Multi-Symmetric Schur Functions

Given a commutative ring R and free variables x,y define the isobaric divided difference operator \xi_{x,y}: R[x,y] \rightarrow R[x,y] by

For 1\leq i \leq n-1 we will write \xi_{i}:= \xi_{x_i,x_{i+1}} for the corresponding operator on \mathbb{Q}[x_1,\ldots,x_n].

[Dem_1974] Let n \geq 1. Define the key polynomials to be the unique collection of polynomials \{\mathcal{K}_{\alpha}(x_1,\ldots, x_n) \}_{\alpha \in \mathbb{Z}_{\geq 0}^{n}} \subset \mathbb{Q}[x_1,\ldots,x_n] determined by the following properties:

• If \alpha_1 \geq \ldots \geq \alpha_n then

  \mathcal{K}_{\alpha}(x_1,\ldots, x_n)= x^{\alpha}.

• Whenever \alpha_i > \alpha_{i+1}

  \mathcal{K}_{s_i(\alpha)}(x_1,\ldots, x_n) = \xi_{i}(\mathcal{K}_{\alpha}(x_1,\ldots, x_n)).

[haglund2007combinatorial] For a composition \mu = (\mu_1,\dots,\mu_n), define the column diagram of \mu as

This is represented by a collection of boxes in positions given by dg'(\mu). The augmented diagram of \mu is given by

Visually, to get \widehat{dg}(\mu) we are adding a bottom row of boxes on length n below the diagram dg'(\mu). Given u = (i,j) \in dg'(\mu) define the following:

• \operatorname{leg}(u) := \{(i,j') \in dg'(\mu): j' > j\}

• \operatorname{arm}^{\operatorname{left}}(u) := \{(i',j) \in dg'(\mu): i'<i, \mu_{i'} \leq \mu_i\}

• \operatorname{arm}^{\operatorname{right}}(u):= \{(i',j-1) \in \widehat{dg}(\mu): i'>i, \mu_{i'}<\mu_{i}\}

• \operatorname{arm}(u) := \operatorname{arm}^{\operatorname{left}}(u) \cup \operatorname{arm}^{\operatorname{right}}(u)

• \operatorname{lg}(u):= |\operatorname{leg}(u)| = \mu_i -j

• a(u) := |\operatorname{arm}(u)|.

A filling of \mu is a function \sigma: dg'(\mu) \rightarrow \{1,...,n\} and given a filling there is an associated augmented filling \widehat{\sigma}: \widehat{dg}(\mu) \rightarrow \{1,...,n\} extending \sigma with the additional bottom row boxes filled according to \widehat{\sigma}((j,0)) = j for j = 1,\dots,n. Distinct lattice squares u,v \in \mathbb{N}^2 are said to attack each other if one of the following is true:

• u and v are in the same row

• u and v are in consecutive rows and the box in the lower row is to the right of the box in the upper row.

A filling \sigma: dg'(\mu) \rightarrow \{1,\dots,n\} is non-attacking if \widehat{\sigma}(u) \neq \widehat{\sigma}(v) for every pair of attacking boxes u,v \in \widehat{dg}(\mu). For a box u= (i,j) let d(u) = (i,j-1) denote the box just below u.

[haglund2007combinatorial] Let \sigma: \mu \rightarrow \{1,\ldots, n\} be a non-attacking labeling. A co-inversion triple is a triple of boxes (u,v,w) in the diagram \widehat{dg}(\mu) of one of the following two types

that satisfy the following criteria:

• in Type 1 the column containing u and w is strictly taller than the column containing v

• in Type 2 the column containing u and w is weakly taller than the column containing v

• in either Type 1 or Type 2, \widehat{\sigma}(u)< \widehat{\sigma}(v)< \widehat{\sigma}(w) or \widehat{\sigma}(v)< \widehat{\sigma}(w)< \widehat{\sigma}(u) or \widehat{\sigma}(w)< \widehat{\sigma}(u)< \widehat{\sigma}(v).

Informally, in Type 1 we require the entries to strictly increase clockwise and in Type 2 we require the entries to strictly increase counterclockwise.

Given \alpha \in \mathbb{Z}_{\geq 0}^{n}, denote by \mathcal{L}'(\alpha) the set of non-attacking labellings \sigma: \alpha \rightarrow \{1,\ldots, n\} such that \widehat{\sigma} is weakly increasing upwards and has no co-inversion triples.

