On certain families of sparse numerical semigroups with Frobenius number even

This paper is about sparse numerical semigroups and applications in the Weierstrass semigroups theory. We describe and find the genus of certain families of sparse numerical semigroups with Frobenius number even and we also study the realization of the elements on these families as Weierstrass semig...

Full description

Saved in:
Bibliographic Details
Published in:Algebra and Discrete Mathematics
Date:2019
Main Authors: Tizziotti, G., Villanueva, J.
Format: Article
Language:English
Published: Інститут прикладної математики і механіки НАН України 2019
Online Access:https://nasplib.isofts.kiev.ua/handle/123456789/188426
Tags: Add Tag
No Tags, Be the first to tag this record!
Journal Title:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Cite this:On certain families of sparse numerical semigroups with Frobenius number even / G. Tizziotti, J. Villanueva // Algebra and Discrete Mathematics. — 2019. — Vol. 27, № 1. — С. 99–116. — Бібліогр.: 20 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-188426
record_format dspace
spelling Tizziotti, G.
Villanueva, J.
2023-02-28T19:21:21Z
2023-02-28T19:21:21Z
2019
On certain families of sparse numerical semigroups with Frobenius number even / G. Tizziotti, J. Villanueva // Algebra and Discrete Mathematics. — 2019. — Vol. 27, № 1. — С. 99–116. — Бібліогр.: 20 назв. — англ.
1726-3255
2010 MSC: 20M10, 20M14, 14H55
https://nasplib.isofts.kiev.ua/handle/123456789/188426
This paper is about sparse numerical semigroups and applications in the Weierstrass semigroups theory. We describe and find the genus of certain families of sparse numerical semigroups with Frobenius number even and we also study the realization of the elements on these families as Weierstrass semigroups.
Partially supported by CNPq 446913/2014-6 and FAPEMIG APQ-01607-14. Partially supported by PNPD/CAPES, by Programa de Pós-doutorado da FAMAT, Câmpus Santa Mônica, Universidade Federal de Uberlândia.
en
Інститут прикладної математики і механіки НАН України
Algebra and Discrete Mathematics
On certain families of sparse numerical semigroups with Frobenius number even
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title On certain families of sparse numerical semigroups with Frobenius number even
spellingShingle On certain families of sparse numerical semigroups with Frobenius number even
Tizziotti, G.
Villanueva, J.
title_short On certain families of sparse numerical semigroups with Frobenius number even
title_full On certain families of sparse numerical semigroups with Frobenius number even
title_fullStr On certain families of sparse numerical semigroups with Frobenius number even
title_full_unstemmed On certain families of sparse numerical semigroups with Frobenius number even
title_sort on certain families of sparse numerical semigroups with frobenius number even
author Tizziotti, G.
Villanueva, J.
author_facet Tizziotti, G.
Villanueva, J.
publishDate 2019
language English
container_title Algebra and Discrete Mathematics
publisher Інститут прикладної математики і механіки НАН України
format Article
description This paper is about sparse numerical semigroups and applications in the Weierstrass semigroups theory. We describe and find the genus of certain families of sparse numerical semigroups with Frobenius number even and we also study the realization of the elements on these families as Weierstrass semigroups.
issn 1726-3255
url https://nasplib.isofts.kiev.ua/handle/123456789/188426
citation_txt On certain families of sparse numerical semigroups with Frobenius number even / G. Tizziotti, J. Villanueva // Algebra and Discrete Mathematics. — 2019. — Vol. 27, № 1. — С. 99–116. — Бібліогр.: 20 назв. — англ.
work_keys_str_mv AT tizziottig oncertainfamiliesofsparsenumericalsemigroupswithfrobeniusnumbereven
AT villanuevaj oncertainfamiliesofsparsenumericalsemigroupswithfrobeniusnumbereven
first_indexed 2025-11-25T22:45:20Z
last_indexed 2025-11-25T22:45:20Z
_version_ 1850571096782274560
fulltext “adm-n1” — 2019/3/22 — 12:03 — page 99 — #107 Algebra and Discrete Mathematics RESEARCH ARTICLE Volume 27 (2019). Number 1, pp. 99–116 c© Journal “Algebra and Discrete Mathematics” On certain families of sparse numerical semigroups with Frobenius number even Guilherme Tizziotti1 and Juan Villanueva2 Communicated by V. Mazorchuk Abstract. This paper is about sparse numerical semigroups and applications in the Weierstrass semigroups theory. We describe and find the genus of certain families of sparse numerical semigroups with Frobenius number even and we also study the realization of the elements on these families as Weierstrass semigroups. Introduction Let Z be the set of integers numbers and N0 be the set of non-negative integers. A subset H = { 0 = n0(H) < n1(H) < · · · } of N0 is a numerical semigroup if its is closed respect to addition and its complement N0 \H is finite. The cardinality of the set Gaps(H) := N0 \H is called genus of the numerical semigroup H and is denoted by g = g(H). Note that g(H) = 0 if and only if H = N0. If g(H) > 0 the elements of Gaps(H) are called gaps. The smallest integer c = c(H) such that c+ h ∈ H, for all h ∈ N0 is called the conductor of H. The least positive integer n1 = n1(H) ∈ H is called the multiplicity of H. As N0 \ H is finite, the set Z \ H has a maximum, which is called Frobenius number and will be denoted by ℓg = ℓg(H). A property known of this number is that ℓg(H) 6 2g − 1, see 1Partially supported by CNPq 446913/2014-6 and FAPEMIG APQ-01607-14. 