Hyperbolic families and coloring graphs on surfaces

By Luke Postle and Robin Thomas

Abstract

We develop a theory of linear isoperimetric inequalities for graphs on surfaces and apply it to coloring problems, as follows. Let be a graph embedded in a fixed surface of genus and let be a collection of lists such that either each list has size at least five, or each list has size at least four and is triangle-free, or each list has size at least three and has no cycle of length four or less. An -coloring of is a mapping with domain such that for every and for every pair of adjacent vertices . We prove

if every non-null-homotopic cycle in has length , then has an -coloring,

if does not have an -coloring, but every proper subgraph does (“-critical graph”), then ,

if every non-null-homotopic cycle in has length , and a set of vertices that are pairwise at distance is precolored from the corresponding lists, then the precoloring extends to an -coloring of ,

if every non-null-homotopic cycle in has length , and the graph is allowed to have crossings, but every two crossings are at distance , then has an -coloring,

if has at least one -coloring, then it has at least distinct -colorings.

We show that the above assertions are consequences of certain isoperimetric inequalities satisfied by -critical graphs, and we study the structure of families of embedded graphs that satisfy those inequalities. It follows that the above assertions hold for other coloring problems, as long as the corresponding critical graphs satisfy the same inequalities.

1. Introduction

All graphs in this paper are finite, and have no loops or multiple edges. Our terminology is standard, and may be found in Reference 11 or Reference 18. In particular, cycles and paths have no repeated vertices, and by a coloring of a graph we mean a proper vertex-coloring; that is, a mapping with domain such that whenever are adjacent vertices of . By a surface we mean a (possibly disconnected) compact -dimensional manifold with possibly empty boundary. The boundary of a surface will be denoted by bd. By the classification theorem every surface is obtained from the disjoint union of finitely many spheres by adding handles, crosscaps and removing the interiors of finitely many pairwise disjoint closed disks. In that case the Euler genus of is .

Motivated by graph coloring problems we study families of graphs that satisfy the following isoperimetric inequalities. By an embedded graph we mean a pair , where is a surface without boundary and is a graph embedded in . We will usually suppress the surface and speak about an embedded graph , and when we will need to refer to the surface we will do so by saying that is embedded in a surface . Let be a family of non-null embedded graphs. We say that is hyperbolic if there exists a constant such that if is a graph that is embedded in a surface , then for every closed curve that bounds an open disk and intersects only in vertices, if includes a vertex of , then the number of vertices of in is at most . We say that is a Cheeger constant for .

The hyperbolic families of interest to us arise from coloring problems, more specifically from a generalization of the classical notion of coloring, introduced by Erdős, Rubin, and Taylor Reference 32 and known as list-coloring or choosability. We need a few definitions in order to introduce it. Let be a graph and let be a collection of lists. If each set is non-empty, then we say that is a list-assignment for . If is an integer and for every , then we say that is a -list-assignment for . An -coloring of is a mapping with domain such that for every and for every pair of adjacent vertices . Thus if all the lists are the same, then this reduces to the notion of coloring. We say that a graph is -choosable, also called -list-colorable, if has an -coloring for every -list-assignment . The list chromatic number of , denoted by , also known as choosability, is the minimum such that is -list-colorable. A graph is -critical if is not -colorable, but every proper subgraph of is. To capture the three most important classes of coloring problems to which our theory applies, let us say that a list-assignment for a graph is of type or a type list-assignment if either

is a -list-assignment, or

is a -list-assignment and is triangle-free, or

is a -list-assignment and has no cycle of length four or less.

We are now ready to describe the three most interesting hyperbolic families of graphs.

Theorem 1.1.

The family of all embedded graphs that are -critical for some type list-assignment is hyperbolic.

Theorem 1.1 is a special case of the more general Theorem 3.6, also stated as Theorem 8.8 and proved in Section 8. Additional examples of hyperbolic families are given in Section 5. We study hyperbolic families in Section 6. Corollary 6.21 implies the following. A family of embedded graphs is closed under curve cutting if for every embedded graph and every simple closed curve in whose image is disjoint from , if denotes the surface obtained from by cutting open along and attaching disk(s) to the resulting curve(s), then the embedded graph belongs to .

Theorem 1.2.

For every hyperbolic family of embedded graphs that is closed under curve cutting there exists a constant such that every graph embedded in a surface of Euler genus has a non-null-homotopic cycle of length at most .

We prove Theorem 1.2 immediately after Corollary 6.21.

Theorem 1.2 is already quite useful. For instance, by Theorem 1.1 it implies the first result stated in the abstract. Our main theorem about hyperbolic families is Theorem 6.28, which describes their structure. Roughly speaking, it says that every member of a hyperbolic family on a fixed surface can be obtained from a bounded size graph by attaching “narrow cylinders”. The theorem suggests the following strengthening of hyperbolicity. Let be a member of a hyperbolic family  embedded in a surface , and let be a cylinder such that the two boundary components of are cycles of . If for all and the number of vertices of in is bounded by some function of , then we say that is strongly hyperbolic. Theorem 3.6 implies that the families from Theorem 1.1 are, in fact, strongly hyperbolic.

We study strongly hyperbolic families in Section 7. The following is a special case of Theorem 7.11.

Theorem 1.3.

For every strongly hyperbolic family of embedded graphs that is closed under curve cutting there exists a constant such that every graph embedded in a surface of Euler genus has at most vertices.

By applying Theorem 1.3 to the families from Theorem 1.1 we deduce the second result stated in the abstract. To prove the next result in the abstract we need to extend the notion of hyperbolicity to rooted graphs. A rooted graph is a pair , where is a graph and . We extend all the previous terminology to rooted graphs in the natural way, so we can speak about rooted embedded graphs. The definitions of hyperbolicity and strong hyperbolicity extend to rooted graphs with the proviso that the disk and cylinder contain no member of in their interiors. Now we can state a more general special case of Theorem 7.11. A proof is given immediately after Theorem 7.11.

Theorem 1.4.

For every strongly hyperbolic family of rooted embedded graphs that is closed under curve cutting there exist constants such that every rooted graph embedded in a surface of Euler genus has at most vertices, and either has a non-null-homotopic cycle of length at most , or some two vertices in are at a distance at most , or .

By applying Theorem 1.4 to the rooted versions of the families from Theorem 1.1 we deduce the following theorem, which we state and formally prove in a slightly more general form as Theorem 3.19.

Theorem 1.5.

There exist constants such that the following holds. Let be a rooted graph embedded in a surface , and let be a type list-assignment for such that every non-null-homotopic cycle in has length at least and every two vertices in are at a distance at least . Then every -coloring of extends to an -coloring of .

Theorem 1.5 proves the third result stated in the abstract. The fourth result follows similarly (formally we prove it as Theorem 3.21), but it requires a generalization of the above two theorems to embedded graphs with a set of distinguished cycles. That is why we will introduce the notion of embedded graphs with rings in Section 3, where a ring is a cycle or a complete graph on at most two vertices. The last result mentioned in the abstract needs the strong hyperbolicity of a different family of embedded graphs. It follows from Theorem 3.22.

The paper is organized as follows. In Section 2 we survey results on list coloring graphs on surfaces. In Section 3 we formulate Theorem 3.8, our main coloring result, and derive a number of consequences from it, assuming a generalization of Theorem 1.1, stated as Theorem 3.6, which we do not prove until Section 8. Theorem 3.8 is an easy consequence of Theorem 7.11, our main result about strongly hyperbolic families. In the short Section 4 we prove a lemma about exponentially many extensions of a coloring of a facial cycle in a planar graph. In Section 5 we present examples of hyperbolic families. We investigate their structure in depth in Section 6, where the main result is Theorem 6.28, giving structural information about members of a hyperbolic family. In Section 7 we give examples of strongly hyperbolic families and investigate their structure. The main theorem is Theorem 7.11, but there is also the closely related Theorem 7.12. The latter eliminates an outcome of the former at the expense of a worse bound. In the final Section 8 we complete the proof of the strong hyperbolicity of the three main families of interest (a.k.a. a generalization of Theorem 1.1), and prove Theorem 3.8. We also formulate and prove the closely related Theorem 8.12, which has an identical proof, using Theorem 7.12 instead of Theorem 7.11.

2. Survey of coloring graphs on surfaces

In this section we survey the main results about coloring graphs on surfaces in order to place our work in a historical context. However, familiarity with this section is not required in order to understand the rest of the paper.

2.1. Coloring graphs on surfaces

The topic of coloring graphs on surfaces dates back to 1852, when Francis Guthrie formulated the Four Color Conjecture, which has since become the Four Color Theorem Reference 7Reference 8Reference 54. Even though this theorem, its history and ramifications are of substantial interest, we omit any further discussion of it, because our primary interest lies in surfaces other than the sphere. Thus we skip to the next development, which is the well-known Heawood formula from 1890, asserting that a graph embedded in a surface of Euler genus can be colored with at most colors. This formula is actually a fairly easy concequence of Euler’s formula. It turned out that the bound it gives is best possible for every surface except the Klein bottle, but that was not established until the seminal work of Ringel and Youngs Reference 53 in the 1960s. Franklin Reference 33 proved that every graph embeddable in the Klein bottle requires only six colors, whereas Heawood’s bound predicts seven. Dirac Reference 19 and Albertson and Hutchinson Reference 3 improved Heawood’s result by showing that every graph in is actually -colorable, unless it has a subgraph isomorphic to the complete graph on vertices.

Thus the maximum chromatic number for graphs embeddable in a surface has been determined for every surface. However, a more modern approach to coloring graphs on surfaces, initiated by Thomassen in the 1990s, is based on the observation that only very few graphs embedded in a fixed surface have chromatic number close to ; in fact, as we are about to see, most have chromatic number at most five. To make this assertion precise, we need a definition. Let be an integer. We say that a graph is -critical if it is not -colorable, but every proper subgraph of is -colorable. Using Euler’s formula, Dirac Reference 20 proved that for every and every surface there are only finitely many -critical graphs that embed in . By a result of Gallai Reference 36, this can be extended to . We will see in a moment that this extends to by a deep result of Thomassen. First however, let us mention a predecessor of this theorem, also due to Thomassen Reference 55.

Theorem 2.1.

For every surface there exists an integer such that if is a graph embedded in such that every non-null-homotopic cycle in has length at least , then is -colorable.

The following related result was obtained by Fisk and Mohar Reference 35.

Theorem 2.2.

There exists an absolute constant such that if is a graph embedded in a surface of Euler genus such that every non-null-homotopic cycle in  has length at least , then

is -colorable, and

if is triangle-free, then it is -colorable, and

if has no cycles of length five or less, then it is -colorable.

The other theorem of Thomassen Reference 60 reads as follows.

Theorem 2.3.

For every surface , there are (up to isomorphism) only finitely many -critical graphs that embed in .

It is clear that Theorem 2.3 implies Theorem 2.1, but we state them separately for two reasons. One is historical, and the other is that we will be able to improve the bound on in Theorem 2.1 to an asymptotically best possible value, and that version is no longer a consequence of Theorem 2.3. Theorem 2.3 immediately implies a polynomial-time algorithm for deciding whether a graph on a fixed surface is -colorable. By a result of Eppstein Reference 31 there is, in fact, a linear-time algorithm, as follows.

Corollary 2.4.

For every surface there exists a linear-time algorithm that decides whether an input graph embedded in is -colorable.

Corollary 2.4 guarantees the existence of an algorithm. To construct an algorithm we would need to find the list of all the -critical graphs that can be embedded in . Such lists are known only for the projective plane Reference 3, the torus Reference 56, and the Klein bottle Reference 14Reference 43. However, the proof of Theorem 2.3 can be converted to a finite-time algorithm that will output the desired list; thus the algorithm, even though not explicitly known, can be constructed in finite-time.

Theorem 2.3 is best possible as it does not extend to -critical graphs. Indeed, Thomassen Reference 60, using a construction of Fisk Reference 34, constructed infinitely many -critical graphs that embed in any given surface other than the sphere. The next logical question then is whether Corollary 2.4 holds for - or -coloring. For -coloring the answer is no, unless P=NP, because -colorability of planar graphs is one of the original NP-complete problems of Garey and Johnson Reference 37. For -coloring there is a trivial algorithm when is the sphere by the Four Color Theorem, and for all other surfaces it is an open problem. Given the difficulties surrounding the known proofs of the Four Color Theorem, the prospects for a resolution of that open problem in the near future are not very bright.

A classical theorem of Grötzsch Reference 39 asserts that every triangle-free planar graph is -colorable. Thomassen Reference 58Reference 59Reference 61 found three reasonably short proofs, and extended the theorem to the projective plane and the torus. However, for the extensions to hold one has to assume that the graph has no cycles of length at most four. Those results prompted Thomassen to formulate the following research program. First, let us recall that the girth of a graph is the maximum integer (or infinity) such that every cycle has length at least . Thomassen asked for which pairs of integers and is it the case that for every surface there are only finitely many -critical graphs of girth at least in . If the answer is positive, then -colorability of graphs of girth at least in can be tested in linear time. If the answer is negative, then there is still the question whether the -colorability of graphs of girth at least in can be tested in polynomial-time.

These questions have now been resolved, except for the already mentioned case and . Let us briefly survey the results. It is fairly easy to see that for every surface there are only finitely many -critical graphs of girth at least in whenever either , or and , or . Theorem 2.3 states that this also holds when , and the following deep theorem, also due to Thomassen Reference 62, says that it also holds when and .

Theorem 2.5.

For every surface there are only finitely many -critical graphs of girth at least five in .

Dvořák, Král’, and Thomas strengthened this theorem by giving an asymptotically best possible bound on the size of the -critical graphs, as follows.

Theorem 2.6.

There exists an absolute constant such that every -critical graph of girth at least five embedded in a surface of Euler genus has at most vertices.

Thus the only case we have not yet discussed is and ; that is, -coloring triangle-free graphs on a fixed surface. There are infinitely many -critical triangle-free graphs on any surface other than the sphere, but, nevertheless, -colorability of triangle-free graphs on any fixed surface can be tested in polynomial time Reference 24Reference 28. However, this algorithm requires different methods. The theory we develop in this paper does not seem to apply to -coloring triangle-free graphs.

2.2. List-coloring graphs on surfaces

Our main topic of interest is list coloring, a generalization of ordinary vertex-coloring, introduced in the Introduction. List coloring differs from regular coloring in several respects. One notable example is that the Four Color Theorem does not generalize to list-coloring. Indeed, Voigt Reference 66 constructed a planar graph that is not -choosable. On the other hand Thomassen Reference 57 proved the following remarkable theorem with an outstandingly short and beautiful proof.

Theorem 2.7.

Every planar graph is -choosable.

The following theorem was also proven by Thomassen, first in Reference 59 and later he found a shorter proof in Reference 61.

Theorem 2.8.

Every planar graph of girth at least five is -choosable.

The corresponding theorem that every planar graph of girth at least four is -choosable is an easy consequence of Euler’s formula; a slightly stronger version of it is part of Theorem 2.14 below. For later reference we need the following small generalization of the last three results mentioned. It follows easily from known results. A graph has crossing number at most one if it can be drawn in the plane such that at most one pair of edges cross.

Theorem 2.9.

Let be a graph of crossing number at most one and let be a type list-assignment for . Then is -colorable.

Proof.

Assume first that is triangle-free and that is a -list-assignment. The graph embeds in the projective plane. By Euler’s formula applied to a projective planar embedding of we deduce that has a vertex of degree at most three, and the theorem follows by induction by deleting such vertex. For the other two cases we may assume, by Theorems 2.7 and 2.8, that is not planar. Thus has a planar drawing where exactly two edges cross, say and . Since is not planar, the vertices are pairwise distinct. Let . When is a -list-assignment, the theorem follows from Reference 64, Theorem 2 applied to the graph obtained from by adding the edges , , , and , except for those that are already present, because Reference 64, Theorem 2 guarantees that we can precolor the subgraph of induced by the vertices arbitrarily. Thus we may assume that has girth at least five. Let us assume first that some two of the vertices are adjacent in , say and are. By Reference 59, Theorem 2.1 the graph has an -coloring, where , , , , , , and for every other vertex of . Such an -coloring of is an -coloring of , as desired. We may therefore assume that no two of the vertices are adjacent in . Let be obtained from by adding two new vertices and three edges so that will become a path in . Let and . By Reference 61, Theorem 2.1 applied to the graph there exists an -coloring of such that , , , and . Such an -coloring of is an -coloring of , as desired.

Let us recall that if is a list-assignment for a graph , then we say that is -critical if does not have an -coloring but every proper subgraph of does. Thomassen Reference 60, Theorem 4.4 proved the following theorem.

Theorem 2.10.

Let be a graph embedded in a surface of Euler genus . Let  be a list-assignment for and let be a set of vertices in such that for each . If is -critical, then .

Naturally then, Thomassen Reference 60, Problem 5 asked whether Theorem 2.3 generalizes to list-coloring, and in Reference 63, Conjecture 6.1 he conjectured that it indeed does.

Conjecture 2.11.

For every surface there are (up to isomorphism) only finitely many graphs such that embeds in and is -critical for some -list-assignment .

A proof of Conjecture 2.11 was announced by Kawarabayashi and Mohar Reference 44, but no proof appeared so far. We give a proof of Conjecture 2.11 in Theorem 3.11 below. Meanwhile, DeVos, Kawarabayashi, and Mohar Reference 17 generalized Theorem 2.1 to list-coloring.

Theorem 2.12.

For every surface there exists an integer such that if is a graph embedded in such that every non-null-homotopic cycle in has length at least , then is -list-colorable.

In Theorem 3.9 we give an independent proof of Theorem 2.12 with an asymptotically best possible bound on .

Kawarabayashi and Thomassen Reference 45 proved that Theorem 2.7 “linearly extends” to higher surfaces, as follows.

Theorem 2.13.

For every graph embedded in a surface of Euler genus there exists a set of size at most such that is -choosable.

We give an alternate proof of Theorem 2.13 (admittedly with a worse constant) and extend it to other coloring problems in Theorem 3.17.

2.3. Extending precolored subgraphs

An important proof technique is to extend a coloring of a subgraph to the entire graph. We will need the following result.

Theorem 2.14.

Let be a plane graph, let be a path in of length at most three such that some face of is incident with every edge of , and let be a type list-assignment for . Then every -coloring of extends to an -coloring of .

Proof.

When is a -list-assignment, the theorem follows from Reference 64, Theorem 2, and when has girth at least five, it follows from Reference 61, Theorem 2.1. Thus we may assume that is triangle-free and that is a -list-assignment. In that case Euler’s formula implies that has a vertex in of degree at most three, and the theorem follows by induction by deleting such vertex.

Theorem 2.14 obviously does not extend to arbitrarily long paths. However, in Reference 51 we have shown the following. By an outer cycle in a plane graph we mean the cycle bounding the outer face (the unique unbounded face).

Theorem 2.15.

Let be a plane graph with outer cycle , let be a -list-assignment for , and let be a minimal subgraph of such that every -coloring of that extends to an -coloring of also extends to an -coloring of . Then has at most vertices.

Earlier versions of this theorem were proved for ordinary coloring by Thomassen Reference 60, Theorem 5.5, who proved it with replaced by , and by Yerger Reference 67, who improved the bound to . If every vertex of has degree at least five and all its neighbors in belong to , then the only graph  satisfying the conclusion of Theorem 2.15 is the graph itself. It follows that the bound in Theorem 2.15 is asymptotically best possible.

Theorem 2.15 was a catalyst that led to this paper, because it motivated us to introduce the notion of a hyperbolic family of embedded graphs and to develop the theory presented here. It is an easy consequence of Theorem 2.15, and we show it formally in Theorem 5.3, that the family of embedded graphs that are -critical for some -list-assignment is hyperbolic.

An analog of Theorem 2.15 for graphs of girth at least five and -list-assignments was obtained by Dvořák and Kawarabayashi Reference 22, Theorem 5.

Theorem 2.16.

Let be a plane graph of girth at least five, let be the outer cycle of , and let be a -list-assignment for . Let be a minimal subgraph of  such that every -coloring of that extends to an -coloring of also extends to an -coloring of . Then has at most vertices.

Let us remark that, unlike the previous two results, the corresponding theorem for graphs of girth at least four and -list-assignments is a fairly easy consequence of Euler’s formula. We state it here for future reference. It follows from Lemma 5.10.

Theorem 2.17.

Let be a triangle-free plane graph, let be the outer cycle of , and let be a -list-assignment for . Let be a minimal subgraph of such that every -coloring of that extends to an -coloring of also extends to an -coloring of . Then has at most vertices.

Thomassen conjectured Reference 60 that if is a set of vertices of a planar graph and any two distinct members of are at least some fixed distance apart, then any precoloring of extends to a -coloring of . Albertson Reference 2 proved this in 1998, and conjectured that it generalizes to -list-coloring. Albertson’s conjecture was recently proved by Dvořák, Lidický, Mohar, and Postle Reference 30, as follows.

Theorem 2.18.

There exists an integer such that the following holds: If is a plane graph with a -list-assignment and is such that every two distinct vertices of are at distance at least in , then any -coloring of extends to an -coloring of .

Albertson and Hutchinson Reference 4 have generalized Albertson’s result to other surfaces. They proved that if the graph is locally planar, then any precoloring of vertices far apart extends.

Theorem 2.19.

Let be a graph embedded in a surface of Euler genus such that every non-null-homotopic cycle of has length at least . If is such that every two distinct vertices of are at distance at least in , then any -coloring of extends to a -coloring of .

Meanwhile, Dean and Hutchinson Reference 16 have proven that if is a graph embedded in a surface , is an -list-assignment for , and such that all distinct vertices have pairwise distance at least four, then any -coloring of extends to an -coloring of . They also asked whether their result extends to lists of other sizes. We restate their question as follows.

Question 2.20.

For which do there exist such that if is a graph embedded in a surface such that every non-null-homotopic cycle in has length at least , is a -list-assignment for and is such that all distinct vertices have pairwise distance at least , then any -coloring of  extends to an -coloring of ?

In Theorem 3.19 we give an independent proof of a common generalization of Theorems 2.18 and 2.19, which also answers Question 2.20.