Let n \geq 1. Define \operatorname{\mathrm{B}}_n to be the Borel subgroup of upper-triangular matrices in \operatorname{\mathrm{GL}}_n and let \operatorname{\mathrm{H}}_n denote the group of diagonal matrices in \operatorname{\mathrm{GL}}_n. Let \mathfrak{b}_n denote the Lie algebra of \operatorname{\mathrm{B}}_n i.e. the set of upper triangular n\times n matrices over \mathbb{C} with the usual commutator. Let \mathcal{U}(\mathfrak{b}_n) denote the universal enveloping algebra of \mathfrak{b}_n. We will write \mathfrak{S}_n for the group of permutations of \{1,\ldots,n\}. For a dominant (i.e. weakly decreasing) integral weight \lambda \in \mathbb{Z}_{\geq 0}^n, let \mathcal{V}^{\lambda} denote the corresponding highest weight representation of \operatorname{\mathrm{GL}}_n.

Given a finite dimensional polynomial representation V of \operatorname{\mathrm{H}}_n, we will denote by \operatorname{\mathrm{char}}(V) \in \mathbb{Z}[x_1,\ldots,x_n] the formal character of V as

where we denote by \alpha the corresponding character of \operatorname{\mathrm{H}}_n.

[Dem_1974] Given a dominant integral weight \lambda \in \mathbb{Z}_{\geq 0}^{n} and \sigma \in \mathfrak{S}_n, define the Demazure module \mathcal{V}^{\lambda}_{\sigma(\lambda)} to be the \operatorname{\mathrm{B}}_n-module

where v \in \mathcal{V}^{\lambda} is any weight vector with weight \sigma(\lambda).

Let \operatorname{\mathbb{Y}} denote the set of all partitions. We will write \ell(\lambda) for the length of the partition \lambda and |\lambda| for the size of the partition. Given a commutative ring R and an infinite set of free variables X = \{x_1,x_2,x_3,\ldots\} we define \Lambda_R(X) to be the ring of symmetric functions over R in the variables X. When R= \mathbb{Q}, we simply write \Lambda(X):= \Lambda_{\mathbb{Q}}(X). We will denote by m_{\lambda}(X) and s_{\lambda}(X) respectively the monomial and Schur symmetric functions respectively. Given a collection of infinite sets of free variables we will write

for the ring of r multi-symmetric functions over R. If R=\mathbb{Q} we write \Lambda(X_1|\cdots|X_r):= \Lambda_{\mathbb{Q}}(X_1|\cdots | X_r). Similarly, we write

and

for the r multi-symmetric monomial, and Schur functions respectively.

We define the plethystic exponential as the element of the graded completion of \Lambda_{\mathbb{Z}}(X) given by

Let \operatorname{\mathcal{P}}_{(n_1|\cdots|n_r)}:= \bigotimes_{i=1}^{r} \mathbb{Q}[x_{i,1},\ldots,x_{i,n_i}]. We let \mathfrak{S}^{(1)}_{n_1}\times \cdots \times \mathfrak{S}^{(r)}_{n_r} act on \operatorname{\mathcal{P}}_{(n_1|\cdots|n_r)} via permutations of the variables x_{i,1},\ldots,x_{i,n_i} where each \mathfrak{S}^{(i)}_{n_i} is a copy of \mathfrak{S}_{n_i}. We will write \Lambda(X_{1}^{(n_1)}|\cdots|X_{r}^{(n_r)}):= \operatorname{\mathcal{P}}_{(n_1|\cdots|n_r)}^{\mathfrak{S}^{(1)}_{n_1}\times \cdots \times \mathfrak{S}^{(r)}_{n_r}} where each X_{i}^{(n_i)} denotes a symmetrized variable set X_{i}^{(n_i)} = x_{i,1}+\ldots+x_{i,n_i}. Let (\lambda^{(1)}|\ldots|\lambda^{(r)}) be an r-tuple of partitions. For all (n_1|\cdots|n_r) with n_i \geq \ell(\lambda^{(i)}) define the multi-symmetric Schur polynomial

For 1\leq i \leq r and n_i \geq 1 we define the map \Xi^{(n_i)}_i: \operatorname{\mathcal{P}}_{(n_1|\cdots|n_r)} \rightarrow \operatorname{\mathcal{P}}_{(n_1|\cdots|n_{i-1}|n_i-1|n_{i+1}|\cdots|n_r)} by

For (\lambda^{(1)}|\cdots |\lambda^{(r)}) \in \operatorname{\mathbb{Y}}^r, define the multi-symmetric Schur function

Given (\lambda^{(1)}|\cdots |\lambda^{(r)}) \in \operatorname{\mathbb{Y}}^r and (n_1|\cdots|n_r) with n_i \geq \ell(\lambda^{(i)}) for all 1\leq i \leq r, define the diagrams

We will require multiple copies of the usual labels \{1,2,3,\ldots\}. We consider labels of the form i_j for i,j \geq 1 which we order by i_j \leq k_{\ell} if either j < \ell or j = \ell and i < k. We naturally extend the definition of co-inversion triples to any lattice diagram with labels of the form i_j for i,j \geq 1 using the ordering just described. Given any such labeling \sigma we define the monomial

We denote by [n_1|\ldots|n_r] the set \{1_1,\ldots,{(n_1)}_1\} \cup \ldots \cup \{1_r,\ldots,{(n_r)}_r \}. Given a labeling

and define

by extending \sigma to the basement boxes via

We will denote by \operatorname{\mathcal{L}}^{(n_1|\cdots|n_r)}(\lambda^{(1)}|\cdots|\lambda^{(r)}) the set of non-attacking labellings

such that \widehat{\sigma} is weakly decreasing upwards and has no co-inversion triples.

