Abstract
The problem of the existence of optimal binary cyclically permutable constant-weight (CPCW) codes has been completely solved for codeword weights . We consider the smallest open cases, namely and . We present such codes for small values of the code length v and derive necessary conditions for the existence of optimal CPCW codes. These necessary conditions can be used to construct such codes, as well as to show that optimal codes with some parameters do not exist. In particular, we use them to prove that an optimal CPCW code does not exist.
Keywords:
coding theory; binary cyclically permutable constant-weight code; optical orthogonal code; cyclic design MSC:
94B60; 94B65
1. Introduction
1.1. Definitions
Denote by the additive group of integers modulo v. Let be a k-element subset of . Then, is the multiset of differences of C, and denotes a t-translate of C, where for .
Definition 1.
A binary cyclically permutable constant-weight (CPCW) code of size s can be defined as a collection of k-subsets of (codewords) such that any two distinct translates of a codeword share at most one element, and any two translates of two distinct codewords also share at most one element:
Everywhere in the text that follows, the abbreviation CPCW means “binary cyclically permutable constant weight”. Originally, a CPCW code was defined as a collection of sequences, but, for some purposes, it is easier to present the codewords as subsets of , and that is why the above definition is widely used too. Condition (1) is called the auto-correlation property and (2) the cross-correlation property. We denote by s the size of , namely the number of its codewords.
Consider a codeword . The auto-correlation property means that all the differences of a codeword of a CPCW code are different, and the cross-correlation property means that for two codewords and . Since , the size of a CPCW code cannot exceed
CPCW codes that reach this bound are called optimal [1,2].
If the code size is exactly , the optimal CPCW code is perfect; that is, all nonzero differences are covered.
CPCW codes are related to many other combinatorial structures. Definition 1, for instance, holds for a optical orthogonal code (OOC) too [1]. In this paper, we will use the relation to partial designs.
Definition 2.
Let be a finite set of points and be a finite collection of k-element subsets of V, called blocks. is a design (partial design) with parameters 2-(v,k,1) if any 2-subset of V is contained in exactly (at most) one block of . A 2-(v,k,1) design is also called a Steiner system and denoted by .
Partial designs are also known as packings [3] or packing designs [4]. We call them partial designs following [5]. An automorphism of a 2- partial design D is a permutation of its points that maps each block of D to a block of D.
Definition 3.
A 2-(v,k,1) design (partial design) is cyclic if it has an automorphism α permuting its points in one cycle, and it is strictly cyclic if each block orbit under this automorphism is of length v (no short orbits).
A cyclic partial design with the maximum possible number of blocks is optimal.
Definition 4.
A circulant matrix (circulant) of order v is a (0,1) square matrix with v rows and columns such that , .
Perfect CPCW codes are equivalent to strictly cyclic 2- designs. An optimal CPCW code is equivalent to an optimal 2- strictly cyclic partial design. The incidence matrix of a 2- strictly cyclic partial design contains submatrices, which are circulant matrices of order v. Each such circulant matrix corresponds to the translates of one of the codewords of the corresponding optimal CPCW code. An example illustrating this equivalence is presented in Figure 1.
We are not going to define more related structures here. We shall only mention that optimal CPCW codes correspond to difference packings, and that perfect CPCW codes are equivalent to cyclic difference families. An interested reader can find more information, for instance, in [6,7].
1.2. Known Results and the Contribution of the Present Paper
Optimal CPCW codes are studied in [2,8,9,10]. There are also many results on CPCW codes that were obtained for related combinatorial objects, like optical orthogonal codes [11,12,13,14,15,16,17,18,19,20,21], difference families [6,22,23,24,25,26], and combinatorial designs [27,28,29,30].
It was proved in [12] that an optimal CPCW code exists for all v except for and or . Direct constructions with explicit codewords are presented in [21] to show the existence of an optimal (v,4,1) CPCW (OOC) code for any positive integer . The existence of an optimal CPCW code for is still an open problem.
In this paper, we consider and 6.
