Guarding Polyominoes under k-hop Visibility or Minimum k ...
Enumerative Combinatorics with Fillings of Polyominoes
Transcript of Enumerative Combinatorics with Fillings of Polyominoes
Enumerative Combinatorics with Fillings of Polyominoes
Catherine Yan Texas A&M Univesrity GSU, October, 2014
TexPoint fonts used in EMF. Read the TexPoint manual before you delete this box.: AAAAA
2
Outline 1. Symmetry of the longest chains ◦ Subsequences in permutations and words ◦ Crossings and nestings in matchings and graphs ◦ A new model: fillings of moon polyominoes
2. Combinatorics of Fillings of Moon polyominoes ◦ Northeast and southeast chains ◦ Forbidden patterns ◦ Transformations ◦ Connections to other objects
3
Part I: Symmetry of the longest chains � Permutations: 732816549 (increasing subsequence) 732816549 (decreasing subsequence) is(w) = | longest i. s.| = 3 ds(w)= | longest d. s.| = 4
� [Deift, Baik & Johansson’99] Asymptotic distribution of is(w) and ds(w).
� is(w) and ds(w) are symmetric.
4
Crossings and nestings in matchings of [2n]
� (cr2, ne2) are symmetric! e.g. cr2 ne2
2 0 1 1 1 1 0 2
# noncrossing matchings of [2n] = # nonnesting matchings of [2n] = nth Catalan number
5
k-crossings/nestings
Conjecture: #Matchings of [2n] with no k-crossings = # Matchings of [2n] with no k-nestings
Theorem [Chen, Deng, Du] # matchings of [2n] with no 3-crossings = # matching of [2n] with no 3-nestings = # pairs of noncrossing Dyck paths
6
Crossing and nesting number
� For a matching M, cr(M)=max{ k: M has a k-crossing} ne(M)=max{k: M has a k-nesting }
Goal: symmetry between cr and ne
# k-crossing and # k-nesting: not symmetric How about “maximal crossing” and “maximal nesting”?
7
Main result on Matchings Theorem [Chen, Deng, Du, Stanley & Y, 07] The pair (cr(M), ne(M)) has a symmetric joint distribution over all matchings on [2n].
Corollary.
# matchings with no k-crossing = # matchings with no k-nesting
A = B
8
Idea:
� Oscillating tableau: a sequence of Ferrers diagrams ;=λ0, λ1, …, λ2n =; s.t.
λi= λi-1 +/ -
;
;
9
Theorem [Stanley & Sundaram’90] There is a bijection between matchings of [2n] and oscillating tableaux of length 2n.
� It is realized by using standard Young tableaux and applying the RSK algorithm.
Theorem [CDDSY] Taking conjugation in the tableaux exchanges cr(M) and ne(M).
10
Set Partitions of [n]
� A graphical representation π= {1,4, 5, 7} {2,6} {3} � Theorem. [CDDSY] (cr(¼), ne(¼)) has a symmetric distribution over all partitions of [n].
1 2 3 4 5 6 7
11
Filling of the triangular board
1 2 3 4 5 6 7 8
8 7 6 5 4 3 2
Crossing: anti-identity submatrix (NE-chain) Nesting: identity submatrix (SE-chain)
1
12
An extension to Ferrers diagram
01-filling of any Ferrers diagram F Every row/column has at most one 1. NE-chain Jk SE-chain Ik
1
1 1
1
1
1
1 1
13
Ferrers diagram
NE(F) = |longest NE chain| SE (F) = |longest SE chain|
[Krattenthaler’06] Given a Ferrers diagram F and an integer n, then (NE(F), SE(F)) has a symmetric distribution over 01-fillings of F with n 1’s..
1
1
1
1
1
14
Generalized triangulation of n-gon
1 2
3
4
5 6
7
8
1
2 3 4 5 6 7
7 6 5 4 3 2 8
k-triangulation: no k+1 diagonals that are mutually intersecting
15
Results about k-triangulation � [Capoyleas & Pach’92] k-triangulations of an n-gon has at most k(2n-2k-1) lines. � [Dress, Koolen & Moulton’02] maximal k-triangulation always has k(2n-2k-1) lines � [Jonsson’05] #maximal k-triangulations =a determinant of Catalan numbers.
16
Catalan number implies symmetry!
try to avoid
17
Stack polyominoes
[Jonsson’05, Jonsson & Welker’07]: F01(L, n, ne < k ) = F01 (L, n, se < k) where n is the number of ones in the filling.
18
Moon polyominoes
[Rubey’11]: F(M, n, ne<k ) = F(M, n, se <k ) And F01(M, n, ne< k) = F01 (M, n, se <k)
19
The General Model: fillings of moon polyominoes
� Polyomino: a finite set of square cells
� Moon polyomino: ◦ Convex ◦ intersection-free (no skew shape)
20
Fillings of moon polyominoes
� Assign an integer to each square
1 1
1
1
1
1
11
1
1
Permuta-tions Words Matchings Set partitions
Graphs Ferrers diagram Stack polyomino Moon polyomino
21
Part II: Combinatorics of fillings of moon polyominoes
◦ Northeast and southeast chains ◦ Forbidden patterns ◦ Transformations ◦ Connections to other objects
22
The model is general: Example 1. Chains of length 2 Permutation: inversion and coinversion
¼=624153 � inversion: {(i - j): i > j } coinversion: {(i - j): i < j } inv(¼)= 9 : { 62, 64, 61, 65, 63, 21, 41, 43, 53}
coinv(¼)=6: { 24, 25, 23, 45, 15, 13 }
P¼ p
i n v (¼) qcoi n v (¼) =A =P
i BiA+B = cA+B = CA = B
A = B
where [k]p.q is the (p,q)-integer pk-1+pk-2q + … + pqk-2 + qk-1.
