Asymptotics of Pattern Classes of Set Partition and Permutation d - - PowerPoint PPT Presentation

asymptotics of pattern classes of set partition and
SMART_READER_LITE
LIVE PREVIEW

Asymptotics of Pattern Classes of Set Partition and Permutation d - - PowerPoint PPT Presentation

Asymptotics of Pattern Classes of Set Partition and Permutation d -tuple Avoidance Benjamin Gunby Harvard Department of Mathematics June 2017 Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation This research is joint


slide-1
SLIDE 1

Asymptotics of Pattern Classes of Set Partition and Permutation d-tuple Avoidance

Benjamin Gunby

Harvard Department of Mathematics

June 2017

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-2
SLIDE 2

This research is joint work with D¨

  • r P´

alv¨

  • lgyi.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-3
SLIDE 3

This research is joint work with D¨

  • r P´

alv¨

  • lgyi.

Paper hopefully on arXiv soon!

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-4
SLIDE 4

Definitions

Let n ∈ Z+. Definition A set partition of [n] is a collection of sets B1, B2, . . . , Bm, pairwise disjoint, with B1 ∪ · · · ∪ Bm = [n]. The order of sets does not matter.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-5
SLIDE 5

Definitions

Let n ∈ Z+. Definition A set partition of [n] is a collection of sets B1, B2, . . . , Bm, pairwise disjoint, with B1 ∪ · · · ∪ Bm = [n]. The order of sets does not matter. We will call these sets blocks of the partition.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-6
SLIDE 6

Definitions

Let n ∈ Z+. Definition A set partition of [n] is a collection of sets B1, B2, . . . , Bm, pairwise disjoint, with B1 ∪ · · · ∪ Bm = [n]. The order of sets does not matter. We will call these sets blocks of the partition. Since the order of sets is irrelevant, we will order the Bi in increasing order of smallest element, and denote the set partition with slashes between the blocks. (e.g. {5, 2, 3} ∪ {4, 6, 1} is denoted 146/235).

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-7
SLIDE 7

Definitions

Let n, k ∈ Z+, and π, π′ be set partitions of n, k respectively.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-8
SLIDE 8

Definitions

Let n, k ∈ Z+, and π, π′ be set partitions of n, k respectively. Definition We say that π contains (respectively avoids) π′ if there exists (respectively does not exist) an increasing injection f : [k] → [n] such that for all i, j ∈ [k] the following are equivalent: i and j are in the same set in π′. f (i) and f (j) are in the same set in π.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-9
SLIDE 9

Definitions

Let n, k ∈ Z+, and π, π′ be set partitions of n, k respectively. Definition We say that π contains (respectively avoids) π′ if there exists (respectively does not exist) an increasing injection f : [k] → [n] such that for all i, j ∈ [k] the following are equivalent: i and j are in the same set in π′. f (i) and f (j) are in the same set in π. In other words, π contains π′ if and only if we can restrict π to a k-element subset of [n], so that the resulting set partition is

  • rder-isomorphic to π.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-10
SLIDE 10

Definitions

Let n, k ∈ Z+, and π, π′ be set partitions of n, k respectively. Definition We say that π contains (respectively avoids) π′ if there exists (respectively does not exist) an increasing injection f : [k] → [n] such that for all i, j ∈ [k] the following are equivalent: i and j are in the same set in π′. f (i) and f (j) are in the same set in π. In other words, π contains π′ if and only if we can restrict π to a k-element subset of [n], so that the resulting set partition is

  • rder-isomorphic to π.

(Example: 146/235 contains the partition 12/34, as we can see by restricting 146/235 to the set {2, 3, 4, 6}.)

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-11
SLIDE 11

Asymptotic Results

Question What are the possible growth rates of a pattern class of set partitions? What pattern classes give which growth rates?

