Abstract
A Steiner Triple System () of order v is a hypergraph that is uniform of rank 3 and has order v, such that every 2-subset of X has degree 1. In this paper, we present some constructions of of type and , examining all possible cases.
MSC:
05C51; 05C65
1. Introduction
A Steiner system is a pair , where X is a v-set and is a family of k-subsets of X such that every h-subset of X is contained in exactly members of [1,2]. In hypergraph theory, a Steiner system is a hypergraph of order v, uniform of rank k, such that every h-subset Y of X has degree [3]. An is a Steiner system with . A Steiner Triple System (STS) is a system .
For fixed , the main problem in Steiner systems theory is to determine the spectrum, which is the set of all the possible values v for which an exists. This problem, in general, has been open since 1844, when Woolhouse formulated it [4]. But, until today, very few cases have been solved. In 1847, T. Kirkman [5] and J. Steiner [6] independently showed that an STS(v) exists if and only if or .
In the literature, there are many constructions to obtain an , starting from a given and indicated by , but not covering all the spectrum. Among them, and are well known. Also, the constructions of Bose and of Skolem are important, because they allow the entire spectrum of STS(v) to be determined.
In this paper, we give constructions of type , in the cases and , examining all the possible cases. For some recent further problems, see [7].
In what follows, we use the terminology and concepts of graph theory and hypergraphs, for which we refer to [3,8,9,10,11] and also [1,2]. For further definitions, properties and results of factorizations, see [12,13], and of cycle decompositions, see [14,15,16,17,18,19].
2. Construction
It is well known that it is always possible to construct an starting from an . The proof can be found in [3]. Here, we briefly outline the technique, which will be useful later, without going into detail.
Theorem 1.
If is an , then there exists an embedding Σ.
Proof.
Let be an STS(v) defined on .
Further, let be a set of cardinality (even number, because or ) such that and is a factorization of the complete graph defined in Y, is any bijection from X into .
Define the hypergraph , where and , with
It follows that is an . Indeed, is a hypergraph of order , uniform of rank 3.
We see that for every , there exists exactly one triple of containing .
At first, note that and , from which the blocks of are either all contained in X or have one vertex in X and two vertices in Y. At this point, consider that
- (1)
- If , there exists, in , exactly one block containing them;
- (2)
- If , there exists in exactly one factor containing them, and therefore there exists in the only block containing them;
- (3)
- If , , then there exists in exactly one pair containing y, which we define as , and therefore there exists in exactly the only block containing (see [3]).
□
3. Factorization of , by Differences
In this section, we want to show how to create a factorization of a set with an even number of elements focusing on the differences between them.
Let , where u is an even number; therefore, the set of differences is .
Theorem 2.
If is coprime to u, then there exist two disjoint factors of , , containing all the pairs having difference a.
Proof.
Consider the complete graph defined in . Since a is coprime to u and if and only if , then there exists in a unique hamiltonian cycle of length u, which can be described as follows:
and it is immediately seen that it can be decomposable in the following two factors:
;
. □
Theorem 3.
Let and (a is not coprime to u): if is even, then there exist two disjoint factors of , , containing all the pairs having difference a.
Proof.
In this case, since if and only if , then in the complete graph defined in , there exist h disjoint cycles of length , described as follows:
;
;
.
which is decomposable in the following factors:
The two factors are built using, for each cycle, the following technique:
- -
- First pair in ;
- -
- Second pair in ;
- -
- Third pair in ;
- -
- And so on.
We need to be even in order to have an even number of pairs for each hamiltonian cycle and to perfectly split those in two subsets. □
- Caseodd
If is odd, each hamiltonian cycle would consist of an odd number of pairs having a difference equal to a.
The last one, , would belong to , making the element i appear twice in the same factor.
Therefore, this method does not always guarantee complete factorization, at least without using some workaround that we will see during the construction.
Theorem 4.
If , then there exists a unique factor H containing all the pairs having difference a.
Proof.
If , we can immediately verify that in the complete graph , defined in , there exists the following unique factor:
. □
Note that this is the technique that we use in the next section to construct factors.