Given a labeling \sigma of a lattice diagram, we say that \sigma is inverse semi-standard if \sigma weakly decreases upwards and strictly increasing rightwards along rows. We will write

We define \mathbb{N}(r)_{\infty} by extending the poset \mathbb{N}(r) by adding in elements \omega_1,\ldots,\omega_r and formal sums \omega_i + j for 1\leq i \leq r and j \geq 0 which we order as follows:

for any a,b,c,d,e,f \in \{1,2,3\ldots\}. Note the subscripts here denote which alphabet those values belong in. Informally, everything within each extended alphabet is ordered as one would expect and each extended alphabet is mutually less than the next one. Given a labeling \sigma: dg'(\operatorname{rev}(\lambda^{(1)})*\ldots * \operatorname{rev}(\lambda^{(r)})) \rightarrow \mathbb{N}(r) define \sigma^{\star}:\widehat{dg}(\operatorname{rev}(\lambda^{(1)})*\cdots * \operatorname{rev}(\lambda^{(r)})) by extending \sigma and defining \sigma^{\star} on the basement boxes via

for 1\leq i \leq r and 1 \leq j \leq \ell(\lambda^{(i)}). Define \operatorname{\mathcal{L}}(\lambda^{(1)}|\cdots |\lambda^{(r)}) to be the set of labellings \sigma: \operatorname{rev}(\lambda^{(1)})*\ldots * \operatorname{rev}(\lambda^{(r)}) \rightarrow \mathbb{N}(r) such that \sigma^{\star} is non-attacking, weakly decreasing upwards, and has no co-inversion triples.

We define the multi-symmetric Kostka coefficients K_{(\mu^{(1)}|\cdots|\mu^{(r)})}^{(\lambda^{(1)}|\cdots|\lambda^{(r)})} for (\mu^{(1)}|\cdots|\mu^{(r)}),(\lambda^{(1)}|\cdots|\lambda^{(r)}) \in \operatorname{\mathbb{Y}}^r by

Given (\mu^{(1)}|\cdots|\mu^{(r)}),(\lambda^{(1)}|\cdots|\lambda^{(r)}) \in \operatorname{\mathbb{Y}}^r define \operatorname{\mathcal{L}}_{(\mu^{(1)}|\cdots|\mu^{(r)})}(\lambda^{(1)}|\cdots|\lambda^{(r)}) to be the set of labellings \sigma \in \operatorname{\mathcal{L}}(\lambda^{(1)}|\cdots|\lambda^{(r)}) such that for all 1\leq j \leq r and 1 \leq i \leq \ell(\mu^{(j)})

For \alpha,\beta \in \mathbb{Z}^n_{\geq 0} define the reverse lexicographic ordering <_{RL} by \beta <_{RL} \alpha if there exists some 1\leq k \leq n such that \beta_i = \alpha_i if i > k and \beta_k < \alpha_{k}. For all n_1,\ldots,n_r \geq 1 and (\alpha^{(1)}|\ldots | \alpha^{(r)}),(\beta^{(1)}|\ldots | \beta^{(r)}) \in \mathbb{Z}_{\geq 0}^{n_1}\times \cdots \times \mathbb{Z}_{\geq 0}^{n_r} we write (\beta^{(1)}|\ldots | \beta^{(r)}) \leq_{RL} (\alpha^{(1)}|\ldots | \alpha^{(r)}) if \beta^{(1)}*\cdots * \beta^{(r)} \leq_{RL} \alpha^{(1)}*\cdots * \alpha^{(r)} or equivalently there exists some 1 \leq k \leq r such that for all i > k, \alpha^{(i)} = \beta^{(i)} and \alpha^{(k)} \leq_{RL} \beta^{(k)}.

