∗
The Compositional Delta Conjecture
Alessandro Iraci
joint work with Michele D’Adderio and Anna Vanden Wyngaerd 2 september, 2019
Diagonal Harmonics
LetR:=C[x1, . . . , xn, y1, . . . , yn]on whichSn acts diagonally:
σf(x1, . . . , xn, y1, . . . , yn) :=f(xσ1, . . . , xσn, yσ1, . . . , yσn)
LetI be the ideal generated by the constant free invariants of this action.
DHn:=R/I
R is bi-graded,I is homogeneous and invariant so DHn is a bi-graded Sn-module.
Diagonal Harmonics
LetR:=C[x1, . . . , xn, y1, . . . , yn]on whichSn acts diagonally:
σf(x1, . . . , xn, y1, . . . , yn) :=f(xσ1, . . . , xσn, yσ1, . . . , yσn) LetI be the ideal generated by the constant free invariants of this action.
DHn:=R/I
R is bi-graded,I is homogeneous and invariant so DHn is a bi-graded Sn-module.
Diagonal Harmonics
LetR:=C[x1, . . . , xn, y1, . . . , yn]on whichSn acts diagonally:
σf(x1, . . . , xn, y1, . . . , yn) :=f(xσ1, . . . , xσn, yσ1, . . . , yσn) LetI be the ideal generated by the constant free invariants of this action.
DHn:=R/I
R is bi-graded,I is homogeneous and invariant so DHn is a bi-graded Sn-module.
Diagonal Harmonics
LetR:=C[x1, . . . , xn, y1, . . . , yn]on whichSn acts diagonally:
σf(x1, . . . , xn, y1, . . . , yn) :=f(xσ1, . . . , xσn, yσ1, . . . , yσn) LetI be the ideal generated by the constant free invariants of this action.
DHn:=R/I
R is bi-graded,I is homogeneous and invariant so DHn is a bi-graded Sn-module.
We have
Diagonal Harmonics
LetR:=C[x1, . . . , xn, y1, . . . , yn]on whichSn acts diagonally:
σf(x1, . . . , xn, y1, . . . , yn) :=f(xσ1, . . . , xσn, yσ1, . . . , yσn) LetI be the ideal generated by the constant free invariants of this action.
DHn:=R/I
R is bi-graded,I is homogeneous and invariant so DHn is a bi-graded Sn-module.
We dene its bi-graded Frobenius characteristic Fq,t(DHn) := X
V⊆DHn irreducible
qdegx(V)tdegy(V)sλ(V) ∈ΛQ(q,t)
The nabla operator
TheMacdonald polynomials,{Hµ}µ`d form a basis of the ringΛ(d)Q(q,t)
I Generalisation of Schur, Hall-Littlewood, Jack . . . I Schur positive.
I Bi-graded Frobenius characteristic of submodules ofDHn (Garsia-Haiman).
Denition
Thenabla operator is the linear operator dened by
∇: ΛQ(q,t)→ΛQ(q,t) Hµ7→e|µ|[Bµ]Hµ Theorem (Haiman)
The bi-graded Frobenius characteristic ofDHn is ∇en.
The nabla operator
TheMacdonald polynomials,{Hµ}µ`d form a basis of the ringΛ(d)Q(q,t) I Generalisation of Schur, Hall-Littlewood, Jack . . .
I Schur positive.
I Bi-graded Frobenius characteristic of submodules ofDHn (Garsia-Haiman).
Denition
Thenabla operator is the linear operator dened by
∇: ΛQ(q,t)→ΛQ(q,t) Hµ7→e|µ|[Bµ]Hµ Theorem (Haiman)
The bi-graded Frobenius characteristic ofDHn is ∇en.
The nabla operator
TheMacdonald polynomials,{Hµ}µ`d form a basis of the ringΛ(d)Q(q,t) I Generalisation of Schur, Hall-Littlewood, Jack . . .
I Schur positive.
I Bi-graded Frobenius characteristic of submodules ofDHn (Garsia-Haiman).
