A new sufficient condition for a Digraph to be Hamiltonian-A proof of Manoussakis Conjecture
Samvel Kh. Darbinyan

TL;DR
This paper proves Manoussakis's conjecture that a 2-strongly connected digraph with a specific degree sum condition is Hamiltonian, and additionally shows such digraphs contain cycles of all lengths under certain degree constraints.
Contribution
It confirms Manoussakis's conjecture and establishes the existence of cycles of all lengths in certain degree-satisfying digraphs.
Findings
Proves the conjecture that degree sum conditions imply Hamiltonicity.
Shows existence of cycles of all lengths in specific digraphs.
Provides a new sufficient condition for Hamiltonian digraphs.
Abstract
Y. Manoussakis (J. Graph Theory 16, 1992, 51-59) proposed the following conjecture. \noindent\textbf{Conjecture}. {\it Let be a 2-strongly connected digraph of order such that for all distinct pairs of non-adjacent vertices , and , , we have . Then is Hamiltonian.} In this paper, we confirm this conjecture. Moreover, we prove that if a digraph satisfies the conditions of this conjecture and has a pair of non-adjacent vertices such that , then contains cycles of all lengths .
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Graph Theory Research · graph theory and CDMA systems · Interconnection Networks and Systems
\publicationdetails
2220204126086
A new sufficient condition
for a digraph to be HamiltonianA proof of Manoussakis conjecture
Samvel Kh. Darbinyan
Yerevan, Armenia
Institute for Informatics and Automation Problems of NAS RA
(2020-02-11; 2020-07-21,2020-11-05; 2020-11-05)
Abstract
Y. Manoussakis (J. Graph Theory 16, 1992, 51-59) proposed the following conjecture.
Conjecture. Let be a 2-strongly connected digraph of order such that for all distinct pairs of non-adjacent vertices , and , , we have . Then is Hamiltonian.
In this paper, we confirm this conjecture. Moreover, we prove that if a digraph satisfies the conditions of this conjecture and has a pair of non-adjacent vertices such that , then contains cycles of all lengths .
keywords:
digraph, hamiltonian cycle, strong digraph, pancyclic digraph
1 Introduction
In this paper, we consider finite digraphs (directed graphs) without loops and multiple arcs. Every cycle and path are assumed simple and directed; its length is the number of its arcs. A digraph is Hamiltonian if it contains a cycle passing through all the vertices of . There are many conditions that guarantee that a digraph is Hamiltonian (see, e.g., Bang-Jensen and Gutin (Springer-Verlag, London, 2000), Bermond and Thomassen (1981), Kühn and Osthus (2012), Manoussakis (1992), Meyniel (1973)). Manoussakis (1992) the following theorem was proved.
Theorem 1.1 (Manoussakis (1992)). *Let be a strong digraph of order . Suppose that satisfies the following condition for every triple such that and are non-adjacent: If there is no arc from to , then . If there is no arc from to , then . Then is Hamiltonian.
Definition 1.2. *Let be a digraph of order . We say that satisfies condition when for all distinct pairs of non-adjacent vertices and .
Manoussakis (1992) proposed the following conjecture. This conjecture is an extension of Theorem 1.1.
Conjecture 1.3 (Manoussakis (1992)). *Let be a 2-strong digraph of order such that for all distinct pairs of non-adjacent vertices , and , we have . Then is Hamiltonian.
Manoussakis (1992) gave an example, which showed that if this conjecture is true, then the minimum degree condition is sharp. Notice that another examples can be found in a paper by Darbinyan (1983), where for any two integers and , the author constructed a family of -strong digraphs of order with minimum degree , which are not Hamiltonian. This result improves a conjecture of Thomassen (see Bermond and Thomassen (1981) Conjecture 1.4.1: Every 2-strong -regular digraph of order , except and , is Hamiltonian). Moreover, when , then from these digraphs we can obtain -strong non-Hamiltonian digraphs of order with minimum degree equal to and the minimal semi-degrees equal to . Thus, if in Conjecture 1.3 we replace with , then for every there are many digraphs of order with high connectivity and high semi-degrees, for which Conjecture 1.3 is not true.
