Approximate Steiner $(r-1,r,n)$-systems without $3$ blocks on $r+2$ points
Alexander Sidorenko

TL;DR
This paper investigates the maximum size of certain hypergraphs avoiding specific small configurations, providing asymptotic bounds for these extremal problems in combinatorics.
Contribution
It establishes asymptotic bounds for the maximum number of edges in hypergraphs avoiding particular small subgraphs, advancing understanding of approximate Steiner systems.
Findings
Asymptotic maximum edges are approximately (1/r) * binomial(n, r-1).
Provides bounds for hypergraphs avoiding small configurations.
Advances extremal combinatorics in hypergraph theory.
Abstract
For a family of -graphs, let denote the maximum number of edges in an -free -graph on vertices. Let denote the family of all -graphs with edges and at most vertices. We prove that .
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.
Approximate Steiner -systems
without blocks on points
Alexander Sidorenko
Abstract
For a family of -graphs, let denote the maximum number of edges in an -free -graph on vertices. Let denote the family of all -graphs with edges and at most vertices. We prove that .
Let be a family of -graphs, (that is, -uniform hypergraphs). An -graph is called -free if it does not have a subgraph isomorphic to a member of . Let denote the maximum number of edges in an -free -graph on vertices. Let denote the family of all -graphs with edges and at most vertices. In 1973, Brown, Erdős and Sós initiated the study of . The case had been studied before. In particular, Erdős [4] proved . Cases when the ratio is not integer were studied in [1, 6, 7, 11, 12, 13, 15]. In the case when is an integer, Brown, Erdős and Sós [3] were able to prove
[TABLE]
The case is equivalent to the packing problem and was asymptotically solved by Rödl [10] who proved
[TABLE]
Recently, the first case with , was solved by Glock [9]: . His result was generalized by Shangguan and Tamo [14] to
[TABLE]
Shangguan and Tamo proved the upper bound
[TABLE]
which in the case , becomes
[TABLE]
A lower bound, obtained from the probabilistic construction of Brown, Erdős, and Sós [3], was explicitly calculated in [14]:
[TABLE]
We will prove a much better lower bound which almost matches the upper one:
[TABLE]
Answering a question of Erdős [5], two sets of authors in [2] and [8] independently proved that for any ,
[TABLE]
It was suggested in [8, Conjecture 7.2] that a similar statement must hold for all , , namely
[TABLE]
We will prove this statement for by constructing -free (or, in terms of [8], -sparse) partial Steiner -systems with blocks (edges). The case is trivial, because any -free system is also -free. We will provide different constructions for (Theorem 1) and (Theorem 4).
We define the degree of a subset of vertices in an -graph as the number of edges that contain this subset.
Theorem 1**.**
For any even , there exists an -free -graph with vertices and edges.
Proof.
Consider a -graph whose vertices are elements of , and distinct vertices form an edge when . Then the degree of a triple is either (when , or [math]. For each , there are exactly pairs such that , , , and . Hence, the number of triples of degree [math] is , and consequently, the number of edges in is .
The only -graph in without triples of degree is formed by edges such that , , , where are disjoint pairs of vertices. If they were present in , we would get
[TABLE]
which is impossible (as is even). ∎
Proposition 2**.**
When and , there exists an -free -graph with vertices and edges.
Proof.
Consider an -graph whose vertices are elements of , where a set of elements is an edge if the sum of them is zero, and does not contain zero-sum subsets of sizes and . It is easy to see that the degree of any -subset of vertices is at most . An -subset has degree [math] only when it contains a zero-sum subset of size , or . For a fixed , the number of zero-sum subsets of size is , so the number of -subsets that contain a zero-sum subset of size is at most . Hence, the number of -subsets of degree [math] is , and the number of edges in is . Obviously, is -free. We need to show that is -free. Suppose, there exist an -subset and distinct edges . Since , we get , and similarly, . Set , , . Then , and , , are pairwise disjoint, because . Set . Then , , , , and we get
[TABLE]
so is a zero-sum subset of size in edge , a contradiction. ∎
Proposition 3**.**
When and , there exists an -free -graph with vertices and edges.
Proof.
The case is covered by Proposition 2. Assume and consider an -graph whose vertices are elements of , and is an edge in when in and is an edge in -graph from Proposition 2. (We say that is the shadow of in .) The number of edges in is times that of . It is easy to see that is -free. We need to check that it is also -free. Suppose, there exist three distinct edges such that . As is -free, at least two of these three edges must have the same shadow in . Let and . Since and , vectors and must differ in at least two places. Since , the same vectors must differ in at most two places. Therefore, they differ in exactly two places, and . Then must have the same shadow as and : . Similarly, must differ in exactly two places from and from . As , it must be the same places where and differ, but this is impossible. ∎
Theorem 4**.**
When , there exists an -free -graph with vertices and edges as .
Proof.
For a given , select and . Then . By Proposition 3, there exists an -free -graph with vertices and edges. Notice that and as . ∎
Any -free -graph has at most edges (because every -subset of vertices has degree at most ). Hence, Theorems 1 and 4 imply
Corollary 5**.**
When and ,
[TABLE]
Simplicity of our constructions of approximate Steiner -systems (comparing with Rödl’s probabilistic proof of 1) can be explained by the fact that their main parameters, and , differ just by .
Acknowledgments. The author would like to thank two anonymous referees for their valuable comments and suggestions.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] N. Alon and A. Shapira. On an extremal hypergraph problem of Brown, Erdős, and Sós. Combinatorica , 26(6):627–645, 2006. doi:10.1007/s 00493-006-0035-9 . · doi ↗
- 2[2] T. Bohman and L. Warnke. Large girth approximate Steiner triple systems. J. London Math. Soc. , 2019. doi:10.1112/jlms.12242 . · doi ↗
- 3[3] W. G. Brown, P. Erdős, and V. T. Sós, Some extremal problems on r 𝑟 r -graphs. In New directions in the theory of graphs (Proc. Third Ann Arbor Conf., Univ. Michigan, Ann Arbor, Mich, 1971) (ed. F. Harary), pages 53–63. Academic Press, Cambridge, MA, 1973. https://old.renyi.hu/~p_erdos/1973-25.pdf
- 4[4] P. Erdős. Extremal problems in graph theory. In Theory of graphs and its applications (ed. M. Fiedler), pages 29–36. Publising House of the Czechoslovak Academy of Sciences, Prague, 1964. https://old.renyi.hu/~p_erdos/1964-06.pdf
- 5[5] P. Erdős. Problems and results in combinatorial analysis. In Colloquio Internazionale sulle Teorie Combinatorie (Rome, 1973) , Tomo II, pages 3–17. Atti dei Convegni Lincei , No. 17, Accad. Naz. Lincei, Rome 1976. https://old.renyi.hu/~p_erdos/1976-35.pdf
- 6[6] P. Erdős, P. Frankl, and V. Rödl. The asymptotic number of graphs not containing a fixed subgraph and a problem for hypergraphs having no exponent. Graphs Combin. , 2(1):113–121, 1986. doi:10.1007/BF 01788085 . · doi ↗
- 7[7] G. Ge and C. Shangguan. Sparse hypergraphs: new bounds and constructions. ar Xiv:1706.03306 , 2017.
- 8[8] S. Glock, D. Kühn, A. Lo, and D. Osthus. On a conjecture of Erdős on locally sparse Steiner triple systems. ar Xiv:1802.04227 , 2018.