2Partially supported by PNPD/CAPES, by Programa de Pós-doutorado da FAMAT, Câmpus Santa Mônica, Universidade Federal de Uberlândia. 2010 MSC: 20M10, 20M14, 14H55. Key words and phrases: Arf numerical semigroup, numerical semigroup, sparse numerical semigroup, Weierstrass semigroup, weight of numerical semigroup. “adm-n1” — 2019/3/22 — 12:03 — page 100 — #108 100 On certain families of sparse numerical semigroups [15]. In particular, H = N0 if and only if −1 is the Frobenius number of H. Aa a consequence of this fact, from now on we use the notation ℓ0 = ℓ0(H) := −1, for all numerical semigroup H . When g > 0, we denote Gaps(H) = { 1 = ℓ1(H) < · · · < ℓg(H) } . So, c = ℓg(H) + 1 and is clear that c = nc−g(H). For simplicity of notation we shall write ℓi for ℓi(H) and nk for nk(H), for all integers i, k such that 0 6 i 6 g and k > 0, when there is no danger of confusion. More details about numerical semigroups theory, see e.g. [17]. Currently, there are several families that have been of interest in the literature due to their properties and applications. Examples of such families are the sparse semigroups, which were introduced in [14]. A numerical semigroup H = {0 = n0 < n1 < · · · } of genus g > 0 with Gaps(H) = {ℓ1 < · · · < ℓg} is called sparse numerical semigroup if ℓi − ℓi−1 6 2, for all integer i such that 1 6 i 6 g, or equivalently ni − ni−1 > 2, for all integer i such that 1 6 i 6 c − g, where c is the conductor of H. For convenience, we considerer the numerical semigroup N0 as sparse. Thus, the concept of sparse numerical semigroups it’s in a way a generalization of the concept of Arf numerical semigroups, which was introduced in [1]. Among other applications, the study of numerical semigroup is related to Algebraic Geometry in the treatment of algebraic curves and their Weierstrass semigroups. More explicitly, given a numerical semigroup H, does it exist a curve X such that for some point P ∈ X has H = H(P )?, where H(P ) is the Weierstrass semigroup of X at P . If the answer is yes, we say that the numerical semigroup H is Weierstrass. Studies to answer this question have been done for decades, see e.g. [4], [12], [15] and [19]. From a geometrical point of view, sparse numerical semigroups are closely related to Weierstrass semigroups arising in double covering of curves, cf. [19]. Its arithmetical structure is strongly influenced by the parity of ℓg. In this work, we study certain families of sparse numerical semigroups which are examples of Weierstrass semigroups. In addition, we manage to describe and find the genus of the semigroups on these families. These aspects are very important in the study of numerical semigroups theory. It is important to note that in [6] the authors find an upper bound for the genus of sparse numerical semigroups with Frobenius number even. Here, in this paper, we get a better bound for the genus of the semigroups on the families of sparse numerical semigroups studied. This paper is organized as follows. Section 1, contains basic concepts about numerical semigroups and backgrounds for the next sections. In Section 2, we study certain families of sparse numerical semigroups with “adm-n1” — 2019/3/22 — 12:03 — page 101 — #109 G. Tizziotti, J. Villanueva 101 Frobenius number even. As the main results, we describe and find the genus of the semigroups on these families of sparse numerical semigroups. Finally, in Section 3 we study the realization of the sparse numerical semigroups determined in the previous section as Weierstrass semigroups. 1. Preliminaries 1.1. Basic concepts Let H = { 0 = n0 < n1 < · · · } be a numerical semigroup of genus g > 0 and Gaps(H) = {ℓ1 < · · · < ℓg}. For each 1 6 i 6 g, the ordered pair (ℓi−1, ℓi) will be called leap on H (or simply leap). The set of leaps on H will be denoted by V = V(H) := { (ℓi−1, ℓi) : 1 6 i 6 g } . Note that |V | = g. The ordered pair (ℓi−1, ℓi) will be called single leap if ℓi − ℓi−1 = 1 and double leap if ℓi − ℓi−1 = 2. Based on this set, for a positive integer m, let us define the subset Vm = Vm(H) := { (ℓi−1, ℓi) : ℓi − ℓi−1 = m, 1 6 i 6 g } and for an interval [a, b], with −1 6 a < b 6 ℓg, let us define the subset V[a,b] = { (ℓi−1, ℓi) : ℓi−1, ℓi ∈ [a, b], 1 6 i 6 g } . For convenience, we define Vm(N0) := ∅, for all positive integer m. To simplify the notation we will denote the cardinality of the set Vm(H) by vm(H), that is, vm = vm(H) := |Vm(H) |. 