- (a)
- Optimal (43,5,1) CPCW code
| {0, 1, 3, 7, 19} | {1, 2, 3, 4, 6, 7, 12, 16, 18, 19, 24, 25, 27, 31, 36, 37, 39, 40, 41, 42} |
| {0, 5, 13, 22, 33} | {5, 8, 9, 10, 11, 13, 15, 17, 20, 21, 22, 23, 26, 28, 30, 32, 33, 34, 35, 38} |
Uncovered differences: 0, 14, 29
- (b)
- Related optimal partial cyclic 2-(43,5,1) designFigure 1. The corresponding partial design.
First, we will summarize the known results for optimal CPCW codes or related combinatorial objects for and 6.
- An optimal OOC with p a product of primes congruent to 1 modulo 4 and greater than 5 exists according to [18].
- It is shown in [14] that an optimal OOC exists for prime , and that there is an optimal OOC for and prime .
- A proof of the existence of a OOC if v is a product of primes congruent to 1 modulo 4 and s is a non-negative integer is given in [16,17].
- There are also several direct constructions for optimal OOCs where , , and v is a product of primes greater than 5.
- A class of optimal optical orthogonal codes with v a product of primes congruent to 7 modulo 12 and greater than 7 and is obtained in [19].
- Some combinatorial constructions for optimal OOCs using g-regular cyclic packings are derived in [4,20].
Some existence results about CPCW codes with and 6 follow from results about cyclic difference families and combinatorial designs.
- It is shown in [23] that a difference family exists for for prime p.
- According to [26], a difference family exists for except possibly for .
- The existence of a cyclic difference family whenever p and q are primes congruent to is proved in [6].
- Existence of a cyclic difference family for any prime power with the exception of is shown in [22].
- It was proved in [29] that a cyclic combinatorial design exists for all and .
Nonexistence results for optimal OOCs with and for an optimal OOC are presented in [7].
Classification results about CPCW codes for and 6 or about related combinatorial objects can be found in [25,27,28,30,31]. The results obtained in these works show that, when v grows, the number of nonequivalent codes becomes extremely large. This makes the classifications for the next larger lengths practically unusable.
In this work, we consider and 6 and the first values of v for which existence or nonexistence results are not known and try to find an example of an optimal CPCW code for these parameters. We conduct this for and and and . We cannot find an optimal code for 9 of the considered parameter sets. We notice that two of these gaps are for . This motivates us to consider the properties of such codes, and we derive necessary conditions for their existence and prove that an optimal CPCW code does not exist.
2. Methods
The explicit examples of new optimal and CPCW codes are obtained by the algorithm for classification of such codes, which we have used and described in [31]. It implies the construction of an array L with all possible codewords, namely with k-subsets of for which the auto-correlation property holds. We then construct codes by choosing their codewords from L in all possible ways. For that purpose, backtrack search on the elements of L is applied, and some partial solutions are filtered away if there exists an automorphism of that maps the partial solution to a partial solution that has already been constructed.
Our aim in the present work is not to obtain all the codes with these parameters but to find one such code. That is why we let the computation stop after having constructed one code, or after having worked for several days on a personal computer. The fact that we have not succeeded to construct a code with definite parameters does not mean that such a code does not exist.
The necessary conditions for the existence of optimal CPCW codes are derived using a method that is very popular in Design Theory, namely the counting of points and pairs of points in two different ways. The nonexistence of an optimal CPCW code is proved by using these necessary conditions and applying counting of the pairs of points in one more way.
3. Results
3.1. Examples of Optimal and CPCW Codes
We present in Table 1 and Table 2 all the previously known existence and nonexistence results regarding optimal and CPCW codes with and 151, respectively. Normal font is used for the v with theoretical results from [6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,26,29] and italics for the results from the computer classifications in [25,27,28,30,31]. A normal or an italic v is struck through if the corresponding result (theoretic or computer-aided) shows nonexistence of optimal or CPCW codes. A question mark means that nothing is known about the existence of optimal codes with this length.
Table 1.
Optimal CPCW codes before the present study.
Table 2.
Optimal CPCW codes before the present study.
