Positivity conjecture and $q\to0$ crystal-limit identification for the $q$-boson Zamolodchikov-Faddeev representation of the $U_q(A^{(1)}_n)$ zero-range process

Kuniba and Okado constructed a representation of the Zamolodchikov-Faddeev (ZF) algebra, whose structure function is the stochastic $R$ matrix of the quantum affine algebra $U_q(A^{(1)}_n)$, on $q$-boson Fock space, yielding a matrix product formula for the stationary probabilities of the inhomogeneous multispecies $U_q(A^{(1)}_n)$ zero-range process for arbitrary rank $n$. The paper left two explicitly acknowledged unresolved issues, which form the present problem. First, a positivity conjecture: the authors expect, based on exact small examples, that there is a normalization within each sector such that all stationary probabilities are polynomials in $q$ and $-\mu_1,\ldots,-\mu_L$ with nonnegative integer coefficients; the sector normalization is rational in $q$, so the conjecture is nontrivial. Second, at $q=0$ there are two independent matrix-product constructions for the multispecies totally asymmetric zero-range process, originating in the combinatorial $R$ of crystal theory and in the tetrahedron equation; they agree with the $q\to0$ limit of the $q$-boson representation for $n=2$ after adjusting conventions, but their relation for rank $n\ge3$ is unknown. The problem asks for a proof or refutation of the positivity conjecture in full generality, and for a determination of the $n\ge3$ relation between the two crystal-theoretic constructions and the $q\to0$ limit of the ZF representation.

Background

A zero-range process (ZRP) is a Markov chain on a ring of $L$ sites in which particles hop from a site to its right neighbor at a rate (or, in discrete time, with a transition weight) depending only on the content of the departure site (the interaction has 'zero range'). A multispecies (or $n$-species) ZRP carries particle species $1,\ldots,n$ that pass through one another in a priority order, so that the cyclic order of species around the ring is preserved. In the $U_q(A^{(1)}_n)$-ZRP below, a configuration assigns each site an occupation vector $\sigma_i\in\mathbb{Z}_{\ge0}^n$ recording how many particles of each species occupy it; in the totally asymmetric zero-range process (TAZRP) constructions recalled later, a site is instead a word in the species, the word's order recording the priority stack. The stationary state is the unique invariant probability measure in each sector, the sector being fixed by the total numbers $m=(m_1,\ldots,m_n)$ of particles of each species; its probabilities $\mathbb{P}(\sigma_1,\ldots,\sigma_L)$ are the basic observables.

The $U_q(A^{(1)}_n)$-ZRP of Kuniba, Mangazeev, Maruyama and Okado is a discrete-time multispecies ZRP whose local transition weights are matrix elements of a stochastic $R$ matrix: a solution of the Yang-Baxter equation (the factorized-consistency relation underlying integrable systems) that additionally satisfies a sum-to-unity condition, so that the associated row transfer matrix is a Markov operator within a commuting family. Here $U_q(A^{(1)}_n)$ is the quantum affine algebra (quantum group) of type $A_n^{(1)}$, $q$ is its deformation parameter, and the stochastic $R$ matrix originates in the quantum $R$ matrix for the symmetric-tensor representations of $U_q(A^{(1)}_n)$. Kuniba and Okado then proved that the stationary probabilities in every basic sector (non-basic sectors reduce to basic sectors of a smaller-rank model by relabeling the species) admit a matrix product form $\mathbb{P}(\sigma_1,\ldots,\sigma_L)=\mathrm{Tr}(X_{\sigma_1}(\mu_1)\cdots X_{\sigma_L}(\mu_L))$ with site-dependent (inhomogeneity) parameters $\mu_i$, where the operators $X_\alpha(\mu)$ carry a representation of the Zamolodchikov-Faddeev (ZF) algebra $X_\alpha(\mu)X_\beta(\lambda)=\sum_{\gamma,\delta}\mathscr{S}(\lambda,\mu)^{\beta,\alpha}_{\gamma,\delta}X_\gamma(\lambda)X_\delta(\mu)$, whose structure function $\mathscr{S}$ is the stochastic $R$ matrix itself; the ZF algebra originates in the factorized scattering of integrable quantum field theory. The representation acts by $q$-bosons on an infinite-dimensional Fock space: $q$-bosons are the $q$-deformed creation, annihilation and number operators $\mathbf{b}^+$, $\mathbf{c}^-$, $\mathbf{k}$ with $\mathbf{k}\,\mathbf{b}^+\mathbf{k}^{-1}=q\,\mathbf{b}^+$ and $\mathbf{c}^-\mathbf{b}^+=1-q\,\mathbf{k}$. Because the trace runs over the infinite Fock space, the normalization of a sector is naturally a rational function of $q$ (one computes $\mathrm{Tr}(\mathbf{k}^r)=(1-q^r)^{-1}$), even though each ratio of stationary probabilities within a sector is rational in $q$ and the $\mu_i$.