1.2. Arf numerical semigroup A numerical semigroup H = {0 = n0 < n1 < · · · } is called Arf numerical semigroup if ni + nj − nk ∈ H, (1) for all integers i, j, k such that 0 6 k 6 j 6 i, or equivalently ni+nj−nk ∈ H, for all integers i, j, k such that 0 6 k 6 j 6 i 6 c − g, where c is the conductor and g is the genus of H, respectively. The Arf numerical semigroups was introduced in [1]. For more details about this family of numerical semigroups, see e.g. [2], [5], [16] and [18]. If g > 0 and Gaps(H) = {ℓ1 < · · · < ℓg}, the Arf property (1) implies that ℓi − ℓi−1 6 2, (2) “adm-n1” — 2019/3/22 — 12:03 — page 102 — #110 102 On certain families of sparse numerical semigroups for all integer i such that 1 6 i 6 g, or equivalently ni − ni−1 > 2, for all integer i such that 1 6 i 6 c− g (see [14, Corollary 1] and [20, Corollary 2.1.4]). Using this property is not difficult to see that if H is an Arf numerical semigroup, then H is a sparse numerical semigroup. For each non-negative integer g, let Ng := {0} ∪ {n ∈ N : n > g + 1} (in the case g = 0, it notation is itself the N0). It is clear that Ng is a numerical semigroup of genus g. The semigroup Ng is called ordinary nu- merical semigroup and is a canonical example of Arf numerical semigroups and, consequently, a example of sparse numerical semigroups. 1.3. Sparse numerical semigroups Theorem 1.1 ([18], Theorem 2.2). Let H be a numerical semigroup of genus g. (1) H is an sparse numerical semigroup if and only if v1(H)+v2(H) = g. In this case, vm(H) = 0, for all positive integer m > 3. (2) If H is an sparse numerical semigroup, then the Frobenius number of H is ℓg = v1(H) + 2v2(H)− 1. (3) If g > 0 and ℓg(H) = 2g −K, for some positive integer K, then H is an sparse numerical semigroup if and only if v1(H) = K − 1 and v2(H) = g −K + 1. For a numerical semigroup H = {0 = n0 < n1 < · · · }, define M = M(H) := n1 − 1. The parameter M was introduced in [14], where if ℓg(H) = 2g −K, for some positive integer K, we have that 0 6 M 6 K. (3) If g > 1 and Gaps(H) = {1 = ℓ1 < · · · < ℓg}, define SM = SM (H) := ∣ ∣ { (ℓi−1, ℓi) : ℓi − ℓi−1 = 1, M + 1 6 i 6 g }∣ ∣ . (4) Note that SM < v1, since { (ℓi−1, ℓi) : ℓi− ℓi−1 = 1, M +1 6 i 6 g } ( V1. Before the next result, let us remember that a semigroup H is called γ-hyperelliptic if it has exactly γ even gaps. If γ = 0, H is called simply hyperelliptic. Proposition 1.2. Let H = {0 = n0 < n1 < · · · } be a numerical semi- group of genus g and M = M(H) = n1 − 1. Then: (1) 0 6 M 6 g; “adm-n1” — 2019/3/22 — 12:03 — page 103 — #111 G. Tizziotti, J. Villanueva 103 (2) M(H) = 0 if and only if H = N0; (3) M(H) = g if and only if H = Ng; (4) M(H) = 1 if and only if H is hyperelliptic; (5) If g > 0 and Gaps(H) = {ℓ1 < · · · < ℓg}, then M is the largest integer such that ℓM = M . Proof. The assertions (1), (2) and (3) are clear. Item (4), follows from the fact that M = 1 if and only if n1 = 2. Finally, item (5) follows directly from the fact that all integers belong to interval [1, n1 − 1] are gaps. Corollary 1.3. Let H be a sparse numerical semigroup of genus g > 1 with Frobenius number ℓg = 2g − K, for some positive integer K, and Gaps(H) = {ℓ1 < · · · < ℓg}. Let M = M(H) and SM = SM (H) be defined in (4). If M < g, then SM = K −M. Proof. By the definition of SM , since M < g, we have that SM 6= 0 and SM = v1(H) − (M − 1). Now, by Theorem 1.1 item (2), follows that SM = K − 1− (M − 1) = K −M . In Proposition 1.2, we present the semigroups for the cases M = 0, 1 and M = g. The case M = 2 is treated in [14, Proposition 3]. The following result treat of the case M = K. Theorem 1.4. Let H be a numerical semigroup of genus g > 0 with Frobenius number ℓg = 2g−K, for some positive integer K. The following statements are equivalent: (1) H is an sparse numerical semigroup and M = K; (2) H = NK or H = {K + 2i− 1 : 1 6 i 6 g −K} ∪N2g−K ; (3) v1(H) = M − 1 and v2(H) = g −M + 1. In this case, Gaps(H) = {1, . . . , ℓK , . . . , ℓg}, where ℓi = 2i − K, for all integer i such that K 6 i 6 g. Proof. We prove that (1) ⇒ (2) ⇒ (3) ⇒ (1). By [14, Theorem 1 (2)] follows that (1) ⇒ (2). Now, suppose (2). If H = NK , then M = K and by Theorem 1.1 (3) follows that v1(H) = M − 1 and v2(H) = g − M + 1. If H = {K + 2i − 1 : 1 6 i 6 g − K} ∪ N2g−K , then Gaps(H) = {1, . . . ,K, ℓK+1, . . . , ℓg}, where ℓK+i = K + 2i, for all integer i such that 1 6 i 6 g − K. So, M = K, v1(H) = K − 1 = M − 1 and v2(H) = g−K+1 = g−M +1. Thus, we have (2) ⇒ (3). The implication (3) ⇒ (1), follows directly from the Theorem 1.1 items (1) and (3). The assertion on the gaps of H follows as in the proof of the implication (2) ⇒ (3). “adm-n1” — 2019/3/22 — 12:03 — page 104 — #112 104 On certain families of sparse numerical semigroups 2. Sparse numerical semigroups with Frobenius number even For each positive integer k, consider the family H sfe k of sparse numerical semigroups H with genus g = g(H) and Frobenius number even of the form 2g − 2k. That is, H sfe k := {H : H is a sparse numerical semigroup with genus g and ℓg = 2g − 2k}. In this section, we will study the classification of the elements of some proper subsets in H sfe k as well as the cardinality of these subsets. If H ∈ H sfe k in [14, Theorem 2] is proved that g(H) 6 6k − 3 and that the family H sfe k is finite. In [20, Question 2.3.10], was conjectured that g(H) 6 4k − 1. This conjecture was proved by