symmetric group tilings The Waldspurger and Mienrenken Decompositions for Type A
James McKeown October 2016
University of Miami Outline
∙ Background, Waldspurger and Meinrenken Theorems ∙ Geometry ⇒ Permutations ∙ Permutations ⇒ Alternating Sign Matrices ∙ Alternating Sign Matrices ⇒ Geometry
1 Let G ⊂ O(n) be a finite group generated by reflections.
Reflection Groups
An element g ∈ O(n) is a reflection if it sends some nonzero vector α ∈ Rn to its negative and fixes the hyperplane orthogonal to α pointwise.
2 Reflection Groups
An element g ∈ O(n) is a reflection if it sends some nonzero vector α ∈ Rn to its negative and fixes the hyperplane orthogonal to α pointwise.
Let G ⊂ O(n) be a finite group generated by reflections.
2 Theorem (Coxeter): Let A be an arrangement in Rn of reflecting n hyperplanes for the reflection group G. Then G ↷ (R ∖ ∪H∈AH) freely and transitively on the chambers.
We pick one such chamber and call it the “weight cone” denoted CW. The
cone dual to CW we call the “root cone” and denote CR.
The Reflection Representation of Sn, and the Braid Arrangement
3 Figure: S3 ↷ R 2 3 Figure: S3 ↷ R = R /(1, 1, 1)
3 We pick one such chamber and call it the “weight cone” denoted CW. The
cone dual to CW we call the “root cone” and denote CR.
The Reflection Representation of Sn, and the Braid Arrangement
3 Figure: S3 ↷ R 2 3 Figure: S3 ↷ R = R /(1, 1, 1) Theorem (Coxeter): Let A be an arrangement in Rn of reflecting n hyperplanes for the reflection group G. Then G ↷ (R ∖ ∪H∈AH) freely and transitively on the chambers.
3 The Reflection Representation of Sn, and the Braid Arrangement
3 Figure: S3 ↷ R 2 3 Figure: S3 ↷ R = R /(1, 1, 1) Theorem (Coxeter): Let A be an arrangement in Rn of reflecting n hyperplanes for the reflection group G. Then G ↷ (R ∖ ∪H∈AH) freely and transitively on the chambers.
We pick one such chamber and call it the “weight cone” denoted CW. The
cone dual to CW we call the “root cone” and denote CR. 3 A first Example– Weights and Roots
α1
W1
W2 α2
4 Weights and Roots
Fact (Coxeter): Cw is a simplicial cone.
Let w1, w2,..., wn be vectors generating the rays of CW [Jargon: called the “fundamental weights”] Then the dual cone is defined as
n CR ∶= {x ∈ R ∶ (x, y) ≤ 0∀y ∈ CW}
Let α1, α2, . . . , αn be vectors which generate rays of CR [Jargon: called the “simple roots”]
5 The first example A2 (S3)
α1
W1
W2 α2
2 −1 3w1 = −2α2 − 1α1 −α1 = 2w2−1w1 ( ) −1 2 3w2 = −1α2 − 2α1 −α2 = −1w2 + 2w1
6 Lengths of weights are then normalized so that
−1 ⎛ ∣ ∣ ∣ ∣ ⎞ ⎛ ∣ ∣ ∣ ∣ ⎞ ⎜ ⎟ ⎜ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ⎟ ⎜ ⎟ − ⎜ ⎟ = ⎜ ⎟ ⎜α1 α2 . . . αn⎟ ⎜w1 w2 ... wn⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎝ ∣ ∣ ∣ ∣ ⎠ ⎝ ∣ ∣ ∣ ∣ ⎠
The matrix of α’s is called the Cartan Matrix. It gives the coordinates of the simple roots in the basis of fundamental weights.
Cartan Matrices: Roots in terms of Weights
In type A, it is conventional to let αi = ei − ei+1 so (αi, αi) = 2 ∀i
7 The matrix of α’s is called the Cartan Matrix. It gives the coordinates of the simple roots in the basis of fundamental weights.