We say a graph is drawn in a surface if is embedded in except that there are allowed to exist points in where two—but only two—edges cross. We call such a point of and the subsequent pair of edges of , a crossing. Dvořák, Lidický, and Mohar Reference 29 proved that crossings far apart instead of precolored vertices also leads to -list-colorability. The distance between the crossing of the edge with the edge and the crossing of the edge with the edge is the length of the shortest path in with one end in the set and the other end in the set .

Theorem 2.21.

If can be drawn in the plane with crossings pairwise at distance at least , then is -list-colorable.

In Theorem 3.21 we give an independent proof of Theorem 2.21 (but with a worse constant) and generalize it to arbitrary surfaces.

2.4. Coloring with short cycles far apart

Earlier we mentioned the classical theorem of Grötzsch Reference 39 that every triangle-free planar graph is -colorable. Aksenov Reference 1 proved that the same is true even if the graph is allowed to have at most three triangles. Havel Reference 40 asked whether there exists an absolute constant such that if every two triangles in a planar graph are at least distance apart, then the graph is -colorable. Havel’s conjecture was proved in Reference 27. The conjecture cannot be extended to -list-coloring verbatim, because there exist triangle-free planar graphs that are not -list-colorable. However, Dvořák Reference 21 proved the following.

Theorem 2.22.

There exists an absolute constant such that if is a planar graph and every two cycles in of length at most four are at distance in of at least , then is -list-colorable.

We give a different proof of and generalize Dvořák’s result to locally planar graphs on surfaces in Theorem 3.20.

2.5. Exponentially many colorings

It is an easy consequence of the Four Color Theorem that every planar graph has exponentially many -colorings. Birkhoff and Lewis Reference 9 obtained an optimal bound, as follows.

Theorem 2.23.

Every planar graph on vertices has at least distinct -colorings, and if it has exactly distinct -colorings, then it is obtained from a triangle by repeatedly inserting vertices of degree three inside facial triangles.

Thomassen Reference 63 proved that a similar bound holds for -colorable graphs embedded in a fixed surface.

Theorem 2.24.

For every surface there exists a constant such that every -colorable graph on vertices embedded in has at least distinct -colorings.

In Reference 63, Theorem 2.1 Thomassen gave a short and elegant proof of Theorem 2.24. The argument also applies to -colorings of triangle-free graphs and -colorings of graphs of girth at least five. In Reference 64, Problem 2 Thomassen asked whether the bound of Theorem 2.23 holds for -list-colorings and proved that a planar graph has exponentially many -list-colorings.

Theorem 2.25.

If is a planar graph and is a -list-assignment for , then has at least distinct -colorings.

Thomassen Reference 64, Problem 3 also asked whether Theorem 2.24 extends to -list-colorings.

Problem 2.26.

Let be a graph embedded in a surface and let be a -list-assignment for . Is it true that if is -colorable, then has at least distinct -colorings, where is a constant depending only on the Euler genus of ?

In Theorem 3.22 we prove the weaker statement that the answer is yes if is replaced by . This result was announced by Kawarabayashi and Mohar Reference 44, but no proof appeared so far.

Thomasssen Reference 65 also proved that planar graphs of girth at least five have exponentially many -list-colorings.

Theorem 2.27.

If is a planar graph of girth at least five and is a -list-assignment for , then has at least distinct -colorings.

Theorem 3.22 extends this to graphs on surfaces.

3. Main coloring results

In this section we formulate Theorem 3.8, our main coloring result, and derive a number of consequences from it.

We need to generalize rooted graphs to allow distinguished facial cycles, rather than just distinguished vertices. Hence the following definition.

Definition 3.1.

A ring is a cycle or a complete graph on one or two vertices. A graph with rings is a pair , where is a graph and is a set of vertex-disjoint rings in . We will often say that is a graph with rings , and sometimes we will just say that is a graph with rings, leaving out the symbol for the set of rings.

Definition 3.2.

We say that a graph with rings is embedded in a surface  if the underlying graph is embedded in in such a way that for every ring there exists a component of bd such that is embedded in , no other vertex or edge of is embedded in , and every component of bd includes some ring of . In those circumstances we say that is an embedded graph with rings. Sometimes we will abbreviate this by saying that is an embedded graph with rings. We also say that is a graph with rings embedded in a surface . A vertex is called a ring vertex if it belongs to some ring in .

Thus if is a graph with rings embedded in a surface , then the rings are in one-to-one correspondence with the boundary components of . In particular, if has no rings, then has no boundary.

Definition 3.3.

By a canvas we mean a quadruple , where is an embedded graph with rings and is a list-assignment for such that the subgraph is -colorable.

Let us remark that this definition of a canvas is more general than the one we used in Reference 50 or Reference 51. We need the notion of a critical canvas, which we define as follows.

Definition 3.4.

Let be a graph, let be a subgraph of , and let be a list-assignment for . We say that is -critical with respect to if and for every proper subgraph of that includes as a subgraph there exists an -coloring of that extends to an -coloring of , but does not extend to an -coloring of . We shall abbreviate -critical by -critical. A canvas is critical if is -critical with respect to .

Definition 3.5.

For we define to be the family of all canvases such that for every that does not belong to a ring, and every cycle in of length at most is equal to a ring.

The above-defined families are the three most interesting families of canvases to which our theory applies. More generally, the theory applies to what we call good families of canvases. Those are defined in Definition 8.3, but for now we can treat this notion as a black box. We now state the fact that the families just defined are good. We prove it in Theorem 8.8, using results from other papers.

Theorem 3.6.

The families are good families of canvases.

Definition 3.7.

If is a surface with boundary, let denote the surface without boundary obtained by gluing a disk to each component of the boundary.

The following is our main coloring theorem.

Theorem 3.8.

For every good family of canvases there exist such that the following holds. Let , let be the Euler genus of , let be the total number of ring vertices in , and let be the maximum number of vertices in a ring in . Then has a subgraph such that includes all the rings in , has at most vertices, and for every -coloring of

either does not extend to an -coloring of , or

extends to at least distinct -colorings of .

Furthermore, either

(a)

there exist distinct rings such that the distance in between them is at most , or

every component of satisfies one of the following conditions:

(b)

the graph has a cycle that is not null-homotopic in ; if includes no ring vertex, then the length of is at most , and otherwise it is at most , or

(c)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Theorem 3.8 is an immediate consequence of Theorem 8.11, proved in Section 8, where it is deduced from Theorem 7.11. In the rest of this section we derive consequences of Theorem 3.8, some of which are new and some of which improve previous results. As a first consequence we improve the bound on in Theorems 2.1 and 2.12, and extend Theorem 2.2 to list-coloring, while simultaneously improving upon the first and third outcome.

Theorem 3.9.

There exists an absolute constant such that if is a graph embedded in a surface of Euler genus in such a way such that every non-null-homotopic cycle in has length exceeding , then is -colorable for every type list-assignment .

Proof.

By Theorem 3.6 the family is a good family of canvases. Let be as in Theorem 3.8 applied to the family , let , and be as stated, and let . Then . Let be a subgraph of whose existence is guaranteed by Theorem 3.8. Since and every non-null-homotopic cycle in has length exceeding , the only way can satisfy one of the conditions (a)–(c) is that is the null graph. It follows that has an -coloring, as desired.

In fact, the proof shows the following strengthening.

Theorem 3.10.

There exist absolute constants such that if is a graph embedded in a surface of Euler genus in such a way such that every non-null-homotopic cycle in has length exceeding and is a type list-assignment for , then has at least distinct -colorings.

The bound in Theorems 3.9 and 3.10 is asymptotically best possible, because by Reference 10, Theorem 4.1 there exist non--colorable graphs on vertices with girth , and the Euler genus of a graph on vertices is obviously at most .

The next consequence of Theorem 3.8 settles Conjecture 2.11, improves the bound on the size of -critical graphs in Theorem 2.3, and extends Theorem 2.6 to list-critical graphs.

Theorem 3.11.

There exists an absolute constant such that if is a graph embedded in a surface of Euler genus and there exists a type list-assignment  for such that is -critical, then .

Proof.

By Theorem 3.6 the family is a good family of canvases. Let be as in Theorem 3.8 applied to the family , let , and be as stated, and let . Then . Let be a subgraph of , whose existence is guaranteed by Theorem 3.8. Then , and either is not -colorable or is -colorable. Since is -critical it follows that , and hence the theorem holds.

The bound in Theorem 3.11 is asymptotically best possible. For -list-assignments it can be seen by considering the graphs obtained from copies of by applying Hajos’ construction. For - and -list-assignments we replace by a -critical graph of girth four and a -critical graph of girth five, respectively.

We have the following immediate corollary. If is an integer, then we say that a graph is -list-critical if is not -list-colorable but every proper subgraph of is.

Corollary 3.12.

There exists an absolute constant such that if and is a -list-critical graph of girth at least embedded in a surface of Euler genus , then .

Proof.

Let be a -list-critical graph. Then is not -colorable for some -list-assignment , and hence has an -critical subgraph . Thus is not -colorable, and hence not -list-colorable. The -list-criticality of implies that . Thus we have shown that is -critical, and the corollary follows from Theorem 3.11.

By the result of Eppstein Reference 31 mentioned earlier we have the following algorithmic consequence.

Corollary 3.13.

For every surface and every there exists a linear-time algorithm to decide whether an input graph of girth at least embedded in  is -list-colorable.

We can also deduce the following algorithm for -coloring.

Corollary 3.14.

For every surface there exists a linear-time algorithm to decide whether given an input graph embedded in and a type list-assignment for there exists an -coloring of .

Proof.

For simplicity we present the proof for -list-assignments, but it clearly extends to the other two types of list-assignment. By Theorem 3.11 there exists a finite list of pairs , where is a graph embedded in and is a -list-assignment for , such that an input graph embedded in is -colorable if and only if there is no such that , where means that there exists a subgraph isomorphism and a mapping such that for every . We may actually assume without loss of generality that for all the pairs under consideration and every element , the set of vertices induces a connected subgraph of .

The corollary now follows from the fact that, under the assumption we just made, for fixed it can be tested in linear-time whether . This can be done by using Eppstein’s algorithm Reference 31, which works in this more general setting, even though it is not stated that way, or it can be deduced more directly from some of the generalizations of Eppstein’s algorithm to relational structures, such as Reference 25.

Definition 3.15.

A graph with rings embedded in a surface is -cell embedded in if every face of is simply connected.

Let be a graph embedded in a surface , and let be the surface obtained by attaching a disk to every facial walk of the embedded graph . Then is -cell embedded in , and we say that the Euler genus of is the combinatorial genus of the embedded graph . We need the following well-known result. We include a proof for convenience.

Lemma 3.16.

Let be a graph embedded in a surface of Euler genus , let be pairwise disjoint subgraphs of , and for let be the combinatorial genus of . Then .

Proof.

We actually prove the stronger statement that if and is the combinatorial genus of , then . It suffices to prove this statement for and we may assume that both and are connected. There exists a simple closed curve whose image is disjoint from , and either  is non-separating, or it separates from . But cannot be non-separating, given that is the combinatorial genus of ; thus it separates from and the lemma follows.

Another consequence of Theorem 3.11 is the following, which gives an independent proof and extension of Theorem 2.13 (but with a worse constant).

Theorem 3.17.

There exists an absolute constant such that for every graph embedded in a surface of Euler genus and every type list-assignment for there exists a set of size at most such that is -colorable.

Proof.

Let be as in Theorem 3.11. We claim that satisfies the conclusion of the theorem. Let be embedded in a surface of Euler genus , and let be a type list-assignment for . Let be a maximal family of pairwise disjoint -critical subgraphs of , and let . Then is -colorable by the maximality of the family. For let be the Euler genus of . By Theorem 3.11 and Lemma 3.16

as desired.

The proof of Theorem 3.11 actually implies the following stronger form, which also generalizes Theorem 2.10 to -list-coloring (albeit with replaced by a larger constant).

Theorem 3.18.

There exists an absolute constant with the following property. Let be a graph without rings embedded in a surface of Euler genus , let , and let be a type list-assignment for . Then there exists a subgraph of such that , and every -coloring of  either

(1)

does not extend to an -coloring of , or

(2)

extends to an -coloring of .

Proof.

By Theorem 3.6 the family is a good family of canvases. Let be as in Theorem 3.8 applied to the family , let , and be as stated, let consist of as isolated vertices, and let be obtained from by deleting, for every , an open disk disjoint from whose closure intersects in . Then . Let be a subgraph of , whose existence is guaranteed by Theorem 3.8. Then , and every -coloring of either does not extend to an -coloring of , or extends to an -coloring of , as desired.

The next theorem implies the former Albertson’s conjecture (Theorem 2.18) and generalizes it to other surfaces, it generalizes Theorem 2.19 to list-coloring, it answers Question 2.20, and implies another result which we will discuss next. Let us remark that the difference between and is that the former graph includes edges of with both ends in , including those that do not belong to .

Theorem 3.19.

There exist absolute constants such that the following holds. Let and assume that every cycle in that is non-null-homotopic in has length at least , where is the Euler genus of . If each ring in has at most four vertices and every two distinct members of are at distance in of at least , then every -coloring of extends to an -coloring of .

Proof.

By Theorem 3.6 the family is a good family of canvases. Let  be as in Theorem 3.8 applied to the family , let , and let . We claim that and satisfy the conclusion of the theorem. To prove that let be as stated. By Theorem 3.8 there exists a subgraph of such that includes all the rings in and for every -coloring of , either does not extend to an -coloring of , or extends to an -coloring of ; and satisfies one of (a)–(c) of Theorem 3.8. But does not satisfy (a) or (b) by hypothesis, and hence it satisfies (c). It follows from Theorem 2.14 that every -coloring of extends to an -coloring of . Consequently every -coloring of extends to an -coloring of , as desired.

The following theorem follows immediately from Theorem 3.19. The first assertion gives an independent proof of and generalizes Theorem 2.22.

Theorem 3.20.

There exist absolute constants such that the following holds. Let be a graph without rings embedded in a surface (without boundary) of Euler genus in such a way that every non-null-homotopic cycle in has length at least . Then

if every two cycles in of length at most four are at distance at least in , then is -list-colorable, and

if every two triangles in are at distance at least in , then is -list-colorable.

We obtain the following generalization and independent proof of Theorem 2.21. For -list-assignments it follows from Theorem 3.19 by replacing each crossing by a cycle of length four, but for the other two cases we need a more careful argument.

Theorem 3.21.

There exist absolute constants and such that the following holds. Let be a graph without rings drawn with crossings in a surface (without boundary) of Euler genus , and let be a type list-assignment for . Assume that every edge crosses at most one other edge. Let be the graph embedded in  obtained from by repeatedly replacing a pair of crossing edges by a degree four vertex adjacent to the ends of and . If every non-null-homotopic cycle in  has length at least and every pair of the new vertices of are at distance at least  in , then is -colorable.

Proof.

By Theorem 3.6 the family is a good family of canvases. Let and be as in Theorem 3.8, when the latter is applied to the family . Let and . We will show that and satisfy the conclusion of the theorem.

We construct a graph with rings embedded in a surface as follows. For every pair of crossing edges and we do the following. First of all, we may assume that the vertices are pairwise distinct, for otherwise we could eliminate the crossing by redrawing one of the crossing edges. We add new vertices and new edges in such a way that and the new vertices will form a facial cycle , the vertices will appear on in the order listed, and will be at distance at least four apart on . We remove the face bounded by  from the surface and we declare to be a ring. By repeating this construction for every pair of crossing edges we arrive at a graph with rings embedded in a surface . For let , and for let be an arbitrary set of size five. It follows that .

By Theorem 3.8 there exists a subgraph of that includes all the rings in such that satisfies one of the conditions (a)–(c) of Theorem 3.8, and every -coloring of that extends to an -coloring of also extends to an -coloring of . The choice of and implies that does not satisfy (a) or (b), and hence it satisfies (c).

We now construct an -coloring of . Let . Let be the unique component of containing . By (c) it includes no other ring. Let be the corresponding subgraph of ; that is, is obtained from by deleting the new vertices and adding the two crossing edges with ends in . By Theorem 2.9 the graph has an -coloring. Let the restriction of to be obtained by taking the restriction of one such -coloring to the four ends of the crossing edges, and extending it to arbitrarily. This completes the construction of the -coloring . It follows from the construction that extends to an -coloring of , and hence it extends to an -coloring of , which by construction is also an -coloring of , as desired.

Finally we prove a weaker version of Problem 2.26. It is implied by the following more general theorem by setting .

Theorem 3.22.

There exist constants such that the following holds. Let , let be the Euler genus of , and let be the total number of ring vertices. If is an -coloring of such that extends to an -coloring of , then extends to at least distinct -colorings of .

Proof.

By Theorem 3.6 the family is a good family of canvases. Let be as in Theorem 3.8 applied to the family , let be as stated, and let be a subgraph of as in Theorem 3.8. It follows that if an -coloring of extends to an -coloring of , then extends to at least distinct -colorings of .

4. Exponentially many colorings

The main result of this short section is Lemma 4.5, which strengthens Theorem 2.15 by saying that if an -coloring of extends to , then it has exponentially many extensions.

When Thomassen proved Theorem 2.25 in Reference 64, Theorem 4, he proved a stronger statement, which we state in a slightly stronger form.

Theorem 4.1.

Let be a plane graph, let be the set of all vertices of that are incident with the outer face, let be a subpath of of length one or two with every vertex and edge incident with the outer face, and let be a list-assignment for such that for every and for every . Let be the number of vertices such that . If has an -coloring, then it has at least distinct -colorings, unless and there exists such that and is adjacent to all the vertices of .

Proof.

This is proved in Reference 64, Theorem 4 under the additional assumption that is a near-triangulation. If the outer face of is bounded by a cycle, then the theorem follows from Reference 64, Theorem 4 by adding edges. Otherwise can be written as , where , and the theorem follows by induction applied to and .

We use Theorem 4.1 to deduce the following corollary.

Corollary 4.2.

Let be a plane graph with outer cycle of length at most four, let  be a -list-assignment for , let be an -coloring of , and let be the number of -colorings of that extend . If , then , unless and there exists a vertex not in adjacent to all the vertices of .

Proof.

Let . Let , , and for let if is a neighbor of and otherwise. Apply Theorem 4.1 to , and . It follows that there are at least distinct -colorings of , unless and there exists with and is adjacent to all vertices of . But then and is adjacent to all the vertices of .

Definition 4.3.

Let us recall that the outer face of a plane graph is the unique unbounded face; all other faces of are called internal. We denote the set of internal faces of by . If is a face of a -connected plane graph , then we let denote the length of the cycle bounding . Likewise, if is a cycle in , then we denote its length by . If is a subgraph of and , then we denote by the subgraph of consisting of all vertices and edges drawn in the closure of . If is a -connected plane graph with outer cycle , then we define

Let us recall that -critical graphs were defined in Definition 3.4. When we proved Theorem 2.15 in Reference 51, we have actually proven the following stronger result Reference 51, Theorem 6.1, which we will now need.

Theorem 4.4.

Let be a plane graph with outer cycle , and let be a -list-assignment for . If is -critical with respect to , then

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

Lemma 4.5.

Let be a plane graph with outer cycle , let be a -list-assignment for , and let be an -coloring of that extends to an -coloring of . Then , where is the number of extensions of to .

Proof.

We proceed by induction on the number of vertices of . It follows from Corollary 4.2 that we may assume that is -connected and there does not exist a separating triangle in , and that if there exists a separating -cycle in , then there must exist a vertex adjacent to all the vertices of , as otherwise the theorem follows by induction. We may assume that by Corollary 4.2.

First suppose there exists a vertex with at least three neighbors on , and let . Let us first handle the case when has at least four neighbors on . Then , as is easily seen. Let be an extension of to that extends to an -coloring of . For all , letting denote the cycle bounding , it follows by induction that has at least extensions into . Thus

where the second inequality holds because . The lemma follows.

Let us assume next that has exactly three neighbors on . Let be distinct colors not equal to for all neighbors of that belong to . For let and for all . If both and extend to -colorings of , it follows by induction applied to the faces of , using the same calculation as in the previous paragraph and the fact that , that . Yet ; hence and the lemma follows.

So we may suppose without loss of generality that does not extend to an -coloring of . Hence there exists such that has a subgraph that is -critical with respect to . Let . Then

where the first inequality follows from Theorem 4.4. As extends to an -coloring of , extends to an -coloring of that extends to . For all , letting denote the cycle bounding , it follows by induction that has at least extensions into . Thus

and the lemma follows.

We may therefore assume that every vertex in has at most two neighbors in . Let us redefine to be the graph . For let be obtained from by removing for every neighbor of that belongs to , and let . Thus every vertex in has exactly two neighbors in . Let us define an auxiliary multigraph with vertex-set and edge-set by saying that the ends of an edge are the two neighbors of in . Then is an outerplanar multigraph.

Let be adjacent vertices of . We claim that there are at most three edges of with ends . Indeed, suppose that say are pairwise distinct and all have ends . Then for all distinct integers the vertices form a cycle of length four. For one such pair this cycle includes the two vertices of in its interior, contrary to what we established at the beginning of the proof. Thus at most three edges of have ends . Furthermore, we claim that if are adjacent in , then at most one edge of has ends . Indeed, suppose that say distinct both have ends . Then for the vertices form a cycle of length three, contrary to there not existing a separating triangle as established at the beginning of the proof.

The results of the previous paragraph imply that . Since we deduce that . By Theorem 4.1 applied to , and an arbitrarily chosen one-edge path we obtain

as desired.

5. Definition and examples of hyperbolic families

In this section we define hyperbolic families in full generality and we give examples of families of embedded graphs with rings that are hyperbolic. Let us recall that embedded graphs with rings were defined at the beginning of Section 3.

Definition 5.1.

Let be a family of non-null embedded graphs with rings. We say that is hyperbolic if there exists a constant such that if is a graph with rings that is embedded in a surface , then for every closed curve that bounds an open disk and intersects only in vertices, if includes a vertex of , then the number of vertices of in is at most . We say that is a Cheeger constant for . Let us point out that since is an open disk, it does not include the vertices of that belong to the image of .

The next lemma shows that if a planar graph with rings belongs to a hyperbolic family, then it has at least one ring.

Lemma 5.2.

Let be a hyperbolic family of embedded graphs with rings, and let be a Cheeger constant for . Let be a graph embedded in a surface of Euler genus zero with rings and let be the total number of ring vertices. Then and if , then has at most non-ring vertices.

Proof.

