Fully Packed Loop Configurations and Littlewood–Richardson coefficients Philippe Nadeau Faculty of Mathematics, University of Vienna FPSAC 22, San Francisco, August 4th 2010.
Rough Outline Fully Packed Loops in a square grid
Rough Outline Fully Packed Loops in a square grid From the square to the triangle
Rough Outline Fully Packed Loops in a square grid From the square to the triangle Fully Packed Loops in a triangle
FPL configurations : Definition Start with the square grid G n with n 2 vertices and 4 n external edges, and pick every other edge on the boundary (starting with the topmost on the left). 14 13 12 1 11 2 10 3 9 4 8 5 6 7
FPL configurations : Definition Start with the square grid G n with n 2 vertices and 4 n external edges, and pick every other edge on the boundary (starting with the topmost on the left). 14 13 12 1 11 A Fully Packed Loop (FPL) 2 10 configuration of size n is a subgraph of G n with exactly 2 3 9 edges incident to each vertex. 4 8 5 6 7
FPL configurations : Enumeration FPL configurations are in simple bijection with numerous objects : alternating sign matrices (ASMs), height matrices, configurations of the six vertex model, Gog triangles,... 14 13 12 1 11 2 10 3 9 4 8 5 6 7
FPL configurations : Enumeration FPL configurations are in simple bijection with numerous objects : alternating sign matrices (ASMs), height matrices, configurations of the six vertex model, Gog triangles,... FPL of size n ASMs of size n
FPL configurations : Enumeration FPL configurations are in simple bijection with numerous objects : alternating sign matrices (ASMs), height matrices, configurations of the six vertex model, Gog triangles,... 14 13 12 1 11 FPL of size n 2 10 3 9 ASMs of size n 4 8 5 6 7
FPL configurations : Enumeration FPL configurations are in simple bijection with numerous objects : alternating sign matrices (ASMs), height matrices, configurations of the six vertex model, Gog triangles,... 14 13 12 1 11 n − 1 (3 i + 1)! � | FPL n | = A n = ( n + i )! 2 10 i =0 3 9 [Zeilberger ’96, Kuperberg ’96] 4 8 5 6 7
FPL configurations : Enumeration FPL configurations are in simple bijection with numerous objects : alternating sign matrices (ASMs), height matrices, configurations of the six vertex model, Gog triangles,... 14 13 12 1 11 n − 1 (3 i + 1)! � | FPL n | = A n = ( n + i )! 2 10 i =0 3 9 [Zeilberger ’96, Kuperberg ’96] 4 8 5 6 7 Why study FPLs rather than ASMs ?
FPL configurations : Refined enumeration Every FPL configuration determines a link pattern on the external edges of the grid G n , where link pattern = set of n noncrossing chords between 2 n labeled points on a disk. 14 13 12 1 11 2 10 3 9 4 8 5 6 7
FPL configurations : Refined enumeration Every FPL configuration determines a link pattern on the external edges of the grid G n , where link pattern = set of n noncrossing chords between 2 n labeled points on a disk. 14 13 12 13 14 12 1 11 1 11 2 10 2 10 3 9 4 7 8 3 5 6 9 4 8 � 2 n � 1 | LP n | = C n := 5 6 7 n + 1 n
FPL configurations : Refined enumeration Main problem : given a link pattern X , how many FPL configurations induce the link pattern X ? We note A X this number. 6 6 6 1 1 1 5 5 5 If X = A X = 2 2 4 2 2 4 4 3 3 3
FPL configurations : Refined enumeration Main problem : given a link pattern X , how many FPL configurations induce the link pattern X ? We note A X this number. 6 6 6 1 1 1 5 5 5 If X = A X = 2 2 4 2 2 4 4 3 3 3 Wieland’s rotation Given a link pattern X , consider the rotated pattern r ( X ) obtained by { i, j } �→ { i + 1 , j + 1 } . Then A X = A r ( X ) . X r ( X )
The Razumov-Stroganov correspondence For i = 1 . . . 2 n let e i act on link patterns by { i, j } , { i + 1 , k } ∈ X → { i, i + 1 } , { j, k } ∈ e i ( X ) . j i j i i + 1 i + 1 e i k k
The Razumov-Stroganov correspondence For i = 1 . . . 2 n let e i act on link patterns by { i, j } , { i + 1 , k } ∈ X → { i, i + 1 } , { j, k } ∈ e i ( X ) . j i j i i + 1 i + 1 e i k k RS correspondence [RS ’01, Cantini and Sportiello ’10] : � ∀ X, 2 nA X = A Y ( i,Y ) ,e i ( Y )= X
The Razumov-Stroganov correspondence For i = 1 . . . 2 n let e i act on link patterns by { i, j } , { i + 1 , k } ∈ X → { i, i + 1 } , { j, k } ∈ e i ( X ) . j i j i i + 1 i + 1 e i k k RS correspondence [RS ’01, Cantini and Sportiello ’10] : � ∀ X, 2 nA X = A Y ( i,Y ) ,e i ( Y )= X These relations completely characterize the A X . Di Francesco and Zinn Justin had previously obtained results on the solutions of these relations → these are now applicable to the quantities A X .