Table 3 and Table 4 are organized in a similar way to Table 1 and Table 2, but in them the existence results from the present paper are implemented too, and bold font is used for them. The question marks mean that we could not find optimal codes for these values of v and the question of the existence of such codes remains open. In particular, we have not found examples of optimal CPCW codes for , and 167, and of optimal CPCW codes for , and 126. These are CPCW codes with relatively small c. The latter is also small for the optimal codes for which there are no theoretical nonexistence results, but they have been proved not to exist with the help of a computer [31], namely an optimal CPCW code and optimal CPCW codes for , and 92. This motivated our study of the properties of optimal CPCW codes.
Table 3.
New optimal CPCW codes.
Table 4.
New optimal CPCW codes.
Table A1 and Table A2 in Appendix A contain all the examples of new optimal and CPCW codes. Each codeword contains the element , and therefore this element is not presented.
3.2. Properties of Optimal CPCW Codes
The optimal CPCW codes of length correspond to 2- designs. They are perfect because their codewords cover all differences. Optimal CPCW codes of length , however, have remarkable combinatorial properties too because they do not cover only one difference. We derive here some necessary conditions for the existence of such codes.
Lemma 1.
The element cannot be covered by the differences of the codewords of a CPCW code of even length. For even v, the element can be obtained as the difference of different pairs .
Proof.
Let C be a codeword of a CPCW code of even length. Let . Then, and must be two different differences from . Suppose , and then , and appears twice in , but this is impossible for a CPCW code. Since , the element can be obtained as the difference of the different pairs , . □
Lemma 2.
The codewords of an optimal CPCW code leave uncovered only the difference .
Proof.
Since is even, by Lemma 1, the possible differences are . The size of such a code is
and each codeword covers differences. Then, all the codewords cover differences; that is, only one difference remains uncovered. □
An optimal CPCW code corresponds to an optimal partial cyclic 2- design. The translates of each codeword correspond to v blocks of this partial design and to a circulant submatrix of the incidence matrix of the partial design. We next consider some properties of this partial design.
Lemma 3.
The pairs of points are not contained in the blocks of an optimal partial cyclic 2- design corresponding to an optimal CPCW code. Each of the remaining pairs of points is in exactly one block of this partial design.
Proof.
The pairs of points that are not contained in the blocks of the design correspond to the different pairs from Lemma 1. □
Lemma 4.
Denote the rows/columns of a circulant M of an even order v by the numbers . Let φ be a permutation that rearranges the rows and the columns of M in such a way that they follow each other as . Then, the resulting matrix is composed of 4 circulants A, B, C, and D of order .
Denote by , , , and the number of ones in a row/column of the matrices A, B, C, and D, respectively. Then, and = .
Proof.
Cutting off the even rows and even columns of M leads to a matrix A of dimension . Let be the element in row i and column j of M and be the element in row i and column j of A. Then, .
We want to show that A is a circulant matrix. Addition of the indices of A is modulo , and of M modulo v. Since , the addition of the even indices of M can be considered modulo too. Then, . M is a circulant, and therefore . It follows that ; that is, A is a circulant matrix.
Similarly, the circulants B, C, and D are obtained by cutting off the even rows and odd columns, the odd rows and even columns, and the odd rows and odd columns, respectively.
The number of ones in each row and column of is the same, and therefore and = . □
Example 1.
We illustrate the action of φ on two arbitrary circulants of even orders 10 and 8.
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
Lemma 5.
Let M be the incidence matrix of an optimal partial cyclic 2- design D corresponding to an optimal CPCW code. If the permutation φ from Lemma 4 is applied to the rows of M and to the columns of each of its t circulant submatrices, a matrix with circulant submatrices of order is obtained. If , the columns of each of the two rows of submatrices cover all possible pairs of points, and the missing pairs of points in the partial design have one point from the first and one from the second row of submatrices. If , there are pairs of points that are not covered by the columns of each row of submatrices, and all possible pairs of points having one point from the first and one from the second row of submatrices are covered.
Proof.