As is non-null, as otherwise we may take a closed curve not intersecting any vertices of that bounds a disk containing all vertices of , a contradiction to the definition of hyperbolic family. If , then let be a closed curve obtained from a closed curve tracing the ring by pushing it slightly into the interior of in such a way that the image of under will intersect the boundary of only in . By the definition of hyperbolic family applied to the curve we deduce that has at most non-ring vertices, as desired.

5.1. Examples arising from (list-)coloring

Theorem 2.15 implies that embedded -list-critical graphs form a hyperbolic family. To obtain a better bound on the Cheeger constant we actually use the stronger Theorem 4.4. More generally, we have the following.

Theorem 5.3.

Let be the family of all embedded graphs with rings such that is -critical with respect to some -list-assignment. Then is hyperbolic with Cheeger constant .

Proof.

Let be a graph with rings embedded in a surface of Euler genus such that is -critical with respect to a -list-assignment , let be the total number of ring vertices, and let be a closed curve that bounds an open disk and intersects only in vertices. To avoid notational complications we will assume that is a simple curve; otherwise we split vertices that visits more than once to reduce to this case. We may assume that includes at least one vertex of , for otherwise there is nothing to show. Let be the set of vertices of intersected by . Then by Theorem 2.14.

Let be the subgraph of consisting of all vertices and edges drawn in the closure of . Let be obtained from as follows. For every pair of vertices that are consecutive on the boundary of we do the following. If and  are adjacent in , then we re-embed the edge so that it will coincide with a segment of . If are not adjacent, then we introduce a new vertex and join it by edges to both and , embedding the new edges in a segment of . Thus  has a cycle embedded in the image of , and hence may be regarded as a plane graph with outer cycle . For let , and for  let be an arbitrary set of size five. It follows that is -critical with respect to . Since every vertex is incident with an internal face of length at least four, Theorem 4.4 implies that the number of vertices of in is at most , as desired.

For our next example, we need the following lemma. The degree of a vertex in a graph will be denoted by , or if the graph is understood from the context.

Lemma 5.4.

Let be a graph embedded in the closed unit disk with vertices embedded on the boundary and vertices embedded in the interior of the disk. If every vertex embedded in the interior of the disk has degree in of at least seven, then .

Proof.

We may assume that every vertex on the boundary of the disk has a neighbor in the interior, for otherwise we may delete the boundary vertex and apply induction. Let us first assume that , and let be the planar graph obtained from by joining by an edge every pair of non-adjacent consecutive vertices on the boundary of the disk, and then adding one new vertex in the complement of the disk and joining it by an edge to every vertex on the boundary. It follows that every vertex on the boundary of the disk has degree at least four in . Since is planar, we have

and the lemma follows. If , then by the planarity of

a contradiction, since .

Example 5.5.

Let be the family of all embedded graphs with rings such that every vertex of that does not belong to a ring has degree at least seven. Then is hyperbolic with Cheeger constant by Lemma 5.4.

Example 5.6.

The family from the previous example includes all embedded graphs with rings such that is -critical with respect to some -list-assignment. That gives an alternate proof that the latter family is hyperbolic, and gives a better Cheeger constant than Theorem 5.3.

For our next example, we need the following lemma. The size of a face of an embedded graph, denoted by , is the sum of the lengths of the walks bounding .

Lemma 5.7.

Let be a graph embedded in the closed unit disk with vertices embedded on the boundary and vertices embedded in the interior of the disk. Assume that every vertex embedded in the interior of the disk has degree in of at least six, and if it has degree exactly six, then it is either incident with a face of size at least four, or it is adjacent to a vertex of degree at least seven, or it is adjacent to a vertex embedded on the boundary of the disk. Then .

Proof.

We may assume that every vertex on the boundary of the disk has a neighbor in the interior, for otherwise we may delete the boundary vertex and apply induction. Let us first assume that , and let be the planar graph obtained from by joining by an edge every pair of non-adjacent consecutive vertices on the boundary of the disk, and then adding one new vertex in the complement of the disk and joining it by an edge to every vertex on the boundary. It follows that every vertex on the boundary of the disk has degree at least four in .

Let be the set of vertices embedded in the interior of the disk with degree six, and let be the set of vertices embedded in the interior of the disk that have degree at least seven. Let be the subset of consisting of vertices adjacent to a vertex embedded on the boundary of the disk, let be the subset of consisting of vertices incident with a face of size at least four, and let be the subset of consisting of vertices adjacent to a vertex of degree at least seven. By assumption, . Since is planar, we have

Hence

However, is at most the sum of the sizes of faces of sizes at least four. That is,

Similarly, is at most the sum of degrees of vertices of degree at least seven. That is,

Hence . Yet since is planar,

Hence . Thus

as desired. If , then a similar calculation applied to the original graph gives a contradiction.

Example 5.8.

Let be the family of all embedded graphs with rings such that every vertex of that does not belong to a ring has degree at least six, and if it has degree exactly six, then it is either incident with a face of size at least four, or it is adjacent to a vertex of degree at least seven, or it is adjacent to a vertex that belongs to a ring. Then is hyperbolic with Cheeger constant by Lemma 5.7.

Example 5.9.

The family from the previous example includes all embedded graphs with rings such that is -critical with respect to some -list-assignment. That gives an alternate proof that the latter family is hyperbolic, and gives a better Cheeger constant than Theorem 5.3.

For our next example, we need the following lemma.

Lemma 5.10.

Let be an embedded graph with rings such that every vertex that does not belong to a ring has degree at least four, every face of of size three is incident with an edge of , and every face of size exactly four is incident with either a vertex of degree at least five, or a vertex of degree at least one that belongs to a ring. Let be the Euler genus of and let be the total number of ring vertices. Then .

Proof.

For the purpose of this proof we will regard as a graph without rings embedded in the surface . Thus every cycle in bounds a face. We may assume that every ring vertex has degree at least one, for otherwise we may delete it and apply induction. Let be the set of all faces of . We begin by giving each vertex a charge of and every face a charge of . By Euler’s formula the sum of the charges is at most . Next we add four units of charge to every ring vertex of degree one, three units of charge to every ring vertex of degree at least two, and three units of charge to every face bounded by a cycle in . and an additional one unit of charge to every ring. Now the sum of charges is at most , and every vertex and every face has non-negative charge, except for triangles which have charge . In the next step every vertex of degree at least five and every ring vertex of degree at least two will send of a unit of charge to every incident face of length four, and every ring vertex of degree at least two will send an additional units of charge to every incident face of length three. Since no ring vertex of degree one is incident with a face of length four, this redistribution of the charges guarantees that every vertex and every face has non-negative charge. Furthermore, every face will have charge of at least , and every ring vertex will have charge of at least . Finally, every face will send of a unit of charge to every incident vertex. Then every face will end up with non-negative charge and every vertex will end up with a charge of at least . Thus , and the lemma follows.

Example 5.11.

Let be the family of all embedded graphs with rings such that every vertex of that does not belong to a ring has degree at least four, every face of of size three is incident with an edge of , and every face of size exactly four is incident with either a vertex of degree at least five, or a vertex of degree at least one that belongs to a ring. Then is hyperbolic with Cheeger constant  by Lemma 5.10. (The Cheeger constant is indeed and not , because we count vertices in the open disk bounded by .)

Lemma 5.12.

The family of all embedded graphs with rings such that every triangle in is not null-homotopic and is -critical with respect to some -list-assignment is hyperbolic with Cheeger constant .

Proof.

This follows by a similar argument as Theorem 5.3. Let and be as in the proof of that theorem, except that we do not introduce new vertices; instead we make adjacent every pair of vertices of that are consecutive on the boundary of . Then the graph with one ring embedded in the disk belongs to the family of Example 5.11. To see that let first . If has degree at most three, then every -coloring of (and one exists by the -criticality of ) can be extended to an -coloring of , a contradiction. Thus has degree at least four. Now let be a face of of size four, let be the set of (four) vertices incident with , and let us assume that every vertex in has degree four and does not belong to . The subgraph of induced by is a cycle, because this subgraph is contained in and no triangle in is null-homotopic. Thus every -coloring of (and one exists by the -criticality of ) can be extended to an -coloring of , a contradiction. This proves our claim that belongs to the family of Example 5.11, and hence satisfies the hyperbolicity condition with Cheeger constant , as desired.

Lemma 5.13.

Let be the family of embedded graphs with rings such that every cycle in of length at most four is not null-homotopic and is -critical with respect to some -list-assignment. Then is hyperbolic with Cheeger constant .

Proof.

This follows from Theorem 2.16 similarly as the proof of Theorem 5.3, the difference being that if the vertices are not adjacent we need to introduce two new vertices forming a path of length three with ends and .

A -coloring of a graph is acyclic if every two color classes induce an acyclic graph. Borodin Reference 12 proved that every planar graph has an acyclic -coloring.

Problem 5.14.

Let be an integer, and let be the class of all embedded graphs with no rings such that does not admit an acyclic -coloring, but every proper subgraph of does. Is hyperbolic for ?

Our next example is related to the following conjecture of Steinberg from 1976 Reference 42, Problem 2.9.

Conjecture 5.15.

Every planar graph with no cycles of length four or five is -colorable.

While Steinberg’s conjecture was recently disproved by Cohen-Addad, Hebdige, Král’, Li, and Salgado Reference 15, Borodin, Glebov, Montassier, and Raspaud Reference 13 showed that every planar graph with no cycles of length four, five, six, or seven is -colorable. That leads us to the following example.

Example 5.16.

Let be an integer, and let be the class of all embedded graphs with rings such that every cycle of length at least four and at most is a ring, and is -critical with respect to the list-assignment , where for every . Yerger Reference 67 proved that is hyperbolic for all . It seems plausible that the techniques developed by Borodin, Glebov, Montassier, and Raspaud Reference 13 can be used to extend this argument to , but the case is wide open.

Example 5.17.

Let be a graph, let be a list-assignment such that for every , and let be an integer. By a --coloring of we mean a mapping such that for every , and for every , every component of the subgraph of induced by vertices with has at most vertices. Thus is a --coloring if and only if it is an -coloring.

Let be the class of all embedded graphs with rings such that for every proper subgraph of that contains all the rings there exists a --coloring of the rings such that extends to a --coloring of but not to a --coloring of . The proof of Theorem 2.15 can be modified to show that is hyperbolic for every integer .

5.2. Examples pertaining to exponentially many colorings

Definition 5.18.

Let and . Let be a graph with rings embedded in a surface of Euler genus , let be the total number of ring vertices, and let be a list-assignment for . We say that is -exponentially-critical with respect to if and for every proper subgraph of that includes all the rings there exists an -coloring of such that there exist at least distinct -colorings of extending , but there do not exist at least distinct -colorings of extending .

Theorem 5.19.

Let and . Then the family of embedded graphs with rings that are -exponentially-critical with respect to some -list-assignment is hyperbolic with Cheeger constant (independent of and ).

Proof.

Let be a graph with rings embedded in a surface of Euler genus such that is -exponentially-critical with respect to a -list-assignment , let be the total number of ring vertices, and let be a closed curve that bounds an open disk and intersects only in vertices. To avoid notational complications we will assume that is a simple curve; otherwise we split vertices that visits more than once to reduce to this case. We may assume that includes at least one vertex of , for otherwise there is nothing to show. Let be the set of vertices of intersected by . Then by Corollary 4.2, since .

Let , and be defined as in the proof of Theorem 5.3. We may assume for a contradiction that . Let be the smallest subgraph of such that includes as a subgraph and every -coloring of  that extends to an -coloring of also extends to an -coloring of . Then  is -critical with respect to , and hence

by Theorem 4.4. (Let us recall that denotes the set of internal faces of .) Let . Thus is a proper subgraph of , and, in fact,

As is -exponentially-critical with respect to , there exists an -coloring of such that extends to at least distinct -colorings of , but does not extend to at least distinct -colorings of .

Let be an -coloring of that extends , let , let be the subgraph of drawn in the closure of , and let be the cycle bounding . Let  be defined as minus the number of vertices of incident with . By the definition of , extends to an -coloring of .

We claim that extends to at least distinct -colorings of . If is incident with no vertex of , then this follows from Lemma 4.5, and otherwise we argue as follows. If , then we apply Corollary 4.2, and otherwise we repeat the following construction. Let be incident with , and let be its two neighbors. We delete and either add an edge joining and (if ), or identify and (if ). We repeat this construction for every incident with , and apply Lemma 4.5 to the resulting graph. This proves our claim that extends to at least distinct -colorings of .

Let be the number of extensions of to . We have

and, by Theorem 4.4

and hence

where the first inequality uses the last claim of the previous paragraph, and the last inequality uses the inequality , which we established earlier.

The coloring extends to at least distinct -colorings of , and each such extension extends to at least distinct -colorings of . But then as , there exist at least distinct -colorings of extending , a contradiction.

The following is shown in Reference 49, Theorem 5.10.

Lemma 5.20.

Let and , and let be the family of embedded graphs with rings such that every cycle of length four or less is equal to a ring and is -exponentially-critical with respect to some -list-assignment. If , then the family is hyperbolic with Cheeger constant independent of and .

The following lemma follows from the work of Kelly and Postle Reference 46.

Lemma 5.21.

Let and , and let be the family of embedded graphs with rings such that every triangle is equal to a ring and is -exponentially-critical with respect to some -list-assignment. Then the family is hyperbolic with Cheeger constant .

Proof.

Let be as stated, let , and let be a closed curve that bounds an open disk and intersects only in vertices and includes a vertex of . Then by Theorem 2.14 and Reference 46, Theorem 7 the number of times, , the curve intersects , counting multiplicities, satisfies . By Reference 46, Theorem 8 the number of vertices in plus is at most . Thus the number of vertices in is at most , as desired.

6. The structure of hyperbolic families

In this section we investigate the structure of hyperbolic families of embedded graphs with rings. The main result is Theorem 6.28. We progress to the main result as follows. In Subsection 6.1, we define “frames”, which are essentially a subgraph of an embedded graph with only one face. We also show that every -cell-embedded graph has a frame (Lemma 6.2) and that minimal frames satisfy certain geodesic properties (Lemma 6.6). In Subsection 6.2 we show (Lemma 6.10) that for hyperbolic families the vertices in the interior of a disk have “logarithmic distance” to the boundary of the disk (an improvement over the trivial bound of linear distance as guaranteed by the definition of hyperbolic). Conversely, then, the neighborhood of a vertex contained in a disk experiences “exponential growth” in its diameter (Lemma 6.13).

Combining these ideas in Subsection 6.3, we prove the key Theorem 6.20, which says that large hyperbolic graphs have a short (i.e., as a function of the Cheeger constant) non-null homotopic cycle. Theorem 6.20 is the key to proving the main result, Theorem 6.28. The intuition for the proof of Theorem 6.20 is that if no such cycle exists, then balls around the midpoints of the segments of the frame have exponential growth in the lengths of the segments, while by hyperbolicity, the whole graph is linear in the size of the frame which is equal to the sum of the lengths of its segments. This linear upped bound versus the exponential growth then provides a contradiction.

Finally in Subsection 6.4, we prove the main result (Theorem 6.28), which provides a structural decomposition for hyperbolic families by asserting that all but vertices of a graph in such a family belong to cylinders of small edge-width (which we call “sleeves”). The proof of Theorem 6.28 proceeds inductively by cutting along the short cycles guaranteed by Theorem 6.20.

6.1. Frames

Definition 6.1.

Let be a graph with rings embedded in a surface and let  be a subgraph of that includes every ring of as a subgraph. Thus may be regarded as a graph with rings . We say that is a frame of if for every component of the subgraph of embedded in has exactly one face, this unique face is simply connected, and every vertex of of degree at most one in belongs to a ring in .

Lemma 6.2.

Let be a graph with rings such that is -cell embedded in a surface . Then has a frame.

Proof.

By applying the forthcoming argument to every component of we may assume that is connected. For a connected surface we proceed by induction on three times the number of faces of plus the number of vertices of degree one that belong to no ring. Since is connected and is -cell embedded, it follows that  is connected. If has one face and no vertex of degree one that belongs to no ring, then is a frame of itself. If is a vertex of of degree one that belongs to no ring, then is a graph with the same set of rings that is -cell embedded in . By induction, has a frame which is then a frame of . So we may suppose that has at least two faces. But then there is an edge of such that the two faces incident with are distinct. It follows that belongs to no ring. Thus is -cell embedded in  with the same set of rings and it has fewer faces than , and at most two more vertices of degree one than . By induction, has a frame which is then a frame of .

Definition 6.3.

Let be a graph with rings. We say that a vertex is smooth if it has degree two in and belongs to no ring. By a segment of we mean either a cycle in such that every vertex of the cycle except possibly one is smooth, or a path in such that every internal vertex of is smooth, the ends of are not smooth, and if has no edges, then its unique vertex is an isolated vertex of . It follows that every graph with rings is an edge-disjoint union of its segments. An internal vertex of a segment of is a vertex of that is smooth. Thus if is a path, then every vertex of other than its ends is internal. It follows that if is a segment, then either is a subset of the edge-set of a ring, or is disjoint from the edge-sets of all rings. In the latter case we say that is a non-ring segment.

Our next lemma shows that the number of non-ring segments of a frame is bounded by a function of the Euler genus and the number of rings.

Lemma 6.4.

Let be a graph with rings such that is -cell embedded in a surface of Euler genus . Let us assume that no component of is the sphere (without boundary) and that has components. If is a frame of , then the number of non-ring segments of is at most .

Proof.

It suffices to prove the lemma for . Thus is connected. We contract each ring of to a vertex, called a new vertex, and consider the natural embedding of the resulting multigraph in the surface obtained from by capping off each boundary component with a disk. Since is not the sphere it follows that has at least one vertex. If has exactly one vertex and no edges, then is the disk, , there are no non-ring segments, and hence the lemma holds. If is a cycle containing no ring vertex, then is the projective plane, , there is exactly one non-ring segment, and, again, the lemma holds.

Thus we may assume that has at least two vertices, and that it is not a cycle containg no ring vertex. Let be the multigraph obtained from by suppressing all vertices of degree two that are not new. Then the number of edges of is equal to the number of non-ring segments of . By Euler’s formula, , because has exactly one face. We have , because every vertex of that is not new has degree at least three, and every new vertex has degree at least one. Thus , and hence , as desired.

Definition 6.5.

Let be a graph with rings embedded in a surface , and let be a frame of . We say that the frame is optimal if the following two conditions are satisfied for every segment of :

(O1)

For every two distinct vertices and every path in with ends and and otherwise disjoint from , if is a subpath of with ends and and every internal vertex smooth, then .

(O2)

For every internal vertex of , every vertex and every path in with ends and and otherwise disjoint from , if are the two subpaths of with one end and the other end not smooth, then .

Let us remark that if is a cycle, then in condition (O1) there may be two choices for the path ; otherwise is unique. In condition (O2) the existence of the path implies that and belong to the same component of , and hence includes at least one vertex that is not smooth; therefore, the paths exist.

Lemma 6.6.

Let be a graph with rings that is -cell embedded in a surface . If is a frame of with minimum, then is an optimal frame.

Proof.

Let be as stated. To prove that it satisfies (O1) let , and be as in that condition. Then , for otherwise replacing by produces a frame with fewer edges than , a contradiction.

To prove that satisfies (O2) let , and be as in (O2). Now divides a simply connected face of into two faces and . When tracing the boundary of the face we encounter the segment twice. We may assume that are numbered in such a way that when tracing the boundary of starting on and moving toward , we encounter next. It follows that there is an index such that when tracing the boundary of we encounter twice and once. We deduce that by replacing by we obtain a frame. The minimality of implies that , as desired.

6.2. Separations and growth rates

Definition 6.7.

A separation of a graph is a pair such that and no edge of has one end in and the other in . If is a graph with rings that is embedded in a surface , then we say that a separation of  is flat if there exists a closed disk such that  includes and the interior of  includes and every edge of incident with a vertex of (regarded as a point set not including its ends). It follows that no ring vertex belongs to the interior of , and hence includes all ring vertices.

If belongs to a hyperbolic family, then, as we show next, the same linear bound holds for closed curves that bound “pinched” disks, or, equivalently, it holds for flat separations. But first we need the following lemma, which says that a flat separation is determined by the faces of a certain multigraph.

Lemma 6.8.

Let be a connected graph with rings embedded in a surface , let be a flat separation of , and let be as in the definition of flat separation. Then there exists a multigraph such that

,

is embedded in ,

the point set of intersects the point set of in ,

every vertex of has positive even degree,

every face of includes an edge of , and

all edges of that belong to the same face of either are all incident with a (possibly different) vertex of or all have both ends in .

Proof.

We proceed by induction on . If a vertex has no neighbor in , then we can remove from and apply induction. Similarly, if a non-ring vertex has no neighbor in , then we can remove from and apply induction. We may therefore assume that every vertex in has a neighbor in , and if it does not belong to a ring, then it also has a neighbor in . If or , then the null graph satisfies the conclusion of the lemma. We may therefore assume that and . In particular, .

We construct the graph in two steps. In the first step we embed, for every vertex , a set of “half-edges” incident with . In the second step we partition the set of all half-edges into pairs, and we merge each pair to form an edge of . To begin the first step let , and let be incident with , consecutive in the cyclic ordering of edges incident with determined by the embedding of in , and such that has its other end in and has its other end in . For every such pair we insert a half-edge incident with into the “angle” between and in such a way that the half-edge will be disjoint from . We also insert a half-edge into the “angle” between and , if is a component of the boundary of , , the other end of belongs to , and no other edge of is a subset of the “angle” between and . This process results in the insertion of a positive even number of half-edges incident with . In the second step we look at an arbitrary face of and one of its boundary walks , where a boundary walk is a sequence of vertices and edges or subsets of bd with the usual properties. Let be the set of half-edges inserted into the face that are incident with a vertex of . The elements of divide into subwalks, which we shall refer to as segments. For every segment either

has an internal vertex and all its internal vertices belong to , or

all internal vertices of belong to ,

and these types of segment alternate along . (To see that consider a vertex such that we inserted an element of incident with . It follows that is incident with an edge whose other end belongs to , and either an edge whose other end belongs to or a subset of the boundary of . Thus belongs to a segment of the first kind and or the subset of the boundary of belongs to a segment of the second kind. Since changes between segments only occur at vertices as above, the statement follows.) It follows that is even. For every segment delimited by half-edges and whose internal vertices belong to we merge and into an edge of , embedding the edge in the interior of in a close proximity to . This completes the construction of . It follows from the construction that has the desired properties. (For the last condition to hold we need that is connected.)