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-12
SLIDE 12

Asymptotic Results

Question What are the possible growth rates of a pattern class of set partitions? What pattern classes give which growth rates? Permutations Set Partitions (Marcus-Tardos Theorem) (G., P´ alv¨

  • lgyi)

f (n) < cn f (n) < cn

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-13
SLIDE 13

Asymptotic Results

Question What are the possible growth rates of a pattern class of set partitions? What pattern classes give which growth rates? Permutations Set Partitions (Marcus-Tardos Theorem) (G., P´ alv¨

  • lgyi)

f (n) < cn f (n) < cn f (n) = n! f (n) = Bn

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-14
SLIDE 14

Asymptotic Results

Question What are the possible growth rates of a pattern class of set partitions? What pattern classes give which growth rates? Permutations Set Partitions (Marcus-Tardos Theorem) (G., P´ alv¨

  • lgyi)

f (n) < cn f (n) < cn f (n) = n! f (n) = Bn cnn

n 2 < f (n) < c′nn n 2 Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-15
SLIDE 15

Asymptotic Results

Question What are the possible growth rates of a pattern class of set partitions? What pattern classes give which growth rates? Permutations Set Partitions (Marcus-Tardos Theorem) (G., P´ alv¨

  • lgyi)

f (n) < cn f (n) < cn f (n) = n! f (n) = Bn cnn

n 2 < f (n) < c′nn n 2

cnn

2n 3 < f (n) < c′nn 2n 3 Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-16
SLIDE 16

Asymptotic Results

Question What are the possible growth rates of a pattern class of set partitions? What pattern classes give which growth rates? Permutations Set Partitions (Marcus-Tardos Theorem) (G., P´ alv¨

  • lgyi)

f (n) < cn f (n) < cn f (n) = n! f (n) = Bn cnn

n 2 < f (n) < c′nn n 2

cnn

2n 3 < f (n) < c′nn 2n 3

cnn

3n 4 < f (n) < c′nn 3n 4

. . .

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-17
SLIDE 17

Main Result

Theorem (G., P´ alv¨

  • lgyi)

Let P be a pattern class of set partitions. Then one of the following cases holds.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-18
SLIDE 18

Main Result

Theorem (G., P´ alv¨

  • lgyi)

Let P be a pattern class of set partitions. Then one of the following cases holds.

1 P consists of all set partitions Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-19
SLIDE 19

Main Result

Theorem (G., P´ alv¨

  • lgyi)

Let P be a pattern class of set partitions. Then one of the following cases holds.

1 P consists of all set partitions 2 Pn is empty for all sufficiently large n Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-20
SLIDE 20

Main Result

Theorem (G., P´ alv¨

  • lgyi)

Let P be a pattern class of set partitions. Then one of the following cases holds.

1 P consists of all set partitions 2 Pn is empty for all sufficiently large n 3 There exists d ∈ Z+ and constants c′ > c > 0 such that

cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-21
SLIDE 21

Main Result

Theorem (G., P´ alv¨

  • lgyi)

Let P be a pattern class of set partitions. Then one of the following cases holds.

1 P consists of all set partitions 2 Pn is empty for all sufficiently large n 3 There exists d ∈ Z+ and constants c′ > c > 0 such that

cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n

Question Which pattern classes fall into which growth rates? That is, which d corresponds to a given P?

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-22
SLIDE 22

Permutability

Definition Given n, d ∈ Z+, and σ1, . . . , σd ∈ Sn, we can construct a set partition of [(d + 1)n] as follows: there will be n blocks B1, . . . , Bn, with Bi = {i, n + σ1(i), 2n + σ2(i), . . . , dn + σd(i)}. We call this set partition [σ1, . . . , σd].

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-23
SLIDE 23

Permutability

Definition Given n, d ∈ Z+, and σ1, . . . , σd ∈ Sn, we can construct a set partition of [(d + 1)n] as follows: there will be n blocks B1, . . . , Bn, with Bi = {i, n + σ1(i), 2n + σ2(i), . . . , dn + σd(i)}. We call this set partition [σ1, . . . , σd]. Example: [132, 321] = 149/268/357

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-24
SLIDE 24

Permutability

Definition Let π be a set partition. Then the permutability of π, denoted pm(π) is the minimum positive integer d such that π is contained in a set partition of the form [σ1, . . . , σd], for some m ∈ Z+ and σ1, . . . , σd ∈ Sm.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-25
SLIDE 25

Permutability

Definition Let π be a set partition. Then the permutability of π, denoted pm(π) is the minimum positive integer d such that π is contained in a set partition of the form [σ1, . . . , σd], for some m ∈ Z+ and σ1, . . . , σd ∈ Sm. Example: 12/34 has permutability 2. (Contained in 145/236 = [21, 12], not in [σ] for any σ.)

