CaROMaD, Volume 01 (2016) : 53 − 61 http://www.laromad.usthb.dz Cahiers de Recherche Opérationnelle et Mathématiques de la Décision Identifying codes of Cartesian product of two cliques Anissa Hissoum & Ahmed Semri LaROMad, Fac. Maths, USTHB, PO 32, 16111 Bab Ezzouar, Algeria [email protected], [email protected] Abstract: An identifying code in a graph is a subset of vertices with the property that the closed neighborhood of any vertex has a unique and nonempty intersection with the code. The vertices of a code are called codewords.The notion of identifying code was initially introduced by Karpovsky, Chakrabarty and Levitin in 1998 to model a fault-detection problem in multiprocessor systems.The problem to determine the smallest number of codewords in identifying codes, if existing, in a graph is N P -hard. In this paper, we study identifying code problem in the Cartesian product of two cliques with different sizes. Key Words: Graph theory, Domination, Identifying code, clique, Cartesian product of two graphes. MSC(2010): 05C85, 68R10 Résumé : Les codes identifiants ont été définis pour la première fois par KARPOVSKY et al. en 1998 pour modéliser un problème de détection et de localisation de processeurs défectueux dans des réseaux multiprocesseurs. Depuis, d’autres applications ont été développées comme dans les systèmes de localisation et de détection dans les environnements fermés munis de capteurs sans fil. Un code identifiant dans un graphe est un sous ensemble de sommets tel que deux sommets quelconques du graphe ont leurs ensembles de voisinage fermé différents et non vides dans le code. Trouver la cardinalité minimum d’un code identifiant dans un graphe, lorsqu’il existe, est un problème N P -difficile. Ce papier traite le problème des codes identifiants dans deux classes de graphes : l’hypercube et le produit cartésien de deux cliques de dimensions différentes. Mots clés : Graphes, Domination, Code identifiant, Graphe de Hamming, Hypercube, clique, Produit Cartésien des graphes. 54 1 Anissa Hissoum & Ahmed Semri Introduction We consider a connected graph G = (V, E) with a vertex set V and an edge set E. We refer to a distinguished subset of V as a code, and to its elements as codewords. An identifying code in G is a nonempty subset of the vertex set V , such that all the vertices of V have different nonempty neighborhoods in the code, ie a dominating and separating code at the same time. The initial application for identifying codes was to fault diagnosis in multiprocessor system in 1998 by Karpovsky, Chakrabarty and Levitin [?]; and has been since fundamentally connected to a wide range of applications, including location detection in hostile environments [?], to energy balancing of such systems [?] and environmental monitoring [?]