The cycle factor in a digraph is a collection of pairwise vertex disjoint cycles such that . It is clear that the existence of a cycle factor in a digraph is a necessary condition for a digraph to be Hamiltonian. The following theorem gives a necessary and sufficient condition for the existence of a cycle factor in a digraph.
Theorem 1.4 (Yeo (1999)). *Let be a digraph. Then has a cycle factor if and only if cannot be partitioned into subsets , , , such that , and is an independent set.
Using theorem Theorem 1.4, it is not difficult to construct 2-strong digraphs satisfying the condition that for every distinct pairs , of non-adjacent vertices, but these digraphs do not even contain a cycle factor.
Thomassen suggested (see Bermond and Thomassen (1981)) the following two conjectures:
1. Conjecture 1.6.7. Every 3-strong digraph of order and with minimum degree at least is strongly Hamiltonian-connected.
2. Conjecture 1.6.8. Let be a 4-strong digraph of order such that the sum of the degrees of any pair of non-adjacent vertices is at least . Then is strongly Hamiltonian-connected.
Investigating these conjectures, Darbinyan (1990) disproved the first conjecture (proving that for every integer there exists a 3-strong non-strongly Hamiltonian-connected digraph of order with the minimum degree at least ) and for the second proved the following two theorems.
Theorem 1.5. Any -strong () digraph of order satisfying the condition that the sum of degrees of any pair of non-adjacent vertices at least , where is some vertex in , is Hamiltonian if and only if any -strong digraph of order satisfying the condition that the sum of degrees of any pair of non-adjacent vertices at least is strongly Hamiltonian-connected.
Theorem 1.6. *Let be a strong digraph of order . Suppose that for every pair of non-adjacent vertices , where is some vertex of . Then contains a cycle of length at least .
It is easy to see that if a digraph satisfies the condition , then it contains at most one pair of non-adjacent vertices such that . From this and Theorem 1.6, the following corollary immediately follows.
Corollary 1.7. *Let be a strong digraph of order satisfying condition . Then contains a cycle of length at least (in particular, contains a Hamiltonian path).
Corollary 1.7 was also later proved by Ning (2015).
It is worth to note that Darbinyan (2017), Darbinyan (2015) and Darbinyan and Karapetyan (2015) studied some properties in digraphs with the conditions of Theorem 1.1. They obtained the following results (in first two results is a digraph of order satisfying the degree condition of Theorem 1.1).
(i) (Darbinyan and Karapetyan (2015)). If is strong, then it contains a cycle of length or is isomorphic to the complete bipartite digraph .
(ii) (Darbinyan (2015)). If is strong, then it contains a Hamiltonian path in which the initial vertex dominates the terminal vertex or is isomorphic to one tournament of order 5.
(iii) (Darbinyan (2017)). Let be a digraph of order and let be a non-empty subset of . Suppose that for every triple of the vertices such that and are non-adjacent: If there is no arc from to , then . If there is no arc from to , then . If there is a path from to and a path from to in for every pair of distinct vertices , then has a cycle which contains at least vertices of .
The last result is best possible in some situations and gives an answer to a question posted by Li et al. (2007).
Theorem 1.8 (Meyniel (1973)). *Let be a strong digraph of order . If for all pairs of non-adjacent vertices in , then is Hamiltonian.
For a short proof of Theorem 1.8, see Bondy and Thomassen (1977). Darbinyan (1985) characterized those digraphs which satisfy Meyniel’s condition, but are not pancyclic. Before stating the main result obtained by Darbinyan (1985), we need to define a family of digraphs.
Definition 1.9. *For integers and , , let denote the set of digraphs , which satisfy the following conditions: (i) ; (ii) is a Hamiltonian cycle in ; (iii) for each , , the vertices and are not adjacent; (iv) whenever and (v) the sum of degrees for any two distinct non-adjacent vertices is at least .
