Counting numerical semigroups of a given genus by using - - PowerPoint PPT Presentation

counting numerical semigroups of a given genus by using
SMART_READER_LITE
LIVE PREVIEW

Counting numerical semigroups of a given genus by using - - PowerPoint PPT Presentation

Counting numerical semigroups of a given genus by using -hyperelliptic semigroups Matheus Bernardini 1 University of Campinas / Federal Institute of S ao Paulo (Joint work in progress with Fernando Torres 1 ) International Meeting on


slide-1
SLIDE 1

Counting numerical semigroups of a given genus by using γ-hyperelliptic semigroups

Matheus Bernardini1 University of Campinas / Federal Institute of S˜ ao Paulo

(Joint work in progress with Fernando Torres1)

International Meeting on Numerical Semigroups with Applications Levico Terme, July 5 2016

1Both authors are partially supported by CNPq Matheus Bernardini IMNS 2016 1 / 34

slide-2
SLIDE 2

1 Introduction 2 A brief survey about counting numerical semigroups by genus 3 Our approach 4 A related problem Matheus Bernardini IMNS 2016 2 / 34

slide-3
SLIDE 3

Introduction

Let S be a numerical semigroup. G(S) := N0 \ S - set of gaps of S; g(S) := #G(S) - genus of S; ng := #{S : g(S) = g}.

Examples

n0 = 1 − N0 n1 = 1 − N0 \ {1} n2 = 2 − N0 \ {1, 2} and N0 \ {1, 3} n3 = 4 − N0 \ {1, 2, 3}, N0 \ {1, 2, 4}, N0 \ {1, 2, 5} and N0 \ {1, 3, 5}

Matheus Bernardini IMNS 2016 3 / 34

slide-4
SLIDE 4

Interest

Studying the behavior of ng.

Main Goal (but still not solved)

ng ≤ ng+1, for all g.

Matheus Bernardini IMNS 2016 4 / 34

slide-5
SLIDE 5

Interest

Studying the behavior of ng.

Main Goal (but still not solved)

ng ≤ ng+1, for all g.

Matheus Bernardini IMNS 2016 4 / 34

slide-6
SLIDE 6

A brief survey

First Bound

If g(S) = g, then 2g + N0 ⊂ S. Hence, ng ≤ 2g − 1 g

  • .
  • M. Bras-Amor´
  • s and A. de Mier - 2007

ng ≤ Cg = 1 g + 1 2g g

  • .

Matheus Bernardini IMNS 2016 5 / 34

slide-7
SLIDE 7

A brief survey

First Bound

If g(S) = g, then 2g + N0 ⊂ S. Hence, ng ≤ 2g − 1 g

  • .
  • M. Bras-Amor´
  • s and A. de Mier - 2007

ng ≤ Cg = 1 g + 1 2g g

  • .

Matheus Bernardini IMNS 2016 5 / 34

slide-8
SLIDE 8

From now on, let ϕ = 1+

√ 5 2

.

  • M. Bras-Amor´
  • s - 2006/2008 (Conjecture)

1

lim

g→∞

ng+1 ng = ϕ; lim

g→∞

ng+1 + ng ng+2 = 1.

2 ng + ng+1 ≤ ng+2, for all g. Matheus Bernardini IMNS 2016 6 / 34

slide-9
SLIDE 9
  • M. Bras-Amor´
  • s - 2009

Let (Fn)n≥0 = (1, 1, 2, 3, 5, 8, 13, . . .) be the Fibonacci sequence. Then 2Fg ≤ ng ≤ 1 + 3 · 2g−3, ∀g ≥ 3.

  • S. Elizalde - 2010

ag ≤ ng ≤ cg, ∀g ≥ 1 where ag and cg are coefficients of some explicit generating functions.

Matheus Bernardini IMNS 2016 7 / 34

slide-10
SLIDE 10
  • M. Bras-Amor´
  • s - 2009

Let (Fn)n≥0 = (1, 1, 2, 3, 5, 8, 13, . . .) be the Fibonacci sequence. Then 2Fg ≤ ng ≤ 1 + 3 · 2g−3, ∀g ≥ 3.