. A necessary and sufficient conditions for a graph G = (V, E) admits an identifying code is that it does not contain twin vertices (two vertices sharing the same set of neighborhood). In this case, the set of vertices V is always an identifying code of G. The problem therefore is to find the minimum number of codewords required to identify each vertex of G. It has been shown in [?] that this problem was N P -hard. If C is an identifying code of minimum cardinality in a graph with n vertices, then |C| ≥ ⌈log2 (n + 1)⌉ [?, Th1] . This first lower bound comes from the fact that with the elements of C, it is possible to build 2|C| − 1 distinct and non-empty sets, therefore to identify the set of vertices of the graph, it is necessary that n ≤ 2|C| − 1. We provide upper bounds on the minimum cardinality of an identifying code in the Cartesian product of two cliques of different sizes. 2 Identifying Codes of Kn Km, m ≥ n ≥ 2 Given two graphs G and H, the Cartesian product of G and H denoted by GH, is the graph on vertex set V (G) × V (H); and any two vertices (x1 , y1 ) and (x2 , y2 ) are adjacent in GH if and only if either x1 x2 ∈ E(G) and y1 = y2 , or y1 y2 ∈ E(H) and x1 = x2 . The cartesian product of two cliques Kn Km , is a matrix with n ∗ m vertices such that each row Rx , x ∈ {1, ..., n} (resp column Cy , y ∈ {1, ..., m} is a clique Km (resp Kn ). (1,1) (1,m) Km (n,1) (m,n) Kn Figure 1: Cartesian product of two cliques Kn Km In [?, Theorem 1], it was shown that minimum cardinality of an identifying code of the Carte the sian product Kn Kn is equal to 3n 2 . The proof of this theorem is based on the determination 3n of an identifying code of cardinality 2 : D = {(x, x)|x = 1, ...n} et A = {(n − x + 1, x)|x = 1, ..., n−1 2 }, {(n − x + 1, x)|x = 1, ..., n2 }, si n est impair si n est pair 55 Identifying codes of Cartesian product of two cliques The set of vertices D ∪ A is an identifying code of Kn Kn . (1,1) A (1,1) A D D Figure 2: Le code D ∪ A suivant la parité de n We will establish upper bounds on the minimum cardinality of an identifying code in the Cartesian product of two cliques of different sizes Kn Km , m > n. Given two cliques Kn and Km , as m > n geq2 and let a and b, respectively, the quotient and the remainder of the Euclidean division of m by n. 2.1 Case when a ≥ 2 and 0 ≤ b ≤ n − 1 Proposition 1 Let C be a minimum identifying code of Kn Kan+b for all n, a ≥ 2 and 0 ≤ b ≤ n − 1, then : |C| ≤ n(2a − 1) + 2b Preuve. The code C = A ∪ B ∪ P is an identifying code of Kn Kan+b , n, a ≥ 2 and 0 ≤ b ≤ n − 1 with cardinality n(2a − 1) + 2b such that : ∅ if b = 0 {(i, i + an)/i ∈ {1, ..., b}} if b 6= 0 ∅ B = {(i, kn − i + 1)/i ∈ {1, ..., n}, k ∈ {1, ..., a − 1}}∪ {(n − k + 1, (a − 1)n + k)/k ∈ {1, ..., b}} n n n, 2 + jn /j ∈ {0, ..., a − 2} if 0 ≤ b ≤ 2 and n is odd if ⌈ n2 ⌉ ≤ b ≤ n − 1 and n is odd n, ⌈ n2 ⌉ + jn /j ∈ {0, ..., a − 1} P = ∅ if n is even A = {(i, i + kn)/i ∈ {1, ..., n}, k ∈ {0, ..., a − 1}} ∪ if b = 0 if b = 6 0 C is dominating set since it contains an element of each row of Kn Kan+b , b 6= 0, a ≥ 2. Now we check that each pair of vertices (x, y), (x′ , y ′ ) ∈ Kn Kan+b is separated by C. Without loss of generality, we suppose y ≤ y ′ . 