Contiero, Moreira and Veloso in [6, Corollary 3.7]. In Theorem 2.8 and Theorem 2.13, we get a better bound for g(H) for certain H ∈ H sfe k . In the previous section, in particular, we describe all the sparse nu- merical semigroups belongs to H sfe k with M = 0, 1 and M = 2k. The next result describe the case M = 3. Lemma 2.1. Let k > 2 be an integer and H be a numerical semigroup of genus g and Frobenius number ℓg = 2g − 2k. If M = 3, the following statements are equivalent: (1) H is be an Arf numerical semigroup; (2) H is be an sparse numerical semigroup; (3) H = 〈4, 4k − 1, 4k + 1, 4k + 2〉; (4) H is k-hyperelliptic. In this case, g = 3k − 1 and Gaps(H) = {1, . . . , 2g − 2k} \ { 4i : i ∈ N, 1 6 i 6 2g−2k−2 4 } . Proof. We prove that (1) ⇒ (2) ⇒ (3) ⇒ (4) ⇒ (3) ⇒ (1). Implication (1) ⇒ (2) is always true. Before show the next implications, since M = 3, we have that the multiplicity of H is n1 = 4. Let G := {1, . . . , 2g − 2k} \ { 4i : i ∈ N, 1 6 i 6 2g−2k−2 4 } . It is clear that Gaps(H) is a subset of G. Since 4 ∈ H and ℓg = 2g− 2k is even, we have that ℓg = 4i0 +2, for some i0 ∈ N. Note that if ℓ ∈ G is even, then ℓ = 4i+ 2, for some i ∈ N0 with i 6 i0. So, ℓ ∈ G and ℓ even implies that ℓ /∈ H, because otherwise we will have that ℓg ∈ H, since ℓ = 4i+ 2, 4 ∈ H, ℓg = 4i0 + 2 and i 6 i0. “adm-n1” — 2019/3/22 — 12:03 — page 105 — #113 G. Tizziotti, J. Villanueva 105 Now, suppose (2). Let ℓ ∈ G. If ℓ is even, we have seen above that ℓ ∈ Gaps(H). If ℓ ≡ 1 (mod 4), then ℓ − 1 ∈ H and we have that ℓ ∈ Gaps(H), since H is sparse. If ℓ ≡ 3 (mod 4), then ℓ + 1 ∈ H and we also have that ℓ ∈ H, since H is sparse. Thus, Gaps(H) = G, and so H = 〈4, 4k − 1, 4k + 1, 4k + 2〉. Thus, we have (2) ⇒ (3). Now, we prove the implication (3) ⇒ (4). If H = 〈4, 4k−1, 4k+1, 4k+ 2〉, then 2 ∈ Gaps(H) and the number of even gaps on the interval [4, 4k] is equal to 4k−4 4 = k − 1. So, H is k-hyperelliptic. To prove the implication (4) ⇒ (3), first note that G∩2N = Gaps(H)∩ 2N. Suppose that H is k-hyperelliptic. Then, G ∩ 2N = Gaps(H) ∩ 2N and |G ∩ 2N | = g − k − 2g−2k−2 4 implies that g − k − 2g−2k−2 4 = k, that is, g = 3k − 1. Since M = 3, g = 3k − 1 and ℓg = 2g − 2k = 4k − 2, we conclude that Gaps(H) = [1, 4k − 2] \ {4i : i ∈ N, 1 6 i 6 k − 1}, and then H = 〈4, 4k − 1, 4k + 1, 4k + 2〉. Finally, we prove the implication (3) ⇒ (1). If H = 〈4, 4k − 1, 4k + 1, 4k + 2〉, then H = 4N0 ∪ {n ∈ N : n > 4k − 1}. Suppose that H = {0 = n0 < 4 = n1 < n2 < · · · } and let i, j, s integers such that 0 6 s 6 j 6 i. If ni, nj or ns belongs to {n ∈ N : n > 4k − 1}, then it is clear that ni + nj − ns ∈ {n ∈ N : n > 4k − 1} and so ni + nj − ns ∈ H. If ni, nj , ns ∈ 4N0 then it is also clear that ni + nj − ns ∈ 4N0 and so ni + nj − ns ∈ H. Therefore, H is Arf. For H ∈ H sfe k , by Equation (3) and Proposition 1.2, follows that 2 6 M(H) 6 2k. (5) For each integer J such that 0 6 J 6 2k − 2, consider the set Hk 2+J of sparse numerical semigroups H ∈ H sfe k such that M(H) = 2 + J . That is, H k 2+J = { H ∈ H sfe k : M(H) = 2 + J } . (6) Since H sfe k = 2k−2 ⋃ · J=0 H k 2+J , in order to study the cardinality of H sfe k it is enough to study the cardinality of Hk 2+J , for all integer J such that 0 6 J 6 2k − 2. Next we will study the cases: J = 2k − 2, J = 2k − 3 and J = 2k − 4. The following result shows that all numerical semigroups H belongs to Hk 2k are Arf and (g(H)− k)-hyperelliptic. “adm-n1” — 2019/3/22 — 12:03 — page 106 — #114 106 On certain families of sparse numerical semigroups Proposition 2.2. Let H be a numerical semigroup of genus g > 0 with Frobenius number ℓg = 2g − 2k, for some positive integer k. If M = 2k, then the following statements are equivalent: (1) H is an Arf numerical semigroup; (2) H is an sparse numerical semigroup; (3) H = N2k or H = {2k + 2i− 1 : 1 6 i 6 g − 2k} ∪N2g−2k; (4) v1(H) = 2k − 1 and v2(H) = g − 2k + 1; (5) H is (g − k)-hyperelliptic. In this case, 2k 6 g 6 3k and Gaps(H) = {1, . . . , 2k − 1, ℓ2k, . . . , ℓg}, where ℓi = 2(i− k), for all integer i such that 2k 6 i 6 g. Proof. We prove that (1) ⇒ (2) ⇒ (3) ⇔ (4), (3) ⇔ (5) and (3) ⇒ (1). The implication (1) ⇒ (2) is always true. The implication (2) ⇒ (3), the equivalence (3) ⇔ (4) and the assertion on the gaps of H is a particular case from Theorem 1.4, by taking M = K = 2k. Now, suppose (3). Then, ni = 2k + 2i− 1, for all integer i such that 1 6 i 6 g − 2k + 1, Let i, j integers such that 0 6 j 6 i 6 g − 2k + 1. If j = 0 is clear that 2ni − nj ∈ H ; other case 2ni − nj = 2k+4i− 2j − 1 > 2k + 2i − 1 > 2k + 1 > ℓ2k. Since in the interval [ℓ2k, ℓg] the gaps ℓi’s are even numbers and 2ni − nj is odd, follow that 2ni − nj ∈ H. This shows that (3) ⇒ (1). Also, by Theorem 1.4, (3) implies that Gaps(H) = {1, . . . , 2k − 1, ℓ2k, . . . , ℓg}, where ℓi = 2(i− k), for all integer i such that 2k 6 i 6 g, and thus (3) ⇒ (5). Finally, suppose (5). Since in the interval [2, 2g − 2k] there are g − k elements even, these elements should be gaps and so (5) ⇒ (3). Note that, since n1 = 2k + 1, from item (3) follows that 2(2k + 1) > ℓg + 2 = (2g − 2k) + 2. Therefore, we have that g 6 3k. Theorem 2.3. For all