23
On words over { 1n1, 2n2,…, knk }
� A word is an arrangement of 1n1, 2n2, …, knk
� Similar results for
◦ Matchings [de Sainte-Catherine’83] ◦ Set partitions [Kasraoui & Zeng’06] ◦ Linked partitions [Chen, Wu & Y’ 08] ◦ Crossing and alignment for permutations [Corteel’07]
24
Theorem [Kasraoui’10] The pair (ne2, se2) has a symmetric joint distribution over the set of 01-fillings of a moon polyomino with any given column sum.
X
M 2 F (M ;t)
pn e2(M ) qs e2(M ) =Y
i
µhiti
¶
p;q
X
M 2 F (M ;t)
pn e2(M ) qs e2(M ) =Y
i
µhiti
¶
p;q
a+ b = ca+ b = c
inv(¼) coinv(¼)
1 1
1 1
25
various mixed statistics
� Bicolor the rows of M and mixed by the position of the top cell/ bottom cell
1 1 1
1 top-mixed statistic ®(S,M): and
bottom-mixed statistic ¯(S,M):
1 1
1 1 and
26
� Mix by the charge of a corner cell
1 1 1
1 Positive chains and
Negative chains 1 1
1 1 and
27
Symmetry on mixed statistics
Theorem. [Chen, Wang, Y, Zhao’10; Wang &Y’13 ] Let ¸(A) be the number of any of the mixed statistics. (Hence ¸(M-A) is the number of remaining 2-chains. ) Then the joint
distribution of the pair (¸(A), ¸ (M-A)) is always symmetric and independent of the subset A.
Note: (¸(;), ¸(M)) = (se2(M), ne2(M)) (¸(M), ¸(;)) = (ne2(M), se2(M))
Special case for permutations: Chebikin’08.
28
The model is special enough!
Many things happen inside rectangles!
29
Example 2: k-noncrossing vs k-nonnesting Problem: # fillings with no k-crossing = # fillings with no k-nesting Method: Start with a filling with no k-crossing, then replace every appearance of k-nesting by other patterns. � [Backelin, West, Xin’07] for 01-fillings of
Ferrers diagrams � [de Mier’07] for multi-graphs with fixed
degree sequences 30
It applies to other patterns.
� Both papers compared patterns Jk and
� One can get more Wilf-equivalent pairs.
31
Applies to symmetric fillings � [Bousquet-Melou, Steingrimsson’05]
symmetric 01-fillings of symmetric Ferrers diagrams – involution
32
Example 3. The major index
� For a word a1 a2 … an , a descent is a position i such that ai > ai+1.
� maj(w) = ∑ { i : i 2 DES(w) }. � [MacMahon’1916] The major index is
equadistributed to inv(w) over words.
33
Example 3. The major index
� For a word a1 a2 … an , a descent is a position i such that ai > ai+1.
� maj(w) = ∑ { i : i 2 DES(w) }. � [MacMahon’1916] The major index is
equadistributed to inv(w) over words.
[Chen, Poznanovik, Y & Yang’ 10] The major index can be extended to 01-fillings of moon polyominoes, which has the same distribution as ne2.
34
Foata’s map Φ with inv(Φ(w))=maj(w)
� Recursive Definition: ◦ If w has length 1, Φ(w)=w.
◦ Otherwise, w= w’ a, then Φ(w) = γa(Φ (w’)) a
w =w1 …wn-1 a
w1… wn-1 u1… un-‐1 v1… vn-‐1
v1… vn-‐1 a
Φ γa
35
Many transformations!
� [CPYY] Foata-type transformations can be defined on fillings of left-aligned stack polyominoes which carry maj to ne2
36
From polyomino to polyomino
• Bijection f from fillings of M to fillings of N s.t. maj(F) =maj(f(F)) • Bejection g from fillings of M to fillings of N s.t. ne2(F) = ne2(g(F))
37
And more… � Lattice path counting and descents in
Ferrers diagrams � Rook placement with restrictions � Pattern avoidance and appearances � Poset, P-partitions � Simplicial complexes/Schubert
polynomials …
38
Relation to other areas…
� Free probability- noncrossing diagrams
39
� Crossings appear in the combinatorial interpretations of ◦ Mixed moments of random variables ◦ Moments of orthogonal polynomials ◦ Linearization coefficients …
40
� Graph optimization and layout: A partition of the edges into k-sets of non-crossing (non-nesting) edges
41
Stacks and Queues
� Stack-number: minimum k such that there is a total order of the vertices with which G has a k-stack layout
� Queue-number: minimum k such that there is a total order of the vertices with which G has a k-queue layout
42
43
• Combinatorial computational biology: RNA pseudo knot structures
44
45
T H A N K
Y O U
V E R Y
M U C H !