We are now ready to prove that if belongs to a hyperbolic family, then the same linear bound holds for closed curves that bound “pinched” disks, or, equivalently, it holds for flat separations.

Lemma 6.9.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let , and let be a flat separation of . If , then .

Proof.

We proceed by induction on . Let be a flat separation of with , and let be as in the definition of flat separation.

Let be the multigraph as in Lemma 6.8. If is null, then , and every edge of is incident with a vertex of , and hence is contained in , and the lemma follows from Lemma 5.2. It follows that is not null. Every cycle of bounds a closed disk . Let us say that a cycle of is maximal if is not a subset of for a cycle of . It follows that every vertex of is contained in for some maximal cycle of . The maximal cycles form a tree structure; in particular, there exist a maximal cycle and a vertex such that is disjoint from all other maximal cycles. Let be the set of all vertices of contained in the interior of . From the definition of hyperbolicity applied to a closed curve tracing we deduce that . If is the unique maximal cycle, then , and hence , as desired. We may therefore assume that is not the unique maximal cycle. Let and let if is disjoint from all other maximal cycles, and if it is not. By induction applied to the separation we conclude that

as desired.

The next lemma says that every vertex on the “flat” side of a flat separation is at most logarithmic distance away from the middle part of the separation.

Lemma 6.10.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let , and let be a flat separation of . Then every vertex of is at distance at most from .

Proof.

We proceed by induction on . If , then by Lemma 6.9, and the lemma vacuously holds. Thus we may assume that , and that the lemma holds for separations with . Let be a vertex of . For let be the set of all vertices of at distance exactly from . For let and . Then is a flat separation and . There exists an integer such that and , for otherwise , contrary to Lemma 6.9. We may assume that , for otherwise satisfies the conclusion of the lemma. By induction applied to the separation the vertex is at distance at most from . But is at distance at most from , and hence is at distance at most from , as desired.

Corollary 6.11.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , and let be a graph embedded in the disk with one ring . Then every vertex of is at distance at most from .

Proof.

This follows from Lemma 6.10 applied to the separation .

Next we will show that planar neighborhoods in a hyperbolic family exhibit exponential growth, but first we need the following lemma. The lemma formalizes the fact that if a vertex is not near a short non-null-homotopic cycle, then the vertices up to a certain distance from form a planar graph, indeed they lie inside a disk.

Lemma 6.12.

Let be an integer, let be a graph with rings embedded in a surface , and let . If is at distance at least from every ring and there exists no non-null-homotopic cycle in of length at most such that every vertex of is at distance at most from , then there exists a closed disk such that

every vertex of at distance at most from belongs to ,

every vertex of at distance at most from belongs to the interior of , and

every edge of incident with a vertex at distance at most from belongs to the interior of .

Furthermore, if there exists no non-null-homotopic cycle in of length at most such that every vertex of is at distance at most from , then the closed disk may be chosen so that every edge of with both ends at distance at most from belongs to the interior of .

Proof.

Let be the subgraph of consisting of vertices at distance at most from and edges incident with a vertex at distance at most from , and let  be a breadth-first search spanning tree of rooted at . By hypothesis every fundamental cycle of  in bounds a disk. The union of these disks together with  is simply connected, and the required disk can be obtained from this union by “fattening” edges of and vertices of at distance at most from that belong to the boundary of this union. This proves the first assertion, including the three bullet points. The second one follows analogously.

We now show that planar neighborhoods in a hyperbolic family exhibit exponential growth.

Lemma 6.13.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let be embedded in a surface , let , and let be an integer. If is at distance at least from every ring and there exists no non-null-homotopic cycle in of length at most such that every vertex of  is at distance at most from , then has at least vertices at distance exactly from .

Proof.

Let be the set of vertices of at distance at most from and let be the set of vertices of at distance at least from . By Lemma 6.12 the assumptions of the lemma imply that there exists a closed disk such that

every vertex of at distance at most from belongs to ,

every vertex of at distance at most from belongs to the interior of , and

every edge of incident with a vertex at distance at most from belongs to the interior of .

Thus is a flat separation of , and hence by Lemma 6.10. It follows that , as desired.

6.3. Edge-width of hyperbolic families

The next lemma says that every vertex that is at least logarithmic distance away from every ring is contained in a non-null-homotopic cycle of at most logarithmic length.

Lemma 6.14.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let be embedded in a surface , let be an integer with , and let be a vertex of at distance at least from every ring of . Then has a non-null-homotopic cycle of length at most such that every vertex of is at distance at most from .

Proof.

Since is hyperbolic, it follows that , and hence . We may assume for a contradiction that the required cycle does not exist. By Lemma 6.13 there are at least vertices at distance exactly from , a contradiction, because those vertices do not include , since .

Let us recall that if is a surface with boundary, then denotes the surface without boundary obtained by gluing a disk to each component of the boundary.

If is a graph with no rings embedded in a surface, then its edge-width is usually defined as the maximum integer such that every non-null-homotopic cycle in has length at least . We extend this definition to graphs with rings as follows.

Definition 6.15.

Let be an embedded graph with rings that is embedded in a surface , and let be the maximum integer (or infinity) such that

(1)

every cycle in that is non-null-homotopic in has at least edges that do not belong to a ring,

(2)

every non-null-homotopic (in ) cycle in that is disjoint from all rings has length at least , and

(3)

every two vertices that belong to distinct rings of are at distance at least in .

In those circumstances we say that is embedded with edge-width .

Our next lemma says that if an embedded graph with rings belonging to a hyperbolic family is embedded with moderately large edge-width (logarithmic in the number of vertices), then it has a specific structure. First we define the structure.

Definition 6.16.

Let be a graph with rings embedded in a surface , and let be the components of . We say that is locally cylindrical if the components of can be numbered in such a way that for all the graph includes the ring of that is embedded in , and there exists a disk that includes every vertex and edge of .

Lemma 6.17.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let be embedded in a surface , and let be an integer with . If is embedded with edge-width at least , then is locally cylindrical and every vertex is at distance at most from some ring.

Proof.

Let . If is at distance at least from every ring, then by Lemma 6.14 applied to the integer there is a non-null-homotopic cycle in of length at most such that every vertex of the cycle is at distance at most from . Since is at distance at least from every ring, it follows that the cycle is disjoint from all rings, contrary to the second condition in the definition of edge-width. Thus every vertex of is at distance at most from some ring.

Let be the components of , and for let be the subgraph of induced by vertices of that are at distance at most from the ring of that is drawn in . The third condition in the definition of edge-width implies that the graphs are the components of . The first condition in the definition of edge-width implies, by the second assertion of Lemma 6.12 applied to the graph obtained from by contracting the ring contained in to a vertex , surface and integer , that each is contained in a closed disk contained in . Thus is locally cylindrical, as desired.

Let us remark that if is locally cylindrical, then each component of may be regarded as a graph with one ring embedded in a disk, and if this embedded graph belongs to the hyperbolic family , then its number of vertices is bounded by a linear function of the length of the ring by Lemma 5.2.

In what follows we will need the embedded graph to be -cell embedded. If it is not, then there is an obvious way to simplify the surface, but we need to know that the new embedded graph belongs to the same family. Hence the following definition.

Definition 6.18.

Let be a family of embedded graphs with rings, let be embedded in a surface with rings , and let be a non-null-homotopic simple closed curve disjoint from . Let be obtained from by cutting open along and pasting a disk or disks on the resulting one or two new boundary components. Let denote the graph when it is regarded as an embedded graph in with rings . If for every and every as above, then we say that is closed under curve cutting.

In the next example we exhibit a hyperbolic family such that the family obtained from it by curve cutting is not hyperbolic.

Example 6.19.

Let be a fixed integer, and for let denote the Cartesian product of a cycle of length and a path on vertices. We start by considering the obvious embedding of in the sphere . Thus has two faces bounded by cycles of length and all other faces are bounded by cycles of length four. Let be the surface obtained from by removing the face bounded by . Thus is a disk whose boundary is the point set of . Let be obtained from by inserting a crosscap inside the face bounded by (or, equivalently, removing the face bounded by and gluing a Möbius band onto ). Thus has Euler genus and one boundary component, namely the point set of . Let be the family consisting of the embedded graphs with one ring for all . Then is hyperbolic, but the family obtained from by curve cutting is not, because it includes the embedded graphs .

The next theorem improves upon the length of the non-null-homotopic cycle guaranteed by Lemma 6.14 from logarithmic to constant, assuming the graph is not too small. The theorem will be a key tool in proving Theorem 6.28, the main result of this section.

Theorem 6.20.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let have rings and let it be embedded in a surface of Euler genus with components, where no component of is the sphere. Let be the total number of ring vertices. Assume that either is -cell embedded, or that is closed under curve cutting. Let . If , then there exists a non-null-homotopic (in ) cycle that is disjoint from all of the rings and has length at most .

Proof.

We begin by observing that

We proceed by induction on . If is not -cell embedded, then there exists a non-null-homotopic simple closed curve such that the image of under is disjoint from . The theorem then follows by induction applied to the same graph embedded in the surface as in the definition of curve cutting. Thus we may assume that is -cell embedded.

By Lemmas 6.2 and 6.6 there exists an optimal frame of . For every non-ring segment of we choose a maximal set of vertices of such that the vertices in are pairwise at distance in of at least and each vertex of is at distance in of at least from each end of . Thus . Let be the union of over all non-ring segments of . Thus , where is the sum of the lengths of all non-ring segments of and is their number.

For let be the set of all vertices of at distance at most from . Since is an optimal frame, it follows from the choice of the set that the sets are pairwise disjoint and each of them is disjoint from every ring. Let . If the subgraph of induced by has a non-null homotopic cycle that has length at most , then such a cycle is disjoint from all of the rings, and therefore satisfies the conclusion of the lemma. We may therefore assume that such cycle does not exist. From Lemma 6.13 we deduce that . Hence as the sets are pairwise disjoint.

The boundary of the unique face of can be traced by a closed curve. By pushing the curve slightly into the interior of the face of we obtain a closed curve that intersects only in vertices and is otherwise contained in the unique face of . By applying the definition of hyperbolicity to we deduce that .

Combining the upper and lower bounds on gives

Using the inequality observed at the beginning of this proof we obtain

and hence . Substituting this bound into the inequality and using the fact that by Lemma 6.4 (which can be applied because no component of is the sphere) we obtain

a contradiction.

The following corollary is an upgrade of Lemma 6.17.

Corollary 6.21.

Let be a hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant for , let have rings, let it be embedded in a surface of Euler genus , and let be the total number of ring vertices. Let and . If the edge-width of is at least , then is locally cylindrical and every vertex is at distance at most from some ring.

Proof.

We begin by observing that

and hence

If a component of is a sphere, then it includes no vertex or edge of by Lemma 5.2. By deleting such components we may assume that no component of is a sphere. We have

for otherwise by Theorem 6.20 the edge-width of is at most . Thus , using the inequality observed at the beginning of the proof. Lemma 6.17 implies that is locally cylindrical and every vertex of has distance at most from some ring by Corollary 6.11, as desired.

We can now prove Theorem 1.2.

Proof of Theorem 1.2.

Let be a hyperbolic family of embedded graphs with no rings that is closed under curve cutting. Thus no member of is locally cylindrical, and hence every member of has edge-width by Corollary 6.21, as desired.

Our motivation for the study of hyperbolic families is to understand graphs that contain no subgraph isomorphic to a member of the family. The following is the first of several results along those lines. It is a more precise version of Theorem 1.2.

Corollary 6.22.

Let be a hyperbolic family of embedded graphs with no rings, let be closed under curve cutting, let be a Cheeger constant for , and let be a graph with no rings embedded in a surface of Euler genus . Let and . If the edge-width of is at least , then has no subgraph isomorphic to a member of .

Proof.

If has a subgraph isomorphic to a graph , then has no rings, and hence is not locally cylindrical. By Corollary 6.21 the graph has edge-width at most , and hence so does , a contradiction.

6.4. Structure of hyperbolic families

Our objective is to prove Theorem 6.28, but first we need the following notion.

Definition 6.23.

Let be a family of embedded graphs with rings, let be embedded in a surface with rings , and let be a non-null-homotopic cycle in . Let be the surface obtained by cutting open along . Thus has one or two more boundary components than , depending on whether is one-sided or two-sided. Let be the graph obtained from during this operation; thus if is one-sided, then corresponds to a cycle in of twice the length of , and if is two-sided, then it corresponds to two cycles in , each of the same length as . If is one-sided, then let ; otherwise let . We regard as a graph embedded in with rings . If for every and every non-null-homotopic cycle in , then we say that is closed under cycle cutting.

It should be noted that despite the similarity in names, curve cutting and cycle cutting play different roles: while the next lemma shows that closure under cycle cutting can be obtained for free, Example 6.19 shows that that is not the case for curve cutting. On the other hand, all hyperbolic families of embedded graphs of interest to us are indeed closed under curve cutting, and hence we make that assumption whenever it is convenient to do so.

Lemma 6.24.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , and let be the inclusion-wise minimal family of embedded graphs that includes and is closed under cycle cutting. Then is hyperbolic with Cheeger constant .

Proof.

Let , let be a non-null-homotopic cycle in , and let , and be obtained as in the definition of cycle cutting. If is a simple closed curve that bounds an open disk , then can be regarded as an open disk in . It follows that is hyperbolic with the same Cheeger constant.

In the next two definitions we formalize the concept of narrow cylinder and a graph being decomposed into a bounded size graph and narrow cylinders.

Definition 6.25.

Let and let be a graph with two rings embedded in a cylinder . Let be cycles in such that the following conditions hold:

(1)

and ,

(2)

the cycles are pairwise disjoint,

(3)

for , the cycle separates into two cylinders, one containing , and the other containing ,

(4)

for the cycle has length at most , and

(5)

for at most vertices of belong to the cylinder with boundary components and .

In those circumstances we say that is a -sleeve. Let us remark in the last condition the cylinder includes the vertices of .

Definition 6.26.

Now let be a graph with rings embedded in a surface , and let be a -sleeve with rings and embedded in a cylinder . Let be distinct rings of of lengths and , respectively. Let be the graph obtained from by identifying with and with and let be the surface obtained from by identifying the two boundary components that embed and and identifying the two boundary components that embed  and . Thus is a graph with two fewer rings than and it is embedded in . We say that , regarded as an embedded graph with rings, was obtained from  by adjoining the -sleeve .

We say that a graph has a sleeve-decomposition with parameters if it can be obtained from an embedded graph with rings on at most vertices by repeatedly adjoining -sleeves.

The following is an easy bound on the number of sleeves in a sleeve-decomposition.

Lemma 6.27.

Let be an embedded graph with rings that has a sleeve-decomposition with parameters . Then the number of sleeves adjoined in the decomposition is at most .

Proof.

By definition of sleeve-decomposition the graph is obtained from a graph on at most vertices by adjoining -sleeves. Since the rings of are pairwise disjoint, the graph has at most rings that are cycles, and every time a sleeve is adjoined, two of those cycles stop being rings. The lemma follows.

We are ready to prove the main result of this section. It states that a graph in a hyperbolic family has a sleeve-decomposition with parameters , where depend only on the Euler genus , the total number of ring vertices , and the Cheeger constant . Moreover, the number of sleeves is also bounded.

Theorem 6.28.

Let be a hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant for , and let be embedded in a connected surface of Euler genus with rings, and let be the total number of ring vertices. Let , , and . Then has a sleeve-decomposition with parameters that uses at most sleeves, where

(1)

and if and ,

(2)

and if and ,

(3)

and if or .

Proof.

By Lemma 6.24 we may assume that is closed under cycle cutting as well as curve cutting. We proceed by induction on . If , then by Lemma 5.2, and if and , then the same lemma implies that the theorem holds with as in (1).

We now handle the case and . Let and be the boundary components of . Let be a maximal collection of cycles of  such that for each , , is a cycle of length at most and  separates  into two cylinders, one containing , and the other containing , and the cycles in are pairwise disjoint and each is disjoint from  and . Let be the cylinders with boundary components , ; , ; , , respectively.

By Theorem 6.20 applied to , it follows that if there are more than vertices between two cycles of length at most in a cylinder, then there exists a cycle of length at most separating from and disjoint from both. Hence, it follows that the subgraph of drawn in is a -sleeve. We may also assume that as otherwise has the desired sleeve-decomposition (with no sleeves). Analogously by Theorem 6.20, the subgraphs of  drawn in and have at most vertices combined as was chosen maximal. Thus has a sleeve-decomposition with as in (2) and the theorem follows. This completes the case and .

We may assume then that or and we have to prove that the theorem holds with and as in (3). Let be a component of . We denote by the surface obtained from by gluing a disk on . We say that a cycle in is a -cycle if it is non-null-homotopic in , but null-homotopic in .

We first show that if has a non-null-homotopic cycle of length at most that is not a -cycle for any component of the boundary of , then the theorem holds. To see that cut along and declare the resulting cycle(s) to be rings in a new graph with rings embedded in a surface of Euler genus with components and a total number of ring vertices.

Suppose . Then , , and . Let and be the two components of . Let be the subgraph of drawn in . Then is a graph drawn in with rings and a total of ring vertices. For each , either or as is not a -cycle. Thus, for each , by induction has a sleeve-decomposition with parameters and at most sleeves, where

and . Adding, we find that

and . Thus has a sleeve-decomposition with parameters with at most sleeves, as desired.

So we may assume that . Suppose is one-sided, then , , and . Further suppose that or . Then it follows from induction that has a sleeve-decomposition with parameters and at most sleeves, where

and

Thus has a sleeve-decomposition with parameters and at most sleeves, as desired. So we may assume that and . Hence if and if . Note then that . If , by Lemma 5.2. Yet and thus . Hence and has a sleeve-decomposition with parameters and sleeves, as desired since . If , then has a sleeve-decomposition with parameters and by (2). Yet and . Hence has a sleeve-decomposition with parameters and at most sleeves, as desired, since .

So we may assume that is two-sided; then , and . Further suppose that or . Then it follows from induction that has a sleeve-decomposition with parameters and at most sleeves, where

and

Thus has a sleeve-decomposition with parameters and at most sleeves, as desired. So we may assume that and . Hence , , , , and . By induction using (2), has a sleeve-decomposition with parameters with at most sleeve. Yet . Hence has a sleeve-decomposition with parameters and at most sleeves, as desired, since

Thus we may assume that every non-null-homotopic cycle in of length at most is a -cycle for some component of the boundary of . Let be a maximal collection of cycles of such that each member of is a -cycle of length at most for some component of the boundary of , and the cycles in are pairwise disjoint and each is disjoint from every ring.

Let and let be the boundary component of such that is a -cycle. We say that is maximal if the component of that contains also contains every -cycle included in . It follows that if includes a -cycle, then it includes a unique maximal -cycle. Let be the surface obtained from by removing, for every maximal -cycle , the component of that includes . Then  is a surface with boundary; it has the same number of boundary components as . Indeed, let be a component of the boundary of . If includes no -cycle, then is a component of the boundary of ; otherwise is a component of the boundary of , where is the maximal -cycle in .

Let be the subgraph of consisting of vertices and edges drawn in . We will regard as a graph with rings embedded in . By Theorem 6.20,

Let be all the boundary components of such that contains a -cycle. Note that . Let be the graph embedded in the cylinder with boundary components and where is the unique maximal -cycle. We regard as a graph embedded in the cylinder with two rings and . By induction using (2), has a sleeve-decomposition with parameters with at most one sleeve where as .

Thus has a sleeve-decomposition with parameters and at most sleeves. Note that

First suppose . Then is at most

as . Furthermore, and so has a sleeve-decomposition with parameters and at most sleeves, as desired.

So we may assume that . But then by assumption. In this case, since

as and . Furthermore, since and hence and so has a sleeve-decomposition with parameters and at most sleeves, as desired.

By using induction when is not connected we obtain the following easier-to-state bound.

Theorem 6.29.

Let be a hyperbolic family of embedded graphs with rings, let  be closed under curve cutting, let be a Cheeger constant for , let be embedded in a surface of Euler genus with rings and a total number of ring vertices. Let and . Then  has a sleeve-decomposition with parameters , and the decomposition uses at most sleeves.

Proof.

This follows from Theorem 6.28 applied to the components of , using the fact that .

We deduce the following consequence in the spirit of Theorems 2.13 and 3.17, but we need a definition first.

Definition 6.30.

For let be a graph with rings embedded in a surface . We say that the embedded graphs with rings and are isomorphic if there exists an isomorphism of and as abstract graphs and a bijection such that for every the restriction of to is an isomorphism of and , and extends to a homeomorphism .

Let be a graph with rings embedded in a surface . By a subgraph of we mean every embedded graph with rings that can be obtained from by repeatedly applying these operations:

deleting a vertex or edge not belonging to any ring, and

deleting all vertices and edges of a ring , removing from the list of rings, and attaching a disk to the boundary component of containing .

Let be a family of embedded graphs with rings, and let be an embedded graph with rings. We say that is -free if no subgraph of is isomorphic to a member of .

Theorem 6.31.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , and let . If is a graph with no rings embedded in a surface of Euler genus , then there exists a set of size at most such that is -free.

Proof.

We show by induction on that there exists a required set of size at most , where is the maximum number of pairwise disjoint pairwise non-homotopic non-null-homotopic separating cycles in of length at most . Since , this implies the theorem. If , then is -free by Lemma 5.2, and so we may assume that and that the theorem holds for all graphs embedded in surfaces of Euler genus at most . If is -free, then satisfies the conclusion of the theorem, and so we may assume that has a subgraph isomorphic to . By Theorem 6.20 either , where is the combinatorial genus of , or (and hence ) has a non-null-homotopic cycle of length at most . In the former case let and in the latter case let . Let be the combinatorial genus of . We apply induction to the embedded graph . We may do so, because in the former case by Lemma 5.2, and in the latter case either (if is non-separating), or separates into two surfaces of genera strictly smaller than . By induction has a set of size at most such that is -free, where is the maximum number of pairwise disjoint pairwise non-homotopic non-null-homotopic separating cycles in of length at most . Then in the former case by Lemma 3.16, and hence the theorem follows.