Theorem 1.10 (Darbinyan (1979), Darbinyan (1985)). Let be a strong digraph of order . Suppose that for all pairs of distinct non-adjacent vertices , in . Then either (a) is pancyclic or (b) is even and is isomorphic to one of , , where is an arbitrary arc of , or (c) in this case does not contain a cycle of length ).
Later on, Theorem 1.10 was also proved by Benhocine (1986). Darbinyan (2019) investigated the pancyclicity of digraphs with the condition . Using Theorem 1.10 and the Moser theorem for a strong tournament to be pancyclic (see Harary and Moser (1966)), we proved the following theorem.
Theorem 1.11 (Darbinyan (2019)). Let be a 2-strong digraph of order satisfying condition . Suppose that there exists a pair of non-adjacent vertices , in such that . Then contains cycles of all lengths .
In this paper we confirm Conjecture 1.3.
Theorem 1.12. Let be a 2-strong digraph of order satisfying condition . Then is Hamiltonian.
Theorem 1.12 also has the following immediate corollaries.
Corollary 1.13 (Woodall (1972)). A digraph of order is Hamiltonian if, for any two vertices and , either or .
Corollary 1.14 (Nash-Williams (1969)). Let be a digraph of order . If for every vertex , and , then is Hamiltonian.
Note that Corollary 1.14 immediately follows from well-known theorem of Ghouila-Houri Ghouila-Houri (1960).
Corollary 1.15 ( Ore (1960)). Let be a simple graph of order , in which the degree sum of any two non-adjacent vertices is at least . Then is Hamiltonian.
As an immediate corollary of Theorems 1.12 and 1.11, we obtain the following theorem.
Theorem 1.16. Let be a 2-strong digraph of order satisfying condition . Suppose that contains a pair of non-adjacent vertices , such that . Then is pancyclic.
In view of Theorem 1.16, it is natural to set the following problem.
Problem 1.17. Let be a 2-strong connected digraph of order satisfying condition . Suppose that is a pair of non-adjacent vertices in such that . Whether is pancyclic?
2 Terminology and notation
In this paper we consider finite digraphs without loops and multiple arcs. We shall assume that the reader is familiar with the standard terminology on digraphs and refer to the book Bang-Jensen and Gutin (Springer-Verlag, London, 2000) for terminology and notations not defined here. The vertex set and the arc set of a digraph are denoted by and , respectively. The order of is the number of its vertices. For any , we also write if . We use the notations if and if . If , is an out-neighbour of and is an in-neighbour of . If and , we write . Two distinct vertices and are adjacent if or (or both). If there is no arc from to , we shall use the notation .
We let , denote the set of out-neighbours, respectively the set of in-neighbours of a vertex in a digraph . If , then and . The out-degree of is and is the in-degree of . Similarly, and . The degree of the vertex in is defined as (similarly, ). The subdigraph of induced by a subset of is denoted by . If is a vertex of a digraph , then the subdigraph is denoted by .
For integers and , , let denote the set of all integers, which are not less than and are not greater than .
The path (respectively, the cycle) consisting of the distinct vertices () and the arcs , (respectively, , , and ), is denoted by (respectively, ). We say that is a path from to or is an -path. Let and be two distinct vertices of a digraph . Cycle that passing through and in , we denote by .
A cycle (respectively, a path) that contains all the vertices of , is a Hamiltonian cycle (respectively, is a Hamiltonian path). A digraph is Hamiltonian if it contains a Hamiltonian cycle. A digraph is strongly Hamiltonian-connected if, for every ordered pair of distinct vertices of there is a Hamiltonian path from to . A digraph of order is pancyclic if it contains cycles of all lengths , . For a cycle of length , the subscripts considered modulo , i.e., for every and such that . If is a path containing a subpath from to , we let denote that subpath. Similarly, if is a cycle containing vertices and , denotes the subpath of from to . If , then .
