$H$-free subgraphs of dense graphs maximizing the number of cliques and their blow-ups
Noga Alon, Clara Shikhelman

TL;DR
This paper investigates the structure of dense, $H$-free subgraphs that maximize the number of cliques and their blow-ups, revealing their colorability and extending to various forbidden graphs and configurations.
Contribution
It establishes that such extremal subgraphs are $(k-1)$-colorable under certain conditions and extends results to general $H$ and balanced blowups.
Findings
Maximal $H$-free subgraphs are $(k-1)$-colorable in dense graphs.
Results apply to graphs with high minimum degree and specific chromatic properties.
Extensions cover general forbidden graphs and maximization of balanced blowups.
Abstract
We consider the structure of -free subgraphs of graphs with high minimal degree. We prove that for every there exists an so that the following holds. For every graph with chromatic number from which one can delete an edge and reduce the chromatic number, and for every graph on vertices in which all degrees are at least , any subgraph of which is -free and contains the maximum number of copies of the complete graph is -colorable. We also consider several extensions for the case of a general forbidden graph of a given chromatic number, and for subgraphs maximizing the number of copies of balanced blowups of complete graphs.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsLimits and Structures in Graph Theory · Advanced Graph Theory Research · Graph Labeling and Dimension Problems
-free subgraphs of dense graphs maximizing the number of
cliques and their blow-ups
Noga Alon Sackler School of Mathematics and Blavatnik School of Computer Science, Tel Aviv University, Tel Aviv 69978, Israel. Email: [email protected]. Research supported in part by a BSF grant, an ISF grant and a GIF grant.
Clara Shikhelman Sackler School of Mathematics, Tel Aviv University, Tel Aviv 69978, Israel. Email: [email protected]. Research supported in part by an ISF grant.
Abstract
We consider the structure of -free subgraphs of graphs with high minimal degree. We prove that for every there exists an so that the following holds. For every graph with chromatic number from which one can delete an edge and reduce the chromatic number, and for every graph on vertices in which all degrees are at least , any subgraph of which is -free and contains the maximum number of copies of the complete graph is -colorable.
We also consider several extensions for the case of a general forbidden graph of a given chromatic number, and for subgraphs maximizing the number of copies of balanced blowups of complete graphs.
1 Introduction
The well known theorem of Turán ([19]) states that a -free subgraph of the complete graph on vertices with the maximum possible number of edges is -chromatic. Erdős, Stone and Simonovits show in [14], [12] that for general with the maximum possible number of edges in an -free graph on vertices is at most more than the number of edges in a -chromatic graph on vertices. In [4] it is shown that the same holds for -free subgraphs of the complete graph that have the maximum possible number of copies of for a fixed such that .
Looking at subgraphs of general graphs it is clear that a -free subgraph of with the maximum possible number of edges has at least as many edges as the largest -partite subgraph. In [11] Erdős asked for which graphs there is an equality between the two. In [1] it is shown that this is the case for line graphs of bipartite graphs. In a different direction, in [8] it is proved that if a graph has a high enough minimum degree then any subgraph of it which is -free and has the maximum possible number of edges is bipartite. In [7] a stronger bound is given on the minimum degree ensuring this. Before stating a generalization of these theorems we introduce some notation.
For a graph , fixed graphs and and an integer let be a -partite subgraph of with the maximum possible number of copies of and let be the family of subgraphs of that are -free and have the maximum possible number of copies of . Let denote the number of copies of in . Call a graph edge critical if there is an edge whose removal reduces the chromatic number of .
In [3] the following theorem is proved, generalizing the results in [8] and [7]. Throughout the paper we denote by the minimum degree in the graph .
Theorem 1.1** ([3]).**
Let be a graph with . Then there are positive constants and such that if is a graph on vertices with then for every
If is edge critical then 2. 2.
Otherwise,
In the present short paper we prove two theorems for -free subgraphs assuming is edge critical. The first is for subgraphs maximizing the number of copies of and the second for subgraphs maximizing the number of blow-ups of . We also establish a proposition concerning graphs that are not edge critical.