Cartan Matrices: Roots in terms of Weights
In type A, it is conventional to let αi = ei − ei+1 so (αi, αi) = 2 ∀i
Lengths of weights are then normalized so that
−1 ⎛ ∣ ∣ ∣ ∣ ⎞ ⎛ ∣ ∣ ∣ ∣ ⎞ ⎜ ⎟ ⎜ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ⎟ ⎜ ⎟ − ⎜ ⎟ = ⎜ ⎟ ⎜α1 α2 . . . αn⎟ ⎜w1 w2 ... wn⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎝ ∣ ∣ ∣ ∣ ⎠ ⎝ ∣ ∣ ∣ ∣ ⎠
7 Cartan Matrices: Roots in terms of Weights
In type A, it is conventional to let αi = ei − ei+1 so (αi, αi) = 2 ∀i
Lengths of weights are then normalized so that
−1 ⎛ ∣ ∣ ∣ ∣ ⎞ ⎛ ∣ ∣ ∣ ∣ ⎞ ⎜ ⎟ ⎜ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ⎟ ⎜ ⎟ − ⎜ ⎟ = ⎜ ⎟ ⎜α1 α2 . . . αn⎟ ⎜w1 w2 ... wn⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎜ ∣ ∣ ∣ ∣ ⎟ ⎝ ∣ ∣ ∣ ∣ ⎠ ⎝ ∣ ∣ ∣ ∣ ⎠
The matrix of α’s is called the Cartan Matrix. It gives the coordinates of the simple roots in the basis of fundamental weights.
7 Waldspurger Matrices
Consider the reflection representation of the symmetric group
ϕ ∶ Sn Ð→ GLn−1(R)
Let D be the matrix with columns the fundamental weights in basis of the simple roots (i.e. the n − 1 × n − 1 inverse of the Cartan matrix).
W(g) ∶= [ϕ(1) − ϕ(g)]D
expressed in the coordinates of simple roots we will call the Waldspurger Matrix of g.
8 Take away: Waldspurger matrices give a tiling of the root cone!
For G an affine reflection group acting on a Euclidean vector space V, and
˚AW ⊂ V the interior of a fundamental domain for the action of G (sometimes called a fundamental alcove) one has the following decomposition:
V = ⊔(1 − g)˚AW g∈G
Take away: Convex hulls of columns of Waldspurger matrices (and the zero vector) give a tiling of the Rn!
Waldspurger and Mienrenken Theorems
Waldspurger’s Theorem (2005):
For G a finite reflection group acting on a Euclidean vector space V,
CR the (closed) cone over the positive roots, and
˚CW ⊂ V the interior of a fundamental domain for the action of G (sometimes called the weight cone), one has the following decomposition:
CR = ⊔(1 − g)˚CW g∈G
Meinrenken’s Theorem (2007):
9 For G an affine reflection group acting on a Euclidean vector space V, and
˚AW ⊂ V the interior of a fundamental domain for the action of G (sometimes called a fundamental alcove) one has the following decomposition:
V = ⊔(1 − g)˚AW g∈G
Take away: Convex hulls of columns of Waldspurger matrices (and the zero vector) give a tiling of the Rn!
Waldspurger and Mienrenken Theorems
Waldspurger’s Theorem (2005):
For G a finite reflection group acting on a Euclidean vector space V,
CR the (closed) cone over the positive roots, and
˚CW ⊂ V the interior of a fundamental domain for the action of G (sometimes called the weight cone), one has the following decomposition:
CR = ⊔(1 − g)˚CW g∈G Take away: Waldspurger matrices give a tiling of the root cone!
Meinrenken’s Theorem (2007):
9 Take away: Waldspurger matrices give a tiling of the root cone!
For G an affine reflection group acting on a Euclidean vector space V, and
˚AW ⊂ V the interior of a fundamental domain for the action of G (sometimes called a fundamental alcove) one has the following decomposition:
V = ⊔(1 − g)˚AW g∈G
Take away: Convex hulls of columns of Waldspurger matrices (and the zero vector) give a tiling of the Rn!
Waldspurger and Mienrenken Theorems
Waldspurger’s Theorem (2005):
Translation: Cones over columns of Waldspurger matrices give a tiling of the root cone!
Meinrenken’s Theorem (2007):
9 Take away: Waldspurger matrices give a tiling of the root cone!
Take away: Convex hulls of columns of Waldspurger matrices (and the zero vector) give a tiling of the Rn!