Denition
Thenabla operator is the linear operator dened by
∇: ΛQ(q,t)→ΛQ(q,t) Hµ7→e|µ|[Bµ]Hµ Theorem (Haiman)
The bi-graded Frobenius characteristic ofDHn is ∇en.
The nabla operator
TheMacdonald polynomials,{Hµ}µ`d form a basis of the ringΛ(d)Q(q,t) I Generalisation of Schur, Hall-Littlewood, Jack . . .
I Schur positive.
I Bi-graded Frobenius characteristic of submodules ofDHn (Garsia-Haiman).
Denition
Thenabla operator is the linear operator dened by
∇: ΛQ(q,t)→ΛQ(q,t) Hµ7→e|µ|[Bµ]Hµ Theorem (Haiman)
The bi-graded Frobenius characteristic ofDHn is ∇en.
The nabla operator
TheMacdonald polynomials,{Hµ}µ`d form a basis of the ringΛ(d)Q(q,t) I Generalisation of Schur, Hall-Littlewood, Jack . . .
I Schur positive.
I Bi-graded Frobenius characteristic of submodules ofDHn (Garsia-Haiman).
Denition
Thenabla operator is the linear operator dened by
∇: ΛQ(q,t)→ΛQ(q,t) Hµ7→e|µ|[Bµ]Hµ
Theorem (Haiman)
The bi-graded Frobenius characteristic ofDHn is ∇en.
The nabla operator
TheMacdonald polynomials,{Hµ}µ`d form a basis of the ringΛ(d)Q(q,t) I Generalisation of Schur, Hall-Littlewood, Jack . . .
I Schur positive.
I Bi-graded Frobenius characteristic of submodules ofDHn (Garsia-Haiman).
Denition
Thenabla operator is the linear operator dened by
∇: ΛQ(q,t)→ΛQ(q,t) Hµ7→e|µ|[Bµ]Hµ Theorem (Haiman)
The bi-graded Frobenius characteristic ofDHn is∇en.
The Shue Theorem
Theorem (Carlsson-Mellit)
∇en= X
D∈LD(n)
qdinv(D)tarea(D)xD
2 4 5
1 3
3 6
1
The Shue Theorem
Theorem (Carlsson-Mellit)
∇en= X
D∈LD(n)
qdinv(D)tarea(D)xD
2 4 5
1 3
3 6
1
LD(n): labelled Dyck paths of size n
The Shue Theorem
Theorem (Carlsson-Mellit)
∇en= X
D∈LD(n)
qdinv(D)tarea(D)xD
5 1 3
3 6
1
Area: number of whole squares between the path andy =x.
The Shue Theorem
Theorem (Carlsson-Mellit)
∇en= X
D∈LD(n)
qdinv(D)tarea(D)xD
2 4 5
1 3
3 6
1
2
3
Dinv: count the number of pairs I same diagonal,
lower label < upper label (primary dinv)
I lower step one diagonal above upper steplower label > upper label
(secondary dinv)
The Shue Theorem
Theorem (Carlsson-Mellit)
∇en= X
D∈LD(n)
qdinv(D)tarea(D)xD
5 1 3
3 6
1
5 1
Dinv: count the number of pairs I same diagonal,
lower label < upper label (primary dinv)
I lower step one diagonal above upper step
The Shue Theorem
Theorem (Carlsson-Mellit)
∇en= X
D∈LD(n)
qdinv(D)tarea(D)xD
2 4 5
1 3
3 6
1
xD :=
n
Y
i=1
xli(D)
whereli(D) is the label of thei-th vertical step ofD.
x21x2x23x4x5x6
The Compositional Shue Theorem
Theorem (Carlsson-Mellit)
∇Cα= X
D∈LD(α)
qdinv(D)tarea(D)xD
5 1 3
3 6
1 We have
X
αn
Cα =en
LD(α): labelled Dyck paths withdiagonal compositionα.
Combinatorial recursion
Theorem (Zabrocki) SetDq,t(α) = P
D∈D(α)
qdinv(D)tarea(D). We have the recursion
Dq,t(a, α) =ta−1 X
βa−1
q`(α)Dq,t(α, β)
for a >0, with initial conditionsDq,t(∅) = 1.