A digraph is strongly connected (or just strong), if there exists a path from to and a path from to for every pair of distinct vertices . A digraph is -strongly connected (or -strong), where , if and is strongly connected for any subset of at most vertices.
For a pair of disjoint subsets and of , we define and .
3 Auxiliary known results
Lemma 3.1 (Häggkvist and Thomassen (1976)). *Let be a digraph of order containing a cycle of length , . Let be a vertex not contained in this cycle. If , then contains a cycle of length for all *.
It is not difficult to prove the following lemma.
Lemma 3.2. *Let be a digraph of order . Assume that and the vertices , in satisfy the degree condition , where . Then contains at least internally disjoint -paths of length two.
The following results were proved by Darbinyan (2019) and its preliminary version presented at Emil Artin International Conference (Darbinyan (2018)).
Theorem 3.3. Let be a 2-strong digraph of order satisfying condition . Suppose that is a pair of non-adjacent vertices in such that . Then is Hamiltonian if and only if contains a cycle through the vertices and .
Theorem 3.4. *Let be a 2-strong digraph of order . Suppose that contains at most one pair of non-adjacent vertices. Then is Hamiltonian.
Remark. *There is a strong non-Hamiltonian digraph of order , which is not 2-strong and has exactly one pair of non-adjacent vertices.
Using Lemma 3.2, it is not difficult to prove the following lemma.
Lemma 3.5. Let be a 2-strong digraph of order and let , be two distinct vertices in . If contains no cycle through and , then , are not adjacent and there is no path of length two between them. In particular, .
Theorem 3.6. *Let be a 2-strong digraph of order satisfying condition . Suppose that is a pair of non-adjacent vertices in such that . Then is Hamiltonian or contains a cycle of length passing through and avoiding (passing through and avoiding ).
As an immediate corollary of Theorems 3.6, 3.3 and Lemma 3.1, we obtain
Corollary 3.7. *Let be a 2-strong non-Hamiltonian digraph of order satisfying condition . Suppose that is a pair of non-adjacent vertices in such that . Then , and contains at most one cycle of length two passing through () *.
4 Preliminaries
Lemma 4.1. Let be a 2-strong digraph of order satisfying condition . Suppose that is a pair of non-adjacent vertices in such that and is a cycle in passing through and avoiding , where . If the subdigraph contains a cycle passing through and , then is Hamiltonian.
Proof 4.1**.**
Suppose, on the contrary, that contains a cycle passing through , but is not Hamiltonian. Since contains at most one cycle of length two passing through (Corollary 3.7), from it follows that . Let be a cycle through in , where .
By Theorem 3.3 we have that contains no cycle through and . Therefore, for each pair of integers and , where and , (here, and ). This implies that for every we have
[TABLE]
Hence,
[TABLE]
Since there is at most one cycle of length two through () (Corollary 3.7), it follows that for and for every (we may assume that ) the following holds:
[TABLE]
Therefore,
[TABLE]
Combining this with (1), we obtain
[TABLE]
The last inequality together with implies that
[TABLE]
Notice that , , are distinct pairs of non-adjacent vertices. We will consider the cases when is even and is odd separately.
Assume first that is even. Using condition and (2), we obtain
[TABLE]
Therefore, , i.e., . The last inequality is impossible, since .
Assume next that is odd. Then . Since , and by Corollary 3.7 (we may assume that ), from condition it follows that . Now, by condition and (2) we have,
[TABLE]
[TABLE]
Hence,
[TABLE]
This means that , which is a contradiction. This contradiction completes the proof of Lemma 4.1.