56 Anissa Hissoum & Ahmed Semri 1. If (x, y) and (x′ , y ′ ) are not in the same column or in the same row : (a) If y ≤ n : i. if y ′ ≤ n then (x, (a − 1)n + x) ∈ N [(x, y)]\N [(x′ , y ′ )] ii. if y ′ ≥ n + 1 then (x, x) ∈ N [(x, y)]\N [(x′ , y ′ )] (b) If y ≥ n + 1 then (x, x) ∈ N [(x, y)]\N [(x′ , y ′ )] 2. If (x, y) and (x′ , y ′ ) are in the same column : (a) if y ≤ (a − 1)n then (x, (a − 1)n + x) ∈ N [(x, y)]\N (x′ , y) (b) if y ≥ (a − 1)n + 1 then (x, x) ∈ N [(x, y)]\N [(x′ , y)] 3. If (x, y) and (x′ , y ′ ) are in the same row. Given k, k′ ∈ {0, ..., a} and j, j ′ ∈ {0, ..., n − 1} such that y = kn + j and y ′ = k′ n + j ′ (a) if y ≤ (a − 1)n i. if x ≤ n2 A. if j = 0 then (n, y) ∈ N [(x, y)]\N [(x, y ′ )] B. if n is odd and j = n2 then (n, y) ∈ N [(x, y)]\N [(x, y ′ )] C. if j ≤ n2 then (n − j + 1, y) ∈ N [(x, y)]\N [(x, y ′ )] D. if j ≥ n2 + 1 then (j, y) ∈ N [(x, y)]\N [(x, y ′ )] ii. if x ≥ n2 + 1 A. if j = 0 then (1, y) ∈ N [(x, y)]\N [(x, y ′ )] B. if n is odd and j = n2 then ( n2 , y) ∈ N [(x, y)]\N [(x, y ′ )] C. if j ≤ n2 then (j, y) ∈ N [(x, y)]\N [(x, y ′ )] D. if j ≥ n2 + 1 then (n − j + 1, y) ∈ N [(x, y)]\N [(x, y ′ )] (b) if (a − 1)n + 1 ≤ y < y ′ ≤ an. Given j and j ′ such that y = (a − 1)n + j et y ′ = (a − 1)n + j ′ i. if j ′ ≤ x then (j, (a − 1)n + j) ∈ N [(x, y)]\N [(x, y ′ )] ii. if j ′ > x then (j ′ , (a − 1)n + j ′ ) ∈ N [(x, y ′ )]\N [(x, y)] Identifying codes of Cartesian product of two cliques 57 (c) if an + 1 ≤ y < y ′ ≤ an + b. Given j and j ′ such that y = an + j and y ′ = an + j ′ i. if j ′ ≤ x then (j, an + j) ∈ N [(x, y)]\N [(x, y ′ )] ii. if j ′ > x then (j ′ , an + j ′ ) ∈ N [(x, y ′ )]\N [(x, y)] (d) if 1 ≤ y ≤ an and an + 1 ≤ y ′ ≤ an + b. Given k ∈ {0, .., a}, j ∈ {0, .., n − 1} and j ′ ∈ {1, .., b} such that y = kn + j et y’=an+j’ i. if x ≥ b + 1 then (j ′ , y ′ ) ∈ N [(x, y ′ )]\N [(x, y)] ii. if x ≤ b then A. if x ≤ n2 • if j = 0 then (n, y) ∈ N [(x, y)]\N [(x, y ′ )] • if n is odd and j = n2 then (n, y) ∈ N [(x, y)]\N [(x, y ′ )] • if j ≤ n2 then (n − j + 1, y) ∈ N [(x, y)]\N [(x, y ′ )] • if j ≥ n2 + 1 then (j, y) ∈ N [(x, y)]\N [(x, y ′ )] B. if x ≥ n2 + 1 • if j = 0 then (n, y) ∈ N [(x, y)]\N [(x, y ′ )] • if n is odd and j = n2 then ( n2 , y) ∈ N [(x, y)]\N [(x, y ′ )] • if j ≤ n2 then (j, y ′ ) ∈ N [(x, y ′ )]\N [(x, y)] • if j ≥ n2 + 1 then (n − j + 1, y) ∈ N [(x, y)]\N [(x, y ′ )] Figure 3: Identifying code of K5 K25 2.2 Case when a = 1 and b 6= 0 In this case, we propose two identifying codes according to the values of b 58 Anissa Hissoum & Ahmed Semri Figure 4: Identifying code of K6 K24 K4 K9 K4 K10 K4 K11 Figure 5: Identifying code of Kn Kan+b , n = 4, a = 2 and b ∈ {1, 2, 3} 2.2.1 If 1 ≤ b ≤ n 2 Proposition 2 If C is a minimum identifying code of Kn Kn+b , 1 ≤ b ≤ n 2 , then : 3n |C| ≤ +b 2 Preuve. If 1 ≤ b ≤ n2 , then the code C = A ∪ B is an identifying code of Kn Kn+b , n ≥ 2 with cardinality 3n 2 + b such that : A = {(i, i)/i ∈ {1, ..., n}} ∪ {(i, n + i)/i ∈ {1, ..., b}} B = {(n − i + 1, i)/i ∈ {1, ..., n 2 } 59 Identifying codes of Cartesian product of two cliques K5 K11 K5 K12 K5 K13 K5 K14 Figure 6: Identifying code of Kn Kan+b , n = 5, a = 2 and b ∈ {1, 2, 3, 4} K7 K8 K7 K9 K7 K10 Figure 7: Identifying codes of Kn Kn+b , n = 7 and b ∈ {1, 2, 3} 2.2.2 If n 2 +1≤b≤n−1 Proposition 3 If C is a minimum identifying code of Kn Kn+b with then : |C| ≤ n + 2b n 2 + 1 ≤ b ≤ n − 1, Preuve. If n2 + 1 ≤ b ≤ n − 1, then the code C = A ∪ B ∪ P is an identifying code of Kn Kn+b , n ≥ 2 with cardinality n + 2b such that : 60 Anissa Hissoum & Ahmed Semri A = {(i, i)/i ∈ {1, ..., n}} ∪ {(i, n + i)/i ∈ {1, ..., b}} B = {(n − i + 1, i)/i ∈ {1, ..., b} P = { n, ∅ n 2 } if n is odd if n is even K7 K11 K7 K12 K7 K13 Figure 8: Codes identifiants de Kn Kn+b , n = 7 et b ∈ {4, 5, 6} We conjecture that the upper bounds found in the propositions ??, ?? et ?? coincide with the exact values of the minimum cardinality of identifying codes of Kn Kan+b . Conjecture 1 (Anissa HISSOUM, Ahmed SEMRI) Let Kn and Km ,m > n ≥ 2 be two cliques and let a, b be the quotient and the remainder of the Euclidean division of m by n respectively. If C is a minimum identifying code of Kn Km , then : n(2a − 1) + 2b 3n |C| = +b 2 n + 2b 3 if a ≥ 2 and 0 ≤ b ≤ n − 1; if a = 1 and 1 ≤ b ≤ n2 ; if a = 1 and n2 ≤ b ≤ n − 1. Conclusion We have established upper bounds on the minimum cardinality of an identifying code in the Cartesian product of two cliques with different sizes Kn Km following values of de a and b, where m = an + b and b ≤ n − 1. All possible cases are summarized in Table ??. Identifying codes of Cartesian product of two cliques 61 Table 1: Recapitulation a b Upper Bound a≥2 a=1 a=1 a=1 0 ≤ b ≤ n − 1 1 ≤ b ≤ n2 n 2 ≤b≤n−1 b=0 n(2a 3n − 1) + 2b 2 +b n + 3n 2b (exact value[?, Theorem 1] 2 References [1] T. Y. Berger-Wolf, W. E. Hart, and J. Saia. Discrete sensor placement problems in distribution networks. Journal of Mathematical and Computer Modelling, vol. 42, no. 13, pp. 1385–1396, 2005. [2] I. Charon, O. Hudry, and A. Lobstein. Minimizing the Size of an Identifying or LocatingDominating Code in a Graph is NP-Hard. Theorical Computer Science, vol. 290, Issue 3, pp 2109–2120, 2003. [3] S. Gravier, J. Moncel , and A. Semri, Identifying codes of Cartesian product of two cliques of the same size. The Electronic Journal of Combinatorics, 15 N4, 2008. [4] M.G. Karpovsky , K. Chakrabarty, and L.B. Levitin. On a new class of codes for identifying vertices in graphs. IEEE Transactions on Information Theory, Vol. 44(2), pp 599–611, 1998. [5] M. Laifenfeld and A. Trachtenberg. Disjoint identifying codes for arbitrary graphs. IEEE International Symposium on Information Theory, Adelaide Australia, 2005. [6] S. Ray, D. Starobinski, A. Trachtenberg, and R. Ungrangsi, Robust location detection with sensor networks. IEEE Journal on Selected Areas in Comm, vol. 22, no. 6, August 2004.
© Copyright 2026 Paperzz