For \mu,\lambda \in \operatorname{\mathbb{Y}} we define the lexicographic ordering <_L via \mu <_L \lambda if there exists some k \geq 1 such that for all i < k, \mu_i = \lambda_i and \mu_k < \lambda_k where we adjoin 0's to the end of partitions when necessary. For (\mu^{(1)}|\ldots | \mu^{(r)}),(\lambda^{(1)}|\ldots | \lambda^{(r)}) \in \operatorname{\mathbb{Y}}^r we define the ordering \prec via (\mu^{(1)}|\ldots | \mu^{(r)}) \prec (\lambda^{(1)}|\ldots | \lambda^{(r)}) if there exists some 1\leq k \leq r such that for all i > k, \mu^{(i)} = \lambda^{(i)} and \mu^{(k)} <_L \lambda^{(k)}.

[MTY19] For \alpha \in \mathbb{Z}_{\geq 0}^n and 1 \leq i < j \leq n let t_{i,j}(\alpha) be the vector obtained by interchanging \alpha_i and \alpha_j, and let m_{i,j}(\alpha) := \alpha + e_i-e_j where for 1\leq k \leq n we let e_k be the standard basis vector. For \beta \in \mathbb{Z}_{\geq 0}^{n} define \beta <_{\kappa} \alpha if \beta can be obtained from \alpha by a sequence of moves t_{i,j} for \alpha_i < \alpha_j and m_{i,j} for \alpha_i < \alpha_j -1.

We define the partial order \triangleleft on \operatorname{\mathbb{Y}}^r by (\mu^{(1)}|\cdots | \mu^{(r)}) \triangleleft (\lambda^{(1)}|\cdots | \lambda^{(r)}) if (\mu^{(1)}|\cdots | \mu^{(r)}) may be obtained from (\lambda^{(1)}|\cdots | \lambda^{(r)}) by repeatedly applying Move 1, Move 2, or Move 3 from ? above.

Given (n_1|\ldots | n_r) with each n_i \geq 1 define \mathrm{P}(n_1|\ldots|n_r) \leq \mathrm{GL}_{n_1+\ldots+n_r} as the unique parabolic subgroup containing the Borel subgroup \mathrm{B}_n and the Levi subgroup L(n_1|\ldots|n_r):= \mathrm{GL}_{n_1}\times \cdots \times \mathrm{GL}_{n_r} \leq \mathrm{GL}_{n_1+\ldots+n_r}.

For (\lambda^{(1)}|\cdots|\lambda^{(r)}) \in \operatorname{\mathbb{Y}}^r and (n_1|\ldots|n_r) with n_i \geq \ell(\lambda^{(i)}) for all 1\leq i \leq r define the Demazure module

For all (\mu^{(1)}|\cdots |\mu^{(r)}) \in \operatorname{\mathbb{Y}}^r and (n_1|\cdots|n_r) with n_i \geq \ell(\mu^{(i)}) for all 1 \leq i \leq r we let \mathcal{D}^{(n_1|\ldots|n_r)}_{(\mu_1|\cdots|\mu_r)} denote the irreducible L(n_1|\cdots | n_r)-module

Define the (a priori) rational numbers (M_{(\mu^{(1)}|\cdots | \mu^{(r)})}^{(\lambda^{(1)}|\cdots | \lambda^{(r)})})_{(\mu^{(1)}|\cdots | \mu^{(r)}),(\lambda^{(1)}|\cdots | \lambda^{(r)}) \in \operatorname{\mathbb{Y}}^r} by

For the remainder of the paper, the variable z will denote a free variable independent of any other variable sets. We will write \langle z^a \rangle \sum_{n \in \mathbb{Z}} c_n z^n:= c_a for any formal series in z.

To simplify notation, for 1\leq i \leq r and 1\leq j \leq n_{i}-1 we will write

for the divided difference operator \xi^{(i)}_j:= \xi_{x_{i,j},x_{i,j+1}}.

Define the operators D_1^{(n_1|\cdots|n_r)},\ldots,D_{r-1}^{(n_{1}|\cdots | n_r)}: \Lambda(X_{1}^{(n_1)}|\cdots|X_{r}^{(n_r)}) \rightarrow \Lambda(X_{1}^{(n_1)}|\cdots|X_{r}^{(n_r)}) by

Define the operators D_1,\ldots,D_{r-1}: \Lambda(X_1|\cdots|X_r)\rightarrow \Lambda(X_1|\cdots|X_r) by

Given a commutative ring R and an ordered set of free variables (y_1,\ldots,y_n) define the Weyl symmetrizer in the variables (y_1,\ldots,y_n) to be the map W_{y_1,\ldots,y_n}:R[y_1,\ldots,y_n] \rightarrow R[y_1,\ldots,y_n]^{\mathfrak{S}_n} given by

Let R be a commutative ring and let Y=y_1+y_2+\ldots be an infinite set of free variables. For a \in\mathbb{Z}, the Bernstein operator is the map \operatorname{\mathcal{B}}_a:\Lambda_R(Y) \rightarrow \Lambda_R(Y) given by