4. Main Results
Let be an . We want to give a construction of an embedding .
Let Y be a set such that and : the cardinality of Y is even, so we can consider its factorization, having factors. Similarly to the previous constructions, v of these factors can be connected to the elements of X through a bijection ; this bijection generates a set of blocks, which we call , as already seen.
The remaining factors detect pairs that can be arranged to form blocks of the new system, which can be generated as translates of k base blocks. Let be the set containing all of them.
We will see that the hypergraph defined as , , is an .
Note that, for , we have exactly the construction . Further, for , we have a construction about which some cases have been studied in [20].
In this paper, we want to look at the cases and , for each possible value of v and k. For the existence of cycle decompositions of the complete graph , see [21,22].
4.1. Construction ,
Remember that : we have some restrictions on v and k.
Hence, and .
Effectively, this is a construction, as .
Now, we deal with a factorization of , based on the differences.
In relation to the general construction described above, note that in what follows, it will be . Further, it will be , which is the set of all the possible differences between two distinct elements of .
Consider the following factors defined in Y:
We can see that every pair of factors , for , contains all and only the pairs of Y having difference i. Further, contains all and only the pairs having difference 10. This occurs because and
- (1)
- For i being odd, , there is in a cycle of length 20, in the order, decomposable in the two factors ;
- (2)
- For , there is in ten cycles of length 4, decomposable into the two factors ;
- (3)
- For , there are in two cycles of length 10, decomposable into the two factors and , respectively;
- (4)
- For , all pairs with this difference give rise to the only factor .
Now, let , where
- -
- , ;
- -
- , ;
- -
- ;
- -
- are defined as follows:
Theorem 5.
is a factorization of
Proof.
It is sufficient to carry out a direct verification that the factors of are all disjoint and that every pair is contained in exactly one factor of . In particular, it can be observed that
- (1)
- is a set of factors of , ;
- (2)
- For , contains all and only the pairs having difference i;
- (3)
- contains all and only the pairs having difference 10;
- (4)
- All the pairs having difference 1 belong to ;
- (5)
- All the pairs having difference 2 belong to ;
- (6)
- All the pairs having difference 4 belong to ;
- (7)
- All the pairs having difference 8 belong to .
□
Constructions
Remember that : hence, we have four different cases to analyze.
For each of those, consider the following:
- -
- , an , where ;
- -
- with the factorization defined as in the previous paragraph.
Therefore, is an , where
- -
- ;
- -
- .
- (1)
- Case , .
As , we get the construction for an .
- (2)
- Case , .
If is a bijection, we can define an with
- ;
- .
Here, the difference triple is .
As long as the differences 1 and 4 are taken together, translates from the base block collect all of the pairs having those two differences, divided among the factors , , and .
The same tecnique applies to differences 2 and 8.
- (3)
- Case , .
If is a bijection, we can define an with
- ;
- .
Here, the two difference triples are and .
4.2. Construction ,
Remember that : we have some restrictions on v and k.
Hence, and .
Effectively, this is a construction, as .
Now, we deal with a factorization of , based on the differences.
Note that, similarly to the case , in relation to the general construction , in what follows, it will be and , which is the set of all the possible differences between two distinct elements of .
Consider the following factors defined in Y:
We can see that every pair of factors , for every , contains all and only the pairs of Y having difference i. Further, contains all and only the pairs having difference 20. This occurs because and
- (1)
- For i being odd, , there is in a unique cycle of length 40, in the order, decomposable in the two factors ;
- (2)
- For , there are in five cycles of length 8 decomposable in the two factors and , respectively;
- (3)
- For , there are in two cycles of length 20 decomposable in the pairs of factors , respectively;
- (4)
- For , there are in four cycles of length 10 decomposable in the two factors and , respectively;
- (5)
- For , there are in ten cycles of length 4 decomposable in the two factors ;
- (6)
- For , all pairs with this difference give rise to the only factor .
Now, let , where
- -
- , ;
- -
- , ;
- -
- ;
- -
- are defined as follows
It follows that
Theorem 6.
is a factorization of
Proof.
