## 1 Introduction

A graph is *chordal* if every induced cycle is of length three.
A classical result of Dirac [Dir61] states that every chordal graph has a *simplicial* vertex, that is, a vertex which neighbourhood is a clique.
However, not all graphs exhibit the nice structure of chordal graphs, and the statement does not extend to general graphs.

### 1.1 From simplicial vertices to avoidable paths

One way to generalise Dirac’s result is through the following more flexible notion.

###### Definition 1.1 (Avoidable vertex).

A vertex in a graph is *avoidable* if every induced path on three vertices with middle vertex is contained in an induced cycle in .
∎

Note that in a chordal graph, every avoidable vertex is simplicial. The next theorem can be inferred from [OCF76, BB98, ACTV15]; see also [BCG19] for a nice introduction.

###### Theorem 1.2.

Every graph has an avoidable vertex.

Recently in [BCG19], the authors considered a generalisation of the concept of avoidable vertices to edges, and extended theorem 1.2 to that notion.

###### Definition 1.3 (Avoidable edge).

An edge in a graph is *avoidable* if every induced path on four vertices with middle edge is contained in an induced cycle in .
∎

###### Theorem 1.4 (Beisegel et al. [Bcg19]).

Every graph has an avoidable edge.

This notion naturally generalises to paths, as follows.

###### Definition 1.5 (Extension).

Given an induced path in a graph , an *extension* of is an induced path in for some vertices .
∎

###### Definition 1.6 (Failing).

An induced path in a graph is *failing* if there is no induced cycle of containing .
∎

###### Definition 1.7 (Avoidable).

A path in a graph is *avoidable* if it is induced and has no failing extension.
Given a subgraph of , we say that is an avoidable path of in if it is avoidable in and .
∎

A graph is -free if it does not contain a , that is, an induced path on vertices. In [BCG19] the authors conjecture that for every positive integer , every graph either is -free or contains an avoidable path on vertices. This conjecture is motivated by the following result of Chvátal et al. [CRS02], which generalises Dirac’s theorem. A -free graph is a graph where every induced cycle has at most vertices. The -free graphs are exactly the chordal graphs. Unless specified otherwise, we consider cycles to be induced.

###### Theorem 1.8 (Chvátal et al. [Crs02]).

For every positive integer , every -free graph either is -free or contains an avoidable path on vertices.

In fact, theorem 1.8 originally states the existence of a simplicial path in the class of -free graphs.
A *simplicial* path is an induced path with no extension: it is avoidable by vacuity.
Note that these two definitions coincide in such a class, as no cycle on at most vertices can contain the extension of an induced path on vertices.

Here, we confirm the aforementioned conjecture [BCG19, Conjecture 1], as follows.

###### Theorem 1.9.

For every positive integer , every graph either is -free or contains an avoidable .

In fact, we prove theorem 1.9 using a stronger induction hypothesis, in the exact same flavour as [CRS02], see theorem 2.4 in section 2.

### 1.2 Consequences

We point out that the proof of theorem 1.9 is self-sufficient, thus this supersedes the arguments for theorems 1.8, 1.4 and 1.2.

By using ingredients of theorem 2.4 (namely lemma 2.3), we obtain a way to build more than one avoidable .

###### Corollary 1.10.

For every positive integer , graph and subset such that is connected, either is -free or there is an avoidable of in .

###### Corollary 1.11.

For every positive integer and graph , either does not contain two non-adjacent , or it contains two non-adjacent avoidable .

Since corollary 1.11 is not as straightforward as its predecessor, we include a proof.

###### Proof.

Let and be two non-adjacent . By corollary 1.10, either is -free or there is an avoidable of in . The first outcome is ruled out by the existence of . Let be an avoidable of in . We repeat the argument with instead of , and obtain an avoidable of in , call it . The two paths and are two non-adjacent avoidable , as desired. ∎

We can also wonder:

###### Question 1.12.

For every positive integer , does every graph either not contain two disjoint , or contain two disjoint avoidable ?

We know the answer to be positive in the case , due to [BCG19, Theorems 3.3 and 6.4]. The answer turns out to be negative in all other cases, as exhibited in the following counter-example for , which consists of a cycle on vertices with an added vertex adjacent to two consecutive vertices on the cycle (see figure 1 for the case ). This graph contains two disjoint , and it has vertices, so any two disjoint are in fact complementary in the graph. Suppose that it contains two disjoint avoidable , and note that each intersects the triangle (otherwise the complement would not be a path). Since there are three vertices in the triangle, there is an avoidable containing a single vertex in the triangle. This has a failing extension, a contradiction.