integer k > 1, H k 2k = {N2k} ∪ { H(k,r) : r ∈ N, 1 6 r 6 k } , where H(k,r) = {2k + 2i− 1 : i ∈ N, 1 6 i 6 r} ∪N2k+2r, for 1 6 r 6 k. In addition, g ( H(k,r) ) = 2k+r. In particular, the cardinality of Hk 2k is ∣ ∣Hk 2k ∣ ∣ = k + 1. Proof. For each integer r such that 1 6 r 6 k, let H(k,r) = {2k+2i−1 : i ∈ N, 1 6 i 6 r}∪N2k+2r. By the definition, is clear that H(k,r) is a numerical semigroup of genus g ( H(k,r) ) = 2k + r and ℓg ( H(k,r) ) = 2g ( H(k,r) ) − 2k. “adm-n1” — 2019/3/22 — 12:03 — page 107 — #115 G. Tizziotti, J. Villanueva 107 So, by Proposition 2.2, we have that H(k,r) belongs to H k 2k, for all integer r such that 1 6 r 6 k. Also, by Proposition 2.2, N2k is contained in H k 2k. Now, let H ∈ H k 2k of genus g and H 6= N2k. By Proposition 2.2, we have 2k+ 1 6 g 6 3k and H = {2k+ 2i− 1 : i ∈ N, 1 6 i 6 g− 2k} ∪N2g−2k. Let r := g − 2k. Then, 1 6 r 6 k and so H = H(k,r). This completes the proof. A numerical semigroup H of genus g > 0 is called quasi-symmetric if the Frobenius number of H is ℓg = 2g − 2. In [2] and [3], has been proved that there are only two Arf numerical semigroups quasi-symmetric: 〈3, 4, 5〉 and 〈3, 5, 7〉. In [14, Example 4] and [20, Example 2.3.13], was proved that Arf numerical semigroups quasi-symmetric is equivalent to sparse numerical semigroups quasi-symmetric. This result can be obtained directly from the previous theorem by taking k = 1. That is, the family of sparse numerical semigroups quasi-symmetric is H sfe 1 = { N2, {3} ∪N4 } = H 1 2 . The following is a further demonstration of the [14, Example 5] about the classification of sparse numerical semigroups of genus g with Frobenius number ℓg = 2g − 4. Corollary 2.4. The family of sparse numerical semigroups of genus g and Frobenius number 2g − 4 is H sfe 2 = { {3, 6} ∪N8, {3, 6, 9} ∪N10, {4} ∪N6,N4, {5} ∪N6, {5, 7} ∪N8 } . In this case, H 2 2 = { {3, 6} ∪N8, {3, 6, 9} ∪N10 } , H 2 3 = { {4} ∪N6 } , H 2 4 = { N4, {5} ∪N6, {5, 7} ∪N8 } . Proof. First, note that H sfe 2 = H2 2 ∪ H2 3 ∪ H2 4 . From [14, Proposition 3], we have that H2 2 = { {3, 6} ∪N8, {3, 6, 9} ∪N10 } . From the item (3) of Lemma 2.1 follows that H2 3 = { {4} ∪ N6 } and taking k = 2 in the previous theorem follows that H2 4 = { N4, {5} ∪N6, {5, 7} ∪N8 } . Lemma 2.5. Let k > 2 be an integer. For each pair of integers r and αr such that 1 6 r 6 k − 1 and 1 6 αr 6 r, let Hαr (k,r) := {2(k + µ− 1) : 1 6 µ 6 αr}∪ {2k + 2ν + 1 : αr 6 ν 6 r − 1} ∪N2k+2r. “adm-n1” — 2019/3/22 — 12:03 — page 108 — #116 108 On certain families of sparse numerical semigroups Then, Hαr (k,r) ∈ H k 2k−1 and g ( Hαr (k,r) ) = 2k + r, for all pairs (r, αr). Proof. Let r, αr and Hαr (k,r) be as above. By the definition, it is clear that Hαr (k,r) is a numerical semigroup with set of gaps given by Gaps ( Hαr (k,r) ) = {i : 1 6 i 6 2k − 1} ∪· {2k + 2µ− 1 : 1 6 µ 6 αr}∪· {2k + 2ν : αr 6 ν 6 r}. In particular, g ( Hαr (k,r) ) = (2k − 1) + αr + (r − αr + 1) = 2k + r. Also, M ( Hαr (k,r) ) = 2k − 1 and ℓg ( Hαr (k,r) ) = 2k + 2r = 2g − 2k. Therefore, Hαr (k,r) ∈ H k 2k−1, for all pairs (r, αr) as above. Henceforth Hαr (k,r) is the sparse numerical semigroup defined in the Lemma 2.5. Remark 2.6. Let k > 2 be an integer and (r, αr) be as defined in the previous lemma. Then, Gaps ( Hαr (k,r) ) = { ℓ1 ( Hαr (k,r) ) , . . . , ℓ2k+r ( Hαr (k,r) )} , where ℓi ( (Hαr (k,r) ) =      i, if 1 6 i 6 2k − 1; 2(i− k) + 1, if 2k 6 i 6 2k − 1 + αr; 2(i− k), if 2k + αr 6 i 6 2k + r. The following theorem will give us a new bound for the genus of semigroups in Hk 2k−1. Before, we will make an observation that will be very useful in the following. Remark 2.7. Let H be a sparse numerical semigroup of genus g > 1 with Frobenius number ℓg = 2g − K, for some positive integer K, and Gaps(H) = {ℓ1 < · · · < ℓg}. Let r and s integers such that 1 6 r < s 6 g. If V[ℓr,ℓs] ∩ V1 = { (ℓj−1, ℓj) } , for some j ∈ {r + 1, . . . , s}, then ℓv = ℓj−1 − 2(j − v − 1), for all integer v such that r 6 v 6 j − 1, and ℓw = ℓj + 2(w − j), for all integer w such that j 6 w 6 s. In particular, ℓj−1 ≡ ℓv (mod 2), for all integer v such that r 6 v 6 j − 1, and ℓj ≡ ℓw (mod 2), for all integer w such that j 6 w 6 s. “adm-n1” — 2019/3/22 — 12:03 — page 109 — #117 G. Tizziotti, J. Villanueva 109 Theorem 2.8. Let k > 2 be an integer. If H ∈ Hk 2k−1, then 2k + 1 6 g(H) 6 3k − 1. Proof. Let k > 2 be an integer and let H ∈ Hk 2k−1 with g(H) = g and Gaps(H) = {1 = ℓ1 < · · · < ℓg}. Since H is a sparse semigroup and M = 2k − 1, we have that ℓM+1 = 2k + 1. If g 6 2k, then ℓg 6 2k, a contradiction with the value of ℓM+1. So, g > 2k + 1. Now, suppose that g = 3k + s, for some integer s > 0. So, ℓg = 4k + 2s. By the Corollary 1.3, SM = 1, that is, there exist a unique single leap (ℓj−1, ℓj), for some integer j such that 2k+1 = M +2 6 j 6 g. Then, by the Remark 2.7, all the even numbers greater than 2k+ 1 and smaller than ℓj−1 are non-gaps. Moreover all the even numbers from ℓj to ℓg are gaps. More precisely, we have ℓg = ℓj + 2(g − j). So, ℓj = (4k + 2s)− 2(3k + s− j) = 2j − 2k. We conclude the proof by studying the following cases separately: j < 3k and j > 3k. Firstly, assume