So we may assume that the latter case holds. If does not separate the surface, then , and the result follows. Finally, if separates the surface, then and , and, again, the result follows.

6.5. From local freedom to global freedom

In this section we prove a generalization of Corollary 6.22.

Definition 6.32.

Let be a hyperbolic family of embedded graphs with rings, and let be a Cheeger constant for . Let be an embedded graph with rings . Let be the subgraph of induced by vertices such that there is a ring such that is at distance at most from . If is -free, then we say that is locally -free. Let us remark that the definition depends on the choice of the Cheeger constant, which will be implicit whenever we will use this notion.

The following is a generalization of Corollary 6.22 to graphs with rings.

Corollary 6.33.

Let be a hyperbolic family of embedded graphs with rings, let  be closed under curve cutting, let be a Cheeger constant for , and let be a graph with rings and a total of ring vertices embedded in a surface of Euler genus . Let and . If the edge-width of is at least and is locally -free, then is -free.

Proof.

If has a subgraph isomorphic to a graph , then by Corollary 6.21 either has edge-width at most , or is locally cylindrical. But has edge-width at least , because so does , and hence is locally cylindrical. Let be as in the definition of local -freedom. By Corollary 6.11 applied to each component of (which we may, because is closed under curve cutting) we deduce that has a subgraph isomorphic to , and hence is not locally -free, as desired.

7. Strongly hyperbolic families

Let be a hyperbolic family of embedded graphs, and let and be fixed. Are there arbitrarily large members of that are embedded in with a total of  ring vertices? Theorem 6.28 implies that the answer depends on whether there exist arbitrarily large sleeves. That motivates the following definition.

Definition 7.1.

Let be a hyperbolic family of embedded graphs with rings, let  be a Cheeger constant for , and let . We say that  is strongly hyperbolic if there exists a constant such that for every embedded in a surface with rings and for every two disjoint cycles of length at most in , if there exists a cylinder with boundary components and , then includes at most vertices of . We say that is a strong hyperbolic constant for .

We can now answer the question posed at the beginning of this section.

Theorem 7.2.

Let be a strongly hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant and a strong hyperbolic constant for , and let be embedded in a surface of Euler genus with a total of ring vertices. Let . Then has at most vertices.

Proof.

Let and . By Theorem 6.29 the graph can be obtained from a graph on at most vertices by adjoining at most -sleeves. But every sleeve has at most vertices by the definition of strongly hyperbolic family, and hence the theorem follows.

In the definition of strong hyperbolicity we imposed a constant bound on the number of vertices in cylinders with boundary components of bounded length. As the next theorem shows, it follows that this implies a linear bound for cylinders with boundaries of any length.

Theorem 7.3.

Let be a strongly hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant and a strong hyperbolic constant for , and let be a graph with rings embedded in a surface . Let . Let be disjoint cycles in such that there exists a cylinder with boundary components , and let . Then includes at most vertices.

Proof.

Let be the subgraph of consisting of all vertices and edges drawn in , and let be regarded as a graph embedded in with rings and . By Lemma 6.24 the graph satisfies the hyperbolicity condition with Cheeger constant , and by the same argument it satisfies the strong hyperbolicity condition with strong hyperbolic constant . It follows that has at most vertices by Theorem 7.2, as desired.

7.1. Examples

The following theorem is proved in Reference 48. The proof is long and its journal version will be split into multiple papers, the first three of which are Reference 50Reference 51Reference 52.

Theorem 7.4.

For every integer there exists an integer such that the following holds. Let be a graph embedded in a cylinder with rings and , where , let be a -list-assignment for , and assume that is -critical with respect to . Then has at most vertices.

Lemma 7.5.

Let be the family of embedded graphs with rings from Theorem 5.3. Then is strongly hyperbolic.

Proof.

The family is hyperbolic by Theorem 5.3 and strongly hyperbolic by Theorem 7.4.

Lemma 7.6.

Let be the family from Lemma 5.12; that is, the family of all embedded graphs with rings such that no triangle is null-homotopic and is -critical with respect to some -list-assignment. Then is strongly hyperbolic.

Proof.

This follows from Lemma 5.10 in the same way as Lemma 5.12, the only difference being that if is as in the proof of Lemma 5.12, then the subgraph of  induced by is a cycle with at most one chord. The rest of the argument goes through without any changes.

The analog of Theorem 7.4 for graphs of girth five and -list-assignments is shown in Reference 49, Theorem 1.8.

Theorem 7.7.

Let be a graph of girth at least five embedded in a cylinder with rings and , let be a -list-assignment for , and assume that is -critical with respect to . Then has at most vertices.

We deduce the following lemma.

Lemma 7.8.

Let be the family of embedded graphs with rings such that every cycle in of length at most four is equal to a ring and is -critical with respect to some -list-assignment. Then is strongly hyperbolic.

Proof.

The family is hyperbolic by Lemma 5.13 and strongly hyperbolic by Theorem 7.7.

Let us point out a subtlety: the family of Lemma 7.8 is a proper subfamily of the family of Lemma 5.13. In fact, the latter family is not strongly hyperbolic.

Lemma 7.9.

Let be the family from Lemma 5.13; that is, the family of embedded graphs with rings such that every cycle in of length at most four is not null-homotopic and is -critical with respect to some -list-assignment. Then is not strongly hyperbolic.

Proof.

This follows from Figure 4 and the discussion following Lemma 4.6 of Reference 26.

The strong hyperbolicity of the family in the next example follows from Theorem 8.8.

Example 7.10.

There exists such that the following holds. For all with and , the family of graphs that are -exponentially-critical with respect to a type list-assignment is a strongly hyperbolic family. Moreover, the Cheeger constant and the strong hyperbolic constant do not depend on or .

7.2. The structure of strongly hyperbolic families

Let us recall that the combinatorial genus of an embedded graph was defined prior to Lemma 3.16. The following is our main result about strongly hyperbolic families of embedded graphs.

Theorem 7.11.

Let be a strongly hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant for , let be a strong hyperbolic constant for , and let be a graph with rings and a total number of ring vertices embedded in a surface of Euler genus . Let , , , and for a ring let . Then has at most vertices and either

(a)

there exist distinct rings such that the distance in between them is at most , or

for every component of one of the following conditions holds:

(b)

contains no rings and has a non-null-homotopic cycle in of length at most , or

(c)

contains a unique ring , every vertex of is at distance strictly less than from , and has a non-null-homotopic cycle in of length at most , or

(d)

and , where is the combinatorial genus of and is the number of ring vertices in that are a subgraph of , or

(e)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Proof.

Since is closed under curve cutting we may assume that is -cell embedded in . We may also assume that is connected. In particular, this implies that is connected. The graph has at most vertices by Theorem 7.2. To prove that one of the statements (a)–(e) holds let us first assume that . If , then (d) holds, and so we may assume that . But then cannot be locally cylindrical, and hence (b) holds by Corollary 6.21. This proves the theorem when .

We may therefore assume that , and that (a) does not hold. It follows that , for otherwise , contrary to Lemma 5.2. We claim that for some no vertex of is at distance exactly from . To prove this claim suppose to the contrary that for every there exists a vertex at distance exactly from . Let be the set of all vertices of at distance at most from . We have . The sets are pairwise disjoint because (a) does not hold, and hence

a contradiction. This proves our claim that for some no vertex of is at distance exactly from . An immediate consequence is that has exactly one ring, for otherwise is not connected because (a) does not hold. Let be the unique ring of . Thus every vertex of is at distance less than from . Since we may assume that (c) does not hold and is -cell embedded we deduce from Lemma 6.12 applied to the graph obtained from by contracting to a vertex that . It follows from Corollary 6.11 that every vertex of is at distance at most from . Thus (e) holds, as desired.

We can now prove Theorem 1.4.

Proof of Theorem 1.4.

Let be a strongly hyperbolic family of rooted embedded graphs that is closed under curve cutting. We convert every rooted graph to an embedded graph with rings by turning each into a -vertex ring with vertex-set and removing from the surface the interior of an arbitrary closed disk that intersects only in and the intersection belongs to the boundary of the disk. Let be the family of embedded graphs with rings so obtained. Then  is strongly hyperbolic with the same Cheeger and strong hyperbolic constants as . Theorem 7.11 implies that and that one of the conditions (a)–(e) holds. Condition (a) implies that some two distinct vertices of are at distance , condition (b) implies that has a non-null-homotopic cycle of length , condition (c) implies that has a non-null-homotopic cycle of length , and if condition (d) holds, then the graph has a non-null-homotopic cycle, which has length at most . Finally, if condition (e) holds for every component of , then by Lemma 5.2, because is closed under curve cutting.

Theorem 7.11 has the following variation, where we eliminate outcome (d) at the expense of increasing the value of .

Theorem 7.12.

Let be as in Theorem 7.11, and for a ring let . Then has at most vertices and either

(a)

there exist distinct rings such that the distance in between them is at most , or

for every component of one of the following conditions holds:

(b)

includes no rings and has a non-null-homotopic cycle in of length at most , or

(c)

contains a unique ring , every vertex of is at distance strictly less than from , and has a non-null-homotopic cycle in of length at most , or

(d)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Proof.

Since is closed under curve cutting we may assume that is -cell embedded in . The graph has at most vertices by Theorem 7.2. If , then Corollary 6.21 implies that (b) holds. We may therefore assume that , and that (a) and (c) do not hold. We claim that for some no vertex of is at distance exactly from . To prove this claim suppose to the contrary that for every there exists a vertex at distance exactly from . Let be the set of all vertices of at distance at most from . We have . The sets are pairwise disjoint because (a) does not hold, and hence

a contradiction. This proves our claim that for some no vertex of is at distance exactly from . An immediate consequence is that has exactly one ring, for otherwise is not connected because (a) does not hold. Let be the unique ring of . Thus every vertex of is at distance less than from , and since (c) does not hold and is -cell embedded we deduce from Lemma 6.12 applied to the graph obtained from by contracting to a vertex that . It follows from Corollary 6.11 that every vertex of is at distance at most from . Thus (d) holds, as desired.

8. Canvases

In this section we prove Theorem 3.8. It is an immediate consequence of Theorem 8.11. Let us recall that canvases and critical canvases were defined in Definition 3.3.

Definition 8.1.

Let and , and let be a canvas. We say that is -exponentially-critical if is -exponentially-critical with respect to , as defined in Definition 5.18.

Definition 8.2.

Let be a family of canvases. We define crit to be the family of all embedded graphs with rings such that there exists a graph such that is a subgraph of , and is a critical canvas for some list-assignment . Note that need not be closed under taking subgraphs; hence the need for the subgraph .

We say that is critically hyperbolic if crit is closed under curve-cutting, and is strongly hyperbolic.

Now let and . We define crit to be the family of all embedded graphs with rings such that there exists a graph such that is a subgraph of , and is an -exponentially-critical canvas.

We say that is critically exponentially hyperbolic if there exists a real number such that for all real numbers

crit is closed under curve-cutting, and

crit is strongly hyperbolic with Cheeger constant and strong hyperbolic constant that do not depend on .

Definition 8.3.

A family of canvases is good if it is both critically hyperbolic and critically exponentially hyperbolic.

Definition 8.4.

Let be an embedded graph with rings. Let be disjoint cycles in such that there exists a cylinder with boundary components and . Let be the subgraph of consisting of all vertices and edges drawn in , and let be a subgraph of that includes both and as subgraphs. We say that is a cylindrical excision of with boundary components  and .

To prove that certain families of canvases are good we need a lemma. To state the lemma we need to consider the following properties of families of canvases.

Definition 8.5.

Let be a family of canvases. We say that satisfies property (P1) if for every the following holds. Let be a subgraph of , and let be a cycle in . If is planar, then there exist at least five -colorings of that extend to -colorings of .

We say that satisfies property (P2) if for every integer there exist integers and such that the following holds. Let . Then for every cylindrical excision of with boundary cycles of length at most , if the distance between and in is at least , then there exist disjoint subgraphs of such that is a subgraph of , , and if an -coloring of extends to an -coloring of , then it extends to an -coloring of .

The lemma we need is the following.

Lemma 8.6.

Let be a family of canvases that satisfies properties (P1) and (P2), let and , and assume that is closed under curve-cutting and is hyperbolic with Cheeger constant . Let and , and let be as in property (P2) applied to and the integer . Assume further that for every and every . If , then is strongly hyperbolic with strong hyperbolic constant .

Proof.

Suppose for a contradiction that there exists an embedded graph with rings , two cycles in of length at most and a cylinder with boundary components such that includes more than vertices of .

Let be the subgraph of consisting of all vertices and edges drawn in . We regard as a graph embedded in the cylinder with two rings and . Then was obtained from by cycle cutting, as defined in Definition 6.23. By Lemma 6.24 the family of embedded graphs with rings obtained from by cycle cutting is hyperbolic with the same Cheeger constant. By applying Theorem 6.28 to this family we deduce that has a sleeve-decomposition with parameters with at most one sleeve, where . Since has strictly more than vertices, it follows that the sleeve-decomposition uses exactly one sleeve. Thus was obtained from an embedded graph with four rings by adjoining a -sleeve , where . But then can be obtained from some embedded graph with rings by adjoining the -sleeve . Let be cycles in as in the definition of a -sleeve.

Since , there exists a list-assignment for a graph such that is a subgraph of , and is -exponentially-critical. Then is a cylindrical excision of with boundary cycles  and .

Let ; since by the definition of -sleeve, we have . For let , let , and for let denote the subgraph of consisting of all vertices and edges drawn in the cylinder with boundary components and that is a subset of . Then is a cylindrical excision of with boundary components and , and the boundary components and are at distance at least in . By property (P2) there exist disjoint subgraphs and of such that is a subgraph of , is a subgraph of , both and have at most vertices, and if an -coloring of extends to an -coloring of , then it extends to an -coloring of .

Let and . Then . Let be the Euler genus of . As is -exponentially-critical, there exists an -coloring of such that there exist at least distinct -colorings of extending , but there do not exist distinct -colorings of extending . Thus there exists a set of distinct -colorings of extending with such that every extends to an -coloring of since for all .

Let . We claim that extends to at least distinct -colorings of . To see this, we first notice that property (P1) implies that for every there exist at least five -colorings of that extend to -colorings of . Since extends into , property (P2) implies that every choice of the -colorings of as above extends to an -coloring of extending . This proves our claim that extends to at least distinct -colorings of .

We have that

where the first inequality uses , the second inequality uses , and the third inequality uses . By the claim of the previous paragraph there exist at least distinct -colorings of that extend . But

where the first inequality follows by the assumption on , the second inequality uses from above, and the third inequality uses , a contradiction.

Let us recall that the families , and were defined in Definition 3.5. In order to apply Lemma 8.6 we need to show that these families satisfy properties (P1) and (P2).

Lemma 8.7.

The families , and satisfy properties (P1) and (P2).

Proof.

Property (P1) follows from Theorem 2.14 applied to a subpath of of length one. To prove that these families satisfy property (P2) let be the family of embedded graphs with rings such that is critical and belongs to for some list-assignment . Then is strongly hyperbolic by Lemma 7.5, Lemma 7.6, and Lemma 7.8. Let be a Cheeger constant, and let be a strong hyperbolic constant for . Let be given, let , and let be an integer strictly larger than the bound on the number of vertices in in Theorem 7.3, when the latter is applied to the family and cycles of length at most . Let , let be a cylindrical excision of with boundary cycles of length at most , let be the corresponding cylinder, and let the distance between and in be at least .

Let be a minimal subgraph of such that both and are subgraphs of and every -coloring that extends to an -coloring of also extends to an -coloring of . Then and it is a critical canvas. Thus , and by Theorem 7.3 the graph has strictly fewer than vertices, and hence is not connected. It follows that has exactly two components, one containing and one containing . For let be the component of containing . Then is -critical with respect to , and hence by Theorems 2.15, 2.16, and 2.17.

We now prove Theorem 3.6, which we restate.

Theorem 8.8.

The families are good families of canvases.

Proof.

For the families crit and crit are clearly closed under curve-cutting. The family is critically hyperbolic by Lemma 7.8, the family is critically hyperbolic by Lemma 7.6, and the family is critically hyperbolic by Lemma 7.5. The hyperbolicity of crit, crit, and crit follows from Lemma 5.20, Lemma 5.21 and Theorem 5.19, respectively. By Lemma 8.7 the families satisfy properties (P1) and (P2), and hence by Lemma 8.6 all three families are good.

In the next theorem we show that if a coloring of the rings extends for a canvas in a critically exponentially hyperbolic family, then it has exponentially many extensions.

Theorem 8.9.

For every critically exponentially hyperbolic family of canvases there exist such that the following holds. Let , let be the total number of ring vertices in , and let be the Euler genus of . If is an -coloring of such that extends to an -coloring of , then extends to at least distinct -colorings of .

Proof.

Let be such that satisfies the conditions in the definition of exponentially critically hyperbolic family of canvases for every . By Theorem 7.2 there exists a constant that depends on but not on such that for every and every graph with rings embedded in a surface of Euler genus and a total of ring vertices.

We will show that and satisfy the theorem. To prove that let , let and be as in the statement of the theorem, and assume for a contradiction that there exists an -coloring of that extends to an -coloring of , but does not extend to at least distinct -colorings of . Let be a minimal subgraph of such that includes all the rings in and there exists an -coloring of such that extends to an -coloring of , but does not extend to at least distinct -colorings of . Then the canvas is -exponentially-critical.

It follows that . As noted earlier this implies that . As extends to an -coloring of , also extends to an -coloring of . But , and hence extends to at least distinct -colorings of , a contradiction.

The following is our main result about good families of canvases.

Theorem 8.10.

For every critically hyperbolic family of canvases there exist such that the following holds. Let , let be the Euler genus of , let be a Cheeger constant of , and let be the total number of ring vertices in . Then has a subgraph such that includes all the rings in , has at most vertices, and for every -coloring of 

either does not extend to an -coloring of , or

extends to an -coloring of , and if is critically exponentially hyperbolic, then extends to at least distinct -colorings of .

Furthermore, either

(a)

there exist distinct rings such that the distance in between them is at most , or

every component of satisfies one of the following conditions:

(b)

contains no rings and has a non-null-homotopic cycle in of length at most , or

(c)

contains a unique ring , every vertex of is at distance strictly less than from , and has a non-null-homotopic cycle in of length at most , or

(d)

and , where is the combinatorial genus of and is the number of ring vertices in that are a subgraph of , or

(e)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Proof.

Let be the Cheeger constant of crit, and let be as in Theorem 7.11 when the latter is applied to the family crit. If is exponentially critically hyperbolic, then let and be as in Theorem 8.9; otherwise let and be arbitrary. Finally, let be such that . Then and . We claim that satisfy the conclusion of the theorem.

Let , let be the Euler genus of , and let be the total number of ring vertices in . Let be a minimal subgraph of such that includes all the rings in and every -coloring of that extends to an -coloring of also extends to an -coloring of . We will show that satisfies the conclusion of the theorem.

It follows that either ; or the canvas is critical, and hence . It follows from Theorem 7.11 applied to the family that  has at most vertices and that it satisfies one of the conditions (a)–(e) of Theorem 7.11. Thus satisfies one of the conditions (a)–(e) of the present theorem. By the definition of every -coloring of either does not extend to an -coloring of or extends to an -coloring of . In the latter case, if is critically exponentially hyperbolic, then Theorem 8.9 implies that extends to at least distinct -colorings of , as desired.

We can consolidate outcomes (b)-(d) of the previous theorem to obtain the following simpler version, which implies Theorem 3.8.

Theorem 8.11.

For every critically hyperbolic family of canvases there exist such that the following holds. Let , let be the Euler genus of , let be the total number of ring vertices in , let be a Cheeger constant of , and let be the maximum number of vertices in a ring in . Then has a subgraph such that includes all the rings in , has at most vertices, and for every -coloring of

either does not extend to an -coloring of , or

extends to an -coloring of , and if is critically exponentially hyperbolic, then extends to at least distinct -colorings of .

Furthermore, either

(a)

there exist distinct rings such that the distance in between them is at most , or

every component of satisfies one of the following conditions:

(b)

the graph has a cycle that is not null-homotopic in ; if includes no ring vertex, then the length of is at most , and otherwise it is at most , or

(c)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

By applying Theorem 7.12 instead of Theorem 7.11 we obtain the following variation of Theorem 8.10. We omit the almost identical proof.

Theorem 8.12.

For every critically hyperbolic family of canvases there exist such that the following holds. Let , let be the Euler genus of , and let be the total number of ring vertices in . Then has a subgraph such that includes all the rings in , has at most vertices, and for every -coloring of

either does not extend to an -coloring of , or

extends to an -coloring of , and if is critically exponentially hyperbolic, then extends to at least distinct -colorings of .

Furthermore, either

(a)

there exist distinct rings such that the distance in between them is at most , or

every component of satisfies one of the following conditions:

(b)

includes no rings and has a non-null-homotopic cycle in of length at most , or

(c)

contains a unique ring , every vertex of is at distance strictly less than from , and has a non-null-homotopic cycle in of length at most , or

(d)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Acknowledgment

This paper is based on part of the doctoral dissertation Reference 48 of the first author, written under the guidance of the second author.