At $q=0$ the quantum-group formalism degenerates to crystal theory, the combinatorial skeleton of a representation in which the basis vectors and the action of the algebra generators are encoded in a directed graph, whose central object is the combinatorial $R$ matrix acting on tensor products of crystals. Two independent $q=0$ matrix-product constructions of the stationary states of the multispecies totally asymmetric zero-range process (TAZRP, the fully asymmetric limit) are known: one from the combinatorial $R$ via the multiline process, a lifting of ring configurations to arrays of carrier lines whose uniform measure projects onto the stationary state, and one from solutions of the tetrahedron equation, the three-dimensional analogue of the Yang-Baxter equation whose reductions induce two-dimensional $R$ matrices. Kuniba and Okado showed that the $q\to0$ limit of their $q$-boson ZF operators coincides with the homogeneous case of a third construction, the inhomogeneous multispecies TAZRP matrix product of Kuniba-Maruyama-Okado, via an automorphism of the $q$-boson algebra (an invertible redefinition of the generators) followed by a singular $\zeta,q\to0$ limit; and that the combinatorial-$R$ and tetrahedron-equation operators agree with the $q=0$ case of their construction for $n=2$, after adjusting conventions. Whether the same agreement persists for rank $n\ge3$ was left open.

In a worked example (rank $n=3$, system size $L=2$, minimum sector), the exactly computed stationary state has coefficients that are products of polynomials in $q$ and the $-\mu_i$ with nonnegative integer coefficients, and Kuniba and Okado wrote: 'We expect that there is a normalization such that all the stationary probabilities belong to $\mathbb{Z}_{\ge0}[q,-\mu_1,\ldots,-\mu_L]$.' (The source text prints $-\mu_n$, a notational slip for $-\mu_L$, since the process has $L$ sites; the authors' own restatement correctly writes $-\mu_L$.) The expectation is conjectural: it asserts the existence, in every sector, of an overall normalization under which all stationary probabilities simultaneously become polynomials with nonnegative integer coefficients, despite the sector normalization being rational in $q$. Their SIGMA review with Watanabe reissued this as Conjecture 4.2 'based on computer experiments' and flagged the detailed study of the sector normalization function $G_k$ as a future problem.

Problem Statement

The problem comprises the two explicitly acknowledged open issues of the $q$-boson Zamolodchikov-Faddeev representation of Kuniba and Okado, which jointly ask for a complete characterization of the matrix-product representation of the stationary states of the inhomogeneous multispecies $U_q(A^{(1)}_n)$ zero-range process and of its relation to the other integrable constructions of the same process.

(1) Positivity conjecture. Prove or refute: for every rank $n\ge1$, every system size $L$, and every sector, there exists a normalization of the sector's stationary state (an overall factor multiplying all its unnormalized stationary probabilities) such that all stationary probabilities $\mathbb{P}(\sigma_1,\ldots,\sigma_L)$ of the inhomogeneous $U_q(A^{(1)}_n)$-ZRP with site parameters $\mu_1,\ldots,\mu_L$ belong to $\mathbb{Z}_{\ge0}[q,-\mu_1,\ldots,-\mu_L]$, that is, are polynomials in $q$ and $-\mu_1,\ldots,-\mu_L$ with nonnegative integer coefficients. The statement must cover the inhomogeneous case and all sectors; a proof may identify the normalization constructively or establish its existence.