Waldspurger and Mienrenken Theorems
Waldspurger’s Theorem (2005):
Translation: Cones over columns of Waldspurger matrices give a tiling of the root cone!
Meinrenken’s Theorem (2007):
For G an affine reflection group acting on a Euclidean vector space V, and
˚AW ⊂ V the interior of a fundamental domain for the action of G (sometimes called a fundamental alcove) one has the following decomposition:
V = ⊔(1 − g)˚AW g∈G
9 Take away: Waldspurger matrices give a tiling of the root cone!
Waldspurger and Mienrenken Theorems
Waldspurger’s Theorem (2005):
Translation: Cones over columns of Waldspurger matrices give a tiling of the root cone!
Meinrenken’s Theorem (2007):
For G an affine reflection group acting on a Euclidean vector space V, and
˚AW ⊂ V the interior of a fundamental domain for the action of G (sometimes called a fundamental alcove) one has the following decomposition:
V = ⊔(1 − g)˚AW g∈G
Take away: Convex hulls of columns of Waldspurger matrices (and the zero vector) give a tiling of the Rn!
9 The Waldspurger Decomposition for A2 (S3)
(12)
˚ CR = ⊔(1 − g)CW g∈G
(13) (123) α1
id (132) W1
W2 α2
(23)
10 The Waldspurger Decomposition for A2 (S3)
(12)
˚ CR = ⊔(1 − g)CW g∈G
(13) (123) α1
id (132) W1
W2 α2
(23)
10 The Waldspurger Decomposition for A2 (S3)
(12)
˚ CR = ⊔(1 − g)CW g∈G
(13) (123) α1
id (132) W1
W2 α2
(23)
10 The Waldspurger Decomposition for A2 (S3)
(12)
˚ CR = ⊔(1 − g)CW g∈G
(13) (123) α1
id (132) W1
W2 α2
(23)
10 The Waldspurger Decomposition for A2 (S3)
(12)
˚ CR = ⊔(1 − g)CW g∈G
(13) (123) α1
id (132) W1
W2 α2
(23)
10 The Waldspurger Decomposition for A2 (S3)
(12)
˚ CR = ⊔(1 − g)CW g∈G
(13) (123) α1
id (132) W1
W2 α2
(23)
10 The Mienrenken Decomposition for A2 (S3)
˚ V = ⊔(1 − g)AW g∈G
α1
W1
W2 α2
11 The Mienrenken Decomposition for A2 (S3)
˚ V = ⊔(1 − g)AW g∈G
α1
W1
W2 α2
11 The Mienrenken Decomposition for A2 (S3)
˚ V = ⊔(1 − g)AW g∈G
α1
W1
W2 α2
11 The Waldspurger Decomposition for A3 (S4)
000 010
110 011
111
100 001
12 Part 2: Geometry ⇒ Permutations
Meinrenken Tile for A3 = S4
13 Meinrenken Tile for A3 = S4
Part 2: Geometry ⇒ Permutations
13 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 ⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭
The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
14 The Fundamental Transformation
Consider 43512 ∈ S5
1 1 1 0
1 2 2 1
1 1 2 1
0 0 1 1
⎧ R ⎫ ⎪ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞R ⎪ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R ⎪ ⎪ ⎜ 1 ⎟ ⎜ 2 ⎟ ⎜ 2 ⎟ ⎜ 1 ⎟R ⎪ that is, 43512 ↦ ⎨a ⎜ ⎟ + b ⎜ ⎟ + c ⎜ ⎟ + d ⎜ ⎟Ra, b, c, d ∈ R≥ ⎬ ⎪ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟R 0⎪ ⎪ 1 1 2 1 R ⎪ ⎪ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠R ⎪ ⎩ 0 0 1 1 R ⎭ 14 ∙ four copies of
S3 picture
∙ Codimension ↕ # cycles (c(n, k))
∙ NOT A CW- complex!
Classical Permutation Statistics, something weird...
(23)
(132) (234) (1342)
(13) *(13)(24)* (24) (1324) (1423) (134) (142)
(1234) (14) (1432) (123) (243)
(124) (143) (1243)
(12) (12)(34) (34)
15 ∙ Codimension ↕ # cycles (c(n, k))
∙ NOT A CW- complex!
Classical Permutation Statistics, something weird...
