Intersection of Longest Paths in A Graph ?
Intersection of Longest Paths in A Graph ?
Intersection of Longest Paths in A Graph ?
Abstract
In 1966, Gallai asked whether every connected graph has a vertex that is common
to all its longest paths. The answer to this question is negative. We prove that
the answer is positive for outerplanar graphs. Another related question was raised
in 1995 at the British Combinatorial Conference: Do any three longest paths in a
connected graph have a vertex in common? We prove that, in a connected graph in
which all non-trivial blocks are Hamiltonian, any three of its longest paths have a
common vertex. Both of these results strengthen a recent result by Axenovich.
Keywords: longest path, intersection of longest paths, outerplanar graph.
1 Introduction
In 1966, during a coloquium on graph theory, Gallai [4] asked whether every
connected graph has a vertex that appears in all its longest paths. A few years
? This research was partially supported by CNPq.
1
Email: {susanna,cris,yw}@ime.usp.br
2
Email: daniel.martin@ufabc.edu.br
later, Walther [7] answered this question negatively by exhibiting a counter-
example on 25 vertices. Later, Walther [8] and Zamfirescu [9] independently
found a smaller counterexample on 12 vertices (see Figure 1).
On the other hand, it is easy to prove that the answer is positive for trees.
In 1990, Klavzar and Petkovsek [5] proved that the answer is also positive for
split graphs, cacti, and some other classes of graphs. More recently, in 2004,
Balister, Gyori, Lehel, and Schelp [3] established a similar result for the class
of circular arc graphs. In this paper we prove the following result.
Theorem 1.1 If G is a connected outerplanar graph, then G has a vertex
common to all its longest paths.
Since non-empty intersection of all longest paths seems to be a property
exhibited by few classes of graphs, it is natural to consider the intersection
of a smaller number of longest paths. In this case, while it is easy to prove
that every two longest paths share a common vertex, for three longest paths, a
satisfactory answer is not known. In fact, the following problem [1,10] remains
open.
Problem 1.2 Do any three longest paths in a connected graph share a com-
mon vertex?
The only known progress on Problem 1.2, excluding those on all longest
paths mentioned above, was obtained by Axenovich [2], who proved the fol-
lowing result.
Theorem 1.3 (Axenovich, 2009) If G is a connected outerplanar graph,
then any three of its longest paths have a common vertex.
In Section 3, we prove the following more general result.
Theorem 1.4 If G is a connected graph in which all non-trivial blocks are
Hamiltonian, then any three of its longest paths have a common vertex.
Note that both Theorems 1.1 and 1.4 generalize Axenovichs result, each
in a different way.
In this paper, all graphs considered are undirected and simple. A graph is
outerplanar if it has an embedding in the plane such that all vertices belong
to the boundary of its outer face (the unbounded face).
If P is a path, then kP k denotes its length (number of edges), and P 1
denotes the reverse of P . If x and y are vertices of a cycle C embedded in the
plane, we denote by Cx,y the section of C that goes clockwise from x to y.
If B is a non-trivial block of a graph G, then we say that a non-trivial path
in G is a pending path (of B) if it intersects B in precisely one of its extremes,
which we call its origin.
Sketch of the proof of Theorem 1.4. Let P be a set of three longest paths
in G. It can be shown that there is a block B of G that contains at least one
vertex of each path in P. If one of the paths in P contains all vertices of B,
then the result follows easily. Thus we can assume B is non-trivial and all
paths in P use two pending paths of B.
Henceforth, we shall consider only pending paths of B that are contained
in the paths of P. If all these pending paths together intersect B in only two
vertices, then the result is obvious. So, let v1 , v2 and v3 be origins of three
disjoint pending paths that are as long as possible.
A longest pending path of B that has vi as its origin will be called a vi -
pending path. Suppose a vi -pending path is at least as long as a vj -pending
path if i < j. Let C be a Hamiltonian cycle in B embedded in the plane in
such a way that v1 , v2 and v3 appear clockwise in C.
Case 1: There is at most one path in P containing both a v1 -pending path
and a v2 -pending path.
Let Pa be a path in P that contains a v1 -pending path, and its other
pending path is as long as possible. Let Pc be a path in P that does not
contain a v1 -pending path, and let Pb be the remaining path in P.
For i {1, 2, 3}, let Ri be a vi -pending path, and for x {a, b, c}, denote
by Tx and Tx0 the two pending paths contained in Px . Thus, we have
For a path Px , let C(Px ) denote the subgraph obtained from Px by remov-
ing all vertices not in C. Observe that C(Px ) must be a path. Combining the
fact that kPx k = kTx k + kC(Px )k + kTx0 k with (1) and (2), we obtain
References
[1] Research problems, Discrete Mathematics 167/168 (1997), pp. 605615, 15th
British Combinatorial Conference.
[2] Axenovich, M., When do three longest paths have a common vertex?, Discrete
Mathematics, Algorithms and Applications 1 (2009), pp. 115120.
[3] Balister, P., E. Gyori, J. Lehel and R. Schelp, Longest paths in circular arc
graphs, Combin. Probab. Comput. 13 (2004), pp. 311317.
[4] Gallai, T., Problem 4, in: Theory of Graphs (1968), p. 362.
[5] Klavzar, S. and M. Petkovsek, Graphs with non empty intersection of longest
paths, Ars Combin. 29 (1990), pp. 1352.
[6] Skupien, Z., Smallest sets of longest paths with empty intersection, Combin.
Probab. Comput. 5 (1996), pp. 429436.
[7] Walther, H., Uber die Nichtexistenz eines Knotenpunktes, durch den alle
langsten Wege eines Graphen gehen, J. Combinatorial Theory 6 (1969), pp. 16.
[8] Walther, H. and H.-J. Voss, Uber Kreise in Graphen, VEB Deutscher Verlag
der Wissenschaften (1974).
[9] Zamfirescu, T., On longest paths and circuits in graphs, Math. Scand. 38 (1976),
pp. 211239.
[10] Zamfirescu, T., Intersecting longest paths or cycles: a short survey, An. Univ.
Craiova Ser. Mat. Inform. 28 (2001), pp. 19.