  • S. Elizalde - 2010

ag ≤ ng ≤ cg, ∀g ≥ 1 where ag and cg are coefficients of some explicit generating functions.

Matheus Bernardini IMNS 2016 7 / 34

slide-11
SLIDE 11
  • A. Zhai - 2011/2013

1

lim

g→∞

ng+1 ng = ϕ;

2

lim

g→∞

ng+1 + ng ng+2 = 1.

Remark

Zhai’s first item implies that ng < ng+1, for g ≫ 0. Checking if ng ≤ ng+1 for all g is still an open problem (weaker conjecture).

Matheus Bernardini IMNS 2016 8 / 34

slide-12
SLIDE 12
  • A. Zhai - 2011/2013

1

lim

g→∞

ng+1 ng = ϕ;

2

lim

g→∞

ng+1 + ng ng+2 = 1.

Remark

Zhai’s first item implies that ng < ng+1, for g ≫ 0. Checking if ng ≤ ng+1 for all g is still an open problem (weaker conjecture).

Matheus Bernardini IMNS 2016 8 / 34

slide-13
SLIDE 13

m(S) := min{s ∈ S : s = 0} - multiplicity of S; N(m, g) := #{S : g(S) = g and m(S) = m}.

  • N. Kaplan - 2012

If 2g < 3m, then N(m, g) = N(m − 1, g − 1) + N(m − 1, g − 2).

Matheus Bernardini IMNS 2016 9 / 34

slide-14
SLIDE 14

Matheus Bernardini IMNS 2016 10 / 34

slide-15
SLIDE 15

Part of M. Bras-Amor´

  • s’ presentation in last IMNS;

r(S) - ordinarization number of S; ng,r := #{S : g(S) = g and r(S) = r}.

Matheus Bernardini IMNS 2016 11 / 34

slide-16
SLIDE 16

Part of M. Bras-Amor´

  • s’ presentation in last IMNS;

r(S) - ordinarization number of S; ng,r := #{S : g(S) = g and r(S) = r}.

Matheus Bernardini IMNS 2016 11 / 34

slide-17
SLIDE 17
  • M. Bras-Amor´
  • s - 2012

If r > max{g

3 + 1,

  • g+1

2

  • − 14}, then ng,r ≤ ng+1,r.
  • M. Bras-Amor´
  • s - 2012 (Conjecture)

If r > g

3, then ng,r ≤ ng+1,r.

Matheus Bernardini IMNS 2016 12 / 34

slide-18
SLIDE 18
  • M. Bras-Amor´
  • s - 2012

If r > max{g

3 + 1,

  • g+1

2

  • − 14}, then ng,r ≤ ng+1,r.
  • M. Bras-Amor´
  • s - 2012 (Conjecture)

If r > g

3, then ng,r ≤ ng+1,r.

Matheus Bernardini IMNS 2016 12 / 34

slide-19
SLIDE 19

Our approach

γ(S): number of even gaps of S - #[G(S) ∩ 2Z]; γ-hyperelliptic semigroup: numerical semigroup with γ even gaps; Nγ(g) := #{S : g(S) = g and γ(S) = γ}. ng =

g

  • γ=0

Nγ(g)

Matheus Bernardini IMNS 2016 13 / 34

slide-20
SLIDE 20

Our approach

γ(S): number of even gaps of S - #[G(S) ∩ 2Z]; γ-hyperelliptic semigroup: numerical semigroup with γ even gaps; Nγ(g) := #{S : g(S) = g and γ(S) = γ}. ng =

g

  • γ=0

Nγ(g)

Matheus Bernardini IMNS 2016 13 / 34

slide-21
SLIDE 21

Examples

n0 = 1 (N0) and Nγ(0) =

  • 1, if γ = 0

0, if γ ≥ 1. n1 = 1 (N0 \ {1}) and Nγ(1) =

  • 1, if γ = 0

0, if γ ≥ 1. n2 = 2 (N0 \ {1, 2} and N0 \ {1, 3}) and Nγ(2) =      1, if γ = 0 1, if γ = 1 0, if γ ≥ 2.