Combinatorial recursion
Theorem (Zabrocki) SetDq,t(α) = P
D∈D(α)
qdinv(D)tarea(D). We have the recursion
Dq,t(a, α) =ta−1 X
βa−1
q`(α)Dq,t(α, β)
for a >0, with initial conditionsDq,t(∅) = 1.
Combinatorial recursion
Theorem (Zabrocki) SetDq,t(α) = P
D∈D(α)
qdinv(D)tarea(D). We have the recursion
Dq,t(a, α) =ta−1 X
βa−1
q`(α)Dq,t(α, β)
for a >0, with initial conditionsDq,t(∅) = 1.
Combinatorial recursion
Theorem (Zabrocki) SetDq,t(α) = P
D∈D(α)
qdinv(D)tarea(D). We have the recursion
Dq,t(a, α) =ta−1 X
βa−1
q`(α)Dq,t(α, β)
for a >0, with initial conditionsDq,t(∅) = 1.
Combinatorial recursion
Theorem (Zabrocki) SetDq,t(α) = P
D∈D(α)
qdinv(D)tarea(D). We have the recursion
Dq,t(a, α) =ta−1 X
βa−1
q`(α)Dq,t(α, β)
for a >0, with initial conditionsDq,t(∅) = 1.
Combinatorial recursion
Theorem (Zabrocki) SetDq,t(α) = P
D∈D(α)
qdinv(D)tarea(D). We have the recursion
Dq,t(a, α) =ta−1 X
βa−1
q`(α)Dq,t(α, β)
for a >0, with initial conditionsDq,t(∅) = 1.
Super-diagonal coinvariants and ∆
fLetR:=C[x1, . . . , xn, y1, . . . , yn, θ1, . . . , θn]whereθiθj =−θjθi.
Consider again the diagonal action of Sn onR and letI be the ideal generated by the constant free invariants of this action.
Mn:=R/I
R is bi-graded,I is homogeneous and invariant so Mnis a bi-graded Sn-module.
Denition
Forf ∈ΛQ(q,t) we dene theDelta operatorsas
∆f: ΛQ(q,t)→ΛQ(q,t):Hµ7→f[Bµ]Hµ
∆0f: ΛQ(q,t)→ΛQ(q,t):Hµ7→f[Bµ−1]Hµ Conjecture (Zabrocki)
The bi-graded Frobenius characteristic of the submodule of Mn in θ-degreek is∆0en−k−1en.
Super-diagonal coinvariants and ∆
fLetR:=C[x1, . . . , xn, y1, . . . , yn, θ1, . . . , θn]whereθiθj =−θjθi. Consider again the diagonal action of Sn onR and letI be the ideal generated by the constant free invariants of this action.
Mn:=R/I
R is bi-graded,I is homogeneous and invariant so Mnis a bi-graded Sn-module.
Denition
Forf ∈ΛQ(q,t) we dene theDelta operatorsas
∆f: ΛQ(q,t)→ΛQ(q,t):Hµ7→f[Bµ]Hµ
∆0f: ΛQ(q,t)→ΛQ(q,t):Hµ7→f[Bµ−1]Hµ Conjecture (Zabrocki)
The bi-graded Frobenius characteristic of the submodule of Mn in θ-degreek is∆0en−k−1en.
Super-diagonal coinvariants and ∆
fLetR:=C[x1, . . . , xn, y1, . . . , yn, θ1, . . . , θn]whereθiθj =−θjθi. Consider again the diagonal action of Sn onR and letI be the ideal generated by the constant free invariants of this action.
Mn:=R/I
R is bi-graded,I is homogeneous and invariant so Mnis a bi-graded Sn-module.
Denition
Forf ∈ΛQ(q,t) we dene theDelta operatorsas
∆f: ΛQ(q,t)→ΛQ(q,t):Hµ7→f[Bµ]Hµ
∆0f: ΛQ(q,t)→ΛQ(q,t):Hµ7→f[Bµ−1]Hµ
Conjecture (Zabrocki)
The bi-graded Frobenius characteristic of the submodule of Mn in θ-degreek is∆0en−k−1en.