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-26
SLIDE 26

Permutability

Definition Let π be a set partition. Then the permutability of π, denoted pm(π) is the minimum positive integer d such that π is contained in a set partition of the form [σ1, . . . , σd], for some m ∈ Z+ and σ1, . . . , σd ∈ Sm. Example: 12/34 has permutability 2. (Contained in 145/236 = [21, 12], not in [σ] for any σ.) Definition (Alternate) The set partition π of [n] has permutability d if and only if [n] can be divided into d + 1 intervals such that each interval contains at most one element of each block of π, but it cannot be divided into d intervals in that way.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-27
SLIDE 27

More Specific Main Result

Theorem (G., P´ alv¨

  • lgyi)

Let P be a pattern class of set partitions. Then one of the following cases holds.

1 P consists of all set partitions 2 Pn is empty for all sufficiently large n 3 There exists d ∈ Z+ and constants c′ > c > 0 such that

cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-28
SLIDE 28

More Specific Main Result

Theorem (G., P´ alv¨

  • lgyi)

Let P be a pattern class of set partitions. Then one of the following cases holds.

1 P consists of all set partitions 2 Pn is empty for all sufficiently large n 3 There exists d ∈ Z+ and constants c′ > c > 0 such that

cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n

In this case, d is the smallest permutability among set partitions that do not occur in P (except when d = 1; then this may be 0).

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-29
SLIDE 29

More Specific Main Result

Theorem (G., P´ alv¨

  • lgyi)

Let P be a pattern class of set partitions. Then one of the following cases holds.

1 P consists of all set partitions 2 Pn is empty for all sufficiently large n 3 There exists d ∈ Z+ and constants c′ > c > 0 such that

cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n

In this case, d is the smallest permutability among set partitions that do not occur in P (except when d = 1; then this may be 0). Example: if P = Av(π), then d = pm(π).

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-30
SLIDE 30

Lower Bound

Theorem cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n,

where d is the smallest permutability not occurring in P. How many set partitions of [n] have permutability at most d − 1?

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-31
SLIDE 31

Lower Bound

Theorem cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n,

where d is the smallest permutability not occurring in P. How many set partitions of [n] have permutability at most d − 1? This includes all set partitions of the form [σ1, . . . , σd−1], where σi ∈ S n

d , so at least

n d

  • !

d−1

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-32
SLIDE 32

Lower Bound

Theorem cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n,

where d is the smallest permutability not occurring in P. How many set partitions of [n] have permutability at most d − 1? This includes all set partitions of the form [σ1, . . . , σd−1], where σi ∈ S n

d , so at least

n d

  • !

d−1 > n ed d−1

d n

, proving the lower bound of the theorem.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-33
SLIDE 33

Upper Bound

Theorem cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n,

where d is the smallest permutability not occurring in P. The upper bound is more complicated; we will describe an important lemma.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-34
SLIDE 34

Upper Bound

Theorem cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n,

where d is the smallest permutability not occurring in P. The upper bound is more complicated; we will describe an important lemma. If P is a pattern class not containing some π of permutability d, then P ⊂ Av(π) ⊂ Av([σ1, . . . , σd]) for some permutations σ1, . . . , σd.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-35
SLIDE 35

Upper Bound

Theorem cnn(1− 1

d )n < |Pn| < c′nn(1− 1 d )n,

where d is the smallest permutability not occurring in P. The upper bound is more complicated; we will describe an important lemma. If P is a pattern class not containing some π of permutability d, then P ⊂ Av(π) ⊂ Av([σ1, . . . , σd]) for some permutations σ1, . . . , σd. Thus it suffices to show that Avn([σ1, . . . , σd]) < c′nn(1− 1

d )n for

some c′.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-36
SLIDE 36

Hypergraphs

We have a ‘commutative diagram’ of generalizations:

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-37
SLIDE 37

Hypergraphs

We have a ‘commutative diagram’ of generalizations: Permutations 0 − 1 Matrices Set Partitions Ordered Hypergraphs