Matheus Bernardini IMNS 2016 14 / 34

slide-22
SLIDE 22
  • F. Torres - 1997

If γ and g are the number of even gaps and the genus of a numerical semigroup S, respectively, then 3γ ≤ 2g.

Remark

If γ is even, then N0 \ ({2, 4, . . . , 2γ} ∪ {1, 3, . . . , γ − 1}) is a numerical semigroup with genus g = 3γ

2 .

ng = ⌊ 2g

3 ⌋

  • γ=0

Nγ(g)

Matheus Bernardini IMNS 2016 15 / 34

slide-23
SLIDE 23
  • F. Torres - 1997

If γ and g are the number of even gaps and the genus of a numerical semigroup S, respectively, then 3γ ≤ 2g.

Remark

If γ is even, then N0 \ ({2, 4, . . . , 2γ} ∪ {1, 3, . . . , γ − 1}) is a numerical semigroup with genus g = 3γ

2 .

ng = ⌊ 2g

3 ⌋

  • γ=0

Nγ(g)

Matheus Bernardini IMNS 2016 15 / 34

slide-24
SLIDE 24

Theorem 1

Let γ be a nonnegative integer and g ≥ 3γ. Then Nγ(g) = Nγ(3γ). Thus, Nγ(g) = Nγ(g + 1), for all g ≥ 3γ.

Theorem 2

Let γ be a nonnegative integer and g < 3γ. Then Nγ(g) < Nγ(3γ).

Matheus Bernardini IMNS 2016 16 / 34

slide-25
SLIDE 25

Theorem 1

Let γ be a nonnegative integer and g ≥ 3γ. Then Nγ(g) = Nγ(3γ). Thus, Nγ(g) = Nγ(g + 1), for all g ≥ 3γ.

Theorem 2

Let γ be a nonnegative integer and g < 3γ. Then Nγ(g) < Nγ(3γ).

Matheus Bernardini IMNS 2016 16 / 34

slide-26
SLIDE 26

g / γ 1 2 3 4 5 6 7 8 9 10 ng 1 1 1 1 1 2 1 1 2 3 1 2 1 4 4 1 2 4 7 5 1 2 6 3 12 6 1 2 7 12 1 23 7 1 2 7 19 10 39 8 1 2 7 21 32 4 67 9 1 2 7 23 51 33 1 118 10 1 2 7 23 62 91 18 204 11 1 2 7 23 65 142 98 5 343 12 1 2 7 23 68 174 257 59 1 592 13 1 2 7 23 68 192 412 271 25 1001 14 1 2 7 23 68 197 514 678 197 6 1693 15 1 2 7 23 68 200 570 1100 793 92 1 2857

Matheus Bernardini IMNS 2016 17 / 34

slide-27
SLIDE 27

Notice that ng = ⌊ g

3⌋

  • γ=0

Nγ(g) + ⌊ 2g

3 ⌋

  • γ=⌊ g

3⌋+1

Nγ(g). ng+1 = ⌊ g

3⌋

  • γ=0

Nγ(g + 1) +

2(g+1)

3

  • γ=⌊ g

3⌋+1

Nγ(g + 1). Theorem 1 states that Nγ(g) = Nγ(g + 1), for γ ≤ g

3.

Matheus Bernardini IMNS 2016 18 / 34

slide-28
SLIDE 28

Notice that ng = ⌊ g

3⌋

  • γ=0

Nγ(g) + ⌊ 2g

3 ⌋

  • γ=⌊ g

3⌋+1

Nγ(g). ng+1 = ⌊ g

3⌋

  • γ=0

Nγ(g + 1) +

2(g+1)

3

  • γ=⌊ g

3⌋+1

Nγ(g + 1). Theorem 1 states that Nγ(g) = Nγ(g + 1), for γ ≤ g

3.