that j < 3k. Then, ℓj = 2j − 2k < 4k 6 4k + 2s. So, 4k is a gap, a contradiction since 2k is a non-gap. Now, suppose that j > 3k. Then, 2k 6 2j − 4k < 2j − 2k − 1 = ℓj−1. Thus, 2j − 4k ∈ H. Therefore, ℓj = 2j − 2k = 2k + (2j − 4k) ∈ H, a contradiction, and the proof is complete. Theorem 2.9. For all integer k > 2, H k 2k−1 = { Hαr (k,r) : (r, αr) ∈ N 2, 1 6 r 6 k − 1, 1 6 αr 6 r } . Proof. Firstly, by the Lemma 2.5, we have that { Hαr (k,r) : (r, αr) ∈ N 2, 1 6 r 6 k − 1, 1 6 αr 6 r } ⊂ H k 2k−1. Now, let H ∈ Hk 2k−1 with g(H) = g. By Theorem 2.8, we have 2k + 1 6 g 6 3k − 1. Let r := g − 2k. So, 1 6 r 6 k − 1. Since ℓg = 2g − 2k, we have that ℓg = ℓ2k+r = 2k + 2r. Let Gaps(H) = {1 = ℓ1 < · · · < ℓg}. By Corollary 1.3, SM = 1, that is, there exist a unique single leap (ℓj−1, ℓj), for some integer j such that 2k + 1 = M + 2 6 j 6 g = 2k + r. Let αr := j − 2k. By Remark 2.7, ℓv = ℓ2k + 2(v − 2k), for all integer v such that 2k 6 v 6 j − 1 and ℓw = ℓ2k+r + 2 [ w − (2k + r) ] , for all integer w such that j 6 w 6 2k+ r. Since M = 2k− 1, we have ℓ2k−1 = 2k− 1 and ℓ2k = 2k + 1. Thus, we concluded that ℓv = 2(v − k) + 1, for all integer v such that 2k 6 v 6 j− 1 = 2k− 1+αr, and ℓw = 2(w− k), for all integer w such that 2k+αr = j 6 w 6 2k+ r = g. Therefore, by Remark 2.6, we have that H = Hαr (k,r) and follows the result. “adm-n1” — 2019/3/22 — 12:03 — page 110 — #118 110 On certain families of sparse numerical semigroups Corollary 2.10. For all integer k > 2, ∣ ∣Hk 2k−1 ∣ ∣ = ( k 2 ) . Proof. By the Theorem 2.9, we have that H k 2k−1 = k−1 ⋃ · r=1 { Hαr (k,r) : αr ∈ N, 1 6 αr 6 r } . Therefore, ∣ ∣ ∣ H k 2k−1 ∣ ∣ ∣ = k−1 ∑ r=1 r = ( k 2 ) . Lemma 2.11. Let k > 3 be an integer. Let r, s and αs be a triple of integers such that 1 6 r 6 k − 2, 1 6 s 6 r and 1 6 αs 6 r − s+ 1. For r ∈ {k, k + 1}, let s = 1 and αs = k. For each triple (r, s, αs), consider H (s,αs) (k,r) :={2k + 2λ− 3 : 1 6 λ 6 s}∪ {2(k + s+ µ− 1) : 1 6 µ 6 αs − 1}∪ {2k + 2ν + 2s− 1 : αs 6 ν 6 r − s} ∪N2k+2r. Then, H (s,αs) (k,r) ∈ Hk 2k−2 and g ( H (s,αs) (k,r) ) = 2k + r, for all triples (r, s, αs). Proof. Let r, s, αs and H (s,αs) (k,r) be as above. By the definition, it is clear that H (s,αs) (k,r) is a numerical semigroup with set of gaps given by Gaps ( H (s,αs) (k,r) ) = {i : 1 6 i 6 2k − 2} ∪· {2(k + λ− 1) : 1 6 λ 6 s}∪· {2k + 2µ+ 2s− 3 : 1 6 µ 6 αs}∪· {2(k + ν + s) : αs − 1 6 ν 6 r − s}. In particular, g ( H (s,αs) (k,r) ) = (2k−2)+s+αs+(r−s−αs+2) = 2k+r. Also, M ( H (s,αs) (k,r) ) = 2k − 2 and ℓg ( H (s,αs) (k,r) ) = 2k + 2r = 2g − 2k. Therefore, H (s,αs) (k,r) ∈ Hk 2k−2, for all triple (r, s, αs) as above. Henceforth H (s,αs) (k,r) is the sparse numerical semigroup defined in the Lemma 2.11. Remark 2.12. Let k > 3 be an integer and (r, s, αs) be as defined in the previous lemma. Then, Gaps ( H (s,αs) (k,r) ) = { ℓ1 ( H (s,αs) (k,r) ) , . . . , ℓ2k+r ( H (s,αs) (k,r) )} , “adm-n1” — 2019/3/22 — 12:03 — page 111 — #119 G. Tizziotti, J. Villanueva 111 where ℓi ( H (s,αs) (k,r) ) =            i, if 1 6 i 6 2k − 2; 2(i− k + 1), if 2k − 1 6 i 6 2k + s− 2; 2(i− k) + 1, if 2k + s− 1 6 i 6 2k + s+ αs − 2; 2(i− k), if 2k + s+ αs − 1 6 i 6 2k + r. Theorem 2.13. Let k > 3 be an integer. If H ∈ Hk 2k−2, then 2k + 1 6 g(H) 6 3k + 1 and g(H) 6= 3k − 1. Proof. Let k > 3 be an integer and let H ∈ Hk 2k−2 with g(H) = g and Gaps(H) = {1 = ℓ1 < · · · < ℓg}. Since M = M(H) = 2k − 2 and H is a sparse semigroup, we have that ℓ2k−1 = ℓM+1 = 2k and, by the Corollary 1.3, SM = 2. If g < 2k, then ℓg < 2k, a contradiction with the value of ℓM+1. If g = 2k, then ℓg = 2k. So, g = M + 1 = 2k − 1 = g − 1, a contradiction. Therefore, g > 2k+1. It is clear that g 6= 3k− 1, because if g = 3k−1, then ℓg = 4k−2 ∈ Gaps(H), a contradiction, since 2k−1 ∈ H . Now, suppose that H has genus g = 3k + s, for some integer s > 2. So, ℓg = 4k+2s and follows that ℓg/2 = 2k+ s ∈ Gaps(H). Since 2k−1 ∈ H , we have that ℓ := 2k + 2s+ 1 ∈ Gaps(H). We prove that 2k + 2s ∈ H. In fact, suppose that 2k + 2s ∈ Gaps(H). Thus, (ℓ − 1, ℓ) is a single leap. If ℓ2k = 2k + 1, then (2k, 2k + 1) is a single leap. Note that, if ℓ2k = 2k+2, then 2k+1 ∈ H and so 2k+2s−1 ∈ Gaps(H) (since s > 2), or equivalently, (ℓ − 2, ℓ − 1) is a single leap. On the other hand, since SM = 2, by Remark 2.7, follows that (2k + 2s+ 1) ≡ (4k + 2s) (mod 2), a contradiction. Therefore, 2k + 2s ∈ H. Then, 2k + 2s − 1 ∈ Gaps(H) and 4k+2s− 1 = (2k− 1)+ (2k+2s) ∈ H . Thus, ℓg−1 = 4k+2s− 2 and follows that 2k+ s− 1 ∈ Gaps(H). So, (2k+ s− 1, 2k+ s) is a single leap. Therefore, since 2k + 2s+ 1 is odd and 4k + 2s− 2 is even, we have that ∣ ∣V[2k+2s+1,4k+2s−2] ∩ V1(H) ∣ ∣ = 1. We conclude the proof by studying the following cases separately: s = 2 and s > 3. Firstly, assume that s = 2. In this case, (2k + 1, 2k + 2) and (2k + 2, 2k + 3) are both single leaps. So, SM = 3, a contradiction. Now, suppose that s > 3. Then, 2k + s− 2 ∈ H and consequently 4k+2s− 4 ∈ H . So, ℓg−2 = 4k+2s− 3, a contradiction, since 4k + 2s− 3 = (2k − 1) + (2k + 2s− 2) ∈ H. Theorem 2.14. For all integer k > 3, H k 2k−2 = { H (s,αs) (k,r) : (r, s, αs) ∈ N 3, 1 6 r 6 k − 2, 1 6 s 6 r, 1 6 αs 