--- title: Tight Cylindric Partitions url: https://www.emergentmind.com/topics/tight-cylindric-partitions type: topic --- # Tight Cylindric Partitions Tight cylindric partitions are cylindric partitions subject to an additional exclusion condition: for every integer \(j \ge 1\), there exists at least one row in which the part \(j\) does not occur. Equivalently, no part value is present in all rows simultaneously. For a profile \(c=(c_1,\dots,c_r)\in \mathbb Z_{\ge 0}^r\) of rank \(r\ge 2\) and level \(\ell=c_1+\cdots+c_r\), the tight class \(T_c\) refines the ordinary cylindric partition class \(C_c\), and its bivariate generating function \(T_c(z,q)\) records the maximum part and the total weight. In the modern formulation, tight cylindric partitions are characterized by exact principal specializations of \(\mathfrak{sl}_r\)-characters, while earlier work on cylindric plane partitions and cylindric partitions supplied the product formulas, \((q,t)\)-weights, path models, asymptotics, and partition-analysis methods from which the tight theory emerged [2508.15113], [1204.4583]. ## 1. Definitions, profiles, and basic statistics Fix \(r\ge 2\) and a profile \(c=(c_1,c_2,\dots,c_r)\in \mathbb Z_{\ge 0}^r\) with level \(\ell=c_1+\cdots+c_r\ge 1\). A cylindric partition of profile \(c\) is an \(r\)-tuple of partitions \(\pi=(\pi^{(1)},\dots,\pi^{(r)})\) satisfying the periodic dominance conditions \[ \pi^{(i)}_j \ge \pi^{(i+1)}_{j+c_{i+1}} \qquad (i=1,\dots,r-1), \] together with the wrap-around condition \[ \pi^{(r)}_j \ge \pi^{(1)}_{j+c_1}. \] Parts beyond the length of a partition are regarded as \(0\). The rank is \(r\), the level is \(\ell\), the largest part is \(\max(\pi)\), and the weight is \[ \mathrm{wt}(\pi)=\text{the sum of all parts across the }r\text{ rows}. \] The associated dominant integral highest weight for \(\mathfrak{sl}_r\) is \[ \Lambda(c)=c_r\Lambda_0+c_1\Lambda_1+\cdots+c_{r-1}\Lambda_{r-1}. \] A cylindric partition is tight if for every integer \(j\ge 1\) there exists at least one row \(i\) such that \(j\) does not occur as a part of \(\pi^{(i)}\). The set of all cylindric partitions of profile \(c\) is denoted \(C_c\), and the tight subset is denoted \(T_c\). The bivariate generating functions are \[ C_c(z,q)=\sum_{\pi\in C_c} z^{\max(\pi)}q^{\mathrm{wt}(\pi)}, \qquad T_c(z,q)=\sum_{\pi\in T_c} z^{\max(\pi)}q^{\mathrm{wt}(\pi)}. \] The literature summarized here uses several profile formalisms. In the cylindric plane partition framework, one fixes a binary string \(\pi=(\pi_1,\dots,\pi_T)\in\{0,1\}^T\) and studies periodic interlacing sequences \((\mu^0,\mu^1,\dots,\mu^T)\) with \(\mu^0=\mu^T\). In the asymptotic framework, one fixes a sign profile \(\delta=(\delta_1,\dots,\delta_h)\) with \(\delta_i\in\{1,-1\}\). A plausible implication is that the subject is organized around several equivalent periodic-interlacing encodings, each optimized for a different problem: exact products, asymptotics, or representation-theoretic interpretation. ## 2. Earlier cylindric models and the emergence of tightness Cylindric plane partitions may be thought of as a natural generalization of reverse plane partitions. For a binary profile \(\pi=(\pi_1,\dots,\pi_T)\), a cylindric plane partition is a periodic interlacing sequence \[ (\mu^0,\mu^1,\ldots,\mu^T), \qquad \mu^0=\mu^T, \] such that for each \(k=1,\dots,T\), - if \(\pi_k=1\), then \(\mu^k/\mu^{k-1}\) is a horizontal strip; - if \(\pi_k=0\), then \(\mu^{k-1}/\mu^k\) is a horizontal strip. The weight is \[ |\mathfrak c|=|\mu^1|+\cdots+|\mu^T|. \] This framework also has a non-intersecting lattice-path model on a cylinder, with a precise bijection between cylindric plane partitions and path families. In this earlier setting, the paper does not define “tight” explicitly. The closest notion is the “minimal family of non-intersecting lattice paths.” In the