Matheus Bernardini IMNS 2016 18 / 34

slide-29
SLIDE 29

Notice that ng = ⌊ g

3⌋

  • γ=0

Nγ(g) + ⌊ 2g

3 ⌋

  • γ=⌊ g

3⌋+1

Nγ(g). ng+1 = ⌊ g

3⌋

  • γ=0

Nγ(g + 1) +

2(g+1)

3

  • γ=⌊ g

3⌋+1

Nγ(g + 1). Theorem 1 states that Nγ(g) = Nγ(g + 1), for γ ≤ g

3.

Matheus Bernardini IMNS 2016 18 / 34

slide-30
SLIDE 30

Corollary

ng ≤ ng+1 if, and only if, ⌊ 2g

3 ⌋

  • γ=⌊ g

3⌋+1

Nγ(g) ≤

2(g+1)

3

  • γ=⌊ g

3⌋+1

Nγ(g + 1).

Matheus Bernardini IMNS 2016 19 / 34

slide-31
SLIDE 31

g / γ 1 2 3 4 5 6 7 8 9 10 ng 1 1 1 1 1 2 1 1 2 3 1 2 1 4 4 1 2 4 7 5 1 2 6 3 12 6 1 2 7 12 1 23 7 1 2 7 19 10 39 8 1 2 7 21 32 4 67 9 1 2 7 23 51 33 1 118 10 1 2 7 23 62 91 18 204 11 1 2 7 23 65 142 98 5 343 12 1 2 7 23 68 174 257 59 1 592 13 1 2 7 23 68 192 412 271 25 1001 14 1 2 7 23 68 197 514 678 197 6 1693 15 1 2 7 23 68 200 570 1100 793 92 1 2857

Matheus Bernardini IMNS 2016 20 / 34

slide-32
SLIDE 32

Conjecture

Let γ be a non-negative integer. Then Nγ(g) ≤ Nγ(g + 1), ∀g.

Remark

If this Conjecture holds, then ng ≤ ng+1 for all g.

Matheus Bernardini IMNS 2016 21 / 34

slide-33
SLIDE 33

Conjecture

Let γ be a non-negative integer. Then Nγ(g) ≤ Nγ(g + 1), ∀g.

Remark

If this Conjecture holds, then ng ≤ ng+1 for all g.

Matheus Bernardini IMNS 2016 21 / 34

slide-34
SLIDE 34

Construction of a γ-hyperelliptic semigroup of genus g

γ - positive integer T = N0 \ {q1, . . . , qγ} numerical semigroup S = 2 · T ∪ (2 · N0 + 1) \ {“suitable choice”of g − γ odd numbers}

Matheus Bernardini IMNS 2016 22 / 34

slide-35
SLIDE 35

“suitable choice”ensures that the final set is closed under addition. S is a γ-hyperelliptic semigroup: even gaps and even non-gaps are determined by T. S has genus g if, and only if, the number of green points choosen as gaps is g − γ − k.

Lemma

Let S be a γ-hyperelliptic semigroup of genus g and O the first odd number in S. Then 2g − 4γ + 1 ≤ O ≤ 2g − 2γ + 1.

Matheus Bernardini IMNS 2016 23 / 34

slide-36
SLIDE 36

“suitable choice”ensures that the final set is closed under addition. S is a γ-hyperelliptic semigroup: even gaps and even non-gaps are determined by T. S has genus g if, and only if, the number of green points choosen as gaps is g − γ − k.

Lemma

Let S be a γ-hyperelliptic semigroup of genus g and O the first odd number in S. Then 2g − 4γ + 1 ≤ O ≤ 2g − 2γ + 1.

Matheus Bernardini IMNS 2016 23 / 34

slide-37
SLIDE 37

Proof (Thm 1)

Numbers qi must be gaps! (otherwise, S ∋ qi + qi = 2qi / ∈ S) g ≥ 3γ ⇒

  • O≥2g−4γ+1

O ≥ 2γ + 1. Hence,

O > qγ ≥ qi, for all i sum of odd elements of S is greater than 4γ + 2.