6 r − s+ 1 } ∪ { H (1,k) (k,r) : r ∈ {k, k + 1} } . “adm-n1” — 2019/3/22 — 12:03 — page 112 — #120 112 On certain families of sparse numerical semigroups Proof. Firstly, by the Lemma 2.11, we have that { H (s,αs) (k,r) : (r, s, αs) ∈ N 3, 1 6 r 6 k− 2, 1 6 s 6 r, 1 6 αs 6 r− s+1 } ∪ { H (1,k) (k,r) : r ∈ {k, k + 1} } ⊂ Hk 2k−2. Now, let H ∈ Hk 2k−2 with g(H) = g. By Theorem 2.13, we have 2k + 1 6 g 6 3k + 1 and g 6= 3k − 1. Let r := g − 2k, then 1 6 r 6 k − 2 or r ∈ {k, k + 1}. Since ℓg = 2g − 2k, we have that ℓg = ℓ2k+r = 2k + 2r. Let Gaps(H) = {1 = ℓ1 < · · · < ℓg} be the gaps set of H. By Corollary 1.3, SM = 2. That is, there exists exactly two single leaps (ℓi−1, ℓi) and (ℓj−1, ℓj), for some integers i and j such that 2k 6 i < j 6 g = 2k + r. We affirm that if r ∈ {k, k + 1}, then i = 2k and j = 3k. In fact, first we will prove that (2k, 2k + 1) is a single leap. Indeed, since M = 2k − 2, we have that ℓ2k−2 = 2k − 2 and ℓ2k−1 = 2k. In particular, 2k − 1 ∈ H. Suppose that 2k + 1 ∈ H. Thus, we have the following: if r = k, then ℓg = 4k = (2k − 1) + (2k + 1) ∈ H; and if r = k + 1, then ℓg = 4k + 2 = 2(2k + 1) ∈ H. Therefore, in both cases, we get a contradiction. Thus, (2k, 2k + 1) is a single leap and so i = 2k. Now, we will prove that j = 3k. Since, 4k − 2 = 2(2k − 1) ∈ H, follows that 4k − 1 is a gap. Then, by Remark 2.7, 2k + 2 ∈ H, since (2k + 2) 6≡ (4k − 1) (mod 2). Thus, 4k + 1 = (2k − 1) + (2k + 2) ∈ H and so 4k is a gap. Consequently, (4k − 1, 4k) is a single leap. Note that, if r = k, then ℓ3k = ℓg = 4k, and if r = k + 1, then ℓ3k = ℓg−1 = 4k. Therefore, in both cases, j = 3k. For r ∈ {1, . . . , k−2}∪{k, k+1}, let s := i−2k+1 and αs := j−i. Note that if r ∈ {1, . . . , k− 2}, then 1 6 s 6 r and 1 6 αs = (j− 2k)− s+1 6 r − s+ 1, and if r ∈ {k, k + 1}, then s = 1 and αs = k. We will prove that H = H (s,αs) (k,r) , with s and αs as above. In fact, by Remark 2.7, we have that ℓu = ℓi−1 − 2(i− u− 1), for all integer u such that 2k − 1 6 u 6 i − 1, ℓv = ℓi + 2(v − i), for all integer v such that i 6 v 6 j−1, and ℓw = ℓj+2(w−j), for all integer w such that j 6 w 6 g. Since ℓg = 2g−2k, it is not hard conclude that ℓj = 2(j−k) and, therefore, ℓj−1 = 2(j−k)−1. In the same way we can conclude that ℓi = 2(i−k)+1 and ℓi−1 = 2(i− k), and that ℓu = 2(u− k+1), for all integer u such that 2k − 1 6 u 6 i− 1 = 2k + s− 2, ℓv = 2(v − k) + 1, for all integer v such that 2k + s− 1 = i 6 v 6 j − 1 = 2k + s+ αs − 2 and ℓw = 2(w − k), for all integer w such that 2k + s+ αs − 1 = j 6 w 6 g = 2k + r. Therefore, by Remark 2.12, we have that H = H (s,αs) (k,r) . Corollary 2.15. For all integer k > 3, ∣ ∣Hk 2k−2 ∣ ∣ = ( k 3 ) + 2. “adm-n1” — 2019/3/22 — 12:03 — page 113 — #121 G. Tizziotti, J. Villanueva 113 Proof. By the Theorem 2.14, we have that H k 2k−2 = k−2 ⋃ · r=1 r ⋃ · s=1 { H (s,αs) (k,r) : αs ∈ N, 1 6 αs 6 r − s+ 1 } ∪· { H (1,k) (k,r) : r ∈ {k, k + 1} } . Therefore, ∣ ∣ ∣ H k 2k−2 ∣ ∣ ∣ = k−2 ∑ r=1 r ∑ s=1 (r − s+ 1) + 2 = k−2 ∑ r=1 ( r + 1 2 ) + 2 = ( k 3 ) + 2. 3. On Weierstrass semigroup Let X be a non-singular, projective, irreducible, algebraic curve of genus g > 1 over a field K. Let K(X ) be the field of rational functions on X and for f ∈ K(X ), (f)∞ will denote the divisor of poles of f . Let P be a point on X . The set H(P ) := {n ∈ N0 : there exist f ∈ K(X ) with (f)∞ = nP}, is a numerical semigroup called Weierstrass semigroup of X at P . Given a numerical semigroup H , does it exist a curve X such that for some point P ∈ X has H = H(P )? If the answer is yes, we say that the numerical semigroup H is Weierstrass. Studies to answer this question have been done for decades, see e.g. [4], [8], [10], [11], [12], [13], [15] and [19]. In addition to the genus g(H), Frobenius number ℓg(H) and multiplicity n1(H), an important concept in this study is the weight of a numerical semigroup H. If Gaps(H) = {1 = ℓ1 < · · · < ℓg} be the gaps set of H, the weight of H is w(H) = g ∑ i=1 (ℓi − i). As a particular result, a numerical semigroup H is Weierstrass if the following condition hold: either w(H) 6 g(H)/2, or g(H)/2 < w(H) 6 g(H)− 1 and 2n1(H) > ℓg(H) (∗) (see Eisenbud-Harris [7], Komeda [9]). Next, we will see which of the semigroups in the families studied in the previous section are Weierstrass. “adm-n1” — 2019/3/22 — 12:03 — page 114 — #122 114 On certain families of sparse numerical semigroups Lemma 3.1. Let k > 2 be an integer. If H = Hαr (k,r) ∈ Hk 2k−1, then w(H) = αr + r(r+1) 2 and 2n1(H) > ℓg(H). Proof. Since H = Hαr (k,r) ∈ Hk 2k−1, by Remark 2.6, we have that w(H) = 2k−1+αr ∑ i=2k (i−2k+1)+ 2k+r ∑ i=2k+αr (i−2k) = r ∑ i=1 i+αr = r(r + 1) 2 +αr. On the other hand, by Lemma 2.5, ℓg(H) = 2k + 2r 6 4k − 2 < 4k = 2(2k) = 2n1(H), since r 6 k − 1. Proposition 3.2. Let k > 2 be an integer and H = Hαr (k,r) ∈ Hk 2k−1. If αr + r(r−1) 2 6 2k − 1, then H is Weierstrass. In particular, if k > 3 and r ∈ { 1, . . . , ⌊ −1+ √ 16k−7 2 ⌋} , then H is Weierstrass. Proof. By previous lemma, 2n1(H) > ℓg(H). So, from the condition (∗) above, follows that H is Weierstrass if w(H) 6 g(H)−1. On the other hand, by Lemma 2.5, g(H) = 2k + r, and, by Lemma 