By Lemma 4, the permutation transforms each of the t circulants of order v to 4 circulants of order . That is why the incidence matrix of the partial design has circulant submatrices of order . If , then is odd. That is why, if a is odd, is even and vice versa. Since the first row of submatrices of contains the even rows of M and the second the odd rows, the missing pairs of points (Lemma 3) have one point from the first row of submatrices and one from the second. If , then is even, a and are both even or both odd, and the missing pairs of points consist of two points from one and the same row of submatrices. □
With respect to the number of ones in the four submatrices, can transform a circulant of order v to types of matrices such that has two circulant submatrices with i ones in each row/column and two with ones in each row/column.
Example 2.
Types of matrices for and 6
For , the types of matrices are
| 0 5 | 1 4 | 2 3 |
| 5 0 | 4 1 | 3 2 |
For , the types of matrices are
| 0 6 | 1 5 | 2 4 | 3 3 |
| 6 0 | 5 1 | 4 2 | 3 3 |
Theorem 1.
For an optimal CPCW code with and , the following equations hold:
Proof.
We obtain the equations by counting in different ways the pairs of points in . Denote by the number of matrices of type that are contained in . Their number is , and this is Equation (3). Equation (4) counts the pairs of points from one and the same row of submatrices. By Lemma 5, their number is . On the other hand, each submatrix of a matrix of type covers pairs of points. Equation (5) counts the pairs in which the two points are from different rows of submatrices. By Lemma 5, all but the missing pairs are covered; that is, all such pairs are . On the other hand, each submatrix of type covers such pairs of points. □
Theorem 2.
For an optimal CPCW code with and , the following equations hold:
Proof.
The proof is similar to that of Theorem 1, but, by Lemma 5, the pairs of points from one and the same row of submatrices are . All the pairs in which the two points are from different rows of submatrices are covered. That is why their number is . □
Corollary 1.
An optimal CPCW code can only exist if t is even.
Proof.
In this case, , and Equation (5) gives
That is why should be divisible by 8. □
Corollary 1 is not a new result. It is part of the nonexistence results presented in [7], Theorem 4.6. The next example shows the application of Theorem 1 on an optimal CPCW code with even t.
Example 3.
For an optimal CPCW code by Theorem 1:
There exists a unique integer solution: , and the optimal CPCW codes comply with it.
3.3. Nonexistence of an Optimal CPCW Code
Multiple nonexistence results on optimal CPCW codes are obtained in [7]. They include the nonexistence of an optimal code. We do not know other theoretical nonexistence results for optimal CPCW codes. Our computer-aided classification in [31] shows that an optimal CPCW code does not exist. Using Theorem 2, we shall now prove this without a computer.
Proposition 1.
An optimal CPCW code does not exist.
Proof.
By Theorem 2:
There exists a unique integer solution:
It follows that the number of ones in is distributed as
We can see from (15) that the submatrices of dimension have , or 5 ones in each row/column. Since 46 is even and is odd, we can use the same ideology as in Lemma 5 and Theorem 1 on the variable-length code corresponding to the first 46 rows of ; that is, we apply to the first 46 rows of a permutation such that after its application these rows follow each other as , and we apply the same permutation to the columns of the submatrices of dimension 46.
A circulant with 1, 3, or 5 ones in each row/column can be transformed by to one, two, or three types of matrices, respectively. We denote them ,, , , , and . The matrix is of dimension and has four submatrices of dimension , two of them with j and two with ones in each row/column.
Denote by the number of matrices of type that are contained in the first 23 rows of . Since 23 is odd, we can apply Theorem 1 and write an analogue of Equation (5) for ; thus, we obtain
where . It follows that
The same result can be obtained by counting the pairs of points such that one is from the first 23 rows of and the other from the second 23 rows. The missing 23 pairs of points are among them, and therefore
4. Discussion
The presented examples of optimal and CPCW codes might be useful in relevant applications, as well as in future research on the topic. The results for small parameters show that nonexistence of some optimal CPCW codes with small c can be expected. The solutions of the equations from Theorems 1 and 2 might be useful in further theoretical or computer-aided research on optimal CPCW codes whose existence remains undecided.