Matheus Bernardini IMNS 2016 24 / 34

slide-38
SLIDE 38

Proof (Thm 1)

Numbers qi must be gaps! (otherwise, S ∋ qi + qi = 2qi / ∈ S) g ≥ 3γ ⇒

  • O≥2g−4γ+1

O ≥ 2γ + 1. Hence,

O > qγ ≥ qi, for all i sum of odd elements of S is greater than 4γ + 2.

Matheus Bernardini IMNS 2016 24 / 34

slide-39
SLIDE 39

Proof (Thm 1)

Sγ(g) := {S : g(S) = g and γ(S) = γ} For a fixed g ≥ 3γ, we find a bijection between Sγ(g) and Sγ(3γ)

Matheus Bernardini IMNS 2016 25 / 34

slide-40
SLIDE 40

Proof (Thm 1)

Given S(g) ∈ Sγ(g), let O(g) be the first odd number of S(g) Let M := g − 3γ. Making a translation by −2M only on the odd numbers higher than or equal to O(g), we obtain a NS S, such that O(3γ) = O(g) − 2M ≥ 2γ + 1 (and this is the first odd number of S) The even gaps of S(g) and S are the same, as the odd gaps of S(g) and S lower than O(3γ). The odd gaps of S(g) and S higher than O(3γ) are translated by −2M Under this construction, we have g(S) = g − M = 3γ. Hence, S ∈ Sγ(3γ) and #Sγ(g) ≤ #Sγ(3γ) Similarly (by making a translation by +2M), we can verify the other inequality and the result follows.

Matheus Bernardini IMNS 2016 26 / 34

slide-41
SLIDE 41

A related problem

γ non-negative integer For g ≥ 3γ, the sequence Nγ(g) is constant and equal to Nγ(3γ) A natural task is about the behavior of fγ := Nγ(3γ)

Lemma

Let γ be a non-negative integer and Mγ := 2γ γ

2 + 1

  • − 1. Then

Mγ + (nγ − γ) · (γ + 1) ≤ fγ ≤ Mγ + (nγ − γ) · 2γ.

Matheus Bernardini IMNS 2016 27 / 34

slide-42
SLIDE 42

A related problem

γ non-negative integer For g ≥ 3γ, the sequence Nγ(g) is constant and equal to Nγ(3γ) A natural task is about the behavior of fγ := Nγ(3γ)

Lemma

Let γ be a non-negative integer and Mγ := 2γ γ

2 + 1

  • − 1. Then

Mγ + (nγ − γ) · (γ + 1) ≤ fγ ≤ Mγ + (nγ − γ) · 2γ.

Matheus Bernardini IMNS 2016 27 / 34

slide-43
SLIDE 43

Theorem 3

Let ǫ > 0. Then lim

γ→∞

fγ (2ϕ + ǫ)γ = 0 and lim

γ→∞

fγ 2γ = ∞. It suggests that the asymptotic behavior of fγ is exponencial of order βγ, where 2 < β ≤ 2ϕ.

Matheus Bernardini IMNS 2016 28 / 34

slide-44
SLIDE 44

Theorem 3

Let ǫ > 0. Then lim

γ→∞

fγ (2ϕ + ǫ)γ = 0 and lim

γ→∞

fγ 2γ = ∞. It suggests that the asymptotic behavior of fγ is exponencial of order βγ, where 2 < β ≤ 2ϕ.

