Bitableau bases for Garsia-Haiman modules of hollow type (Q958722)

Garsia-Haiman modules \(\mathbb C[X_n,Y_n]/\mathcal I_{\gamma}\) are quotient rings in the variables \(X_n=\{x_1,\dots,x_n\}\) and \(Y_n=\{y_1,\dots,y_n\}\) that generalize the quotient ring \(\mathbb C[X_n]/\mathcal I\), where \(\mathcal I\) is the ideal generated by the elementary symmetric polynomials \(e_j(X_n)\), \(1\leq j\leq n\). The ideals \(\mathcal I_{\gamma}\) are defined via determinants as follows: Each \(\alpha_i=(\alpha_{i,1},\alpha_{i,2})\in\mathbb N^2\) is identified with the unit square in the first quadrant having \(\alpha_i\) as its vertex nearest the origin. (In this article \(\mathbb N\) includes \(0\).) A \textit{lattice diagram} \(L[\pmb\alpha]=(\alpha_1,\dots,\alpha_n)\) is a sequence of such unit squares and has an associated determinant \(\Delta_{L[\pmb\alpha]}=\Delta_{L[\pmb\alpha]}(X_n,Y_n)=\text{det}(x_{j}^{\alpha_{i,1}}y_{j}^{\alpha_{i,2}})_{ij}\). Any polynomial \(P(X_n,Y_n)\) has a corresponding polynomial of differential operators \(P(\partial_x,\partial_y)=P(\partial_{x_1},\dots,\partial_{x_n},\partial_{y_1},\dots,\partial_{y_n})\), and we define the ideal \(\mathcal I_{L[\pmb\alpha]}=\{P(X_n,Y_n)\in\mathbb C[X_n,Y_n]:P(\partial_x,\partial_y)\Delta_{L[\pmb\alpha]}=0\}\). The quotients \(\mathbb C[X_n,Y_n]/\mathcal I_{L[\pmb\alpha]}\) are called \textit{Garsia-Haiman modules} and were used to study \(q,t\)-Kostka coefficients. The article under review deals with \textit{hollow} lattice diagrams, which are, roughly, subsets of hook shapes obtained by removing a (perhaps trivial) continuous region of squares from each of the arm and the leg of the hook. A hollow lattice diagram is parametrized by a sequence \(\gamma=(m,k,p)\) with \(m\in\mathbb Z^2_{\geq 1}\) and \(k,p\in\mathbb N^2\). We write \(\mathcal I_\gamma, \Delta_\gamma\) for \(\mathcal I_{L[\pmb\alpha_{\gamma}]}, \Delta _{L[\pmb\alpha_{\gamma}]}\). The rings \(R=\mathbb C[X_n,Y_n]/\mathcal I_{\gamma}\) carry \(S_n\)-representations, so we can define the (bi-graded) character \(\text{ch}(R)\), the Frobenius series \(\mathcal F\text{ch}(R)\) and Hilbert series \(\mathcal H (R)\). The main result is Theorem 26, which gives an explicit combinatorial basis \(\mathcal B \mathcal B_{\gamma}\) for \(R\), and may be described as follows: A \textit{standard bitableau} is a pair \((S,U)\), where \(S\) is a standard \(\mathbb N\)-filled tableau, and \(U\) is a column-strict \(\mathcal A'\)-tableau of the same shape as \(S\). (\(\mathcal A'=\mathbb N^2\) with a certain ordering.) If \(S\) is injective, we define the bideterminant \([S,U,]_{\text{det}}\) and bipermanent \([S,U,]_{\text{per}}\) for the bitableau \((S,U)\). The elements of \(\mathcal B \mathcal B_{\gamma}\) are defined by certain bipermanents and complete symmetric functions parametrized by \(\gamma\). To prove Theorem 26, a sequence of ideals \(\mathcal G_{\gamma}\subset\mathcal H_{\gamma}\subset\mathcal J_{\gamma}\subset\mathcal K_{\gamma}\subset\mathcal I_{\gamma}\) is defined, and for each a basis is constructed for the corresponding quotient space and the Hilbert series is computed. Using the correspondence between the elements of \(\mathcal B \mathcal B_{\gamma}\) and the irreducible characters of \(S_n\), \(\text{ch}(R)\) (Theorem 1) and \(\mathcal F\text{ch}(R)\) (Corollary 2) are deduced. The article is clearly written and includes numerous examples which illustrate important definitions and constructions.