Author Contributions
Conceptualization, S.T. and T.B.; methodology, S.T. and T.B.; software, S.T.; validation, S.T. and T.B.; investigation, S.T. and T.B.; data curation, T.B.; writing—original draft preparation, S.T. and T.B.; writing—review and editing, S.T. and T.B. All authors have read and agreed to the published version of the manuscript.
Funding
The research of both authors was partially supported by the Centre of Excellence in Informatics and ICT under Grant No. BG16RFPR002-1.014-0018-C01, financed by the Research, Innovation and Digitalization for Smart Transformation Programme 2021–2027 and co-financed by the European Union.
Data Availability Statement
All the new results are included in the paper.
Conflicts of Interest
The authors declare no conflicts of interest.
Abbreviations
The following abbreviations are used in this manuscript:
| CPCW Code | Binary Cyclically Permutable Constant-Weight Code |
| OOC | Optical Orthogonal Code |
| CDF | Cyclic Difference Family |
Appendix A
Table A1.
Optimal (v,5,1) CPCW codes.
Table A2.
Optimal (v,6,1) CPCW codes.
References
- Chung, F.R.K.; Salehi, J.A.; Wei, V.K. Optical orthogonal codes: Design, analysis and applications. IEEE Trans. Inf. Theory 1989, 35, 595–604. [Google Scholar] [CrossRef] [Scilit]
- Bitan, S.; Etzion, J. Constructions for optimal constant weight cyclically permutable codes and difference families. IEEE Trans. Inf. Theory 1992, 41, 77–87. [Google Scholar] [CrossRef] [Scilit]
- Stinson, D.R.; Wei, R.; Yin, J. Packings. In The CRC Handbook of Combinatorial Designs, 2nd ed.; Colbourn, C.J., Dinitz, J.H., Eds.; Chapman and Hall/CRC Press: Boca Raton, FL, USA, 2006; pp. 550–556. [Google Scholar]
- Fuji-Hara, R.; Miao, Y. Optical orthogonal codes: Their bounds and new optimal constructions. IEEE Trans. Inf. Theory 2000, 46, 2396–2406. [Google Scholar] [CrossRef] [Scilit]
- Chu, W.; Colbourn, C.J. Optimal (n, 4, 2)-OOC of small order. Discret. Math. 2004, 279, 163–172. [Google Scholar] [CrossRef] [Scilit]
- Buratti, M.; Pasotti, A. Further progress on difference families with block size 4 or 5. Des. Codes Cryptogr. 2010, 56, 1–20. [Google Scholar] [CrossRef] [Scilit]
- Buratti, M.; Stinson, D.R. New results on modular Golomb rulers, optical orthogonal codes and related structures. Ars Math. Contemp. 2020, 1, 1–27. [Google Scholar] [CrossRef] [Scilit]
- Baicheva, T.; Topalova, S. Classification of optimal (v,4,1) binary cyclically permutable constant weight codes and cyclic S(2,4,v) designs with v ≤ 76. Probl. Inf. Transm. 2011, 47, 224–231. [Google Scholar] [CrossRef] [Scilit]
- Moreno, O.; Zhan, Z.; Kumar, P.V.; Zinoviev, V.A. New constructions of optimal cyclically permutable constant weight codes. IEEE Trans. Inf. Theory 1995, 41, 448–455. [Google Scholar] [CrossRef]
- Nguyen, Q.A.; Gyöfri, L.; Massey, J.L. Constructions of binary constant weight cyclic codes and cyclically permutable codes. IEEE Trans. Inf. Theory 1992, 38, 940–949. [Google Scholar] [CrossRef] [Scilit]
- Baicheva, T.; Topalova, S. Classification of maximal optical orthogonal codes of weight 3 and small lengths. Serdica J. Comput. 2015, 9, 83–92. [Google Scholar] [CrossRef] [Scilit]
- Brickell, E.F.; Wei, V.K. Optical orthogonal codes and cyclic block designs. Congr. Numer. 1987, 58, 175–192. [Google Scholar]