Lemma 4.2. Let be a 2-strong digraph of order satisfying condition . Suppose that is a pair of non-adjacent vertices in such that and is a cycle of length passing through and avoiding in . Then either is Hamiltonian or for every , the following holds:
[TABLE]
Proof 4.2**.**
Suppose that is not Hamiltonian. Since is 2-strong, . Then by Theorem 3.3, there is no cycle through and . Therefore, we have that if with , then (for otherwise, , where , is a cycle through and , a contradiction). Let , , and , be chosen so that is minimal and is maximal with these properties. Then
[TABLE]
If and , then by a similar argument as above, we conclude that if with , then . Assume that or . Observe that is a cycle through which does not contain , and because of (3). Therefore by Lemma 4.1, the subdigraph contains no cycle through since is not Hamiltonian. This implies that
[TABLE]
since . From the last equalities it follows that if there are such that and with , then , and is a cycle passing through and , a contradiction. Thus, we may assume that for every pair of integers and , ,
[TABLE]
Now suppose that the theorem is not true. Then is not Hamiltonian and there is an integer such that
[TABLE]
It is easy to see that there are vertices and such that , and
[TABLE]
Then by (4),
[TABLE]
Assume first that . Since is 2-strong, (4) and (7) imply that . Now from (5), (6) and (7) it follows that:
(i) if or , then (respectively)
[TABLE]
or
[TABLE]
(ii) if , then . Thus, in each case we have that is not strong, which contradicts the condition that is 2-strongly connected.
Assume next that . This case is similar to the first case and we omit the details. Lemma 4.2 is proved.
The following lemma is proved by Darbinyan (2019). We present its proof for completeness.
Lemma 4.3. Let be a 2-strong digraph of order satisfying condition . Suppose that is a pair of non-adjacent vertices in such that and is a cycle of length passing through and avoiding in . If and there are integers such that and , then is Hamiltonian.
Proof 4.3**.**
Suppose, on the contrary, that is not Hamiltonian. By Theorem 3.3, contains no cycle through and . Therefore, there are no integers and , , such that (for otherwise, is a cycle through and ). Since the arcs , , , are in and , it is easy to check that:
(i) if with , then ;
(ii) if with , then . Thus, in both cases we have a contradiction. Therefore,
[TABLE]
in particular, and the vertices and ( and ) are not adjacent. The last equality together with the fact that contains at most one cycle of length two passing through (Corollary 3.7) implies that
[TABLE]
Now we consider the vertex . It is not difficult to check that:
(iii) if with , then ;
(iv) if with , then . In both cases we have a contradiction. Therefore, we may assume that
[TABLE]
This implies that
[TABLE]
[TABLE]
Without loss of generality, we may assume that , are chosen as maximal as possible and , are chosen as minimal as possible, i.e.,
[TABLE]
This, since contains at most one cycle of length two passing through , implies that
[TABLE]
[TABLE]
Since and are two distinct pairs of non-adjacent vertices, from (8), (9), the last inequality and condition it follows that
[TABLE]
[TABLE]
which is a contradiction. Lemma 4.3 is proved.
5 Proof of Theorem 1.12
Recall the statement of Theorem 1.12.
Theorem 1.12. Let be a 2-strong digraph of order satisfying condition . Then is Hamiltonian.
Proof 5.1**.**
By Theorem 3.4, the theorem is true if contains at most one pair of non-adjacent vertices. We may therefore assume that contains at least two distinct pairs of non-adjacent vertices. If the degrees sum of any two non-adjacent vertices at least , then by Meyniel’s theorem, the theorem is true. We may therefore assume that contains a pair of non-adjacent vertices, say , such that . By Theorem 3.3, to prove the theorem, it suffices to prove that contains a cycle through and . If , then by Lemma 3.5 we have that contains a cycle trough and , which, in turn, implies that is Hamiltonian (by Theorem 3.3). Thus, we may assume that . By Theorem 3.6 we have that either is Hamiltonian or contains a cycle of length passing through and avoiding (passing through and avoiding ).
