxiv: v2 [math.co] 29 Oct 2017

arxiv: v2 [math.co] 29 Oct 2017 arxiv:1404.3385v2 [math.co] 29 Oct 2017 A proof for a conjecture of Gyárfás, Lehel, Sárközy and Schelp on Berge-cycles G.R. Omidi Department of Mathematical Sciences, Isfahan University of Technology,

More information

On the Chromatic Thresholds of Hypergraphs

On the Chromatic Thresholds of Hypergraphs On the Chromatic Thresholds of Hypergraphs József Balogh Jane Butterfield Ping Hu John Lenz Dhruv Mubayi September 14, 2011 Abstract Let F be a family of r-uniform hypergraphs. The chromatic threshold

More information

Tripartite Ramsey Numbers for Paths

Tripartite Ramsey Numbers for Paths Tripartite Ramsey Numbers for Paths András Gyárfás, Miklós Ruszinkó, Gábor N. Sárközy,,2 and Endre Szemerédi 3 COMPUTER AND AUTOMATION RESEARCH INSTITUTE HUNGARIAN ACADEMY OF SCIENCES BUDAPEST, P.O. BOX

More information

Removal Lemmas with Polynomial Bounds

Removal Lemmas with Polynomial Bounds Removal Lemmas with Polynomial Bounds Lior Gishboliner Asaf Shapira Abstract A common theme in many extremal problems in graph theory is the relation between local and global properties of graphs. One

More information

A Generalized Turán Problem and its Applications

A Generalized Turán Problem and its Applications A Generalized Turán Problem and its Applications Lior Gishboliner Asaf Shapira Abstract The investigation of conditions guaranteeing the appearance of cycles of certain lengths is one of the most well-studied

More information

An approximate version of Hadwiger s conjecture for claw-free graphs

An approximate version of Hadwiger s conjecture for claw-free graphs An approximate version of Hadwiger s conjecture for claw-free graphs Maria Chudnovsky Columbia University, New York, NY 10027, USA and Alexandra Ovetsky Fradkin Princeton University, Princeton, NJ 08544,

More information

Painting Squares in 2 1 Shades

Painting Squares in 2 1 Shades Painting Squares in 1 Shades Daniel W. Cranston Landon Rabern May 1, 014 Abstract Cranston and Kim conjectured that if G is a connected graph with maximum degree and G is not a Moore Graph, then χ l (G

More information