A direct verification is sufficient. In particular, observe that
- (1)
- is a set of factors of , ;
- (2)
- For , contains all and only the pairs having a difference equal to i;
- (3)
- contains all and only the pairs having a difference equal to 20;
- (4)
- All and only the pairs having difference 1 belong to ;
- (5)
- All and only the pairs having difference 2 belong to ;
- (6)
- All and only the pairs having difference 8 belong to ;
- (7)
- All and only the pairs having difference 16 belong to .
□
Constructions
Remember that : hence, we have seven different cases to analyze.
For each of those, consider
- -
- , an , where ;
- -
- with the factorization defined as in the previous section.
Therefore, is an , where
- -
- ;
- -
- .
- (1)
- Case , .
As , we get the construction for an .
- (2)
- Case , .
If is a bijection, we can define an with
- ;
Here, the difference triple is .
- (3)
- Case , .
If is a bijection, we can define an with
- ;
Here, the two difference triples are and .
- (4)
- Case , .
If
is a bijection, we can define an with
- ;
Here, the three difference triples are , and .
- (5)
- Case , .
If
is a bijection, we can define an with
- ;
Here, the four difference triples are , , and .
- (6)
- Case , .
If
is a bijection, we can define an with
- ;
Here, the five difference triples are , , , and .
As long as differences 1 and 8 are taken together, translates from the base block collect all of the pairs having those two differences, divided among the factors , , and .
The same logic applies to differences 2 and 16.
- (7)
- Case , .
If
is a bijection, we can define an with
- ,
Here, the six difference triples are , , , , and .
The three not-taken factors, the ones in , are , and .
Finally, we have the following final theorem:
Theorem 7.
Let be an and , or , with its factorization : assume that both k and v, in each possible case, satisfy the conditions on their values. Let Γ and Δ be defined as before, through and bijection .
Therefore, the hypergraph defined as
- -
- ;
- -
- ,
which is an .
Proof.
Indeed,
- (1)
- It is immediately seen that is a hypergraph of order , uniform of rank 3;
- (2)
- For every , there exists exactly one triple of containing .
To prove (2), consider the following three possible cases:
- (2.1)
- ;
- (2.2.1)
- , and the difference between is associated with a factor of ;
- (2.2.2)
- , and the difference between is associated with a factor of ;
- (2.3)
- .
- (2.1)
- Since is an and do not contain vertices of X, there exists exactly one block in containing .
- (2.2.1)
- The pair appears in exactly one factor among the ones in : each of them detects a difference triple and base block with its translates, containing all of the pairs having the difference we considered. Thus, there is a block in with the pair .
- (2.2.2)
- There exist such that . Let ; therefore, the block .
- (2.3)
- In this case, consider the factor . There exists exactly one pair of the factor F containing y, which we define as . The triple is the triple in question.
So, we have proved that, in general, for every pair , it is . To prove that it is exactly , it is sufficient to prove that , which is the exact number of blocks in any .
Since and are pairwise disjoint, it follows that
while
Hence, . □
5. Conclusions
In this paper, we have introduced some new constructions for . The motivation is of a general nature. Steiner Systems, denoted by , were introduced in 1844 by Woolhouse, and the main problem, from the beginning, was to find for which values of they exist (the spectrum). This immediately presented itself as a very difficult problem and, indeed, it is still generally unsolved. Just think that to date, the spectrum has only been found for and , i.e., for STS, h = 2 and k = 4, h = 2 and k = 5, h = 3 and k = 4. The difficulty lies in making constructions of such systems as a function of v. For this reason, the constructions are the central focus in this theory.
In the field of STS many constructions are known. Some of them use only factorizations; others use only the difference method. Here, we have introduced a construction which uses both: factorizations and the difference method. This is an original idea never used before. In this construction, the difficulty is that they need to be chosen: (1) v factors of factorizations of elements to associate with the v vertices of the given systems; (2) the remaining factors must give pairs which can together define the base block generating the systems.
Author Contributions
Writing—original draft preparation, L.F. and M.G.; writing—review and editing, L.F. and M.G. All authors have read and agreed to the published version of the manuscript.