Suppose that is not Hamiltonian, i.e., contains no cycle through and . Let be a cycle of length in , which does not contain . Let be the maximum integer such that and be the minimum integer such that . Since is 2-strong and contains no cycle passing through and , it follows that and there are some integers , , , such that and
[TABLE]
[TABLE]
Note that if contains a cycle of length two passing trough , then , otherwise , and . Therefore, it is not difficult to see that
[TABLE]
In order to prove the theorem, it is convenient for the digraph and the cycle to prove the following claims.
Claim 5.1.* If , then .*
Proof 5.2**.**
Notice that is a cycle passing through and avoiding . By (10) we have that . Now by Lemma 4.1 , the induced subdigraph contains no cycle through . Then, since , we have
[TABLE]
The first equality together with 2-connectedness of implies that there is an integer such that . The last equality means that if , then . Assume that , i.e., . In this case, we have that if with (respectively, ), then (respectively, ), which is a contradiction. This proves that .
Claim 5.2.* Suppose that and , where and . Then .*
Proof 5.3**.**
Assume that Claim 5.2 is not true. Then for some , . Then, since the arcs , , are in and , we have a cycle , which contradicts our initial supposition.
Claim 5.3.* Suppose that , with and (possibly, ). Then there is an integer such that , , (possibly, ). Moreover, either there is a vertex with such that or there is a vertex with such that .*
Proof 5.4**.**
By Claim 5.2,
[TABLE]
*Since and , obviously there is an integer such that , , possibly ). This together with
implies that*
[TABLE]
Now suppose that the claim is not true. Then
[TABLE]
The second equality of (14) together with and the fact that there is no path of length two between and (Lemma 3.5) implies that the vertices , are not adjacent and
[TABLE]
This together with (13), (10) and the fact that there is at most one cycle of length two through (Corollary 3.7) implies that
[TABLE]
[TABLE]
[TABLE]
[TABLE]
Now consider the vertex . Note that since . Using (12) and the first equality of (14), we obtain
[TABLE]
[TABLE]
Combining the last two inequalities, (Corollary 3.7) and , we obtain
[TABLE]
which contradicts condition , since , are two distinct pairs of non-adjacent vertices. This contradiction completes the proof of Claim 5.3.
Claim 5.4.* If , then *
Proof 5.5**.**
Suppose, on the contrary, that and with and . Let be the maximum with these properties, i.e.,
[TABLE]
Notice that is a cycle in and by (10). Therefore by Lemma 4.1, the subdigraph does not contain a cycle through . In particular,
[TABLE]
and if , then
[TABLE]
By Claim 5.1, we have
[TABLE]
From (17) and (18) it follows that and, if , then . In both cases we have that .
If with , then , a contradiction. We may therefore assume that . This together with (16) implies that
[TABLE]
Applying Lemma 4.2 to the vertex , we obtain that
[TABLE]
Let , where and . Choose maximal with these properties, i.e.,
[TABLE]
*From (15) it follows that , i.e., . If with , then , a contradiction. We may therefore assume that
. This together with implies that*
[TABLE]
In particular, the vertices and are not adjacent, and since (i.e., ). Using Lemma 4.3, we obtain
[TABLE]
Then, since and (20), we have that . This together with (15) implies that
[TABLE]
Therefore, , i.e., . Since , and , are not adjacent, there is an integer such that and . Then, since (21) and we have that and
[TABLE]
This together with (10) implies that
[TABLE]
[TABLE]
From (19), and the fact that there is no path of length two between and (Lemma 3.5) it follows that
[TABLE]
This together with (10), (23) and the fact that there is at most one cycle of length two through (Corollary 3.7) implies that
[TABLE]
[TABLE]
[TABLE]
[TABLE]
Since and (20), it follows that
[TABLE]
In particular, from and (26) it follows that
[TABLE]
*We will consider the cases , separately.
Case 1.* .*
Then by the first equality of (22) we have
[TABLE]
Using the fact that there is no path of length two between and (Lemma 3.5) and (19), we obtain that . This together with (by (26)) and (28) implies that
[TABLE]
[TABLE]
[TABLE]
*Now we divide this case into the following subcases.