In section 3, we present a concise algorithm which follows the proof of theorem 2.4. As discussed there, the algorithm has complexity which, while naive, is the right order of magnitude under ETH.

## 2 A stronger induction hypothesis

All graphs considered in this paper are finite, simple and loopless.
Given a graph , we denote by its set of vertices, and by its set of edges.
Edges are denoted by (or ) instead of .
If is an edge, then we say that and are *adjacent*.
Given a vertex , the neighbourhood of is the set of vertices of that are adjacent to .
The closed neighbourhood of is the set .
If , then we define and .
The subgraph of *induced* by , denoted by , is the graph , and is the graph .
Given two adjacent vertices and of , the graph obtained by *merging* and is the graph obtained from by replacing and with a new vertex such that .
Given a graph and two subsets and of , we say that *dominates* if every vertex of has a neighbour in (equivalently, if ).

We first define two useful properties.

###### Definition 2.1 (Basic property ).

Given a positive integer and a graph , the property holds if either is -free or there is an avoidable in . ∎

###### Definition 2.2 (Refined property ).

Given a positive integer , a graph and a vertex , the property holds if either is -free or there is an avoidable of in .

Given a positive integer and a graph , the property holds if holds for every . ∎

Note that property does not directly imply property . We also emphasise the fact that an avoidable path in a subgraph is not necessarily an avoidable path in the whole graph.

We now prove a form of heredity in .

###### Lemma 2.3.

Let be a positive integer, a graph and an edge of . Let be the graph obtained from by merging the two vertices and into one vertex . If contains a , then implies .

###### Proof.

Suppose contains a , and that holds but not . Since is not -free, there is an avoidable of in . Call it . The path is contained in , so in particular in . Since does not hold, is not an avoidable of . Thus, there is a failing extension of in . Note that , and are all pairwise distinct.

Hence, is an extension of in , and there is an induced cycle in containing the path . If , then the cycle is also an induced cycle in containing , a contradiction. Therefore, . By replacing with either , or the edge as appropriate, we obtain an induced cycle in containing , a contradiction. ∎

We are now ready to prove the main technical result of this paper.

###### Theorem 2.4.

For every positive integer and every graph , both properties and hold.

###### Proof.

Suppose the statement is false and consider a counter-example which is minimal with respect to the number of vertices.

###### Lemma 2.5.

The property holds for every .

###### Proof.

We proceed by contradiction. Suppose that does not hold for some and some vertex , that is, there exists a in , and every in has a failing extension. We prove the following.

0.3in

###### Claim 2.6.

Every in dominates .

###### Proof.

Assume towards a contradiction that there is a in , call it , which is not adjacent to some vertex . Then contains a . Let be the graph obtained from by merging and into a vertex . Since has fewer vertices than , the property holds by minimality of . By lemma 2.3, the property holds, a contradiction.

Let . Then contains a . As contains fewer vertices than , the property holds. Let be an avoidable of . By assumption, is not an avoidable of . So there is a failing extension of in . Since has no failing extension in , we can assume without loss of generality that . It follows that : otherwise the cycle contradicts the fact that is failing. By definition of an extension, is an induced path. Let be the only neighbour of in , and let us now consider the path . It is a , and it does not intersect . However, no vertex in it is adjacent to which lies in , contradicting claim 2.6.

###### Lemma 2.7.

The property holds for every .

###### Proof.

Assume towards a contradiction that for some , the property does not hold. By lemma 2.5, the property holds for every vertex . In other words, the graph contains a but no avoidable , and for every vertex , either is -free or there is an avoidable of in .

We derive the following claim.

###### Claim 2.8.

Every in dominates .

###### Proof.

Suppose there is a , call it , that does not dominate some vertex of . Since holds, either is -free or there is an avoidable of in . The first case contradicts the existence of , and the second contradicts the fact that does not hold.

Since does not hold, contains a , say , that is not avoidable. So it has a failing extension . Let be the only neighbour of in , and consider the path . It is an induced and none of its vertices is adjacent to . This contradicts claim 2.8.

Finally, lemmas 2.7 and 2.5 together contradict being a counter-example. ∎

Theorem 1.9 directly follows from theorem 2.4.