Theorem 1.2**.**
For every two integers and every edge critical graph such that there exist constants and such that the following holds. Let be a graph on vertices with , then for every the graph is -colorable.
For integers and let denote the -blow-up of , that is, the graph obtained by replacing each vertex of by an independent set of size and each edge by a complete bipartite graph between the corresponding independent sets.
Theorem 1.3**.**
For integers and and every edge critical such that there exist constants and such that the following holds. Let be a graph on vertices with , then every is -colorable.
Finally, for graphs which are not edge critical we prove the following.
Proposition 1.4**.**
For every integers and and graph such that there exists and such that the following holds. Let be a graph on vertices with and assume that or , then every can be made colorable by deleting edges.
Theorems 1.2 and 1.3 cannot be directly generalized to graphs that are not edge critical as we can add to any -partite graph an edge without creating a copy of such . On the other hand, we believe that the error term in Proposition 1.4 can be improved to for some .
The rest of this short paper is organized as follows. In Section 2 we state several known results and prove some helpful lemmas. Section 3 contains the proof of of Theorem 1.2. Theorem 1.3 is proved in Section 4 and the proof of Proposition 1.4 appears in Section 5. The final Section 6 contains some concluding remarks and open problems.
2 Preliminary results
We start by stating several results about -free graphs with high degrees and by deducing a corollary. Some of the theorems stated are simplified versions of the original results.
The first result about -free graphs is by Andrásfai, Erdős and Sós.
Theorem 2.1** ([6]).**
Let be a graph on vertice. If is -free and then .
A generalization of Theorem 2.1 proved in [13] is the following.
Theorem 2.2** ([13]).**
Let be a fixed edge critical graph which is not and assume . If is a graph on vertices which is -free and contains a copy of then .
This implies that if is large enough, , and if is -free for some edge critical graph with then it must also be -free. Together with Theorem 2.1 we get the following corollary:
Corollary 2.3**.**
Let be a fixed edge critical graph such that . Let be a graph on vertices which is -free and satisfies , then .
We next state the graph removal lemma as it appears in [9] (see also [2], [18] and [16]) and prove a simple lemma using it. Throughout the paper we denote by the number of vertices in the graph .
Theorem 2.4** (The graph removal lemma).**
For any graph H with vertices and any , there exists a such that any graph on vertices which contains at most copies of can be made -free by removing at most edges.
Throughout the paper, for fixed graphs and and an integer we denote by the maximum possible number of copies of in an -free graph on vertices.
Lemma 2.5**.**
Let be a fixed graph such that and let be an -free graph on vertices, where . Then can be made -free by deleting edges.
Proof.
Note, first, that the number of copies of in is . Indeed, in [4] it is shown that if a graph is a subgraph of a blow-up of a graph then .
Since , is contained in a blow-up of and hence . By the graph removal lemma can be made -free by removing edges, as needed. ∎
We next prove two additional more technical lemmas.
Lemma 2.6**.**
Let be a graph on vertices satisfying for some fixed , and let and be integers.
** 2. 2.
Let then
where .
Proof.
To prove part 1 note that as the number of copies of in is at least
[TABLE]
Randomly partitioning the graph into sets yields a graph in which the expected number of copies of is a least:
[TABLE]
Thus should have at least as many copies. This proves (1).
Similarly to prove part 2 observe that the number of copies of in is at least
[TABLE]
Randomly partitioning into parts gives a graph in which the expected number of copies of is at least
[TABLE]
and thus must have at least as many copies of . ∎
Lemma 2.7**.**
Let be a graph on vertices with for some fixed and let and be integers. For a set satisfying for some fixed let be the maximum number of copies of in a -partite subgraph of . Then there exist constants and such that for every
** 2. 2.
**
Proof.