Perm Mats σ→[σ] 0−1 mat=bip. graph Blocks=Edges

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-38
SLIDE 38

Hypergraphs

We have a ‘commutative diagram’ of generalizations: Permutations 0 − 1 Matrices Set Partitions Ordered Hypergraphs

Perm Mats σ→[σ] 0−1 mat=bip. graph Blocks=Edges

So it makes sense to generalize to ordered hypergraphs to obtain information about asymptotics.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-39
SLIDE 39

Hypergraphs

We have a ‘commutative diagram’ of generalizations: Permutations 0 − 1 Matrices Set Partitions Ordered Hypergraphs

Perm Mats σ→[σ] 0−1 mat=bip. graph Blocks=Edges

So it makes sense to generalize to ordered hypergraphs to obtain information about asymptotics. Definition A d-permutation hypergraph be the hypergraph corresponding to some set partition of the form [σ1, . . . , σd−1] (that is, the image of [σ1, . . . , σd−1] under the bottom map in the diagram above).

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-40
SLIDE 40

Hypergraph Avoidance

We want to define a notion of hypergraph avoidance that extends

  • ur notion of set partition avoidance.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-41
SLIDE 41

Hypergraph Avoidance

We want to define a notion of hypergraph avoidance that extends

  • ur notion of set partition avoidance.

Definition An ordered hypergraph G contains (respectively avoids) an ordered hypergraph H if and only if there exists (respectively does not exist) an order-preserving injection V (H) → V (G) and an injection E(H) → E(G) that are compatible:

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-42
SLIDE 42

Hypergraph Avoidance

We want to define a notion of hypergraph avoidance that extends

  • ur notion of set partition avoidance.

Definition An ordered hypergraph G contains (respectively avoids) an ordered hypergraph H if and only if there exists (respectively does not exist) an order-preserving injection V (H) → V (G) and an injection E(H) → E(G) that are compatible: If E is sent to E ′, then all vertices in E are sent to vertices of E ′ (E may have fewer vertices than E ′).

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-43
SLIDE 43

Hypergraph Avoidance

We want to define a notion of hypergraph avoidance that extends

  • ur notion of set partition avoidance.

Definition An ordered hypergraph G contains (respectively avoids) an ordered hypergraph H if and only if there exists (respectively does not exist) an order-preserving injection V (H) → V (G) and an injection E(H) → E(G) that are compatible: If E is sent to E ′, then all vertices in E are sent to vertices of E ′ (E may have fewer vertices than E ′). For example, G contains the hypergraph H on [4] with edges {1, 3} and {2, 4} if and only if there exist two different edges E1 and E2 of G and vertices v1, v′

1 ∈ E1, v2, v′ 2 ∈ E2 with

v1 < v2 < v′

1 < v′

  • 2. (A single 4-vertex edge would not suffice.)

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-44
SLIDE 44

Main Lemma

Similarly to bounding the number of ones in a 0 − 1 matrix that avoids a permutation matrix, we need a lemma that bounds the number of edges (or similar) in an ordered hypergraph that avoids a d-permutation hypergraph.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-45
SLIDE 45

Main Lemma

Similarly to bounding the number of ones in a 0 − 1 matrix that avoids a permutation matrix, we need a lemma that bounds the number of edges (or similar) in an ordered hypergraph that avoids a d-permutation hypergraph. Lemma (G., P´ alv¨

  • lgyi)

Let H be a d-permutation hypergraph. Then there exists c such that any ordered hypergraph G on [n] that avoids H has

  • E∈E(G)

|E| < cnd−1.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-46
SLIDE 46

Main Lemma

Similarly to bounding the number of ones in a 0 − 1 matrix that avoids a permutation matrix, we need a lemma that bounds the number of edges (or similar) in an ordered hypergraph that avoids a d-permutation hypergraph. Lemma (G., P´ alv¨

  • lgyi)

Let H be a d-permutation hypergraph. Then there exists c such that any ordered hypergraph G on [n] that avoids H has

  • E∈E(G)

|E| < cnd−1. This lemma is interesting in its own right:

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-47
SLIDE 47

Main Lemma

Similarly to bounding the number of ones in a 0 − 1 matrix that avoids a permutation matrix, we need a lemma that bounds the number of edges (or similar) in an ordered hypergraph that avoids a d-permutation hypergraph. Lemma (G., P´ alv¨