## 3 An algorithm for theorem 2.4

By going through the proof and extracting the key ingredients, we obtain a straightforward algorithm verifying both properties (see algorithm 1).

The algorithm uses the subprocedure InducedPath that, given a graph and a positive integer , decides whether contains a .
If it does, the procedure returns one, otherwise it returns null.
The naive algorithm for that (testing all subsets of size ) has complexity .
However, this is nearly optimal.
Indeed, the problem of finding a in a given graph is W[1]-hard^{1}^{1}1see e.g. [CFK15] for definitions around complexity when parametrised by (see [CFK15, Ex. 13.16, p. 460]).
In fact, the hinted reduction has a linear blow-up, so it follows that there is no algorithm under ETH.

Let be a positive integer, and let (resp. ) be the worst case complexity of FindAvoidablePath (resp. FindAvoidablePathRefined) on an -vertex graph with parameter . We have , and . We obtain and . While this may well be improved, the known limitations for finding an induced path on vertices also apply for an induced avoidable path on vertices (by theorem 1.9, if the first exists, then so does the second). Therefore, the order of magnitude of this naive algorithm is correct.

Note that there is a yet more naive algorithm blindly checking for every subset of size if it corresponds to an avoidable path. That algorithm has comparable complexity to ours (though slightly worse, at least at first sight). However, we wanted to emphasise that our proof of theorem 2.4 is constructive and yields an elementary algorithm. Also, we believe that it provides an outline of the proof which might be helpful to the reader.

## 4 Conclusion

Given the discussions in section 1.2, it is tempting to ask when a graph admits three (or more) disjoint (resp. pairwise non-adjacent) avoidable paths. Note that though corollary 1.10 arms us with sufficient conditions for there to be more than two avoidable , we do not believe that the corresponding sufficient conditions are necessary. However, it seems the picture is murky already for chordal graphs.

It is tempting to wonder whether we can obtain another avoidable structure. Though in some cases the very notion of extension becomes unclear (what should an extension of a clique be?), it does not seem like any other structure survives the test of chordal graphs or simple ad hoc constructions—even when allowing a family of graphs instead of fixing a single pattern (like a path on vertices). This motivates us to formulate the following question.

###### Question 4.1.

Does there exist a family of connected graphs, not containing any path, such that any graph is either -free or contains an avoidable element of ?

The notion of avoidability in this context is deliberately left up to interpretation.

## Acknowledgements

We gratefully acknowledge support from Nicolas Bonichon and the Simon family for the organisation of the Pessac Graph Workshop, where this research was done. We are indebted to Michał Pilipczuk for providing helpful references regarding the complexity of finding an induced path of given length. Last but not least, we thank Peppie for her unwavering support during the work sessions.

## References

- [ACTV15] Pierre Aboulker, Pierre Charbit, Nicolas Trotignon, and Kristina Vušković. Vertex elimination orderings for hereditary graph classes. Discrete Mathematics, 338(5):825–834, 2015.
- [BB98] Anne Berry and Jean-Paul Bordat. Separability generalizes Dirac’s theorem. Discrete Applied Mathematics, 84(1-3):43–53, 1998.
- [BCG19] Jesse Beisegel, Maria Chudnovsky, Vladimir Gurvich, Martin Milanič, and Mary Servatius. Avoidable vertices and edges in graphs. In Zachary Friggstad, Jörg-Rüdiger Sack, and Mohammad R Salavatipour, editors, Algorithms and Data Structures, pages 126–139, Cham, 2019. Springer International Publishing.
- [CFK15] Marek Cygan, Fedor V Fomin, Łukasz Kowalik, Daniel Lokshtanov, Dániel Marx, Marcin Pilipczuk, Michał Pilipczuk, and Saket Saurabh. Parameterized algorithms. Springer, 2015.
- [CRS02] Vašek Chvátal, Irena Rusu, and R. Sritharan. Dirac-type characterizations of graphs without long chordless cycles. Discrete Mathematics, 256(1-2):445–448, 2002.
- [Dir61] Gabriel Andrew Dirac. On rigid circuit graphs. Abhandlungen aus dem Mathematischen Seminar der Universität Hamburg, 25(1):71–76, 1961.
- [OCF76] Tatsuo Ohtsuki, Lap Kit Cheung, and Toshio Fujisawa. Minimal triangulation of a graph and optimal pivoting order in a sparse matrix. Journal of Mathematical Analysis and Applications, 54(3):622–633, 1976.