Let . We first prove part 1. Fix a partition of into parts with copies of . By Lemma 2.6, part 1 the number of copies of is at least:
[TABLE]
Averaging we get that there is a vertex, say , so that the number of copies of it takes part in is at least
[TABLE]
Let be the above fixed partition of which has copies of and assume, without loss of generality, that . We add to and bound from below the number of copies of we add by doing this. Let and let be the number of neighbors has in which are not neighbors of . Note that and .
For each we estimate the number of copies of in which takes part that use vertices from that are not neighbors of . There are of those, and in the worst case each such vertex is connected to all of the sets for . Thus the number of copies of that takes part in and does not is at most
[TABLE]
And so by adding the number of copies of added is at least
[TABLE]
The proof of (2) is similar. By Lemma 2.6 part 2, in any partition of into parts in which the number of copies of is , this number is at least
[TABLE]
Let be such a partition. By averaging there must be a vertex, say , such that the number of copies of it takes part in is at least:
[TABLE]
Assume, without loss of generality, that , and let us add to . Let and let be the vertices in that are neighbors of and not of . Then the number of copies of in this partition that takes part in and does not is at most
[TABLE]
where the last inequality is true for some . Thus when adding to the number of copies of added is at least
[TABLE]
as needed. ∎
3 Maximizing the number of cliques
In the proof of Theorem 1.2 we use the following result from [4].
Proposition 3.1** ([4]).**
Let be a graph such that then
[TABLE]
Proof of Theorem 1.2.
Let and be as in the theorem and let . If , as is edge-critical, by Corollary 2.3 and we are done. Thus assume towards contradiction that .
As any partition of into parts is -free, by Lemma 2.6, part 1, the number of copies of in must be at least
[TABLE]
Consider the following iterative process of removing vertices from . Put and . Let be an arbitrarily chosen vertex of satisfying . Define and . For if the minimum degree in satisfies then stop the process, otherwise take a vertex of degree and define and .
We first show that this process must stop after at most steps. To see this note that the number of copies of removed with each deleted vertex is exactly the number of copies of in its neighborhood. By Proposition 3.1 for any -chromatic graph , . As is -free, the neighborhood of any vertex should be -free, where is such that .
Thus at step (starting to count from ), at most copies of have been removed.
As the following equality holds
[TABLE]
one can choose so that . Thus the number of copies of removed at step is no more than:
[TABLE]
Together with the fact that
[TABLE]
we conclude that the number of copies of removed during the first steps is at most
[TABLE]
The graph has at most copies of , and hence the total number of copies of in is at most
[TABLE]
But if is small enough this contradics (1). Thus the process must stop after steps.
As and is edge critical, Corollary 2.3 implies that . Define .
The partite subgraph of with the maximum possible number of copies of has at least as many copies of as . By Lemma 2.7, part 1 we can now add the vertices removed during the steps of the process starting from until , keeping the resulting subgraph -partite, where with each such vertex we add at least copies of . Assuming that is small enough to ensure, say, it follows that in each such step the number of added copies of exceeds the number of copies removed in the corresponding removal step.
When all the vertices are back we obtain a partite subgraph of containing more copies of than . This subgraph is -free, contradicting the maximality of . Thus the inequality must hold and the desired result follows. ∎
4 Maximizing the number of blow-ups of cliques
To prove Theorem 1.3 we first need a good estimate on for satisfying .
Proposition 4.1**.**
For integers and and any fixed graph such that ,
[TABLE]
Proof.
To show that it is enough to take the -sided Turán graph (i.e. the -partite graph with sides of nearly equal size). As it is -free and has copies of .
As for the upper bound, in [4] it is shown that the graph which is free and has the maximum possible number of copies of is a complete multipartite graph. It is not difficult to see that the Turán graph maximizes the number of copies of among these. Thus
[TABLE]
Let be as in the proposition, and let be an -free graph on vertices with the maximum number of copies of . By Lemma 2.5 can be made -free by deleting edges, and with them at most copies of . Let be the graph obtained by removing those edges.
As is -free we get
[TABLE]
and so as needed. ∎
The idea of the proof of Theorem 1.3 is similar to the one of Theorem 1.2 but some of the estimates are more involved.
