Random gluing polygons Sergei Chmutov joint work with Boris Pittel - - PowerPoint PPT Presentation

random gluing polygons
SMART_READER_LITE
LIVE PREVIEW

Random gluing polygons Sergei Chmutov joint work with Boris Pittel - - PowerPoint PPT Presentation

Random gluing polygons Sergei Chmutov joint work with Boris Pittel Ohio State University, Mansfield Stochastic Topology and Thermodynamic Limits Workshop at ICERM Wednesday, October 19, 2016 9:00 9:45 a.m. Sergei Chmutov Random gluing


slide-1
SLIDE 1

Random gluing polygons

Sergei Chmutov joint work with Boris Pittel

Ohio State University, Mansfield

Stochastic Topology and Thermodynamic Limits Workshop at ICERM Wednesday, October 19, 2016 9:00 — 9:45 a.m.

Sergei Chmutov Random gluing polygons

slide-2
SLIDE 2
  • Polygons. Notations.

n := # (oriented) polygons N := total (even) number of sides nj := # j-gons, nj = n, jnj = N [N] := {1, 2, . . . , N} α ∈ SN is a permutation of [N] cyclically permutes edges of polygons according to their orientations. Example.

6 3 2 1 4 5 7 8

α = (1234)(5678) nj equals the number of cycles of α of length j.

Sergei Chmutov Random gluing polygons

slide-3
SLIDE 3

Gluing polygons. Permutations.

β ∈ SN is an involution without fixed points; β has N/2 cycles of length 2.

e

1

β(e1) α ( β ( e1 ) ) = : e2 β(e

2

) α(β(e2))=:e3

γ := αβ

e1 e2 e3

Σα,β # vertices of Σα,β = # cycles of γ. # connected components of Σα,β = # orbits of the subgroup generated by α and β.

Sergei Chmutov Random gluing polygons

slide-4
SLIDE 4

Gluing polygons. Example.

n = 2, N = 8, α = (1234)(5678)

6 3 2 1 4 5 7 8 8 7 6 5 1 2 3 4 8 2 6 4 3 7 5 1

β = (15)(28)(37)(46) γ = (16)(25)(38)(47)

Sergei Chmutov Random gluing polygons

slide-5
SLIDE 5

Gluing polygons. Example.

n = 2, N = 8, α = (1234)(5678)

3 2 4 1 5 7 8 6 1 5 3 7 6 8 2 4

β = (15)(24)(37)(68) γ = (1652)(3874)

Sergei Chmutov Random gluing polygons

slide-6
SLIDE 6

Gluing polygons. Example.

n = 2, N = 8, α = (1234)(5678)

1 3 2 4 5 7 8 6 4 1 2 3 8 5 6 7

β = (13)(24)(57)(68) γ = (1432)(5876)

Sergei Chmutov Random gluing polygons

slide-7
SLIDE 7

Gluing polygons. Example.

n = 2, N = 8, α = (1234)(5678)

4 1 2 3 5 7 8 6 5 6 7 8 3 2 1 4

β = (13)(24)(56)(78) γ = (1432)(57)(6)(8)

Sergei Chmutov Random gluing polygons

slide-8
SLIDE 8

Gluing polygons. Example.

n = 2, N = 8, α = (1234)(5678)

8 4 5 6 7 5 7 8 6 3 2 1 1 2 3 4

β = (12)(34)(56)(78) γ = (13)(2)(4)(57)(6)(8)

Sergei Chmutov Random gluing polygons

slide-9
SLIDE 9

Gluing polygons. Example.

n = 2, N = 8, α = (1234)(5678) There are 7!! = 105 possibilities for choosing β. surface Σα,β S2 T 2 2T 2 T 2 + S2 2S2 # gluings 36 60 1 4 4

Sergei Chmutov Random gluing polygons

slide-10
SLIDE 10

Random gluing.

n := {nj} is a partition of n = nj. Let Cn be the conjugacy class of α, all permutations in SN with the cycle structure n. Let CN/2 be the conjugacy class of β, all permutations in SN with all cycles length 2. A random surface is the surface Σα,β obtained by gluing according to the permutations α and β that are independently chosen uniformly at random from the conjugacy classes Cn and CN/2 respectively.

Sergei Chmutov Random gluing polygons

slide-11
SLIDE 11

Harer-Zagier formula. n = 1.

n = 1, α = (123 . . . N). Example: N = 6

6 1 2 3 4 5 6 1 2 3 4 5 1 1 2 3 4 5 6 2 4 3 5 6

Sergei Chmutov Random gluing polygons

slide-12
SLIDE 12

Harer-Zagier formula. n = 1, N = 6.

n = 1, N = 6 Vn =# vertices of Σα,β. |CN/2| = 5!! = 15. Vn = 4 Vn = 2 Generating function: TN(y) :=

β

yVn. T6(y) = 5y4 + 10y2.