3.1, w(H) = αr + r(r+1) 2 . Thus, H is Weierstrass if αr + r(r+1) 2 6 2k+ r− 1, that is, if αr + r(r−1) 2 6 2k − 1. Now, if r ∈ { 1, . . . , ⌊ −1+ √ 16k−7 2 ⌋} , then r + r(r−1) 2 6 2k − 1. Also, since k > 3, we have that r 6 k − 1. So, since αr 6 r, the required result follows. Lemma 3.3. Let k > 3 be an integer. If H = H (s,αs) (k,r) ∈ Hk 2k−2, then w(H) = 2s−1+αs+ r(r+1) 2 . In addition, if r 6∈ {k, k+1}, then 2n1(H) > ℓg(H). Proof. Since H = H (s,αs) (k,r) ∈ Hk 2k−2, by Remark 2.12, we have that w(H) = 2k+s−2 ∑ i=2k−1 (i− 2k + 2) + 2k+s+αs−2 ∑ i=2k+s−1 (i− 2k + 1) + 2k+r ∑ i=2k+s+αs−1 (i− 2k) = r ∑ i=−1 i+ 2s+ αs = −1 + r(r + 1) 2 + 2s+ αs. On the other hand, by Lemma 2.11, ℓg(H) = 2k + 2r 6 4k − 2 = 2(2k − 1) = 2n1(H), since r 6 k − 2. “adm-n1” — 2019/3/22 — 12:03 — page 115 — #123 G. Tizziotti, J. Villanueva 115 Proposition 3.4. Let k > 3 be an integer and H = H (s,αs) (k,r) ∈ Hk 2k−2. If r 6∈ {k, k + 1} and 2s + αs + r(r−1) 2 6 2k, then H is Weierstrass. In particular, if k > 4 and r ∈ { 1, . . . , ⌊ −3+ √ 16k+1 2 ⌋} , then H is Weierstrass. Proof. Using the lemmas 2.11 and 3.3, the proof of the first statement is analogous to the proof of the Proposition 3.2. Now, if r ∈ { 1, . . . , ⌊ −3+ √ 16k+1 2 ⌋} , then 2r + r(r−1) 2 6 2k − 1. Also, since k > 4, we have that r 6 k − 2. So, since s 6 r and αs 6 r − s+ 1, the required result follows. We observe that, for r ∈ {k, k+ 1}, the condition (∗) above cannot be used to conclude that the semigroups H (1,k) (k,r) ’s are Weierstrass. References [1] C. Arf, Une interpretation algébrique de la suite des ordres de multiplicité dúne branche algébrique, Proc. London Math. Soc., 50 (1949), 256–287. [2] V. Barucci, D. E. Dobbs and M. Fontana, Maximality properties in numerical semi- groups and applications to one-dimensional analytically irreducible local domains, Mem. Amer. Math. Soc. 125, 1997. [3] M. Bras-Amorós, Acute semigroups, the order bound on the minimum distance, and the Feng-Rao improvement, IEEE Trans. Inform. Theory 50 (6), (2004), 1282–1289. [4] R.O. Buchweitz On Zariski’s criterion for equisingularity and nonsmoothable monomial curves, These, Paris VII, 1981. [5] A. Campillo, J. I. Farrán and C. Munuera, On the parameters of algebraic-geometry codes related to Arf semigroups, IEEE Trans. Inform. Theory 46 (7) (2000), 2634– 2638. [6] A. Contiero, C. G. T. A. Moreira and P. M. Veloso, On the structure of numerical sparse semigroups and applications to Weierstrass points, Journal of Pure and Applied Algebra 219 (2015), 3946-–3957. [7] D. Eisenbud and J. Harris, Existence, decomposition and limits of certain Weier- strass points, Invent. Math. 87 (1987), 495-515. [8] J. Komeda, On Weierstrass points whose first non-gaps are four, J. Reine Angew. Math. 341 (1983), 237-270. [9] J. Komeda, On primitive Schubert indices of genus g and weight g − 1, J. Math. Soc. Japan 43(3) (1991), 437-445. [10] J. Komeda, On the existence of Weierstrass points whose first non-gaps are five, Manuscripta Math. 76(2) (1992), 193-211. [11] J. Komeda, On the existence of Weierstrass gap sequences on curves of genus 6 8, J. Pure Appl. Algebra 97 (1994), 51-71. “adm-n1” — 2019/3/22 — 12:03 — page 116 — #124 116 On certain families of sparse numerical semigroups [12] J. Komeda, Double coverings of curves and non-Weierstrass semigroups, Commun. Algebra 41, (2013), 312-324. [13] C. Maclachlan, Weierstrass points on compact Riemann surfaces, J. London Math. Soc. (2) 3 (1971), 722-724. [14] C. Munuera, F. Torres and J. Villanueva, Sparse Numerical Semigroups, Lecture Notes in Computer Science: Applied Algebra, Algebraic Algorithms and Error- Correcting Codes, 5527, 23–31, Springer-Verlag Berlin Heilderberg (2009). [15] G. Oliveira, Weierstrass semigroups and the canonical ideal of non-trigonal curves, Manuscripta Math., 71 (1991), 431–450. [16] J. C. Rosales, P. A. García-Sánchez, J. I. García-García and M. B. Branco, Arf Numerical Semigroups, Journal of Algebra 276 (2004), 3–12. [17] J. C. Rosales and P. A. García-Sánchez, Numerical Semigroups, Developments in Mathematics, Springer-Verlag New York Inc., United States (2012). [18] G. Tizziotti and J. Villanueva, On κ-sparse numerical semigroups, Journal of Algebra and Its Applications, 17 (11), 2018, 1850209-1–1850209-13. [19] F. Torres, Weierstrass points and double coverings of curves with applications: Sym- metric numerical semigroups which cannot be realized as Weierstrass semigroups, Manuscripta Math. 83 (1994), 39–58. [20] J. Villanueva, “Semigrupos fracamente de Arf e pesos de semigrupos", Ph. D. Thesis, UNICAMP, 2008. Contact information G. Tizziotti Universidade Federal de Uberlândia, Câmpus Santa Mônica, Faculdade de Matemática, Av. João Naves de Ávila 2.160, Santa Mônica, 38.408-100, Uberlândia - MG, Brasil E-Mail(s): guilhermect@ufu.br J. Villanueva Universidade Federal de Mato Grosso, Câmpus Universitário do Araguaia, Instituto de Ciências Exatas e da Terra, Av. Senador Valdon Varjão 6.390, Setor Industrial, 78.600-000, Barra do Garças - MT, Brasil E-Mail(s): vz_juan@yahoo.com.br Received by the editors: 20.07.2017 and in final form 22.11.2017.