Proof of Theorem 1.3.
Let be a graph with and let . If then by Corollary 2.3, , as is edge critical and we are done.
Assume towards contradiction that . Consider the following iterative process, similar to the one in the proof of Theorem 1.2. Put and . At step if satisfies then stop the process, otherwise take of degree and define and . We show that the process must stop after at most steps.
To bound the number of removed at each step we take care of two cases. If in a copy of there is a vertex in the same color class of that is a neighbor of in , call this copy dense. If all of the vertices in the color class of are non-neighbors of it in call the copy sparse.
First we estimate the number of dense copies. Let be the graph obtained by taking and adding to it a vertex that is connected to all of the other vertices. The number of dense copies of containing is at most .
As is edge critical there is a vertex such that , let . By a result in [4] if is contained in a blow-up of then . As the neighborhood of must be -free and as is contained in a blow-up of , it follows that . Thus the number of dense copies of in containing is .
As for the sparse copies, let be the number of sparse copies of in containing . Let and , and let for such that . Using Proposition 4.1 we obtain the following bound on the number of sparse copies of containing
[TABLE]
To bound this quantity consider the following function . Note that is a polynomial in , for and . Furthermore
[TABLE]
Thus for , , and is positive in . It follows that between [math] and obtains its global maximum at the single values for which , and it is increasing in .
In our case and as it follows that . We conclude that . Plugging this value it follows that
[TABLE]
Next we bound . As the following holds:
[TABLE]
For an appropriate , indeed such a exists as and for .
Therefore, the number of copies of (both dense and sparse) removed at step is at most
[TABLE]
If the process continued for steps, as the total number of copies of removed is at most
[TABLE]
By proposition 4.1 in the graph the number of copies of is at most
[TABLE]
Thus the number of copies of in is at most
[TABLE]
in contradiction to the maximality of . And so the process must stop after steps.
Assume that we have stopped at step and let . By Corollary 2.3 must be -partite, thus the -partite subgraph of with the maximum possible number of copies of has at least as many copies of as .
By Lemma 2.7, part 2, we can return the vertices removed in the process in a reverse order (starting from until ) keeping the graph -partite and adding with each vertex at least copies of . Assuming that is small enough to ensure, say, it follows that with each vertex we add more copies of than were removed at the corresponding step.
Thus when all vertices are returned we obtain an -partite graph with more copies of than . As an -partite graph is -free this contradicts the definition of . Thus it must be that and is -partite. ∎
5 Forbidding graphs that are not edge critical
The proofs of Theorems 1.2 and 1.3 actually give a stronger result than stated, as follows
Lemma 5.1**.**
Let and be integers, let be a graph on vertices such that , where is sufficiently small, and let . Assume that or . Then for every -free subgraph of on the same set of vertices, say , at least one of the following holds:
* for some .* 2. 2.
* can be made -chromatic by deleting edges.*
Proof.
If then by Theorem 2.1 is -chromatic and hence case (2) holds and we are done. If we consider a similar process to the one in the proofs of Theorem 1.2 and 1.3. For step of the process define , for steps let be a vertex of minimum degree in and define . The process stops when either or when for small enough.
The calculations in the proofs of Theorems 1.2 and 1.3 (see equation (2) and (4)) yield that when removing a vertex of degree less than the number of copies of removed with it is at most
[TABLE]
Assume that the process stops at step . If then by Theorem 2.1 the graph is -chromatic. In these steps vertices were deleted and with them no more than edges, thus case (2) holds.
If for some define the graph as follows. If take to be a -chromatic subgraph of on the vertices of with the maximum possible number of copies of . It must be that for an appropriate which tends to [math] as tends to [math]. If take , by Theorem 2.1 this graph is -chromatic.
As is -chromatic in both cases we can apply Lemma 2.7 to it and add back the vertices removed in the process, starting from to , while keeping the graph -chromatic. We get that the number of copies of added with each vertex is at least
[TABLE]
Let be the graph obtained after adding back all the vertices.