Matheus Bernardini IMNS 2016 28 / 34

slide-45
SLIDE 45

g / γ 1 2 3 4 5 6 7 8 9 10 ng 1 1 1 1 1 2 1 1 2 3 1 2 1 4 4 1 2 4 7 5 1 2 6 3 12 6 1 2 7 12 1 23 7 1 2 7 19 10 39 8 1 2 7 21 32 4 67 9 1 2 7 23 51 33 1 118 10 1 2 7 23 62 91 18 204 11 1 2 7 23 65 142 98 5 343 12 1 2 7 23 68 174 257 59 1 592 13 1 2 7 23 68 192 412 271 25 1001 14 1 2 7 23 68 197 514 678 197 6 1693 15 1 2 7 23 68 200 570 1100 793 92 1 2857

Matheus Bernardini IMNS 2016 29 / 34

slide-46
SLIDE 46

γ fγ fγ/fγ−1 n2γ fγ/n2γ 1 1 1 1 2 2 2 1 2 7 3,5 7 1 3 23 3,285714 23 1 4 68 2,956522 67 1,015 5 200 2,941176 204 0,981 6 615 3,075 592 1,039 7 1764 2,868292 1693 1,042 8 5060 2,868480 4806 1,053 9 14626 2,890514 13467 1,086 10 41785 2,856899 37396 1,117 11 117573 2,813761 103246 1,139 12 332475 2,827818 282828 1,176 13 933891 2,808905 770832 1,212 14 2609832 2,794579 2091030 1,248

Matheus Bernardini IMNS 2016 30 / 34

slide-47
SLIDE 47

Conjecture

lim

γ→∞

fγ fγ−1 = ϕ2 ≈ 2, 618 and lim

γ→∞

fγ n2γ = C, where C is a constant.

Remark

Firts item is a consequence of second item.

Remark

There is a relation between the sequence fγ and the conjecture proposed by M. Bras-Amor´

  • s (12). In fact, if fγ is an increasing sequence, then the

conjecture is also true.

Matheus Bernardini IMNS 2016 31 / 34

slide-48
SLIDE 48

Conjecture

lim

γ→∞

fγ fγ−1 = ϕ2 ≈ 2, 618 and lim

γ→∞

fγ n2γ = C, where C is a constant.

Remark

Firts item is a consequence of second item.

Remark

There is a relation between the sequence fγ and the conjecture proposed by M. Bras-Amor´

  • s (12). In fact, if fγ is an increasing sequence, then the

conjecture is also true.

Matheus Bernardini IMNS 2016 31 / 34

slide-49
SLIDE 49

References I

  • M. Bernardini and F. Torres, Counting numerical semigroups of a

given genus via even gaps, In progress.

  • M. Bras-Amor´
  • s, Bounds on the number of numerical semigroups of a

given genus, Journal of Pure and Applied Algebra 213:6 (2009), 997–1001.

  • M. Bras-Amor´
  • s, Fibonacci-like behavior of the number of numerical

semigroups of a given genus, Semigroup Forum 76 (2008), 379–384.

  • M. Bras-Amor´
  • s, The ordinarization transform of a numerical

semigroup and semigroups with a large number of intervals, Journal of Pure and Applied Algebra 216 (2012), 2507–2518.

  • M. Bras-Amor´
  • s and A. de Mier, Representation of numerical

semigroups by Dyck Paths, Semigroup Forum 75 (2007), 676–681.

Matheus Bernardini IMNS 2016 32 / 34

slide-50
SLIDE 50

References II

  • S. Elizalde, Improved bounds on the number of numerical semigroups
  • f a given genus, Journal of Pure and Applied Algebra 214 (2010),

1862–1873.

  • N. Kaplan, Counting numerical semigroups by genus and some cases
  • f a question of Wilf, Journal of Pure and Applied Algebra 216

(2012), 1016–1032.

  • F. Torres, On γ-hyperelliptic numerical semigroups, Semigroup Forum

55 (1997), 364-379.

  • F. Torres, On n-sheeted Covering of Curves and Semigroups which

Cannot Be Realized as Weierstrass Semigroups, Communications in Algebra 23 (11) (1995), 4211-4228.

  • A. Zhai, Fibonacci-like growth of numerical semigroups of a given

genus, Semigroup Forum 86 (2013), 634–662.

Matheus Bernardini IMNS 2016 33 / 34

slide-51
SLIDE 51

Thank You!

Matheus Bernardini IMNS 2016 34 / 34