Subcase 1.1.* The vertices and are not adjacent.*
Then and are two distinct pairs of non-adjacent vertices. Since , , and , combining (25), (24) and (29), we obtain
[TABLE]
[TABLE]
*which contradicts condition .
Subcase 1.2.* The vertices and are adjacent.*
Then . Therefore by Lemma 3.5 and (19), the vertices and are not adjacent. Since (because of and Corollary 3.7) and , from (25) and (29) it follows that
[TABLE]
[TABLE]
*which contradicts condition . The discussion of Case 1 is completed.
Case 2.* .*
*We divide this case into the following subcases.
Subcase 2.1.* .*
Then since . Then by the definition of and . Recall that the vertices , are not adjacent by (19) and Lemma 3.5. Now it is easy to see that . Since with and , by Claim 5.2 we have that . This together with (27) and implies that
[TABLE]
[TABLE]
This together with (24) and , we obtain
[TABLE]
[TABLE]
*which is a contradiction since and are two distinct pairs of non-adjacent vertices.
Subcase 2.2.* .*
From , and (23) it follows that
[TABLE]
in particular, the vertices and are not adjacent. Observe that is a cycle in passing through , avoiding and . By Lemma 4.1, the induced subdigraph contains no cycle through . In particular, this means that
[TABLE]
*for otherwise, if with and , then is a cycle in through , a contradiction.
Subcase 2.2.1.* There is an integer such that and*
[TABLE]
*Then , and because of the first equality of (31). Recall that . Hence, . If with , then , a contradiction. We may therefore assume that . This together with *
* and the fact that there is no path of length two between and implies that*
[TABLE]
Combining this, (10) and (30), we obtain
[TABLE]
[TABLE]
[TABLE]
[TABLE]
For the vertex , using (32) and the second equality of (22), we obtain
[TABLE]
[TABLE]
[TABLE]
This together with (33), (24), , and implies that
[TABLE]
[TABLE]
which contradicts condition since and are two distinct pairs of non-adjacent vertices.
Subcase 2.2.2.* There is no such that .*
Then . This together with the second equality of (22) implies that
[TABLE]
[TABLE]
[TABLE]
[TABLE]
Combining this, , (24) and , we obtain
[TABLE]
[TABLE]
which contradicts condition . In each case we obtain a contradiction and hence the discussion of Case 2 is completed. This completes the proof of Claim 5.4.
Now we are ready to complete the proof of the main result.
By Claim 5.4, if , then Similarly, if , then . Using Lemma 4.3, we obtain From the last three equalities it follows that
[TABLE]
From (34) and Lemma 4.2 it follows that . Applying Lemma 4.2 to the vertices and , we obtain
[TABLE]
Let and with , , and . Choose maximal and minimal with these properties, i.e.,
[TABLE]
From (34) it follows that and , i.e., and . If , then , a contradiction. We may therefore assume that , which in turn implies that . By Lemma 4.2, . Let , where and . Choose maximal with this property, i.e.,
[TABLE]
From (35) it follows that and , i.e., and . We may assume that (recall that , ) is chosen so that
[TABLE]
*We consider the cases and separately.
Case 1.* .*
For this case, it is not difficult to check that the conditions of Claim 5.3 hold. Therefore, there is an integer such that , , (possibly, ), and either there is a vertex with such that or there is a vertex with such that .
Assume first that . Then, since the arcs , , , , , are in and , we have that , or when or when respectively. In each case we have a contradiction.
Assume next that . By Lemma 4.2, . Let , where and . Choose maximal with this property, i.e.,
[TABLE]
From (36) (respectively, from (35)) it follows that , i.e., (respectively, , i.e., ). If , then , a contradiction. We may therefore assume that . By Lemma 4.2,
[TABLE]
Let , where and . Choose maximal with this property, i.e.,
[TABLE]
From (38) (respectively, from (35)) it follows that , i.e., (respectively, , i.e., ).
