Morris II Identity: Combinatorics & Flow Polytopes

• Morris II is a fundamental constant-term identity in combinatorics, algebra, and geometry defined through Gamma-product formulas and linked to Catalan and Narayana numbers.
• It interprets multivariate constant terms as volumes and lattice-point counts in flow polytopes, bridging algebraic formulas with geometric combinatorics.
• Its refined forms employ recurrence relations and bijective proofs to generalize classical results and demonstrate inherent symmetry in combinatorial structures.

Morris II, also known as the “second Morris identity,” is a fundamental constant-term identity in combinatorics, algebra, and related fields. It admits interpretations in terms of multivariate constant terms, volumes and lattice-point enumerations of flow polytopes, and combinatorial refinements related to Catalan and Narayana numbers. Its algebraic structure is explored through recurrence relations, product formulas, and bijective combinatorics, with deep connections to flow polytope theory and polyhedral subdivisions.

1. Definition and Classical Formulation

For positive integers $n, a, b$ and integer $c \ge 0$, the Morris II identity is expressed as: $$M_n(a,b,c)\;:=\; \operatorname{CT}_{x_1,\dots,x_n} \left( \prod_{i=1}^n x_i^{-a+1}(1-x_i)^{-b} \prod_{1\le i<j\le n}(x_j-x_i)^{-c} \right)$$ where $\operatorname{CT}$ denotes extraction of the constant term in each $x_i$. Morris's theorem, equivalent to a Selberg-type integral, provides a closed-form: $$M_n(a,b,c) = \prod_{j=0}^{n-1} \frac{ \Gamma(a-1+b+(n-1+j)\frac{c}{2})\; \Gamma(\frac{c}{2} +1) }{ \Gamma(a+j\frac{c}{2})\; \Gamma(b+j\frac{c}{2})\; \Gamma((j+1)\frac{c}{2} +1) }$$ The prototypical case, $M_n(1,1,1)$, reduces to the product of the first $n-1$ Catalan numbers,

$M_n(1,1,1) = \prod_{i=1}^{n-1} C_i$

where $C_i$ is the $i$th Catalan number (Morales et al., 2021).

2. Refined Formulation and Narayana Numbers

Morales–Shi introduced a two-parameter refinement: $$\Psi_n(k, a, b, c) := \operatorname{CT}_{x_1,\dots, x_n} [t^k] \prod_{i=1}^n x_i^{-a+1}(1-x_i)^{-b} \left(1 + t\,\frac{x_i}{1-x_i}\right)\, \prod_{1\le i<j\le n}(x_j-x_i)^{-c}$$ where $[t^k]$ denotes extraction of the $t^k$ coefficient. The identity $\Psi_n(0,a,b,c)=M_n(a,b,c)$ holds, and: $M_n(a, b+1, c) = \sum_{k=0}^n \Psi_n(k, a, b, c)$ The principal product formula is

$$\Psi_n(k, a, b, c) = \binom{n}{k}\; M_n(a, b, c)\; \prod_{j=1}^k (a-1 + (n-j)\tfrac{c}{2})\, (b + (j-1)\tfrac{c}{2})$$

In the Catalan/Narayana regime $(a, b, c) = (1, 1, 1)$,

$$\Psi_n(k, 1, 1, 1) = N(n, k+1)\, \prod_{i=1}^{n-1} C_i$$

where $N(n,r)$ is the Narayana number (Morales et al., 2021).

3. Combinatorial and Geometric Interpretations

(i) Volumes of Flow Polytopes and Integer Flows

The Morris II constant term $M_n(a, b, c)$ enumerates (via volumes or Kostant partition functions) flow polytopes associated with a digraph $k_{n+2}^{a,b,c}$ on $\{0,1,\dots, n+1\}$, having $a$ parallel edges $0\to i$, $b$ parallel edges $i\to n+1$, and $c$ parallel edges $i\to j$ for $1\le i<j\le n$. For any acyclic $G$ with source $0$, sink $n+1$, and in-degrees $d_i$: $$\operatorname{Vol}( \mathcal{F}_G(1, 0, \dots, 0, -1) ) = K_G(0, d_1, \dots, d_n, -\sum d_i)$$ where $K_G(\mathbf{a})$ is the Kostant partition function. Specifically, Corteel–Kim–Morales showed

$$\operatorname{Vol}( \mathcal{F}_{k_{n+2}^{a,b,c}}(1, 0, \ldots, 0, -1) ) = M_n(a, b, c)$$