  • lgyi)

Let H be a d-permutation hypergraph. Then there exists c such that any ordered hypergraph G on [n] that avoids H has

  • E∈E(G)

|E| < cnd−1. This lemma is interesting in its own right: The d = 2 case is a prior theorem of (independently) Klazar-Marcus and Balogh-Bollobas-Morris.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-48
SLIDE 48

Main Lemma

Similarly to bounding the number of ones in a 0 − 1 matrix that avoids a permutation matrix, we need a lemma that bounds the number of edges (or similar) in an ordered hypergraph that avoids a d-permutation hypergraph. Lemma (G., P´ alv¨

  • lgyi)

Let H be a d-permutation hypergraph. Then there exists c such that any ordered hypergraph G on [n] that avoids H has

  • E∈E(G)

|E| < cnd−1. This lemma is interesting in its own right: The d = 2 case is a prior theorem of (independently) Klazar-Marcus and Balogh-Bollobas-Morris. The case where G is also a d-permutation hypergraph is a theorem of Klazar-Marcus.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-49
SLIDE 49

Permutation-Tuple Avoidance

Getting back to permutations:

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-50
SLIDE 50

Permutation-Tuple Avoidance

Getting back to permutations: Given a d-tuple (σ1, . . . , σd) of permutations, we’ve constructed a set partition: [σ1, . . . , σd].

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-51
SLIDE 51

Permutation-Tuple Avoidance

Getting back to permutations: Given a d-tuple (σ1, . . . , σd) of permutations, we’ve constructed a set partition: [σ1, . . . , σd]. This gives a notion of avoidance on d-tuples of permutations of the same size, which is fairly natural on its own:

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-52
SLIDE 52

Permutation-Tuple Avoidance

Getting back to permutations: Given a d-tuple (σ1, . . . , σd) of permutations, we’ve constructed a set partition: [σ1, . . . , σd]. This gives a notion of avoidance on d-tuples of permutations of the same size, which is fairly natural on its own: Definition Let σ1, . . . , σd ∈ Sn and σ′

1, . . . , σ′ d ∈ Sm be permutations. Then

the d-tuple (σ1, . . . , σd) contains (respectively avoids) the d-tuple (σ1, . . . , σd) if there exist (respectively do not exist) indices i1, . . . , im ∈ [n] with i1 < · · · < im satisfying the property that for any j, σj(i1) · · · σj(im) has the same relative ordering as σ′

j(1) · · · σ′ j(m).

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-53
SLIDE 53

Permutation-Tuple Avoidance

Getting back to permutations: Given a d-tuple (σ1, . . . , σd) of permutations, we’ve constructed a set partition: [σ1, . . . , σd]. This gives a notion of avoidance on d-tuples of permutations of the same size, which is fairly natural on its own: Definition Let σ1, . . . , σd ∈ Sn and σ′

1, . . . , σ′ d ∈ Sm be permutations. Then

the d-tuple (σ1, . . . , σd) contains (respectively avoids) the d-tuple (σ1, . . . , σd) if there exist (respectively do not exist) indices i1, . . . , im ∈ [n] with i1 < · · · < im satisfying the property that for any j, σj(i1) · · · σj(im) has the same relative ordering as σ′

j(1) · · · σ′ j(m).

In other words, a d-tuple T1 contains another d-tuple T2 if and

  • nly if each permutation in T1 contains the corresponding

permutation in T2 at the same location.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-54
SLIDE 54

Permutation-Tuple Avoidance

Example: d = 2, (σ′

1, σ′ 2) = (21, 12).

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-55
SLIDE 55

Permutation-Tuple Avoidance

Example: d = 2, (σ′

1, σ′ 2) = (21, 12).

(σ1, σ2) avoids (21, 12) if and only if any inversion in σ1 is also an inversion of σ2.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-56
SLIDE 56

Permutation-Tuple Avoidance

Example: d = 2, (σ′

1, σ′ 2) = (21, 12).