Table of Contents

  1. Abstract
  2. 1. Introduction
    1. Theorem 1.1.
    2. Theorem 1.2.
    3. Theorem 1.3.
    4. Theorem 1.4.
    5. Theorem 1.5.
  3. 2. Survey of coloring graphs on surfaces
    1. 2.1. Coloring graphs on surfaces
    2. Theorem 2.1.
    3. Theorem 2.2.
    4. Theorem 2.3.
    5. Corollary 2.4.
    6. Theorem 2.5.
    7. Theorem 2.6.
    8. 2.2. List-coloring graphs on surfaces
    9. Theorem 2.7.
    10. Theorem 2.8.
    11. Theorem 2.9.
    12. Theorem 2.10.
    13. Conjecture 2.11.
    14. Theorem 2.12.
    15. Theorem 2.13.
    16. 2.3. Extending precolored subgraphs
    17. Theorem 2.14.
    18. Theorem 2.15.
    19. Theorem 2.16.
    20. Theorem 2.17.
    21. Theorem 2.18.
    22. Theorem 2.19.
    23. Question 2.20.
    24. Theorem 2.21.
    25. 2.4. Coloring with short cycles far apart
    26. Theorem 2.22.
    27. 2.5. Exponentially many colorings
    28. Theorem 2.23.
    29. Theorem 2.24.
    30. Theorem 2.25.
    31. Problem 2.26.
    32. Theorem 2.27.
  4. 3. Main coloring results
    1. Definition 3.1.
    2. Definition 3.2.
    3. Definition 3.3.
    4. Definition 3.4.
    5. Definition 3.5.
    6. Theorem 3.6.
    7. Definition 3.7.
    8. Theorem 3.8.
    9. Theorem 3.9.
    10. Theorem 3.10.
    11. Theorem 3.11.
    12. Corollary 3.12.
    13. Corollary 3.13.
    14. Corollary 3.14.
    15. Definition 3.15.
    16. Lemma 3.16.
    17. Theorem 3.17.
    18. Theorem 3.18.
    19. Theorem 3.19.
    20. Theorem 3.20.
    21. Theorem 3.21.
    22. Theorem 3.22.
  5. 4. Exponentially many colorings
    1. Theorem 4.1.
    2. Corollary 4.2.
    3. Definition 4.3.
    4. Theorem 4.4.
    5. Lemma 4.5.
  6. 5. Definition and examples of hyperbolic families
    1. Definition 5.1.
    2. Lemma 5.2.
    3. 5.1. Examples arising from (list-)coloring
    4. Theorem 5.3.
    5. Lemma 5.4.
    6. Example 5.5.
    7. Example 5.6.
    8. Lemma 5.7.
    9. Example 5.8.
    10. Example 5.9.
    11. Lemma 5.10.
    12. Example 5.11.
    13. Lemma 5.12.
    14. Lemma 5.13.
    15. Problem 5.14.
    16. Conjecture 5.15.
    17. Example 5.16.
    18. Example 5.17.
    19. 5.2. Examples pertaining to exponentially many colorings
    20. Definition 5.18.
    21. Theorem 5.19.
    22. Lemma 5.20.
    23. Lemma 5.21.
  7. 6. The structure of hyperbolic families
    1. 6.1. Frames
    2. Definition 6.1.
    3. Lemma 6.2.
    4. Definition 6.3.
    5. Lemma 6.4.
    6. Definition 6.5.
    7. Lemma 6.6.
    8. 6.2. Separations and growth rates
    9. Definition 6.7.
    10. Lemma 6.8.
    11. Lemma 6.9.
    12. Lemma 6.10.
    13. Corollary 6.11.
    14. Lemma 6.12.
    15. Lemma 6.13.
    16. 6.3. Edge-width of hyperbolic families
    17. Lemma 6.14.
    18. Definition 6.15.
    19. Definition 6.16.
    20. Lemma 6.17.
    21. Definition 6.18.
    22. Example 6.19.
    23. Theorem 6.20.
    24. Corollary 6.21.
    25. Corollary 6.22.
    26. 6.4. Structure of hyperbolic families
    27. Definition 6.23.
    28. Lemma 6.24.
    29. Definition 6.25.
    30. Definition 6.26.
    31. Lemma 6.27.
    32. Theorem 6.28.
    33. Theorem 6.29.
    34. Definition 6.30.
    35. Theorem 6.31.
    36. 6.5. From local freedom to global freedom
    37. Definition 6.32.
    38. Corollary 6.33.
  8. 7. Strongly hyperbolic families
    1. Definition 7.1.
    2. Theorem 7.2.
    3. Theorem 7.3.
    4. 7.1. Examples
    5. Theorem 7.4.
    6. Lemma 7.5.
    7. Lemma 7.6.
    8. Theorem 7.7.
    9. Lemma 7.8.
    10. Lemma 7.9.
    11. Example 7.10.
    12. 7.2. The structure of strongly hyperbolic families
    13. Theorem 7.11.
    14. Theorem 7.12.
  9. 8. Canvases
    1. Definition 8.1.
    2. Definition 8.2.
    3. Definition 8.3.
    4. Definition 8.4.
    5. Definition 8.5.
    6. Lemma 8.6.
    7. Lemma 8.7.
    8. Theorem 8.8.
    9. Theorem 8.9.
    10. Theorem 8.10.
    11. Theorem 8.11.
    12. Theorem 8.12.
  10. Acknowledgment

Mathematical Fragments

Theorem 1.1.

The family of all embedded graphs that are -critical for some type list-assignment is hyperbolic.

Theorem 1.2.

For every hyperbolic family of embedded graphs that is closed under curve cutting there exists a constant such that every graph embedded in a surface of Euler genus has a non-null-homotopic cycle of length at most .

Theorem 1.3.

For every strongly hyperbolic family of embedded graphs that is closed under curve cutting there exists a constant such that every graph embedded in a surface of Euler genus has at most vertices.

Theorem 1.4.

For every strongly hyperbolic family of rooted embedded graphs that is closed under curve cutting there exist constants such that every rooted graph embedded in a surface of Euler genus has at most vertices, and either has a non-null-homotopic cycle of length at most , or some two vertices in are at a distance at most , or .

Theorem 1.5.

There exist constants such that the following holds. Let be a rooted graph embedded in a surface , and let be a type list-assignment for such that every non-null-homotopic cycle in has length at least and every two vertices in are at a distance at least . Then every -coloring of extends to an -coloring of .

Theorem 2.1.

For every surface there exists an integer such that if is a graph embedded in such that every non-null-homotopic cycle in has length at least , then is -colorable.

Theorem 2.2.

There exists an absolute constant such that if is a graph embedded in a surface of Euler genus such that every non-null-homotopic cycle in  has length at least , then

is -colorable, and

if is triangle-free, then it is -colorable, and

if has no cycles of length five or less, then it is -colorable.

Theorem 2.3.

For every surface , there are (up to isomorphism) only finitely many -critical graphs that embed in .

Corollary 2.4.

For every surface there exists a linear-time algorithm that decides whether an input graph embedded in is -colorable.

Theorem 2.6.

There exists an absolute constant such that every -critical graph of girth at least five embedded in a surface of Euler genus has at most vertices.

Theorem 2.7.

Every planar graph is -choosable.

Theorem 2.8.

Every planar graph of girth at least five is -choosable.

Theorem 2.9.

Let be a graph of crossing number at most one and let be a type list-assignment for . Then is -colorable.

Theorem 2.10.

Let be a graph embedded in a surface of Euler genus . Let  be a list-assignment for and let be a set of vertices in such that for each . If is -critical, then .

Conjecture 2.11.

For every surface there are (up to isomorphism) only finitely many graphs such that embeds in and is -critical for some -list-assignment .

Theorem 2.12.

For every surface there exists an integer such that if is a graph embedded in such that every non-null-homotopic cycle in has length at least , then is -list-colorable.

Theorem 2.13.

For every graph embedded in a surface of Euler genus there exists a set of size at most such that is -choosable.

Theorem 2.14.

Let be a plane graph, let be a path in of length at most three such that some face of is incident with every edge of , and let be a type list-assignment for . Then every -coloring of extends to an -coloring of .

Theorem 2.15.

Let be a plane graph with outer cycle , let be a -list-assignment for , and let be a minimal subgraph of such that every -coloring of that extends to an -coloring of also extends to an -coloring of . Then has at most vertices.

Theorem 2.16.

Let be a plane graph of girth at least five, let be the outer cycle of , and let be a -list-assignment for . Let be a minimal subgraph of  such that every -coloring of that extends to an -coloring of also extends to an -coloring of . Then has at most vertices.

Theorem 2.17.

Let be a triangle-free plane graph, let be the outer cycle of , and let be a -list-assignment for . Let be a minimal subgraph of such that every -coloring of that extends to an -coloring of also extends to an -coloring of . Then has at most vertices.

Theorem 2.18.

There exists an integer such that the following holds: If is a plane graph with a -list-assignment and is such that every two distinct vertices of are at distance at least in , then any -coloring of extends to an -coloring of .

Theorem 2.19.

Let be a graph embedded in a surface of Euler genus such that every non-null-homotopic cycle of has length at least . If is such that every two distinct vertices of are at distance at least in , then any -coloring of extends to a -coloring of .

Question 2.20.

For which do there exist such that if is a graph embedded in a surface such that every non-null-homotopic cycle in has length at least , is a -list-assignment for and is such that all distinct vertices have pairwise distance at least , then any -coloring of  extends to an -coloring of ?

Theorem 2.21.

If can be drawn in the plane with crossings pairwise at distance at least , then is -list-colorable.

Theorem 2.22.

There exists an absolute constant such that if is a planar graph and every two cycles in of length at most four are at distance in of at least , then is -list-colorable.

Theorem 2.23.

Every planar graph on vertices has at least distinct -colorings, and if it has exactly distinct -colorings, then it is obtained from a triangle by repeatedly inserting vertices of degree three inside facial triangles.

Theorem 2.24.

For every surface there exists a constant such that every -colorable graph on vertices embedded in has at least distinct -colorings.

Theorem 2.25.

If is a planar graph and is a -list-assignment for , then has at least distinct -colorings.

Problem 2.26.

Let be a graph embedded in a surface and let be a -list-assignment for . Is it true that if is -colorable, then has at least distinct -colorings, where is a constant depending only on the Euler genus of ?

Definition 3.3.

By a canvas we mean a quadruple , where is an embedded graph with rings and is a list-assignment for such that the subgraph is -colorable.

Definition 3.4.

Let be a graph, let be a subgraph of , and let be a list-assignment for . We say that is -critical with respect to if and for every proper subgraph of that includes as a subgraph there exists an -coloring of that extends to an -coloring of , but does not extend to an -coloring of . We shall abbreviate -critical by -critical. A canvas is critical if is -critical with respect to .

Definition 3.5.

For we define to be the family of all canvases such that for every that does not belong to a ring, and every cycle in of length at most is equal to a ring.

Theorem 3.6.

The families are good families of canvases.

Theorem 3.8.

For every good family of canvases there exist such that the following holds. Let , let be the Euler genus of , let be the total number of ring vertices in , and let be the maximum number of vertices in a ring in . Then has a subgraph such that includes all the rings in , has at most vertices, and for every -coloring of

either does not extend to an -coloring of , or

extends to at least distinct -colorings of .

Furthermore, either

(a)

there exist distinct rings such that the distance in between them is at most , or

every component of satisfies one of the following conditions:

(b)

the graph has a cycle that is not null-homotopic in ; if includes no ring vertex, then the length of is at most , and otherwise it is at most , or

(c)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Theorem 3.9.

There exists an absolute constant such that if is a graph embedded in a surface of Euler genus in such a way such that every non-null-homotopic cycle in has length exceeding , then is -colorable for every type list-assignment .

Theorem 3.10.

There exist absolute constants such that if is a graph embedded in a surface of Euler genus in such a way such that every non-null-homotopic cycle in has length exceeding and is a type list-assignment for , then has at least distinct -colorings.

Theorem 3.11.

There exists an absolute constant such that if is a graph embedded in a surface of Euler genus and there exists a type list-assignment  for such that is -critical, then .

Lemma 3.16.

Let be a graph embedded in a surface of Euler genus , let be pairwise disjoint subgraphs of , and for let be the combinatorial genus of . Then .

Theorem 3.17.

There exists an absolute constant such that for every graph embedded in a surface of Euler genus and every type list-assignment for there exists a set of size at most such that is -colorable.

Theorem 3.19.

There exist absolute constants such that the following holds. Let and assume that every cycle in that is non-null-homotopic in has length at least , where is the Euler genus of . If each ring in has at most four vertices and every two distinct members of are at distance in of at least , then every -coloring of extends to an -coloring of .

Theorem 3.20.

There exist absolute constants such that the following holds. Let be a graph without rings embedded in a surface (without boundary) of Euler genus in such a way that every non-null-homotopic cycle in has length at least . Then

if every two cycles in of length at most four are at distance at least in , then is -list-colorable, and

if every two triangles in are at distance at least in , then is -list-colorable.

Theorem 3.21.

There exist absolute constants and such that the following holds. Let be a graph without rings drawn with crossings in a surface (without boundary) of Euler genus , and let be a type list-assignment for . Assume that every edge crosses at most one other edge. Let be the graph embedded in  obtained from by repeatedly replacing a pair of crossing edges by a degree four vertex adjacent to the ends of and . If every non-null-homotopic cycle in  has length at least and every pair of the new vertices of are at distance at least  in , then is -colorable.

Theorem 3.22.

There exist constants such that the following holds. Let , let be the Euler genus of , and let be the total number of ring vertices. If is an -coloring of such that extends to an -coloring of , then extends to at least distinct -colorings of .

Theorem 4.1.

Let be a plane graph, let be the set of all vertices of that are incident with the outer face, let be a subpath of of length one or two with every vertex and edge incident with the outer face, and let be a list-assignment for such that for every and for every . Let be the number of vertices such that . If has an -coloring, then it has at least distinct -colorings, unless and there exists such that and is adjacent to all the vertices of .

Corollary 4.2.

Let be a plane graph with outer cycle of length at most four, let  be a -list-assignment for , let be an -coloring of , and let be the number of -colorings of that extend . If , then , unless and there exists a vertex not in adjacent to all the vertices of .

Theorem 4.4.

Let be a plane graph with outer cycle , and let be a -list-assignment for . If is -critical with respect to , then

Lemma 4.5.

Let be a plane graph with outer cycle , let be a -list-assignment for , and let be an -coloring of that extends to an -coloring of . Then , where is the number of extensions of to .

Lemma 5.2.

Let be a hyperbolic family of embedded graphs with rings, and let be a Cheeger constant for . Let be a graph embedded in a surface of Euler genus zero with rings and let be the total number of ring vertices. Then and if , then has at most non-ring vertices.

Theorem 5.3.

Let be the family of all embedded graphs with rings such that is -critical with respect to some -list-assignment. Then is hyperbolic with Cheeger constant .

Lemma 5.4.

Let be a graph embedded in the closed unit disk with vertices embedded on the boundary and vertices embedded in the interior of the disk. If every vertex embedded in the interior of the disk has degree in of at least seven, then .

Lemma 5.7.

Let be a graph embedded in the closed unit disk with vertices embedded on the boundary and vertices embedded in the interior of the disk. Assume that every vertex embedded in the interior of the disk has degree in of at least six, and if it has degree exactly six, then it is either incident with a face of size at least four, or it is adjacent to a vertex of degree at least seven, or it is adjacent to a vertex embedded on the boundary of the disk. Then .

Lemma 5.10.

Let be an embedded graph with rings such that every vertex that does not belong to a ring has degree at least four, every face of of size three is incident with an edge of , and every face of size exactly four is incident with either a vertex of degree at least five, or a vertex of degree at least one that belongs to a ring. Let be the Euler genus of and let be the total number of ring vertices. Then .

Example 5.11.

Let be the family of all embedded graphs with rings such that every vertex of that does not belong to a ring has degree at least four, every face of of size three is incident with an edge of , and every face of size exactly four is incident with either a vertex of degree at least five, or a vertex of degree at least one that belongs to a ring. Then is hyperbolic with Cheeger constant  by Lemma 5.10. (The Cheeger constant is indeed and not , because we count vertices in the open disk bounded by .)

Lemma 5.12.

The family of all embedded graphs with rings such that every triangle in is not null-homotopic and is -critical with respect to some -list-assignment is hyperbolic with Cheeger constant .

Lemma 5.13.

Let be the family of embedded graphs with rings such that every cycle in of length at most four is not null-homotopic and is -critical with respect to some -list-assignment. Then is hyperbolic with Cheeger constant .

Definition 5.18.

Let and . Let be a graph with rings embedded in a surface of Euler genus , let be the total number of ring vertices, and let be a list-assignment for . We say that is -exponentially-critical with respect to if and for every proper subgraph of that includes all the rings there exists an -coloring of such that there exist at least distinct -colorings of extending , but there do not exist at least distinct -colorings of extending .

Theorem 5.19.

Let and . Then the family of embedded graphs with rings that are -exponentially-critical with respect to some -list-assignment is hyperbolic with Cheeger constant (independent of and ).

Lemma 5.20.

Let and , and let be the family of embedded graphs with rings such that every cycle of length four or less is equal to a ring and is -exponentially-critical with respect to some -list-assignment. If , then the family is hyperbolic with Cheeger constant independent of and .

Lemma 5.21.

Let and , and let be the family of embedded graphs with rings such that every triangle is equal to a ring and is -exponentially-critical with respect to some -list-assignment. Then the family is hyperbolic with Cheeger constant .

Lemma 6.2.

Let be a graph with rings such that is -cell embedded in a surface . Then has a frame.

Lemma 6.4.

Let be a graph with rings such that is -cell embedded in a surface of Euler genus . Let us assume that no component of is the sphere (without boundary) and that has components. If is a frame of , then the number of non-ring segments of is at most .

Lemma 6.6.

Let be a graph with rings that is -cell embedded in a surface . If is a frame of with minimum, then is an optimal frame.

Lemma 6.8.

Let be a connected graph with rings embedded in a surface , let be a flat separation of , and let be as in the definition of flat separation. Then there exists a multigraph such that

,

is embedded in ,

the point set of intersects the point set of in ,

every vertex of has positive even degree,

every face of includes an edge of , and

all edges of that belong to the same face of either are all incident with a (possibly different) vertex of or all have both ends in .

Lemma 6.9.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let , and let be a flat separation of . If , then .

Lemma 6.10.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let , and let be a flat separation of . Then every vertex of is at distance at most from .

Corollary 6.11.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , and let be a graph embedded in the disk with one ring . Then every vertex of is at distance at most from .

Lemma 6.12.

Let be an integer, let be a graph with rings embedded in a surface , and let . If is at distance at least from every ring and there exists no non-null-homotopic cycle in of length at most such that every vertex of is at distance at most from , then there exists a closed disk such that

every vertex of at distance at most from belongs to ,

every vertex of at distance at most from belongs to the interior of , and

every edge of incident with a vertex at distance at most from belongs to the interior of .

Furthermore, if there exists no non-null-homotopic cycle in of length at most such that every vertex of is at distance at most from , then the closed disk may be chosen so that every edge of with both ends at distance at most from belongs to the interior of .

Lemma 6.13.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let be embedded in a surface , let , and let be an integer. If is at distance at least from every ring and there exists no non-null-homotopic cycle in of length at most such that every vertex of  is at distance at most from , then has at least vertices at distance exactly from .

Lemma 6.14.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let be embedded in a surface , let be an integer with , and let be a vertex of at distance at least from every ring of . Then has a non-null-homotopic cycle of length at most such that every vertex of is at distance at most from .

Lemma 6.17.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let be embedded in a surface , and let be an integer with . If is embedded with edge-width at least , then is locally cylindrical and every vertex is at distance at most from some ring.

Example 6.19.

Let be a fixed integer, and for let denote the Cartesian product of a cycle of length and a path on vertices. We start by considering the obvious embedding of in the sphere . Thus has two faces bounded by cycles of length and all other faces are bounded by cycles of length four. Let be the surface obtained from by removing the face bounded by . Thus is a disk whose boundary is the point set of . Let be obtained from by inserting a crosscap inside the face bounded by (or, equivalently, removing the face bounded by and gluing a Möbius band onto ). Thus has Euler genus and one boundary component, namely the point set of . Let be the family consisting of the embedded graphs with one ring for all . Then is hyperbolic, but the family obtained from by curve cutting is not, because it includes the embedded graphs .

Theorem 6.20.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , let have rings and let it be embedded in a surface of Euler genus with components, where no component of is the sphere. Let be the total number of ring vertices. Assume that either is -cell embedded, or that is closed under curve cutting. Let . If , then there exists a non-null-homotopic (in ) cycle that is disjoint from all of the rings and has length at most .

Corollary 6.21.

Let be a hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant for , let have rings, let it be embedded in a surface of Euler genus , and let be the total number of ring vertices. Let and . If the edge-width of is at least , then is locally cylindrical and every vertex is at distance at most from some ring.

Corollary 6.22.

Let be a hyperbolic family of embedded graphs with no rings, let be closed under curve cutting, let be a Cheeger constant for , and let be a graph with no rings embedded in a surface of Euler genus . Let and . If the edge-width of is at least , then has no subgraph isomorphic to a member of .

Definition 6.23.

Let be a family of embedded graphs with rings, let be embedded in a surface with rings , and let be a non-null-homotopic cycle in . Let be the surface obtained by cutting open along . Thus has one or two more boundary components than , depending on whether is one-sided or two-sided. Let be the graph obtained from during this operation; thus if is one-sided, then corresponds to a cycle in of twice the length of , and if is two-sided, then it corresponds to two cycles in , each of the same length as . If is one-sided, then let ; otherwise let . We regard as a graph embedded in with rings . If for every and every non-null-homotopic cycle in , then we say that is closed under cycle cutting.

Lemma 6.24.

Let be a hyperbolic family of embedded graphs with rings, let be a Cheeger constant for , and let be the inclusion-wise minimal family of embedded graphs that includes and is closed under cycle cutting. Then is hyperbolic with Cheeger constant .

Theorem 6.28.

Let be a hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant for , and let be embedded in a connected surface of Euler genus with rings, and let be the total number of ring vertices. Let , , and . Then has a sleeve-decomposition with parameters that uses at most sleeves, where

(1)

and if and ,

(2)

and if and ,

(3)

and if or .

Theorem 6.29.

Let be a hyperbolic family of embedded graphs with rings, let  be closed under curve cutting, let be a Cheeger constant for , let be embedded in a surface of Euler genus with rings and a total number of ring vertices. Let and . Then  has a sleeve-decomposition with parameters , and the decomposition uses at most sleeves.

Theorem 7.2.

Let be a strongly hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant and a strong hyperbolic constant for , and let be embedded in a surface of Euler genus with a total of ring vertices. Let . Then has at most vertices.

Theorem 7.3.

Let be a strongly hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant and a strong hyperbolic constant for , and let be a graph with rings embedded in a surface . Let . Let be disjoint cycles in such that there exists a cylinder with boundary components , and let . Then includes at most vertices.

Theorem 7.4.

For every integer there exists an integer such that the following holds. Let be a graph embedded in a cylinder with rings and , where , let be a -list-assignment for , and assume that is -critical with respect to . Then has at most vertices.

Lemma 7.5.

Let be the family of embedded graphs with rings from Theorem 5.3. Then is strongly hyperbolic.

Lemma 7.6.