(23)
∙ four copies of
(132) (234) S3 picture (1342)
(13) *(13)(24)* (24) (1324) (1423) (134) (142)
(1234) (14) (1432) (123) (243)
(124) (143) (1243)
(12) (12)(34) (34)
15 ∙ NOT A CW- complex!
Classical Permutation Statistics, something weird...
(23)
∙ four copies of
(132) (234) S3 picture (1342)
∙ Codimension (13) *(13)(24)* (24) ↕ (1324) (1423) (134) (142) # cycles (c(n, k))
(1234) (14) (1432) (123) (243)
(124) (143) (1243)
(12) (12)(34) (34)
15 Classical Permutation Statistics, something weird...
(23)
∙ four copies of
(132) (234) S3 picture (1342)
∙ Codimension (13) *(13)(24)* (24) ↕ (1324) (1423) (134) (142) # cycles (c(n, k))
(1234) (14) (1432) (123) (243) ∙ NOT A CW- complex! (124) (143) (1243)
(12) (12)(34) (34)
15 ∙ Unimodal Motzkin Paths. ∙ Elements of the root lattice inside a certain polytope ∙ Abelian ideals in the nilradical of sln ∙ Young diagrams with hooklength less than n.
Interesting Bijections:
UM Vectors
Theorem: n There are exactly 2 possible columns of Waldspurger matrices of type An. We will call them UM vectors.
16 ∙ Unimodal Motzkin Paths. ∙ Elements of the root lattice inside a certain polytope ∙ Abelian ideals in the nilradical of sln ∙ Young diagrams with hooklength less than n.
UM Vectors
Theorem: n There are exactly 2 possible columns of Waldspurger matrices of type An. We will call them UM vectors.
Interesting Bijections:
16 ∙ Elements of the root lattice inside a certain polytope ∙ Abelian ideals in the nilradical of sln ∙ Young diagrams with hooklength less than n.
UM Vectors
Theorem: n There are exactly 2 possible columns of Waldspurger matrices of type An. We will call them UM vectors.
Interesting Bijections: ∙ Unimodal Motzkin Paths.
16 ∙ Abelian ideals in the nilradical of sln ∙ Young diagrams with hooklength less than n.
UM Vectors
Theorem: n There are exactly 2 possible columns of Waldspurger matrices of type An. We will call them UM vectors.
Interesting Bijections: ∙ Unimodal Motzkin Paths. ∙ Elements of the root lattice inside a certain polytope
16 ∙ Young diagrams with hooklength less than n.
UM Vectors
Theorem: n There are exactly 2 possible columns of Waldspurger matrices of type An. We will call them UM vectors.
Interesting Bijections: ∙ Unimodal Motzkin Paths. ∙ Elements of the root lattice inside a certain polytope ∙ Abelian ideals in the nilradical of sln
16 UM Vectors
Theorem: n There are exactly 2 possible columns of Waldspurger matrices of type An. We will call them UM vectors.
Interesting Bijections: ∙ Unimodal Motzkin Paths. ∙ Elements of the root lattice inside a certain polytope ∙ Abelian ideals in the nilradical of sln ∙ Young diagrams with hooklength less than n.
16 Warning: Note that WT (M) may be “over-determined” on the diagonal. In the case where M is a permutation matrix, but in general this need not be the case.
If an n × n matrix M has this property, we will say it is sum-symmetric
M ∈ SSn The map is linear and surjective, with kernel the diagonal matrices.
WT ∶ SSn ↠ Matn−1
The Waldspurger Transform applied to other Matrices
From an n × n matrix M, define the n − 1 × n − 1 matrix, WT (M) where ⎧ ⎪∑ M i ≤ j ⎪ a,b ⎪a≤i ⎪ > WT (M) ∶= ⎨b j . i,j ⎪ ≥ ⎪∑ Ma,b i j ⎪ ⎪a>i ⎩b≤j
17 If an n × n matrix M has this property, we will say it is sum-symmetric
M ∈ SSn The map is linear and surjective, with kernel the diagonal matrices.