Assume first that . Then it is not difficult to see that or when or when , respectively. In each case we have a contradiction.
Continuing this process, we finally conclude that for some , (here, ) since all the vertices are distinct and in }. We already have constructed a cycle when . Assume that . By the above arguments we have that:
If is odd, then .
*If is even, then or when or when , respectively. In all cases we have a cycle through and , which contradicts our supposition and hence the discussion of Case 1 is completed.
Case 2.* .*
*Then . Recall that , and , where , and . Note that , are two distinct pairs of non-adjacent vertices.
Subcase 2.1.* .*
Since and , we have that is a cycle in and . If , then is a cycle in passing through , which contradicts Lemma 4.1. We may therefore assume that , i.e., .
Assume first that . Then and . From the first equality of (35) it follows that . This equality together with (37) implies that
[TABLE]
[TABLE]
[TABLE]
This together with (11) and implies that
[TABLE]
[TABLE]
which contradicts condition .
Assume that , i.e., . We may assume that is chosen so that . This and (37) imply that
[TABLE]
[TABLE]
[TABLE]
Since , it is not difficult to check that if with , then , which is a contradiction. Therefore, . This together with and the fact that there is no path of length two between and implies that
[TABLE]
Using this and (10), we obtain
[TABLE]
[TABLE]
[TABLE]
Combining this, (11) and (39), we obtain
[TABLE]
[TABLE]
*which contradicts condition and hence the discussion of Subcase 2.1 is completed.
Subcase 2.2.* .*
Then since .
Assume first that . Then . By Lemma 4.2,
[TABLE]
Let , where and . From the second equality of (35) it follows that , i.e., . By (36) we have that , i.e., . Therefore, , a contradiction.
Assume next that . From the maximality of and it follows that . This last equality together with (37) implies that
[TABLE]
[TABLE]
[TABLE]
This together with (11), and implies that
[TABLE]
[TABLE]
which contradicts condition , since , . The discussion of Case 2 is completed. Theorem 1.12 is proved.
Acknowledgements.
The author would like to thank the anonymous referees for thoroughly review and many helpful comments and suggestions which improved substantially the rewriting of this paper. We also thank PhD P. Hakobyan for formatting the manuscript of this paper.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Bang-Jensen and Gutin (Springer-Verlag, London, 2000) J. Bang-Jensen and G. Gutin. Digraphs: Theory, Algorithms and Applications , Springer-Verlag, London, 2000.
- 2Benhocine (1986) A. Benhocine. Pancyclism and Meyniel’s conditions. Discrete Math. , 58:113–120, 1986.
- 3Bermond and Thomassen (1981) J.-C. Bermond and C. Thomassen. Cycles in digraphs − - a survey. J. Graph Theory , 5(1):1–43, 1981.
- 4Bondy and Thomassen (1977) J. Bondy and C. Thomassen. A short proof of Meyniel’s theorem. Discrete Math. , 19:195–197, 1977.
- 5(5) S. Darbinyan. On Hamiltonian bypasses in digraphs with the condition of Y. Manoussakis. In Procceedings Conference Computer Science and Information Tecnologies, (CSIT 2015) . Yerevan, 2015, pp. 53-63. 10.1109/CSI Technol. 2015.7358250 . · doi ↗
- 6Darbinyan (1979) S. Darbinyan. On pancyclic digraphs. Preprint of the Computing Center of Akademy of Sciences of Armenia , 1979.
- 7Darbinyan (1983) S. Darbinyan. Disproof of a conjecture of Thomassen. Akad. Nauk Armyan. SSR Dokl. , 76(2):51–54, 1983.
- 8Darbinyan (1985) S. Darbinyan. Pancyclicity of digraphs with the Meyniel condition. Studia Sci. Math. Hungar., (Ph.D. Thesis, Institute Mathematici Akad. Nauk BSSR, Minsk, 1981) , 20(1-4):95–117, 1985.