path model, paths live on a triangular lattice wrapped on a cylinder of period \(T\), and a minimal family is intended to have no extraneous vertical gaps between paths beyond those forced by non-intersection. The summary further states that, in the literature intuition, “tight” often means configurations minimizing gaps, and a natural specialization is obtained by fixing an alternating profile \(\pi\), such as \[ \pi=(1,0,1,0,\dots,1,0), \] and imposing minimal vertical spacing between paths at each \(x\), so that all surface cubes have level \(1\). Under these constraints, the family is “tight” in the sense of minimal separation; locally, steps alternate, eliminating \(UU/DD\) patterns. In terms of the interlacing sequence, this corresponds to alternately adding and removing horizontal strips, forcing the thinnest possible evolution [1204.4583]. This interpretation yields a concrete small case. For \(T=2\) and \(\pi=(1,0)\), the Schur specialization gives \[ \sum_{\mathfrak c\in \mathrm{CPP}(1,0)} z^{|\mathfrak c|} = \prod_{m\ge 1}\frac{1}{1-z^m}, \] the ordinary partition generating function. In the Hall–Littlewood specialization \(q=0\), tightness forces only arm-length-zero, level-\(1\) peak/valley contributions to remain. This does not yet constitute the modern definition of tight cylindric partitions, but it isolates the minimal-gap geometry that the later theory formalizes. ## 3. Character identities and product formulas The defining structural feature of tight cylindric partitions is the disappearance of the extra factor that occurs for ordinary cylindric partitions. It is well known that \[ \sum_{\pi\in C_c} q^{\mathrm{wt}(\pi)} = (q^r;q^r)_\infty^{-1}\cdot \Pr(L_{\mathfrak{sl}_r}(\Lambda(c))), \] where \(\Pr(\cdot)\) denotes principal specialization. By contrast, tight cylindric partitions satisfy the exact identity \[ T_c(1,q)=\Pr(L_{\mathfrak{sl}_r}(\Lambda(c))). \] Thus the tight class matches the principal specialization with no extra factor. Let \(m=r+\ell\) and \(\theta(a;q):=(a;q)_\infty(q/a;q)_\infty\). The univariate generating function has the periodic product form \[ T_c(1,q)=\Pr(L_{\mathfrak{sl}_r}(\Lambda(c))) = (q^r;q^r)_\infty (q^m;q^m)_\infty^{r-1} (q;q)_\infty^r \cdot \prod_{1\le i0\}. \] For any nonempty subset \(J\subseteq I_c\), let \(c(J)\) be the composition determined by \[ c_i(J)= \begin{cases} c_i-1, & \text{if } i\in J \text{ and } i-1\notin J,\\ c_i+1, & \text{if } i\notin J \text{ and } i-1\in J,\\ c_i, & \text{otherwise}, \end{cases} \] with \(c_0:=c_r\) and indices modulo \(r\). The tight analog of the Corteel–Welsh recurrences is \[ T_c(z,q)+\frac{zq^r}{1-zq^r}\,T_c(zq^r,q) = \sum_{\varnothing\ne J\subseteq I_c} (-1)^{|J|-1}\, \frac{T_{c(J)}(zq^{|J|},q)}{1-zq^{|J|}}. \] The term \[ \frac{zq^r}{1-zq^r}\,T_c(zq^r,q) \] accounts for partitions whose maximum appears in the first position of all \(r\) rows and is later deleted; the right-hand side is an inclusion–exclusion sum over “addboxes” operations. For two rows, one obtains “diamond relations.” If \(1\le i\le \ell-1\), then \[ T_{(\ell-i,i)}(z,q)+\frac{zq^2}{1-zq^2}\,T_{(\ell-i,i)}(zq^2,q) = \frac{T_{(\ell-i+1,i-1)}(zq,q)}{1-zq} + \frac{T_{(\ell-i-1,i+1)}(zq,q)}{1-zq} - \frac{T_{(\ell-i,i)}(zq^2,q)}{1-zq^2}. \] There is also a recurrence without inclusion–exclusion: \[ T_{(c_1,c_2)}(z,q) = 1+ \sum_{a=0,\ a\ne c_2}^{\ell} \frac{zq^{|a-c_2|}}{1-zq^{|a-c_2|}} \,T_{(\ell-a,a)}(zq^{|a-c_2|},q). \] These recurrences are solved explicitly in the two-row case. For \(0\le b\le \lfloor \ell/2\rfloor\), \[ T_{(\ell-b,b)}(z,q)=T_{(b,\ell-b)}(z,q) \] equals each of the following multisums. First, in terms of independent \(n_j\ge 0\), with \(N_j:=n_j+n_{j+1}+\cdots+n_\ell\), \[ T_{(\ell-b,b)}(z,q) = \sum_{n_1,\dots,n_\ell\ge 0} \frac{ z^{N_1} q^{(N_1^2+\cdots+N_\ell^2)/2 + (n_1+n_3+\cdots+n_{2b-1})/2 + (N_{2b+1}+\cdots+N_\ell)/2} \cdot (q)_{N_1} }{ (zq;q)_{N_1}\cdot (q)_{n_1}\cdots (q)_{n_\ell} }. \] Second, in terms of nonincreasing \(N_1\ge \cdots \ge N_\ell\ge 0\), \[ T_{(\ell-b,b)}(z,q) = \sum_{N_1,\dots,N_\ell\ge 0} \frac{ z^{N_1} q^{\sum_{i=1}^{\ell}\binom{N_i+1}{2}-\sum_{i=1}^{b}N_{2i}} \cdot (q)_{N_1} }{ (zq;q)_{N_1}\cdot (q)_{N_1-N_2}\cdots (q)_{N_{\ell-1}-N_\ell}\cdot (q)_{N_\ell} }. \] At \(z=1\), these specialize to known parity refinements of Andrews–Gordon and Andrews–Bressoud multisums, up to the standard \(q\)-dilation noted in the paper. For level \(1\), more generally, \[ T_c(z,q) = 1+\sum_{n\ge 1} z^n \left[ \frac{(q^r;q^r)_n}{(q)_n} - \frac{(q^r;q^r)_{n-1}}{(q)_{n-1}} \right], \] and for \(r=2\) this begins \[ T_c(z,q)=1+zq+z^2(q^2+q^3)+z^3(q^3+q^4+q^5+q^6)+\cdots \] [2508.15113]. ## 5. Combinatorial realizations: paths, abaci, DHK partitions, and partition analysis The tight condition has several combinatorial realizations. In the cylindric plane partition model, the \((q,t)\)-Macdonald weight can be expressed plethystically as \[ W_{\mathfrak c}(q,t)=\Omega\!\left[(q-t)\,\mathcal D_{\mathfrak c}(q,t)\right], \] where \[ \mathcal D_{\mathfrak c}(q,t) = \sum_{s\in \mathrm{peak}(\mathfrak c)} q^{a_{\mathfrak c}(s)}t^{\ell_{\mathfrak c}(s)}- \sum_{s\in \mathrm{valley}(\mathfrak c)} q^{a_{\mathfrak c}(s)}t^{\ell_{\mathfrak c}(s)}. \] Here “peak” means a local \(UD\) pattern, “valley” means \(DU\), the leg length is the level in the path model, and the arm length counts inversion gaps. Under the tight specialization described in that paper—alternating profile and minimal vertical separation—every surface cube has level \(1\), and the weight simplifies because all contributing peaks and valleys occur at level \(1\) and with minimal arm lengths [1204.4583]. In the dedicated tight theory for two rows, the relevant combinatorial model is the 2-string abacus. A 2-string abacus of type \(A_{\ell-a,a}\) has yoked pairs of beads, and a yoke has shape \(s\in\{0,1,\dots,\ell\}\) if the top bead is \(s\) positions to the right of the bottom bead. Tightness means that no yoke can be shifted further left; equivalently, yokes are as close together as possible. There is a bijection between 2-rowed cylindric partitions of profile \((\ell-a,a)\) and 2-string abaci of type \(A_{\ell-a,a}\), and tight cylindric partitions correspond exactly to tight abaci. The same two-row tight objects are also in bijection with the colored partitions studied by Dousse, Hardiman, and Konan. A DHK partition of type \((a,\ell)\) is a weakly decreasing colored sequence \[ \lambda_1\ge \lambda_2\ge \cdots \ge \lambda_s=0 \] with colors \(u_i\in\{0,1,\dots,\ell\}\) satisfying \[ \lambda_i-\lambda_{i+1}=|u_i-u_{i+1}|, \] with the final \(0\) colored \(a\) and exactly one part equal to \(0\). If \[ D_{a,\ell}(z,q)=\sum_{\lambda\in \mathrm{DHK}_{a,\ell}} z^{\#(\lambda)}q^{\mathrm{wt}(\lambda)}, \] then the same recurrence as for the two-row tight generating function implies \[ T_{(\ell-a,a)}(z,q)=D_{a,\ell}(z,q). \] At \(z=1\), \[ D_{a,\ell}(1,q) = \Pr(L_{\mathfrak{sl}_2}((\ell-a)\Lambda_0+a\Lambda_1)) = \frac{(q^{a+1},q^{\ell-a+1},q^{\ell+2};q^{\ell+2})_\infty}{(q;q)_\infty\cdot (q;q^2)_\infty} \] [2508.15113]. A distinct but related two-row approach is MacMahon’s partition analysis. For \(k=2\) and \(c=(c_1,c_2)\), a cylindric partition is a pair of weakly decreasing rows \(a\) and \(b\) with inter-row inequalities of the form \[ a_i\ge b_{i+c_2}, \qquad b_i\ge a_{i+c_1}. \] That paper does not define “tight cylindric partitions” explicitly. Instead, it states that in the literature “tightness” typically refers to saturating the difference or interlacing constraints, and that within the partition-analysis framework tightness can be imposed by replacing weak inequalities by equalities or fixed minimal differences. The \(\Omega_{\ge}\)-operator elimination rules remain available; for example, \[ \Omega_{\ge}\!