(σ1, σ2) avoids (21, 12) if and only if any inversion in σ1 is also an inversion of σ2. σ1 ≤ σ2 in the Weak Bruhat Order.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-57
SLIDE 57

Permutation-Tuple Asymptotics

If (σ′

1, . . . , σ′ d) avoids (σ1, . . . , σd), with σ1, . . . , σd ∈ Sn, then

[σ′

1, . . . , σ′ d] is a set partition of [(d + 1)n] avoiding [σ1, . . . , σd].

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-58
SLIDE 58

Permutation-Tuple Asymptotics

If (σ′

1, . . . , σ′ d) avoids (σ1, . . . , σd), with σ1, . . . , σd ∈ Sn, then

[σ′

1, . . . , σ′ d] is a set partition of [(d + 1)n] avoiding [σ1, . . . , σd].

Avn((σ1, . . . , σd)) ≤ Av(d+1)n([σ1, . . . , σd]) < c′nnn(1− 1

d )(d+1)

= c′nn

  • d2−1

d

  • n

for some c′.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-59
SLIDE 59

Permutation-Tuple Asymptotics

The lower bound also holds! (As long as σ1, . . . , σd have size > 1)

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-60
SLIDE 60

Permutation-Tuple Asymptotics

The lower bound also holds! (As long as σ1, . . . , σd have size > 1) We can assume σ1, . . . , σd have size 2. We can assume further that they’re all 12.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-61
SLIDE 61

Permutation-Tuple Asymptotics

The lower bound also holds! (As long as σ1, . . . , σd have size > 1) We can assume σ1, . . . , σd have size 2. We can assume further that they’re all 12. A (slightly reinterpreted) result of Brightwell says that Av(12, . . . , 12) > cnn

  • d2−1

d

  • n for some c.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-62
SLIDE 62

Permutation-Tuple Asymptotics

The lower bound also holds! (As long as σ1, . . . , σd have size > 1) We can assume σ1, . . . , σd have size 2. We can assume further that they’re all 12. A (slightly reinterpreted) result of Brightwell says that Av(12, . . . , 12) > cnn

  • d2−1

d

  • n for some c.

Put it all together: Theorem (G., P´ alv¨

  • lgyi)

Let σ1, . . . , σd ∈ Sm, m > 1. There exist constants c′ > c > 0 such that cnn

  • d2−1

d

  • n < Avn((σ1, . . . , σd)) < c′nn
  • d2−1

d

  • n

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-63
SLIDE 63

An Open Question

Question Can we classify pattern classes of permutation d-tuples to within an exponential?

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-64
SLIDE 64

An Open Question

Question Can we classify pattern classes of permutation d-tuples to within an exponential? Our theorem solves the problem for classes with 1 basis element.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-65
SLIDE 65

An Open Question

Question Can we classify pattern classes of permutation d-tuples to within an exponential? Our theorem solves the problem for classes with 1 basis element. The product of a pattern class of d-tuples with one of d′-tuples gives a pattern class of d + d′-tuples.

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-66
SLIDE 66

An Open Question

Question Can we classify pattern classes of permutation d-tuples to within an exponential? Our theorem solves the problem for classes with 1 basis element. The product of a pattern class of d-tuples with one of d′-tuples gives a pattern class of d + d′-tuples. For d-tuples, can obtain within an exponential factor of nαn, for α = d − 1

d1 − · · · − 1 dk , di ∈ Z+, di ≤ d. Can we obtain

anything else?

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-67
SLIDE 67

Other Open Questions

What does the permutability statistic look like?

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-68
SLIDE 68

Other Open Questions

What does the permutability statistic look like? Can we compute the exponential factors in any of these cases? (Analogous to Stanley-Wilf limits)

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-69
SLIDE 69

Other Open Questions

What does the permutability statistic look like? Can we compute the exponential factors in any of these cases? (Analogous to Stanley-Wilf limits) For Av((12, 12)), we should have cn(n!)

3 2 . Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-70
SLIDE 70

Other Open Questions

What does the permutability statistic look like? Can we compute the exponential factors in any of these cases? (Analogous to Stanley-Wilf limits) For Av((12, 12)), we should have cn(n!)

3 2 .

Exponential factor is not currently known! π

2 ?

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation

slide-71
SLIDE 71

Thank You!

Benjamin Gunby Asymptotics of Pattern Classes of Set Partition and Permutation