Super-diagonal coinvariants and ∆
fLetR:=C[x1, . . . , xn, y1, . . . , yn, θ1, . . . , θn]whereθiθj =−θjθi. Consider again the diagonal action of Sn onR and letI be the ideal generated by the constant free invariants of this action.
Mn:=R/I
R is bi-graded,I is homogeneous and invariant so Mnis a bi-graded Sn-module.
Denition
Forf ∈ΛQ(q,t) we dene theDelta operatorsas
∆f: ΛQ(q,t)→ΛQ(q,t):Hµ7→f[Bµ]Hµ
∆0f: ΛQ(q,t)→ΛQ(q,t):Hµ7→f[Bµ−1]Hµ
The Delta Conjecture
Conjecture (Haglund-Remmel-Wilson)
∆0en−k−1en= X
D∈LD(n)∗k
qdinv(D)tarea(D)xD
The Delta Conjecture
Conjecture (Haglund-Remmel-Wilson)
∆0en−k−1en= X
D∈LD(n)∗k
qdinv(D)tarea(D)xD
∗
∗
4 5
1 3
3 6
1
LD(n)∗k: labelled Dyck paths of size n withkdecoratedrises
The Delta Conjecture
Conjecture (Haglund-Remmel-Wilson)
∆0en−k−1en= X
D∈LD(n)∗k
qdinv(D)tarea(D)xD
∗
∗
2 4 5
1 3
3 6
1
Area: number of whole squares between the path andy =x and in rows not containing decorated rises.
The Delta Conjecture
Conjecture (Haglund-Remmel-Wilson)
∆0en−k−1en= X
D∈LD(n)∗k
qdinv(D)tarea(D)xD
∗
∗
4 5
1 3
3 6
1 3
5 1
Dinv,xD: same as for the undecorated case
The Delta Conjecture
Conjecture (Haglund-Remmel-Wilson)
∆0en−k−1en= X
D∈LD(n)∗k
qdinv(D)tarea(D)xD
Denition
Forf ∈ΛQ(q,t) we dene the following operators onΛQ(q,t)
Π:=X
i∈N
(−1)i∆ei Θf :=Πf
X (1−q)(1−t)
Π−1
Theorem (D'Adderio-Iraci-Vanden Wyngaerd) Θek∇en−k = ∆0en−k−1en
The Compositional Delta Conjecture
Conjecture (D'Adderio-Iraci-Vanden Wyngaerd) Θek∇Cα = X
D∈LD(α)∗k
qdinv(D)tarea(D)xD
∗
∗
4 5
1 3
3 6
1 LD(α)∗k: labelled decorated Dyck paths withdiagonal compositionα. Rows containing decorated rises do not count.
α = (4,2)
Combinatorial recursion
∗
∗
∗
∗
∗
∗
∗
Combinatorial recursion
Theorem (D'Adderio-Iraci-Vanden Wyngaerd) We have the combinatorial recursion, fora, k >0,
Dq,t(a, α)∗k =ta−1 X
βa−1
q`(α)Dq,t(α, β)∗k
+ta−1X
βa
q`(α)Dq,t(α, β)∗k−1
with initial conditions Dq,t(∅)∗k=δk,0 and Dq,t(α)∗0 =Dq,t(α).
Towards a Theta Conjecture
I Algebra side: diagonal coinvariants of
R:=C[x1, . . . , xn, y1, . . . , yn, θ1, . . . , θn, η1, . . . , ηn]
whereθiθj =−θjθi andηiηj =−ηjηi inθ-degreek andη-degreel. I Symmetric function side: ΘekΘel∇en−k−l.
I Combinatorial side: labelled Dyck paths with decorated rises and valleys. Statistics?
Thank you for your attention!
t3 t2
t qt q2t 1 q q2 q3
Forf ∈ΛQ(q,t) we setf[Bµ]to be f evaluated in the content of this picture