\left[\frac{1}{(1-A\lambda)(1-B/\lambda)}\right] = \frac{1}{(1-A)(1-AB)}. \] The paper develops explicit generating functions for ordinary two-row cylindric partitions, including \[ CP_{(1,1)}(n)=\frac{(-q;q^2)_n}{(q;q)_{2n}}, \qquad CP_{(2,0)}(n)=\frac{(-q^2;q^2)_{n-1}}{(q;q)_{2n}}, \] and states that the same pipeline should apply after imposing equality or fixed-difference constraints [2501.19272]. ## 6. Asymptotics, profile dependence, and special cases For ordinary cylindric partitions of width \(h\) and sign profile \(\delta=(\delta_1,\dots,\delta_h)\) with \(\delta_i\in\{1,-1\}\), let \[ W_\delta:=\{h\}\cup \{j-i: i\delta_j\}\cup \{h+i-j: i\delta_j}} \Gamma\!\Bigl(\frac{j-i}{h}\Bigr) \prod_{\substack{i\pi_j}} \frac{(t z^{j-i};q)_\infty}{(z^{j-i};q)_\infty}. \] For alternating profile in the tight-path sense, the \(T=2\) Schur case again collapses to the ordinary partition product [1204.4583]. The recent theory also records concrete tight examples. For the two-row tight profile \((3,0)\), \[ \pi^{(1)}=(10,8,4,3,3),\qquad \pi^{(2)}=(9,5,1) \] is tight, with \[ \max(\pi)=10,\qquad \mathrm{wt}(\pi)=43, \] so its contribution to \(T_{(3,0)}(z,q)\) is \(z^{10}q^{43}\). The associated tight abacus has yokes of shapes \(0,1,0,1,1,1,0,1,3,3,2\), and its cumulative-left-vacancy data yields the corresponding DHK partition after reversing [2508.15113]. ## 7. Conceptual significance and open directions Tight cylindric partitions sit at the intersection of three strands of the literature. The first is the exact enumeration of cylindric plane partitions via product formulas, \((q,t)\)-vertex operators, and lattice-path models. The second is the asymptotic theory of cylindric partitions, where profile dependence is visible both in the exponential rate and in the Gamma-product leading constant. The third is the two-row finite-\(q\) theory, where MacMahon partition analysis, Rogers–Ramanujan-type identities, Andrews–Gordon and Bressoud refinements, and Bailey-type hierarchies emerge from explicit recurrences and alternating \(q\)-binomial sums. The 2025 tight theory isolates a class for which the representation-theoretic meaning is especially clean: \(T_c(1,q)\) is exactly the principal specialization of the \(\mathfrak{sl}_r\)-character. In two rows, the theory simultaneously admits functional equations, multisum closed forms, an abacus realization, and a bijection with DHK colored partitions. This positions tight cylindric partitions as a sharpened version of cylindric partition combinatorics in which the global overlap of equal parts across all rows is forbidden rather than merely weighted away. The open directions stated in the literature include a direct proof of the two-row closed forms independent of the \(q\)-difference telescoping method, extension of bivariate analyses to tight \(r\)-rowed cases with \(r\ge 3\), explicit bijections between tight cylindric partitions and Kleshchev multipartitions for \(\mathfrak{sl}_2\) in Rogers–Ramanujan-type identities, investigation of unimodality phenomena for the \(z\)-coefficients of both tight and ordinary cylindric generating functions, clarification of the representation-theoretic meaning of the maximum-part statistic, and analogs of results such as Gessel–Krattenthaler determinantal formulas for the tight setting [2508.15113]. These directions are closely aligned with the methods already present in the surrounding literature. The partition-analysis paper states that equality or fixed-difference constraints can be encoded directly in the crude generating function and then treated by the same \(\Omega_{\ge}\)-elimination machinery, suggesting a systematic route toward tight two-row generating functions and further Bailey-generated hierarchies [2501.19272]. Source: https://www.emergentmind.com/topics/tight-cylindric-partitions