Sergei Chmutov Random gluing polygons

slide-13
SLIDE 13

Harer-Zagier formula.

  • J. Harer and D. Zagier, The Euler characteristic of the moduli

space of curves, Invent. Math. 85 (1986) 457–485. TN(y) :=

β

yVn. Generating function: T(x, y) := 1 + 2xy + 2x

  • k=1

T2k(y) (2k − 1)!!xk. T(x, y) = 1 + x 1 − x y —,B.Pittel, JCTA 120 (2013) 102–110: gN = genus of Σα,β. Asymptotically as N → ∞, gN is normal N((N − log N)/2, (log N)/4).

Sergei Chmutov Random gluing polygons

slide-14
SLIDE 14

Main result.

—,B.Pittel, On a surface formed by randomly gluing together polygonal discs, Advances in Applied Mathematics, 73 (2016) 23–42. Vn =# vertices of Σα,β.

  • Theorem. Vn is asymptotically normal with mean and variance

log N both, Vn ∼ N(log N, log N), as N → ∞, and uniformly on n.

Sergei Chmutov Random gluing polygons

slide-15
SLIDE 15

Previous results.

E[Vn] ∼ log n Var(χ) ∼ log n

  • N. Pippenger, K. Schleich, Topological characteristics of

random triangulated surfaces, Random Structures Algorithms, 28 (2006) 247–288. All polygons are triangles.

  • A. Gamburd, Poisson-Dirichlet distribution for random Belyi

surfaces, Ann. Probability, 34 (2006) 1827–1848. All polygons have the same number of sides, k. 2 lcm(2, k) | kn γ is asymptotically uniform on the alternating group Akn.

Sergei Chmutov Random gluing polygons

slide-16
SLIDE 16

Key Theorem.

Depending on the parities of permutations α ∈ Cn and β ∈ CN/2 the permutation γ = αβ is either even γ ∈ AN or odd γ ∈ Ac

N := SN − AN.

The probability distribution of γ is asymptotically uniform (for N → ∞ uniformly in n) on AN or on Ac

N.

Let Pγ be the probability distribution of γ and let U be the uniform probability measure on AN or on Ac

N.

Let Pγ − U := (1/2)

s∈SN |Pγ(s) − U(s)| be the total

variation distance between Pγ and U.

  • Theorem. Pγ − U = O
  • N−1

.

Sergei Chmutov Random gluing polygons

slide-17
SLIDE 17

Ideas of the proof.

P . Diaconis, M. Shahshahani, Generating a random permutation with random tranpositions, Z. Wahr. Verw. Gebiete, 57 (1981) 159–179. Using the Fourier analysis on finite groups and the Plancherel Theorem: P − U2 ≤ 1 4

  • ρ∈

G, ρ=id

dim(ρ) tr ˆ P(ρ)ˆ P(ρ)∗ ; here G denotes the set of all irreducible representations ρ of G, “id” denotes the trivial representation, dim(ρ) is the dimension

  • f ρ, and ˆ

P(ρ) is the matrix value of the Fourier transform of P at ρ, ˆ P(ρ) :=

g∈G ρ(g)P(g).

Sergei Chmutov Random gluing polygons

slide-18
SLIDE 18

Ideas of the proof.

For G = SN, the irreducible representations ρ are indexed by partitions λ ⊢ N, λ = (λ1 ≥ λ2 ≥ . . . ) of N. Let f λ := dim(ρλ) (given by the hook length formula) and χλ be the character of ρλ. Pγ − U2 ≤ 1 4

  • λ=(N), (1N)
  • χλ(Cn)χλ(CN/2)

f λ 2 . Gamburd used estimate from S. V. Fomin, N. Lulov, On the number of rim hook tableaux, J. Math. Sciences, 87 (1997) 4118–4123, for N = kn, |χλ(CN/k)| = O

  • N1/2−1/(2k)

(f λ)1/k.

Sergei Chmutov Random gluing polygons

slide-19
SLIDE 19

Ideas of the proof.

  • M. Larsen, A. Shalev, Characters of symmetric groups: sharp

bounds and applications, Invent. Math., 174 (2008) 645–687. Extension of the Fomin-Lulov bound for all permutations σ without cycles of length below m, and partitions λ: |χλ(σ)| ≤ (f λ)1/m+o(1), N → ∞. Pγ − U2 = O(N−2).

Sergei Chmutov Random gluing polygons

slide-20
SLIDE 20

Thanks.

THANK YOU!

Sergei Chmutov Random gluing polygons