Funding
This research received no external funding.
Data Availability Statement
No new data were created or analyzed in this study.
Conflicts of Interest
The authors declare no conflicts of interest.
References
- Colbourn, C.; Rosa, A. Triple Systems; Oxford University Press: New York, NY, USA, 1999. [Google Scholar]
- Lindner, C.C.; Rodger, C.A. Design Theory; CRC Press: Boca Raton, FL, USA, 2012. [Google Scholar]
- Gionfriddo, M.; Milazzo, L.; Voloshin, V. Hypergraphs and Designs; Nova Science Publisher: New York, NY, USA, 2015. [Google Scholar]
- Woolhouse, W.S.B. Prize Question 1733; Lady’s and Gentleman’s Diary: London, UK, 1844. [Google Scholar]
- Kirkman, T.P. On a Problem in Combinations. Camb. Dublin Math. J. 1847, 2, 191–204. [Google Scholar]
- Steiner, J. Combinatorische Aufgabe. J. Reine Angew. Math. 1853, 45, 435–438. [Google Scholar]
- Gionfriddo, M. Some Problems and Conjectures about Voloshin Triple Systems. Mathematics 2026, 14, 42. [Google Scholar] [CrossRef] [Scilit]
- Berge, C. Graphs; North Holland: New York, NY, USA, 1985. [Google Scholar]
- Berge, C. Hypergraphs; North Holland: New York, NY, USA, 1989. [Google Scholar]
- Voloshin, V. The Mixed Hypergraphs. Couputer Sci. J. Mold. 1993, 1, 45–52. [Google Scholar]
- Voloshin, V. Introduction to Graph and Hypergraph Theory; Nova Science Publisher: New York, NY, USA, 2009. [Google Scholar]
- Lo Faro, G.; Milici, S.; Tripodi, A. Uniform (Ck, Pk+1)-Factorizations of Kn − I, when k is even. Mathematics 2022, 10, 936. [Google Scholar] [CrossRef] [Scilit]
- Rodger, C.A. Maximal Sets of Factors. Le Mat. 2004, LIX, 325–339. [Google Scholar]
- Alspach, B.; Gavlas, H. Cycle decomposition of Kn and Kn I. J. Combin. Th. Ser. B 2001, 81, 77–99. [Google Scholar] [CrossRef] [Scilit]
- Billington, E. Multipartite graph decompositions: Cycles and close trails. Le Mat. 2001, LIX, 53–72. [Google Scholar]
- Buratti, M.; Del Fra, A. Cyclic hamiltonian cycle systems of the complete graph. Discrete Math. 2004, 279, 107–119. [Google Scholar] [CrossRef] [Scilit]
- Lo Faro, G.; Milici, S.; Tripodi, A. Uniformly Resolvable Decompositions of Kv − I into n-Cycles and n-Stars, for even n. Mathematics 2020, 8, 1755. [Google Scholar] [CrossRef] [Scilit]
- Lindner, C.C.; Rodger, C.A. Decomposition into Cycles: Cycle Systems; Contemporary Design Theory, Wiley Interscience Series; Wiley: New York, NY, USA, 1992; pp. 325–367. [Google Scholar]
- Rodger, C.A. Cycle Systems; Handbook of Combinatorial Designs; Colbourn, C., Dinitz, J., Eds.; CRC Press: Boca Raton, FL, USA, 1996; pp. 266–270. [Google Scholar]
- Bonacini, P.; Gionfriddo, M.; Marino, L. A Construction of Steiner Triple Systems of type v⟶2v + 7. arXiv 2025, arXiv:2511.04823. [Google Scholar]
- Fu, H.; Wu, S. Cyclically decomposing complete graphs into cycles. Discret. Math. 2004, 282, 267–273. [Google Scholar] [CrossRef] [Scilit]
- Sajna, M. Cycle decomposition: Complete graphs and fixed lenght cycles. J. Combin. Des. 2002, 10, 27–78. [Google Scholar] [CrossRef] [Scilit]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2026 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license.