WT ∶ SSn ↠ Matn−1
The Waldspurger Transform applied to other Matrices
From an n × n matrix M, define the n − 1 × n − 1 matrix, WT (M) where ⎧ ⎪∑ M i ≤ j ⎪ a,b ⎪a≤i ⎪ > WT (M) ∶= ⎨b j . i,j ⎪ ≥ ⎪∑ Ma,b i j ⎪ ⎪a>i ⎩b≤j Warning: Note that WT (M) may be “over-determined” on the diagonal. In the case where M is a permutation matrix, but in general this need not be the case.
17 If an n × n matrix M has this property, we will say it is sum-symmetric
M ∈ SSn The map is linear and surjective, with kernel the diagonal matrices.
WT ∶ SSn ↠ Matn−1
The Waldspurger Transform applied to other Matrices
From an n × n matrix M, define the n − 1 × n − 1 matrix, WT (M) where ⎧ ⎪∑ M i ≤ j ⎪ a,b ⎪a≤i WT ( ) ∶= ⎨b>j M i,j ⎪ . ⎪∑ M i ≥ j ⎪ a,b ⎪a>i ⎩b≤j
Proposition: WT (M) is well-defined if and only if the ith row sum of M equals the ith column sum of M, for 1 ≤ i ≤ n.
17 The map is linear and surjective, with kernel the diagonal matrices.
WT ∶ SSn ↠ Matn−1
The Waldspurger Transform applied to other Matrices
From an n × n matrix M, define the n − 1 × n − 1 matrix, WT (M) where ⎧ ⎪∑ M i ≤ j ⎪ a,b ⎪a≤i WT ( ) ∶= ⎨b>j M i,j ⎪ . ⎪∑ M i ≥ j ⎪ a,b ⎪a>i ⎩b≤j
Proposition: WT (M) is well-defined if and only if the ith row sum of M equals the ith column sum of M, for 1 ≤ i ≤ n.
If an n × n matrix M has this property, we will say it is sum-symmetric
M ∈ SSn
17 The Waldspurger Transform applied to other Matrices
From an n × n matrix M, define the n − 1 × n − 1 matrix, WT (M) where ⎧ ⎪∑ M i ≤ j ⎪ a,b ⎪a≤i WT ( ) ∶= ⎨b>j M i,j ⎪ . ⎪∑ M i ≥ j ⎪ a,b ⎪a>i ⎩b≤j
Proposition: WT (M) is well-defined if and only if the ith row sum of M equals the ith column sum of M, for 1 ≤ i ≤ n.
If an n × n matrix M has this property, we will say it is sum-symmetric
M ∈ SSn The map is linear and surjective, with kernel the diagonal matrices.
WT ∶ SSn ↠ Matn−1
17 Part 2: Permutations ⇒ Alternating Sign Matrices
Waldspurger, Meinrenken and ???
18 Waldspurger, Meinrenken and ???
Part 2: Permutations ⇒ Alternating Sign Matrices
18 These matrices generalize permutation matrices and arise naturally when using Dodgson condensation to compute a determinant. n − 1 × n − 1 matrices with UM columns and rows with their maxes on the diagonal are in bijection with n × n ASMs via the WT map!
ASMs
An alternating sign matrix (or ASM) is a square matrix of 0s, 1s, and −1s such that the sum of each row and column is 1 and the nonzero entries in each row and column alternate in sign.
19 n − 1 × n − 1 matrices with UM columns and rows with their maxes on the diagonal are in bijection with n × n ASMs via the WT map!
ASMs
An alternating sign matrix (or ASM) is a square matrix of 0s, 1s, and −1s such that the sum of each row and column is 1 and the nonzero entries in each row and column alternate in sign. These matrices generalize permutation matrices and arise naturally when using Dodgson condensation to compute a determinant.
19 ASMs
An alternating sign matrix (or ASM) is a square matrix of 0s, 1s, and −1s such that the sum of each row and column is 1 and the nonzero entries in each row and column alternate in sign. These matrices generalize permutation matrices and arise naturally when using Dodgson condensation to compute a determinant. n − 1 × n − 1 matrices with UM columns and rows with their maxes on the diagonal are in bijection with n × n ASMs via the WT map!
19 1 1 1 0 0
1 2 2 1 0
1 2 3 2 1
0 1 2 2 1
0 0 1 1 1
Fact (A. Lascoux, M. Schützenberger): Half the entropy of a permutation is its rank in the MacNeille completion of the Bruhat order– a distributive lattice with elements the alternating sign matrices, or ASMs.