Let be the family from Lemma 5.12; that is, the family of all embedded graphs with rings such that no triangle is null-homotopic and is -critical with respect to some -list-assignment. Then is strongly hyperbolic.

Theorem 7.7.

Let be a graph of girth at least five embedded in a cylinder with rings and , let be a -list-assignment for , and assume that is -critical with respect to . Then has at most vertices.

Lemma 7.8.

Let be the family of embedded graphs with rings such that every cycle in of length at most four is equal to a ring and is -critical with respect to some -list-assignment. Then is strongly hyperbolic.

Theorem 7.11.

Let be a strongly hyperbolic family of embedded graphs with rings, let be closed under curve cutting, let be a Cheeger constant for , let be a strong hyperbolic constant for , and let be a graph with rings and a total number of ring vertices embedded in a surface of Euler genus . Let , , , and for a ring let . Then has at most vertices and either

(a)

there exist distinct rings such that the distance in between them is at most , or

for every component of one of the following conditions holds:

(b)

contains no rings and has a non-null-homotopic cycle in of length at most , or

(c)

contains a unique ring , every vertex of is at distance strictly less than from , and has a non-null-homotopic cycle in of length at most , or

(d)

and , where is the combinatorial genus of and is the number of ring vertices in that are a subgraph of , or

(e)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Theorem 7.12.

Let be as in Theorem 7.11, and for a ring let . Then has at most vertices and either

(a)

there exist distinct rings such that the distance in between them is at most , or

for every component of one of the following conditions holds:

(b)

includes no rings and has a non-null-homotopic cycle in of length at most , or

(c)

contains a unique ring , every vertex of is at distance strictly less than from , and has a non-null-homotopic cycle in of length at most , or

(d)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Definition 8.3.

A family of canvases is good if it is both critically hyperbolic and critically exponentially hyperbolic.

Lemma 8.6.

Let be a family of canvases that satisfies properties (P1) and (P2), let and , and assume that is closed under curve-cutting and is hyperbolic with Cheeger constant . Let and , and let be as in property (P2) applied to and the integer . Assume further that for every and every . If , then is strongly hyperbolic with strong hyperbolic constant .

Lemma 8.7.

The families , and satisfy properties (P1) and (P2).

Theorem 8.8.

The families are good families of canvases.

Theorem 8.9.

For every critically exponentially hyperbolic family of canvases there exist such that the following holds. Let , let be the total number of ring vertices in , and let be the Euler genus of . If is an -coloring of such that extends to an -coloring of , then extends to at least distinct -colorings of .

Theorem 8.10.

For every critically hyperbolic family of canvases there exist such that the following holds. Let , let be the Euler genus of , let be a Cheeger constant of , and let be the total number of ring vertices in . Then has a subgraph such that includes all the rings in , has at most vertices, and for every -coloring of 

either does not extend to an -coloring of , or

extends to an -coloring of , and if is critically exponentially hyperbolic, then extends to at least distinct -colorings of .

Furthermore, either

(a)

there exist distinct rings such that the distance in between them is at most , or

every component of satisfies one of the following conditions:

(b)

contains no rings and has a non-null-homotopic cycle in of length at most , or

(c)

contains a unique ring , every vertex of is at distance strictly less than from , and has a non-null-homotopic cycle in of length at most , or

(d)

and , where is the combinatorial genus of and is the number of ring vertices in that are a subgraph of , or

(e)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Theorem 8.11.

For every critically hyperbolic family of canvases there exist such that the following holds. Let , let be the Euler genus of , let be the total number of ring vertices in , let be a Cheeger constant of , and let be the maximum number of vertices in a ring in . Then has a subgraph such that includes all the rings in , has at most vertices, and for every -coloring of

either does not extend to an -coloring of , or

extends to an -coloring of , and if is critically exponentially hyperbolic, then extends to at least distinct -colorings of .

Furthermore, either

(a)

there exist distinct rings such that the distance in between them is at most , or

every component of satisfies one of the following conditions:

(b)

the graph has a cycle that is not null-homotopic in ; if includes no ring vertex, then the length of is at most , and otherwise it is at most , or

(c)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

Theorem 8.12.

For every critically hyperbolic family of canvases there exist such that the following holds. Let , let be the Euler genus of , and let be the total number of ring vertices in . Then has a subgraph such that includes all the rings in , has at most vertices, and for every -coloring of

either does not extend to an -coloring of , or

extends to an -coloring of , and if is critically exponentially hyperbolic, then extends to at least distinct -colorings of .

Furthermore, either

(a)

there exist distinct rings such that the distance in between them is at most , or

every component of satisfies one of the following conditions:

(b)

includes no rings and has a non-null-homotopic cycle in of length at most , or

(c)

contains a unique ring , every vertex of is at distance strictly less than from , and has a non-null-homotopic cycle in of length at most , or

(d)

includes precisely one member of , there exists a disk that includes , and every vertex of is at distance at most from .