- Buratti, M.; Momihara, K.; Pasotti, A. New results on optimal (v, 4, 2, 1) optical orthogonal codes. Des. Codes Cryptogr. 2011, 58, 89–109. [Google Scholar] [CrossRef] [Scilit]
- Chang, Y.; Ji, L. Optimal (4up, 5, 1) optical orthogonal codes. J. Comb. Des. 2004, 12, 346–361. [Google Scholar] [CrossRef] [Scilit]
- Chu, W.; Golomb, S.W. A new recursive construction for optical orthogonal codes. IEEE Trans. Inf. Theory 2003, 49, 3072–3076. [Google Scholar]
- Ma, S.; Chang, Y. A new class of optimal optical orthogonal codes with weight five. IEEE Trans. Inf. Theory 2004, 50, 1848–1850. [Google Scholar] [CrossRef]
- Ma, S.; Chang, Y. Constructions of optimal optical orthogonal codes with weight five. J. Comb. Des. 2005, 13, 54–69. [Google Scholar] [CrossRef] [Scilit]
- Tang, Y.; Yin, J. Combinatorial constructions for a class of optimal OOCs. Sci. China A 2002, em 45, 1268–1275. [Google Scholar]
- Wang, S.; Wang, L.; Wang, J. A new class of optimal optical orthogonal codes with weight six. In Proceedings of the Seventh International Workshop of Signal Design and its Applications in Communications, Bengaluru, India, 14–18 September 2015. [Google Scholar]
- Yin, J. Some combinatorial constructions for optical orthogonal codes. Discrere Math. 1998, 185, 201–219. [Google Scholar] [CrossRef] [Scilit]
- Zhao, C.; Chang, Y.; Feng, F. The existence of optimal (v, 4, 1) optical orthogonal codes achieving the Johnson bound. IEEE Trans. Inf. Theory 2024, 70, 8746–8757. [Google Scholar] [CrossRef] [Scilit]
- Chen, K.; Zhu, L. Existence of (q, 6, 1) difference families with q a prime power. Des. Codes Cryptogr. 1998, 15, 167–173. [Google Scholar] [CrossRef] [Scilit]
- Chen, K.; Zhu, L. Existence of (q, k, 1) difference families with q a prime power and k = 4,5. J. Comb. Des. 1999, 7, 21–30. [Google Scholar] [CrossRef]
- Chen, K.; Wei, R.; Zhu, L. Existence of (q, 7, 1) difference families with q a prime power. J. Comb. Des. 2002, 10, 126–138. [Google Scholar] [CrossRef] [Scilit]
- Hetman, I.; Baicheva, T.; Topalova, S. Construction of (v,k,1) cyclic difference families with small parameters. J. Algebra Comb. Discret. Struct. Appl. 2026, 13, 111–119. [Google Scholar]
- Julian, R.; Abel, R.; Costa, S.; Finizio, N. Directed-ordered whist tournaments and (v,5,1) difference families: Existence results and some new classes of Z-cyclic solutions. Discret. Appl. Math. 2004, 143, 43–53. [Google Scholar] [CrossRef] [Scilit]
- Colbourn, M.J.; Mathon, R.A. On cyclic Steiner 2-designs. Ann. Discret. Math. 1980, 7, 215–253. [Google Scholar]
- Colbourn, C.J.; Colbourn, M.J. On cyclic Steiner systems S(2,6,91). Abstr. Pap. Present. Am. Math. Soc. 1982, 5, 463. [Google Scholar]
- Hanani, H. Balanced incomplete block designs and related designs. Discret. Math. 1975, 11, 255–369. [Google Scholar] [CrossRef] [Scilit]
- Janko, Z.; Tonchev, V.D. Cyclic 2-(91, 6, 1) designs with multiplier automorphisms. Discret. Math. 1991, 97, 265–268. [Google Scholar] [CrossRef] [Scilit]
- Baicheva, T.; Topalova, S. Classification of optimal (v, k, 1) binary cyclically permutable constant weight codes with k = 5, 6 and 7 and small lengths. Des. Codes Cryptogr. 2019, 87, 365–374. [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.