Link patterns X with nice expressions for A X a b a b c i + j + k − 1 X = � � � A X = i + j + k − 2 i =1 j =1 k =1 c [Zinn-Justin, Zuber, Di Francesco]
Link patterns X with nice expressions for A X a b a b c i + j + k − 1 X = � � � A X = i + j + k − 2 i =1 j =1 k =1 c [Zinn-Justin, Zuber, Di Francesco] A X = Complicated determinant formula X = [Caselli and Krattenthaler ’04, Zinn-Justin ’08]
Link patterns X with nice expressions for A X a b a b c i + j + k − 1 X = � � � A X = i + j + k − 2 i =1 j =1 k =1 c [Zinn-Justin, Zuber, Di Francesco] A X = Complicated determinant formula X = [Caselli and Krattenthaler ’04, Zinn-Justin ’08] A X = A n − 1 . X = [Di Francesco and Z-J, Cantini and Sportiello]
Link patterns with nested arches We consider now integers n, m ≥ 0 , and link patterns with m nested arches, and π is a noncrossing matching with n arches. m π X = π ∪ m We will determine an expression for A π ∪ m , based on FPLs in a triangle. ( → The case m = 0 gives the usual numbers A X .)
Link patterns with nested arches We suppose m ≥ 3 n − 1 , and choose k such that 0 ≤ k ≤ m − (3 n − 1) . m π π 4 n − 2 m − 3 n − k + 1 k
Link patterns with nested arches Fixed edges based on a lemma from [de Gier, ’02]. π k 4 n − 2 m − 3 n − k + 1
Link patterns with nested arches Fixed edges based on a lemma from [de Gier, ’02]. Three regions appear R 2 R 1 T π k 4 n − 2 m − 3 n − k + 1
Link patterns with nested arches We can then write, for m ≥ 3 n − 1 and 0 ≤ k ≤ m − (3 n − 1) � |R 1 ( σ, k ) | × t π A π ∪ m = σ,τ × |R 2 ( τ, m − 3 n − k + 1) | σ,τ R 2 R 1 T τ σ π k m − 3 n − k + 1 • σ, τ are words of length 2 n on { 0 , 1 } ; • R 1 ( σ, . ) , R 2 ( τ, . ) are the sets of FPLs in R 1 and R 2 . • t π σ,τ is the number of FPL configurations in the triangle T .
Words and Shapes Let σ = σ 1 . . . σ p be a word in { 0 , 1 } p ; we write | σ | := p . Words ↔ Ferrers shapes in a box. Example : σ = 0101011110 , so | σ | = 10 , | σ | 0 = 4 , | σ | 1 = 6 . σ 0 1
Words and Shapes Let σ = σ 1 . . . σ p be a word in { 0 , 1 } p ; we write | σ | := p . Words ↔ Ferrers shapes in a box. Example : σ = 0101011110 , so | σ | = 10 , | σ | 0 = 4 , | σ | 1 = 6 . σ 0 1 d ( σ ) := the number of boxes in the diagram σ . σ ∗ := (1 − σ p ) · · · (1 − σ 2 )(1 − σ 1 ) In the example, d ( σ ) = 9 and σ ∗ = 1000010101 .
Words and Shapes σ → σ ′ σ ≤ σ ′ At most one more box per column
Words and Shapes σ → σ ′ σ ≤ σ ′ At most one more box per column A semi standard Young tableau of shape σ and entries bounded by N is a filling of the shape σ by integers in { 1 , . . . , N } such that entries are strictly increasing in columns and weakly increasing in rows.
Words and Shapes σ → σ ′ σ ≤ σ ′ At most one more box per column A semi standard Young tableau of shape σ and entries bounded by N is a filling of the shape σ by integers in { 1 , . . . , N } such that entries are strictly increasing in columns and weakly increasing in rows. The number of such tableaux is given by SSY T ( σ, N ) , an 1 H ( σ ) N d ( σ ) . explicit polynomial in N with leading term (Here H ( σ ) is the product of hook lengths of the shape σ .)
Regions R 1 and R 2 Proposition [Caselli,Krattenthaler,Lass,N. ’05] Let σ be a word of length 2 n , and k ∈ N . There is a bijection between FPLs in R 1 ( σ, k ) and semistandard Young tableaux of shape σ and length n + k . R 2 R 1 T τ σ π k m − 3 n − k + 1
Regions R 1 and R 2 Proposition [Caselli,Krattenthaler,Lass,N. ’05] Let σ be a word of length 2 n , and k ∈ N . There is a bijection between FPLs in R 1 ( σ, k ) and semistandard Young tableaux of shape σ and length n + k . So for m ≥ 3 n − 1 (and k = 0 ) we obtain : � |R 1 ( σ, 0) | · t π A π ∪ m = σ,τ · |R 2 ( τ, m − 3 n + 1) | σ,τ � SSY T ( σ, n ) · t π σ,τ · SSY T ( τ ∗ , m − 2 n + 1) = σ,τ
Recommend
More recommend