References

Reference [1]
V. A. Aksenov, The extension of a -coloring on planar graphs (Russian), Diskret. Analiz Vyp. 26 Grafy i Testy (1974), 3–19, 84. MR0543788,
Show rawAMSref \bib{Aks}{article}{ author={Aksenov, V. A.}, title={The extension of a $3$-coloring on planar graphs}, language={Russian}, journal={Diskret. Analiz Vyp. 26 Grafy i Testy}, date={1974}, pages={3--19, 84}, review={\MR {0543788}}, }
Reference [2]
Michael O. Albertson, You can’t paint yourself into a corner, J. Combin. Theory Ser. B 73 (1998), no. 2, 189–194, DOI 10.1006/jctb.1998.1827. MR1632011,
Show rawAMSref \bib{Alb}{article}{ author={Albertson, Michael O.}, title={You can't paint yourself into a corner}, journal={J. Combin. Theory Ser. B}, volume={73}, date={1998}, number={2}, pages={189--194}, issn={0095-8956}, review={\MR {1632011}}, doi={10.1006/jctb.1998.1827}, }
Reference [3]
Michael O. Albertson and Joan P. Hutchinson, The three excluded cases of Dirac’s map-color theorem, Second International Conference on Combinatorial Mathematics (New York, 1978), Ann. New York Acad. Sci., vol. 319, New York Acad. Sci., New York, 1979, pp. 7–17. MR556001,
Show rawAMSref \bib{AlbHutch1}{article}{ author={Albertson, Michael O.}, author={Hutchinson, Joan P.}, title={The three excluded cases of Dirac's map-color theorem}, conference={ title={Second International Conference on Combinatorial Mathematics (New York, 1978)}, }, book={ series={Ann. New York Acad. Sci.}, volume={319}, publisher={New York Acad. Sci., New York}, }, date={1979}, pages={7--17}, review={\MR {556001}}, }
Reference [4]
Michael O. Albertson and Joan P. Hutchinson, Extending colorings of locally planar graphs, J. Graph Theory 36 (2001), no. 2, 105–116, DOI 10.1002/1097-0118(200102)36:2<105::AID-JGT5>3.3.CO;2-8. MR1805498,
Show rawAMSref \bib{AlbHutch2}{article}{ author={Albertson, Michael O.}, author={Hutchinson, Joan P.}, title={Extending colorings of locally planar graphs}, journal={J. Graph Theory}, volume={36}, date={2001}, number={2}, pages={105--116}, issn={0364-9024}, review={\MR {1805498}}, doi={10.1002/1097-0118(200102)36:2<105::AID-JGT5>3.3.CO;2-8}, }
[5]
Michael O. Albertson and Joan P. Hutchinson, Graph color extensions: when Hadwiger’s conjecture and embeddings help, Electron. J. Combin. 9 (2002), no. 1, Research Paper 37, 10. MR1928789,
Show rawAMSref \bib{AlbHutch3}{article}{ author={Albertson, Michael O.}, author={Hutchinson, Joan P.}, title={Graph color extensions: when Hadwiger's conjecture and embeddings help}, journal={Electron. J. Combin.}, volume={9}, date={2002}, number={1}, pages={Research Paper 37, 10}, issn={1077-8926}, review={\MR {1928789}}, }
[6]
Michael O. Albertson and Joan P. Hutchinson, Extending precolorings of subgraphs of locally planar graphs, European J. Combin. 25 (2004), no. 6, 863–871, DOI 10.1016/j.ejc.2003.06.005. MR2079903,
Show rawAMSref \bib{AlbHutch4}{article}{ author={Albertson, Michael O.}, author={Hutchinson, Joan P.}, title={Extending precolorings of subgraphs of locally planar graphs}, journal={European J. Combin.}, volume={25}, date={2004}, number={6}, pages={863--871}, issn={0195-6698}, review={\MR {2079903}}, doi={10.1016/j.ejc.2003.06.005}, }
Reference [7]
K. Appel and W. Haken, Every planar map is four colorable. I. Discharging, Illinois J. Math. 21 (1977), no. 3, 429–490. MR0543792,
Show rawAMSref \bib{4CT1}{article}{ author={Appel, K.}, author={Haken, W.}, title={Every planar map is four colorable. I. Discharging}, journal={Illinois J. Math.}, volume={21}, date={1977}, number={3}, pages={429--490}, issn={0019-2082}, review={\MR {0543792}}, }
Reference [8]
K. Appel, W. Haken, and J. Koch, Every planar map is four colorable. II. Reducibility, Illinois J. Math. 21 (1977), no. 3, 491–567. MR0543793,
Show rawAMSref \bib{4CT2}{article}{ author={Appel, K.}, author={Haken, W.}, author={Koch, J.}, title={Every planar map is four colorable. II. Reducibility}, journal={Illinois J. Math.}, volume={21}, date={1977}, number={3}, pages={491--567}, issn={0019-2082}, review={\MR {0543793}}, }
Reference [9]
G. D. Birkhoff and D. C. Lewis, Chromatic polynomials, Trans. Amer. Math. Soc. 60 (1946), 355–451, DOI 10.2307/1990348. MR0018401,
Show rawAMSref \bib{BirLew}{article}{ author={Birkhoff, G. D.}, author={Lewis, D. C.}, title={Chromatic polynomials}, journal={Trans. Amer. Math. Soc.}, volume={60}, date={1946}, pages={355--451}, issn={0002-9947}, review={\MR {0018401}}, doi={10.2307/1990348}, }
Reference [10]
Béla Bollobás, Extremal graph theory, London Mathematical Society Monographs, vol. 11, Academic Press, Inc. [Harcourt Brace Jovanovich, Publishers], London-New York, 1978. MR506522,
Show rawAMSref \bib{BolExtremal}{book}{ author={Bollob\'as, B\'ela}, title={Extremal graph theory}, series={London Mathematical Society Monographs}, volume={11}, publisher={Academic Press, Inc. [Harcourt Brace Jovanovich, Publishers], London-New York}, date={1978}, pages={xx+488}, isbn={0-12-111750-2}, review={\MR {506522}}, }
Reference [11]
Béla Bollobás, Modern graph theory, Graduate Texts in Mathematics, vol. 184, Springer-Verlag, New York, 1998, DOI 10.1007/978-1-4612-0619-4. MR1633290,
Show rawAMSref \bib{Bollobas}{book}{ author={Bollob\'as, B\'ela}, title={Modern graph theory}, series={Graduate Texts in Mathematics}, volume={184}, publisher={Springer-Verlag, New York}, date={1998}, pages={xiv+394}, isbn={0-387-98488-7}, review={\MR {1633290}}, doi={10.1007/978-1-4612-0619-4}, }
Reference [12]
O. V. Borodin, On acyclic colorings of planar graphs, Discrete Math. 25 (1979), no. 3, 211–236, DOI 10.1016/0012-365X(79)90077-3. MR534939,
Show rawAMSref \bib{Borodin}{article}{ author={Borodin, O. V.}, title={On acyclic colorings of planar graphs}, journal={Discrete Math.}, volume={25}, date={1979}, number={3}, pages={211--236}, issn={0012-365X}, review={\MR {534939}}, doi={10.1016/0012-365X(79)90077-3}, }
Reference [13]
O. V. Borodin, A. N. Glebov, M. Montassier, and A. Raspaud, Planar graphs without 5- and 7-cycles and without adjacent triangles are 3-colorable, J. Combin. Theory Ser. B 99 (2009), no. 4, 668–673, DOI 10.1016/j.jctb.2008.11.001. MR2518199,
Show rawAMSref \bib{BorGleMonRas}{article}{ author={Borodin, O. V.}, author={Glebov, A. N.}, author={Montassier, M.}, author={Raspaud, A.}, title={Planar graphs without 5- and 7-cycles and without adjacent triangles are 3-colorable}, journal={J. Combin. Theory Ser. B}, volume={99}, date={2009}, number={4}, pages={668--673}, issn={0095-8956}, review={\MR {2518199}}, doi={10.1016/j.jctb.2008.11.001}, }
Reference [14]
Nathan Chenette, Luke Postle, Noah Streib, Robin Thomas, and Carl Yerger, Five-coloring graphs on the Klein bottle, J. Combin. Theory Ser. B 102 (2012), no. 5, 1067–1098, DOI 10.1016/j.jctb.2012.05.001. MR2959391,
Show rawAMSref \bib{KB}{article}{ author={Chenette, Nathan}, author={Postle, Luke}, author={Streib, Noah}, author={Thomas, Robin}, author={Yerger, Carl}, title={Five-coloring graphs on the Klein bottle}, journal={J. Combin. Theory Ser. B}, volume={102}, date={2012}, number={5}, pages={1067--1098}, issn={0095-8956}, review={\MR {2959391}}, doi={10.1016/j.jctb.2012.05.001}, }
Reference [15]
Vincent Cohen-Addad, Michael Hebdige, Daniel Král’, Zhentao Li, and Esteban Salgado, Steinberg’s conjecture is false, J. Combin. Theory Ser. B 122 (2017), 452–456, DOI 10.1016/j.jctb.2016.07.006. MR3575214,
Show rawAMSref \bib{CohHebKraLiSal}{article}{ author={Cohen-Addad, Vincent}, author={Hebdige, Michael}, author={Kr\'al', Daniel}, author={Li, Zhentao}, author={Salgado, Esteban}, title={Steinberg's conjecture is false}, journal={J. Combin. Theory Ser. B}, volume={122}, date={2017}, pages={452--456}, issn={0095-8956}, review={\MR {3575214}}, doi={10.1016/j.jctb.2016.07.006}, }
Reference [16]
Alice M. Dean and Joan P. Hutchinson, List-coloring graphs on surfaces with varying list-sizes, Electron. J. Combin. 19 (2012), no. 4, Paper 50, 16. MR3007185,
Show rawAMSref \bib{DeanHutch}{article}{ author={Dean, Alice M.}, author={Hutchinson, Joan P.}, title={List-coloring graphs on surfaces with varying list-sizes}, journal={Electron. J. Combin.}, volume={19}, date={2012}, number={4}, pages={Paper 50, 16}, issn={1077-8926}, review={\MR {3007185}}, }
Reference [17]
Matt DeVos, Ken-ichi Kawarabayashi, and Bojan Mohar, Locally planar graphs are 5-choosable, J. Combin. Theory Ser. B 98 (2008), no. 6, 1215–1232, DOI 10.1016/j.jctb.2008.01.003. MR2462315,
Show rawAMSref \bib{DeVos}{article}{ author={DeVos, Matt}, author={Kawarabayashi, Ken-ichi}, author={Mohar, Bojan}, title={Locally planar graphs are 5-choosable}, journal={J. Combin. Theory Ser. B}, volume={98}, date={2008}, number={6}, pages={1215--1232}, issn={0095-8956}, review={\MR {2462315}}, doi={10.1016/j.jctb.2008.01.003}, }
Reference [18]
Reinhard Diestel, Graph theory, 4th ed., Graduate Texts in Mathematics, vol. 173, Springer, Heidelberg, 2010, DOI 10.1007/978-3-642-14279-6. MR2744811,
Show rawAMSref \bib{Diestel}{book}{ author={Diestel, Reinhard}, title={Graph theory}, series={Graduate Texts in Mathematics}, volume={173}, edition={4}, publisher={Springer, Heidelberg}, date={2010}, pages={xviii+437}, isbn={978-3-642-14278-9}, review={\MR {2744811}}, doi={10.1007/978-3-642-14279-6}, }
Reference [19]
G. A. Dirac, Map-colour theorems, Canadian J. Math. 4 (1952), 480–490. MR0050869,
Show rawAMSref \bib{Dirac1}{article}{ author={Dirac, G. A.}, title={Map-colour theorems}, journal={Canadian J. Math.}, volume={4}, date={1952}, pages={480--490}, issn={0008-414X}, review={\MR {0050869}}, }
Reference [20]
G. A. Dirac, The colouring of maps, J. London Math. Soc. 28 (1953), 476–480, DOI 10.1112/jlms/s1-28.4.476. MR0056904,
Show rawAMSref \bib{Dirac2}{article}{ author={Dirac, G. A.}, title={The colouring of maps}, journal={J. London Math. Soc.}, volume={28}, date={1953}, pages={476--480}, issn={0024-6107}, review={\MR {0056904}}, doi={10.1112/jlms/s1-28.4.476}, }
Reference [21]
Zdeněk Dvořák, 3-choosability of planar graphs with -cycles far apart, J. Combin. Theory Ser. B 104 (2014), 28–59, DOI 10.1016/j.jctb.2013.10.004. MR3132743,
Show rawAMSref \bib{3ChooseFarApart}{article}{ author={Dvo\v r\'ak, Zden\v ek}, title={3-choosability of planar graphs with $(\leq 4)$-cycles far apart}, journal={J. Combin. Theory Ser. B}, volume={104}, date={2014}, pages={28--59}, issn={0095-8956}, review={\MR {3132743}}, doi={10.1016/j.jctb.2013.10.004}, }
Reference [22]
Zdeněk Dvořák and Ken-ichi Kawarabayashi, Choosability of planar graphs of girth , arXiv:1109.2976v1.
[23]
Zdeněk Dvořák and Ken-ichi Kawarabayashi, List-coloring embedded graphs, Proceedings of the Twenty-Fourth Annual ACM-SIAM Symposium on Discrete Algorithms, SIAM, Philadelphia, PA, 2012, pp. 1004–1012. MR3202964,
Show rawAMSref \bib{DvoKawListEmb}{article}{ author={Dvo\v r\'ak, Zden\v ek}, author={Kawarabayashi, Ken-ichi}, title={List-coloring embedded graphs}, conference={ title={Proceedings of the Twenty-Fourth Annual ACM-SIAM Symposium on Discrete Algorithms}, }, book={ publisher={SIAM, Philadelphia, PA}, }, date={2012}, pages={1004--1012}, review={\MR {3202964}}, }
Reference [24]
Zdeněk Dvořák, Daniel Král’, and Robin Thomas, Three-coloring triangle-free graphs on surfaces, Proceedings of the twentieth Annual ACM-SIAM Symposium on Discrete Algorithms (SODA), New York, NY (2009), 120–129.
Reference [25]
Zdeněk Dvořák, Daniel Král’, and Robin Thomas, Testing first-order properties for subclasses of sparse graphs, J. ACM 60 (2013), no. 5, Art. 36, 24, DOI 10.1145/2499483. MR3124685,
Show rawAMSref \bib{DvoKraThofol}{article}{ author={Dvo\v r\'ak, Zden\v ek}, author={Kr\'al', Daniel}, author={Thomas, Robin}, title={Testing first-order properties for subclasses of sparse graphs}, journal={J. ACM}, volume={60}, date={2013}, number={5}, pages={Art. 36, 24}, issn={0004-5411}, review={\MR {3124685}}, doi={10.1145/2499483}, }
Reference [26]
Zdeněk Dvořák, Daniel Král’, and Robin Thomas, Three-coloring triangle-free graphs on surfaces III. Graphs of girth five, arXiv:1402.4710.
Reference [27]
Zdeněk Dvořák, Daniel Král’, and Robin Thomas, Three-coloring triangle-free graphs on surfaces V. Coloring planar graphs with distant anomalies, arXiv:0911.0885.
Reference [28]
Zdeněk Dvořák, Daniel Král’, and Robin Thomas, Three-coloring triangle-free graphs on surfaces VI. -colorability of quadrangulations, arXiv:1509.01013.
Reference [29]
Zdeněk Dvořák, Bernard Lidický, and Bojan Mohar, 5-choosability of graphs with crossings far apart, J. Combin. Theory Ser. B 123 (2017), 54–96, DOI 10.1016/j.jctb.2016.11.004. MR3597095,
Show rawAMSref \bib{Crossings}{article}{ author={Dvo\v r\'ak, Zden\v ek}, author={Lidick\'y, Bernard}, author={Mohar, Bojan}, title={5-choosability of graphs with crossings far apart}, journal={J. Combin. Theory Ser. B}, volume={123}, date={2017}, pages={54--96}, issn={0095-8956}, review={\MR {3597095}}, doi={10.1016/j.jctb.2016.11.004}, }
Reference [30]
Zdeněk Dvořák, Bernard Lidický, Bojan Mohar, and Luke Postle, 5-list-coloring planar graphs with distant precolored vertices, J. Combin. Theory Ser. B 122 (2017), 311–352, DOI 10.1016/j.jctb.2016.06.006. MR3575207,
Show rawAMSref \bib{AlbertsonsConj}{article}{ author={Dvo\v r\'ak, Zden\v ek}, author={Lidick\'y, Bernard}, author={Mohar, Bojan}, author={Postle, Luke}, title={5-list-coloring planar graphs with distant precolored vertices}, journal={J. Combin. Theory Ser. B}, volume={122}, date={2017}, pages={311--352}, issn={0095-8956}, review={\MR {3575207}}, doi={10.1016/j.jctb.2016.06.006}, }
Reference [31]
David Eppstein, Subgraph isomorphism in planar graphs and related problems, J. Graph Algorithms Appl. 3 (1999), no. 3, 27, DOI 10.7155/jgaa.00014. MR1750082,
Show rawAMSref \bib{Eppstein2}{article}{ author={Eppstein, David}, title={Subgraph isomorphism in planar graphs and related problems}, journal={J. Graph Algorithms Appl.}, volume={3}, date={1999}, pages={no. 3, 27}, issn={1526-1719}, review={\MR {1750082}}, doi={10.7155/jgaa.00014}, }
Reference [32]
Paul Erdős, Arthur L. Rubin, and Herbert Taylor, Choosability in graphs, Proceedings of the West Coast Conference on Combinatorics, Graph Theory and Computing (Humboldt State Univ., Arcata, Calif., 1979), Congress. Numer., XXVI, Utilitas Math., Winnipeg, Man., 1980, pp. 125–157. MR593902,
Show rawAMSref \bib{ErdRubTay}{article}{ author={Erd\H os, Paul}, author={Rubin, Arthur L.}, author={Taylor, Herbert}, title={Choosability in graphs}, conference={ title={Proceedings of the West Coast Conference on Combinatorics, Graph Theory and Computing}, address={Humboldt State Univ., Arcata, Calif.}, date={1979}, }, book={ series={Congress. Numer., XXVI}, publisher={Utilitas Math., Winnipeg, Man.}, }, date={1980}, pages={125--157}, review={\MR {593902}}, }
Reference [33]
P. Franklin, A Six Color Problem, J. Math. Phys. 13 (1934), 363–379.
Reference [34]
Steve Fisk, The nonexistence of colorings, J. Combinatorial Theory Ser. B 24 (1978), no. 2, 247–248. MR0491299,
Show rawAMSref \bib{Fisk}{article}{ author={Fisk, Steve}, title={The nonexistence of colorings}, journal={J. Combinatorial Theory Ser. B}, volume={24}, date={1978}, number={2}, pages={247--248}, review={\MR {0491299}}, }
Reference [35]
Steve Fisk and Bojan Mohar, Coloring graphs without short nonbounding cycles, J. Combin. Theory Ser. B 60 (1994), no. 2, 268–276, DOI 10.1006/jctb.1994.1018. MR1271274,
Show rawAMSref \bib{FisMoh}{article}{ author={Fisk, Steve}, author={Mohar, Bojan}, title={Coloring graphs without short nonbounding cycles}, journal={J. Combin. Theory Ser. B}, volume={60}, date={1994}, number={2}, pages={268--276}, issn={0095-8956}, review={\MR {1271274}}, doi={10.1006/jctb.1994.1018}, }
Reference [36]
T. Gallai, Kritische Graphen. II (German, with Russian summary), Magyar Tud. Akad. Mat. Kutató Int. Közl. 8 (1963), 373–395 (1964). MR0188100,
Show rawAMSref \bib{Gallai}{article}{ author={Gallai, T.}, title={Kritische Graphen. II}, language={German, with Russian summary}, journal={Magyar Tud. Akad. Mat. Kutat\'o Int. K\"ozl.}, volume={8}, date={1963}, pages={373--395 (1964)}, review={\MR {0188100}}, }
Reference [37]
Michael R. Garey and David S. Johnson, Computers and intractability, W. H. Freeman and Co., San Francisco, Calif., 1979. A guide to the theory of NP-completeness; A Series of Books in the Mathematical Sciences. MR519066,
Show rawAMSref \bib{GarJoh}{book}{ author={Garey, Michael R.}, author={Johnson, David S.}, title={Computers and intractability}, note={A guide to the theory of NP-completeness; A Series of Books in the Mathematical Sciences}, publisher={W. H. Freeman and Co., San Francisco, Calif.}, date={1979}, pages={x+338}, isbn={0-7167-1045-5}, review={\MR {519066}}, }
[38]
John Gimbel and Carsten Thomassen, Coloring graphs with fixed genus and girth, Trans. Amer. Math. Soc. 349 (1997), no. 11, 4555–4564, DOI 10.1090/S0002-9947-97-01926-0. MR1422897,
Show rawAMSref \bib{GimbelThom}{article}{ author={Gimbel, John}, author={Thomassen, Carsten}, title={Coloring graphs with fixed genus and girth}, journal={Trans. Amer. Math. Soc.}, volume={349}, date={1997}, number={11}, pages={4555--4564}, issn={0002-9947}, review={\MR {1422897}}, doi={10.1090/S0002-9947-97-01926-0}, }
Reference [39]
Herbert Grötzsch, Zur Theorie der diskreten Gebilde. VII. Ein Dreifarbensatz für dreikreisfreie Netze auf der Kugel (German), Wiss. Z. Martin-Luther-Univ. Halle-Wittenberg. Math.-Nat. Reihe 8 (1958/1959), 109–120. MR0116320,
Show rawAMSref \bib{Gro}{article}{ author={Gr\"otzsch, Herbert}, title={Zur Theorie der diskreten Gebilde. VII. Ein Dreifarbensatz f\"ur dreikreisfreie Netze auf der Kugel}, language={German}, journal={Wiss. Z. Martin-Luther-Univ. Halle-Wittenberg. Math.-Nat. Reihe}, volume={8}, date={1958/1959}, pages={109--120}, review={\MR {0116320}}, }
Reference [40]
Ivan Havel, O zbarvitelnosti rovinných grafů třemi barvami, Mathematics (Geometry and Graph Theory), Univ. Karlova, Prague (1970), 89–91.
[41]
Joan P. Hutchinson, On list-coloring extendable outerplanar graphs, Ars Math. Contemp. 5 (2012), no. 1, 171–184. MR2880673,
Show rawAMSref \bib{HutchOuterplanar}{article}{ author={Hutchinson, Joan P.}, title={On list-coloring extendable outerplanar graphs}, journal={Ars Math. Contemp.}, volume={5}, date={2012}, number={1}, pages={171--184}, issn={1855-3966}, review={\MR {2880673}}, }
Reference [42]
T. R. Jensen and B. Toft, Graph Coloring Problems, J. Wiley & Sons, New York, Chichester, Brisbane, Toronto, Singapore 1995.
Reference [43]
Ken-ichi Kawarabayashi, Daniel Kráľ, Jan Kynčl, and Bernard Lidický, 6-critical graphs on the Klein bottle, SIAM J. Discrete Math. 23 (2008/09), no. 1, 372–383, DOI 10.1137/070706835. MR2476836,
Show rawAMSref \bib{KB2}{article}{ author={Kawarabayashi, Ken-ichi}, author={Kr\'a\v l, Daniel}, author={Kyn\v cl, Jan}, author={Lidick\'y, Bernard}, title={6-critical graphs on the Klein bottle}, journal={SIAM J. Discrete Math.}, volume={23}, date={2008/09}, number={1}, pages={372--383}, issn={0895-4801}, review={\MR {2476836}}, doi={10.1137/070706835}, }
Reference [44]
Ken-ichi Kawarabayashi and Bojan Mohar, List-color-critical graphs on a fixed surface, Proceedings of the Twentieth Annual ACM-SIAM Symposium on Discrete Algorithms, SIAM, Philadelphia, PA, 2009, pp. 1156–1165. MR2807558,
Show rawAMSref \bib{KM}{article}{ author={Kawarabayashi, Ken-ichi}, author={Mohar, Bojan}, title={List-color-critical graphs on a fixed surface}, conference={ title={Proceedings of the Twentieth Annual ACM-SIAM Symposium on Discrete Algorithms}, }, book={ publisher={SIAM, Philadelphia, PA}, }, date={2009}, pages={1156--1165}, review={\MR {2807558}}, }
Reference [45]
Ken-ichi Kawarabayashi and Carsten Thomassen, From the plane to higher surfaces, J. Combin. Theory Ser. B 102 (2012), no. 4, 852–868, DOI 10.1016/j.jctb.2012.03.001. MR2927410,
Show rawAMSref \bib{KawThofrom}{article}{ author={Kawarabayashi, Ken-ichi}, author={Thomassen, Carsten}, title={From the plane to higher surfaces}, journal={J. Combin. Theory Ser. B}, volume={102}, date={2012}, number={4}, pages={852--868}, issn={0095-8956}, review={\MR {2927410}}, doi={10.1016/j.jctb.2012.03.001}, }
Reference [46]
Tom Kelly and Luke Postle, Exponentially many 4-list colorings of triangle-free graphs on surfaces, J. Graph Theory 87 (2018), no. 2, 230–238, DOI 10.1002/jgt.22153. MR3742180,
Show rawAMSref \bib{KelPos}{article}{ author={Kelly, Tom}, author={Postle, Luke}, title={Exponentially many 4-list colorings of triangle-free graphs on surfaces}, journal={J. Graph Theory}, volume={87}, date={2018}, number={2}, pages={230--238}, issn={0364-9024}, review={\MR {3742180}}, doi={10.1002/jgt.22153}, }
[47]
Bojan Mohar and Carsten Thomassen, Graphs on surfaces, Johns Hopkins Studies in the Mathematical Sciences, Johns Hopkins University Press, Baltimore, MD, 2001. MR1844449,
Show rawAMSref \bib{MoharThom}{book}{ author={Mohar, Bojan}, author={Thomassen, Carsten}, title={Graphs on surfaces}, series={Johns Hopkins Studies in the Mathematical Sciences}, publisher={Johns Hopkins University Press, Baltimore, MD}, date={2001}, pages={xii+291}, isbn={0-8018-6689-8}, review={\MR {1844449}}, }
Reference [48]
Luke Jamison Postle, 5-list-coloring graphs on surfaces, ProQuest LLC, Ann Arbor, MI, 2012. Thesis (Ph.D.)–Georgia Institute of Technology. MR3152411,
Show rawAMSref \bib{PosPhD}{book}{ author={Postle, Luke Jamison}, title={5-list-coloring graphs on surfaces}, note={Thesis (Ph.D.)--Georgia Institute of Technology}, publisher={ProQuest LLC, Ann Arbor, MI}, date={2012}, pages={253}, isbn={978-1267-89709-1}, review={\MR {3152411}}, }
Reference [49]
Luke Postle, 3 list coloring graphs of girth at least five on surfaces, arXiv:1710.06898v1.
Reference [50]
Luke Postle and Robin Thomas, Five-list-coloring graphs on surfaces I. Two lists of size two in planar graphs, J. Combin. Theory Ser. B 111 (2015), 234–241, DOI 10.1016/j.jctb.2014.08.003. MR3315607,
Show rawAMSref \bib{PosThoTwotwo}{article}{ author={Postle, Luke}, author={Thomas, Robin}, title={Five-list-coloring graphs on surfaces I. Two lists of size two in planar graphs}, journal={J. Combin. Theory Ser. B}, volume={111}, date={2015}, pages={234--241}, issn={0095-8956}, review={\MR {3315607}}, doi={10.1016/j.jctb.2014.08.003}, }
Reference [51]
Luke Postle and Robin Thomas, Five-list-coloring graphs on surfaces II. A linear bound for critical graphs in a disk, J. Combin. Theory Ser. B 119 (2016), 42–65, DOI 10.1016/j.jctb.2015.12.005. MR3486337,
Show rawAMSref \bib{PosThoLinDisk}{article}{ author={Postle, Luke}, author={Thomas, Robin}, title={Five-list-coloring graphs on surfaces II. A linear bound for critical graphs in a disk}, journal={J. Combin. Theory Ser. B}, volume={119}, date={2016}, pages={42--65}, issn={0095-8956}, review={\MR {3486337}}, doi={10.1016/j.jctb.2015.12.005}, }
Reference [52]
Luke Postle and Robin Thomas, Five-list-coloring graphs on surfaces III. One list of size one and one list of size two, J. Combin. Theory Ser. B 128 (2018), 1–16, DOI 10.1016/j.jctb.2017.06.004. MR3725183,
Show rawAMSref \bib{PosThoTwoOne}{article}{ author={Postle, Luke}, author={Thomas, Robin}, title={Five-list-coloring graphs on surfaces III. One list of size one and one list of size two}, journal={J. Combin. Theory Ser. B}, volume={128}, date={2018}, pages={1--16}, issn={0095-8956}, review={\MR {3725183}}, doi={10.1016/j.jctb.2017.06.004}, }
Reference [53]
Gerhard Ringel and J. W. T. Youngs, Solution of the Heawood map-coloring problem, Proc. Nat. Acad. Sci. U.S.A. 60 (1968), 438–445. MR0228378,
Show rawAMSref \bib{RingelYoungs}{article}{ author={Ringel, Gerhard}, author={Youngs, J. W. T.}, title={Solution of the Heawood map-coloring problem}, journal={Proc. Nat. Acad. Sci. U.S.A.}, volume={60}, date={1968}, pages={438--445}, issn={0027-8424}, review={\MR {0228378}}, }
Reference [54]
Neil Robertson, Daniel Sanders, Paul Seymour, and Robin Thomas, The four-colour theorem, J. Combin. Theory Ser. B 70 (1997), no. 1, 2–44, DOI 10.1006/jctb.1997.1750. MR1441258,
Show rawAMSref \bib{4CTRSST}{article}{ author={Robertson, Neil}, author={Sanders, Daniel}, author={Seymour, Paul}, author={Thomas, Robin}, title={The four-colour theorem}, journal={J. Combin. Theory Ser. B}, volume={70}, date={1997}, number={1}, pages={2--44}, issn={0095-8956}, review={\MR {1441258}}, doi={10.1006/jctb.1997.1750}, }
Reference [55]
Carsten Thomassen, Five-coloring maps on surfaces, J. Combin. Theory Ser. B 59 (1993), no. 1, 89–105, DOI 10.1006/jctb.1993.1057. MR1234386,
Show rawAMSref \bib{Tho5colmaps}{article}{ author={Thomassen, Carsten}, title={Five-coloring maps on surfaces}, journal={J. Combin. Theory Ser. B}, volume={59}, date={1993}, number={1}, pages={89--105}, issn={0095-8956}, review={\MR {1234386}}, doi={10.1006/jctb.1993.1057}, }
Reference [56]
Carsten Thomassen, Five-coloring graphs on the torus, J. Combin. Theory Ser. B 62 (1994), no. 1, 11–33, DOI 10.1006/jctb.1994.1052. MR1290628,
Show rawAMSref \bib{ThomTorus}{article}{ author={Thomassen, Carsten}, title={Five-coloring graphs on the torus}, journal={J. Combin. Theory Ser. B}, volume={62}, date={1994}, number={1}, pages={11--33}, issn={0095-8956}, review={\MR {1290628}}, doi={10.1006/jctb.1994.1052}, }
Reference [57]
Carsten Thomassen, Every planar graph is -choosable, J. Combin. Theory Ser. B 62 (1994), no. 1, 180–181, DOI 10.1006/jctb.1994.1062. MR1290638,
Show rawAMSref \bib{ThomPlanar}{article}{ author={Thomassen, Carsten}, title={Every planar graph is $5$-choosable}, journal={J. Combin. Theory Ser. B}, volume={62}, date={1994}, number={1}, pages={180--181}, issn={0095-8956}, review={\MR {1290638}}, doi={10.1006/jctb.1994.1062}, }
Reference [58]
Carsten Thomassen, Grötzsch’s -color theorem and its counterparts for the torus and the projective plane, J. Combin. Theory Ser. B 62 (1994), no. 2, 268–279, DOI 10.1006/jctb.1994.1069. MR1305053,
Show rawAMSref \bib{Thom3Color}{article}{ author={Thomassen, Carsten}, title={Gr\"otzsch's $3$-color theorem and its counterparts for the torus and the projective plane}, journal={J. Combin. Theory Ser. B}, volume={62}, date={1994}, number={2}, pages={268--279}, issn={0095-8956}, review={\MR {1305053}}, doi={10.1006/jctb.1994.1069}, }
Reference [59]
Carsten Thomassen, -list-coloring planar graphs of girth , J. Combin. Theory Ser. B 64 (1995), no. 1, 101–107, DOI 10.1006/jctb.1995.1027. MR1328294,
Show rawAMSref \bib{Thom3ListColor}{article}{ author={Thomassen, Carsten}, title={$3$-list-coloring planar graphs of girth $5$}, journal={J. Combin. Theory Ser. B}, volume={64}, date={1995}, number={1}, pages={101--107}, issn={0095-8956}, review={\MR {1328294}}, doi={10.1006/jctb.1995.1027}, }
Reference [60]
Carsten Thomassen, Color-critical graphs on a fixed surface, J. Combin. Theory Ser. B 70 (1997), no. 1, 67–100, DOI 10.1006/jctb.1996.1722. MR1441260,
Show rawAMSref \bib{ThomCritical}{article}{ author={Thomassen, Carsten}, title={Color-critical graphs on a fixed surface}, journal={J. Combin. Theory Ser. B}, volume={70}, date={1997}, number={1}, pages={67--100}, issn={0095-8956}, review={\MR {1441260}}, doi={10.1006/jctb.1996.1722}, }
Reference [61]
Carsten Thomassen, A short list color proof of Grötzsch’s theorem, J. Combin. Theory Ser. B 88 (2003), no. 1, 189–192, DOI 10.1016/S0095-8956(03)00029-7. MR1974149,
Show rawAMSref \bib{ThoShortlist}{article}{ author={Thomassen, Carsten}, title={A short list color proof of Gr\"otzsch's theorem}, journal={J. Combin. Theory Ser. B}, volume={88}, date={2003}, number={1}, pages={189--192}, issn={0095-8956}, review={\MR {1974149}}, doi={10.1016/S0095-8956(03)00029-7}, }
Reference [62]
Carsten Thomassen, The chromatic number of a graph of girth 5 on a fixed surface, J. Combin. Theory Ser. B 87 (2003), no. 1, 38–71, DOI 10.1016/S0095-8956(02)00027-8. Dedicated to Crispin St. J. A. Nash-Williams. MR1967881,
Show rawAMSref \bib{ThoGirth5}{article}{ author={Thomassen, Carsten}, title={The chromatic number of a graph of girth 5 on a fixed surface}, note={Dedicated to Crispin St. J. A. Nash-Williams}, journal={J. Combin. Theory Ser. B}, volume={87}, date={2003}, number={1}, pages={38--71}, issn={0095-8956}, review={\MR {1967881}}, doi={10.1016/S0095-8956(02)00027-8}, }
Reference [63]
Carsten Thomassen, The number of -colorings of a graph on a fixed surface, Discrete Math. 306 (2006), no. 23, 3145–3153, DOI 10.1016/j.disc.2005.04.027. MR2273145,
Show rawAMSref \bib{ThomExpQuestion}{article}{ author={Thomassen, Carsten}, title={The number of $k$-colorings of a graph on a fixed surface}, journal={Discrete Math.}, volume={306}, date={2006}, number={23}, pages={3145--3153}, issn={0012-365X}, review={\MR {2273145}}, doi={10.1016/j.disc.2005.04.027}, }
Reference [64]
Carsten Thomassen, Exponentially many 5-list-colorings of planar graphs, J. Combin. Theory Ser. B 97 (2007), no. 4, 571–583, DOI 10.1016/j.jctb.2006.09.002. MR2325797,
Show rawAMSref \bib{ThomWheels}{article}{ author={Thomassen, Carsten}, title={Exponentially many 5-list-colorings of planar graphs}, journal={J. Combin. Theory Ser. B}, volume={97}, date={2007}, number={4}, pages={571--583}, issn={0095-8956}, review={\MR {2325797}}, doi={10.1016/j.jctb.2006.09.002}, }
Reference [65]
Carsten Thomassen, Many 3-colorings of triangle-free planar graphs, J. Combin. Theory Ser. B 97 (2007), no. 3, 334–349, DOI 10.1016/j.jctb.2006.06.005. MR2305887,
Show rawAMSref \bib{Thomany}{article}{ author={Thomassen, Carsten}, title={Many 3-colorings of triangle-free planar graphs}, journal={J. Combin. Theory Ser. B}, volume={97}, date={2007}, number={3}, pages={334--349}, issn={0095-8956}, review={\MR {2305887}}, doi={10.1016/j.jctb.2006.06.005}, }
Reference [66]
Margit Voigt, List colourings of planar graphs, Discrete Math. 120 (1993), no. 1-3, 215–219, DOI 10.1016/0012-365X(93)90579-I. MR1235909,
Show rawAMSref \bib{Voigt}{article}{ author={Voigt, Margit}, title={List colourings of planar graphs}, journal={Discrete Math.}, volume={120}, date={1993}, number={1-3}, pages={215--219}, issn={0012-365X}, review={\MR {1235909}}, doi={10.1016/0012-365X(93)90579-I}, }
Reference [67]
Carl Roger Yerger Jr, Color-critical graphs on surfaces, ProQuest LLC, Ann Arbor, MI, 2010. Thesis (Ph.D.)–Georgia Institute of Technology. MR2941743,
Show rawAMSref \bib{YerPhD}{book}{ author={Yerger, Carl Roger, Jr}, title={Color-critical graphs on surfaces}, note={Thesis (Ph.D.)--Georgia Institute of Technology}, publisher={ProQuest LLC, Ann Arbor, MI}, date={2010}, pages={150}, isbn={978-1124-56681-8}, review={\MR {2941743}}, }

Article Information

MSC 2010
Primary: 05C10 (Planar graphs; geometric and topological aspects of graph theory), 05C15 (Coloring of graphs and hypergraphs)
Author Information
Luke Postle
Department of Combinatorics and Optimization, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
lpostle@uwaterloo.ca
MathSciNet
Robin Thomas
School of Mathematics, Georgia Institute of Technology, Atlanta, Georgia 30332-0160
thomas@math.gatech.edu
MathSciNet
Additional Notes

The first author was partially supported by NSERC under Discovery Grant No. 2014-06162, the Ontario Early Researcher Awards program and the Canada Research Chairs program.

The second author was partially supported by NSF under Grant No. DMS-1202640.

Journal Information
Transactions of the American Mathematical Society, Series B, Volume 5, Issue 7, ISSN 2330-0000, published by the American Mathematical Society, Providence, Rhode Island.
Publication History
This article was received on , revised on , and published on .
Copyright Information
Copyright 2018 by the authors under Creative Commons Attribution 3.0 License (CC BY 3.0)
Article References
  • Permalink
  • Permalink (PDF)
  • DOI 10.1090/btran/26
  • MathSciNet Review: 3882883
  • Show rawAMSref \bib{3882883}{article}{ author={Postle, Luke}, author={Thomas, Robin}, title={Hyperbolic families and coloring graphs on surfaces}, journal={Trans. Amer. Math. Soc. Ser. B}, volume={5}, number={7}, date={2018}, pages={167-221}, issn={2330-0000}, review={3882883}, doi={10.1090/btran/26}, }

Settings

Change font size
Resize article panel
Enable equation enrichment

Note. To explore an equation, focus it (e.g., by clicking on it) and use the arrow keys to navigate its structure. Screenreader users should be advised that enabling speech synthesis will lead to duplicate aural rendering.

For more information please visit the AMS MathViewer documentation.