Assume that is small enough to ensure that for some . Let , and note that . Thus the difference in the number of copies of in and is at least
[TABLE]
where and are chosen so that the last equality holds.
As is a -free subgraph of , and thus and case (1) holds, as needed.
∎
The proof of Proposition 1.4 is now a simple corollary of the last lemma.
Proof of Proposition 1.4..
Let be a graph on vertices with and let . By Lemma 2.6 . By Lemma 2.5 there is a graph which is -free and , and thus
[TABLE]
Let . To apply Lemma 5.1 we show that
[TABLE]
By theorems 1.2 and 1.3 is -chromatic, and so it is -free. Together with the fact that is -free, we get
[TABLE]
implying (4).
Thus case (1) in Lemma 5.1 does not hold for , and so case (2) must hold, i.e. can be made -chromatic by deleting edges. As we got from by deleting edges we get the required result. ∎
6 Concluding remarks and open problems
- •
Corollary 2.3 and Theorems 1.2 and 1.3 cannot be directly generalized for graphs which are not edge critical. In [5] the following is shown (a weaker version of this statement is proved in [3])
Theorem 6.1** ([5]).**
Let be a fixed graph on vertices such that and let be an -free graph on vertices with , where is large enough. Then one can delete at most edges from and make it -colorable.
This suggests that a stronger version of Proposition 1.4, stating that any extremal graph as in the proposition can be made -chromatic by deleting edges for some , is likely to be true.
- •
Theorem 1.3 is limited to the case where , and from this we also get the condition in Proposition 1.4. One of the problems in extending it to graphs with higher chromatic number is that of finding an explicit tight bound on for . In [4] it is shown that the extremal graph is -partite. However, it is not difficult to check that for such that and large values of , the parts are not of equal sizes.
- •
Theorems in the same spirit as those proven here may hold for other pairs of graphs and . In [4] it is observed that if is not contained in any blow-up of then . This of course does not mean that the extremal graph is a blow-up of , but in cases it is a similar behavior to that in the results proven here might be expected.
A notable example is the case and . In [15] and independently [17] it is shown that when the extremal graph is the equal sided blow-up of . It might be true that this behavior holds for subgraphs of graphs of high minimum degree and not only for subgraphs of , that is, the extremal subgraphs in this case may be subgraphs of the equal sided blow-up of .
- •
The problem of obtaining the best possible bounds for the minimum degree ensuring that the results stated in Theorems 1.2, 1.3 and Propositon 1.4 hold is also interesting, but appears to be difficult. Even the very special case of Theorem 1.2 with and , conjectured in [7] to be , is open.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] N. Alon, Problems and results in extremal combinatorics, II, Discrete Math. 308, 4460-4472, (2008).
- 2[2] N. Alon, E. Fischer, M. Krivelevich and M. Szegedy, Efficient testing of large graphs. Combinatorica 20(4) , 451-476, (2000).
- 3[3] N. Alon, A. Shapira and B. Sudakov, Additive approximation for edge-deletion problems, Annals of mathematics , 371-411, (2009).
- 4[4] N. Alon and C. Shikhelman, Many T 𝑇 T copies in H 𝐻 H -free graph, Journal of Combinatorial Theory, Series B, 121, 146-172, (2016).
- 5[5] N. Alon, and B. Sudakov, H 𝐻 H -free graphs of large minimum degree, Electronic Journal of Combinatorics, (2006).
- 6[6] B. Andrásfai, P. Erdős and V. T. Sós, On the connection between chromatic number, maximal clique and minimal degree of a graph, Discrete Mathematics 8.3, 205-218, (1974).
- 7[7] J. Balogh, P. Keevash and B. Sudakov, On the minimal degree implying equality of the largest triangle-free and bipartite subgraphs, Journal of Combinatorial Theory, Series B 96, no. 6, 919-932, (2006).
- 8[8] A. Bondy, J. Shen, S. Thomassé and C. Thomassen, Density conditions for triangles in multipartite graphs, Combinatorica 26, no. 2, 121-131, (2006).