The refinement $\Psi_n(k, a, b, c)$ admits a corresponding interpretation via subdivided flow polytopes: $$\Psi_n(k, a, b, c) = \sum_{S \subseteq [n],\, |S| = k}\! \operatorname{Vol}( \mathcal{F}_{k_{n+2}^{a,b,c}(S)}(1, 0, \dots, 0, -1) )$$ where rerouting at $S$ modifies the network structure (Morales et al., 2021).

(ii) Lattice Point Enumeration

Equivalently, $$M_n(a, b, c) = K_{k_{n+2}^{a,b,c}}(0,a_1,\dots,a_n, -\sum a_i)$$ with $a_i = a-1 + c(i-1)$. The refinement $\Psi_n(k, a, b, c)$ enumerates such flows in which $a_i$ is strict for exactly $k$ variables.

4. Proof Techniques and Structural Recurrences

The product form for the refinement follows from a system of four key recurrence relations, satisfied by $\Psi_n(k, a, b, c)$,

$$\begin{aligned} &\Psi_n(n, a, b, c) = \Psi_n(0, a-1, b+1, c)\,,\ &\Psi_n(n-1, 1, b, c) = \Psi_{n-1}(0, c, b+1, c)\,,\ &\Psi_n(0, 1, b, 0) = 1\,,\ &k(b + (k-1)\tfrac{c}{2})\, \Psi_n(k, a, b, c) = (n-k+1)(a-1 + (n-k)\tfrac{c}{2})\, \Psi_n(k-1, a, b, c)\, \quad 1 \leq k \leq n\,, \end{aligned}$$

and the boundary condition $\Psi_1(\cdot) = 1$. The last "three-term" recurrence is deduced by introducing

$$U(x) = \prod_{i=1}^n x_i^{-a}(1-x_i)^{-b} \prod_{i<j}(x_j-x_i)^{-c}$$

and applying differentiation, (anti-)symmetrization, and residue calculus to relate constant terms (Morales et al., 2021).

5. Symmetry and Bijective Proofs via Triangulations

The product formula guarantees $M_n(a, b, c) = M_n(b, a, c)$. Morales–Shi provided a bijective, shelling-based proof using the Danilov–Karzanov–Koshevoy (DKK) unimodular triangulation. Vertices of $G$ are framed by ordering incident edges; maximal coherent cliques of $0\to n+1$ routes index facets. The Postnikov–Morales–Striker bijection matches such cliques with integer flows. By graph reversal and order swapping,

$$\mathcal{F}_G^\mathbb{Z} \xrightarrow{\Omega^{-1}} \{\text{cliques in } G\} \xrightarrow{\text{reverse}} \{\text{cliques in } G^r\} \xrightarrow{\Omega} \mathcal{F}_{G^r}^\mathbb{Z}$$

yields an explicit bijection between lattice points of the corresponding flow polytopes, establishing the symmetry combinatorially (Morales et al., 2021).

6. Context and Relationship to Hypergeometric and Polyhedral Combinatorics

The Morris II identity synthesizes perspectives from hypergeometric-type constant term evaluations, polyhedral geometry (flow polytopes and their subdivisions), and algebraic combinatorics (Catalan, Narayana sequences). Its refinements interpolate between classical enumeration and fine-grained polyhedral subdivisions, mirroring algebraic and geometric properties. The flow polytope framework connects the identity with the structure of Kostant partition functions and localized lattice point counts, while recurrence and triangulation arguments provide both algebraic and combinatorial validation. The interplay between these facets exemplifies contemporary research at the interface of algebra, geometry, and combinatorics (Morales et al., 2021).

7. Summary Table

Mathematical Avatar	Interpretation	Combinatorial Object
$M_n(a,b,c)$ (constant term)	Gamma-product eval	Catalan product for $(1,1,1)$
$\Psi_n(k,a,b,c)$ (refinement)	Narayana factor	Flow polytope subdivision / lattice points
$M_n(a,b,c)=M_n(b,a,c)$	Symmetry	DKK triangulation bijection