Height=Entropy
Theorem:
1 n ∑ WT (g) = ∑(g(i) − i)2 1≤i,j≤n−1 2 i=1
20 Fact (A. Lascoux, M. Schützenberger): Half the entropy of a permutation is its rank in the MacNeille completion of the Bruhat order– a distributive lattice with elements the alternating sign matrices, or ASMs.
Height=Entropy
Theorem:
1 n ∑ WT (g) = ∑(g(i) − i)2 1≤i,j≤n−1 2 i=1
1 1 1 0 0
1 2 2 1 0
1 2 3 2 1
0 1 2 2 1
0 0 1 1 1
20 Height=Entropy
Theorem:
1 n ∑ WT (g) = ∑(g(i) − i)2 1≤i,j≤n−1 2 i=1
1 1 1 0 0
1 2 2 1 0
1 2 3 2 1
0 1 2 2 1
0 0 1 1 1
Fact (A. Lascoux, M. Schützenberger): Half the entropy of a permutation is its rank in the MacNeille completion of the Bruhat order– a distributive lattice with elements the alternating sign matrices, or ASMs. 20 The Bruhat Order
[1 1] 1 1
321 [1 0] [1 1] 1 1 0 1 312 231
[0 0] [1 0] 0 1 0 0 132 213 [0 0] 123 0 0
21 The MacNielle Completion of the Bruhat Order
[1 1] 1 1
321 [1 0] [1 1] 1 1 0 1 312 231 [1 0] 0 1 ? [0 0] [1 0] 0 1 0 0 132 213 [0 0] 123 0 0
22 The MacNielle Completion for S4
23 What does this tells us about the Meinrenken Tile?
Part 3: ASMs ⇒ Geometry
24 A Geometric Realization of the Hasse Diagram of AMS lattice
25 ∙ Is there a good way to study the Meinrenken tile as a polytopal complex? ∙ Is there always a way to rearrange the pieces in the Mienrenken tile to get a convex polytopal complex?
∙ The domain of the WT map is SSn If we restrict ourselves to Gln what is the image? What is its topology? ∙ In experimentation, the WT map seems to preserve both the Birkhoff polytope and the ASM polytope. Is this true in general?
Open Questions
∙ Does this embedding of the Hasse diagram extend to other types?
26 ∙ Is there always a way to rearrange the pieces in the Mienrenken tile to get a convex polytopal complex?
∙ The domain of the WT map is SSn If we restrict ourselves to Gln what is the image? What is its topology? ∙ In experimentation, the WT map seems to preserve both the Birkhoff polytope and the ASM polytope. Is this true in general?
Open Questions
∙ Does this embedding of the Hasse diagram extend to other types? ∙ Is there a good way to study the Meinrenken tile as a polytopal complex?
26 ∙ The domain of the WT map is SSn If we restrict ourselves to Gln what is the image? What is its topology? ∙ In experimentation, the WT map seems to preserve both the Birkhoff polytope and the ASM polytope. Is this true in general?
Open Questions
∙ Does this embedding of the Hasse diagram extend to other types? ∙ Is there a good way to study the Meinrenken tile as a polytopal complex? ∙ Is there always a way to rearrange the pieces in the Mienrenken tile to get a convex polytopal complex?
26 ∙ In experimentation, the WT map seems to preserve both the Birkhoff polytope and the ASM polytope. Is this true in general?
Open Questions
∙ Does this embedding of the Hasse diagram extend to other types? ∙ Is there a good way to study the Meinrenken tile as a polytopal complex? ∙ Is there always a way to rearrange the pieces in the Mienrenken tile to get a convex polytopal complex?
∙ The domain of the WT map is SSn If we restrict ourselves to Gln what is the image? What is its topology?
26 Open Questions
∙ Does this embedding of the Hasse diagram extend to other types? ∙ Is there a good way to study the Meinrenken tile as a polytopal complex? ∙ Is there always a way to rearrange the pieces in the Mienrenken tile to get a convex polytopal complex?
∙ The domain of the WT map is SSn If we restrict ourselves to Gln what is the image? What is its topology? ∙ In experimentation, the WT map seems to preserve both the Birkhoff polytope and the ASM polytope. Is this true in general?
26 Thank You!
27