(2) $q\to0$ crystal-limit comparison for $n\ge3$. Determine the precise relation between the $q\to0$ limit of the $q$-boson ZF-algebra representation and the matrix product operators for the multispecies totally asymmetric zero-range process obtained independently from (i) the combinatorial $R$ matrix of crystal theory (the multiline process construction) and (ii) the tetrahedron equation (the hat-relation construction, an operator identity induced by the tetrahedron equation): prove that, after adjustment of conventions as is known to suffice for $n=2$, these operator sets coincide for all $n\ge3$, or establish that they genuinely differ, identifying the discrepancy.

References

1. Atsuo Kuniba and Masato Okado, "A $q$-boson representation of Zamolodchikov-Faddeev algebra for stochastic $R$ matrix of $U_q(A^{(1)}_n)$", Lett. Math. Phys. 107 (2017) 1111-1130, DOI 10.1007/s11005-016-0934-7, https://arxiv.org/abs/1610.00531
2. Atsuo Kuniba, Masato Okado and Satoshi Watanabe, "Integrable Structure of Multispecies Zero Range Process", SIGMA 13 (2017) 044, DOI 10.3842/SIGMA.2017.044, https://arxiv.org/abs/1701.07279
3. Atsuo Kuniba and Vladimir V. Mangazeev, "Density and current profiles in $U_q(A^{(1)}_2)$ zero range process", Nucl. Phys. B 922 (2017) 148-185, DOI 10.1016/j.nuclphysb.2017.07.001, https://arxiv.org/abs/1705.10979
4. Atsuo Kuniba, Shouya Maruyama and Masato Okado, "Inhomogeneous generalization of a multispecies totally asymmetric zero range process", J. Stat. Phys. 164 (2016) 952-968, DOI 10.1007/s10955-016-1555-3, https://arxiv.org/abs/1602.00764
5. Atsuo Kuniba, Shouya Maruyama and Masato Okado, "Multispecies totally asymmetric zero range process: I. Multiline process and combinatorial $R$", J. Integr. Syst. 1 (2016) xyw002, DOI 10.1093/integr/xyw002, https://arxiv.org/abs/1511.09168
6. Atsuo Kuniba, Shouya Maruyama and Masato Okado, "Multispecies totally asymmetric zero range process: II. Hat relation and tetrahedron equation", J. Integr. Syst. 1 (2016) xyw008, DOI 10.1093/integr/xyw008, https://arxiv.org/abs/1602.04574
7. Atsuo Kuniba, Vladimir V. Mangazeev, Shouya Maruyama and Masato Okado, "Stochastic $R$ matrix for $U_q(A^{(1)}_n)$", Nucl. Phys. B 913 (2016) 248-277, DOI 10.1016/j.nuclphysb.2016.09.016, https://arxiv.org/abs/1604.08304
8. Atsuo Kuniba and Masato Okado, "Matrix product formula for $U_q(A^{(1)}_2)$-zero range process", J. Phys. A: Math. Theor. 50 (2017) 044001, DOI 10.1088/1751-8121/50/4/044001, https://arxiv.org/abs/1608.02779
9. Atsuo Kuniba, Masato Okado and Travis Scrimshaw, "A strange five vertex model and multispecies ASEP on a ring", Algebraic Combinatorics 8 (2026) 1713-1741, DOI 10.5802/alco.457, https://arxiv.org/abs/2408.12092
10. Arvind Ayyer and Atsuo Kuniba, "Multispecies inhomogeneous $t$-PushTASEP with general capacity", preprint (2026), https://arxiv.org/abs/2602.17179
11. Olya Mandelshtam and Travis Scrimshaw, "Twisted multiline queues for the steady states of TASEP and TAZRP", Electron. J. Probab. 30 (2025) 1-31, DOI 10.1214/25-EJP1374, https://arxiv.org/abs/2410.21781
12. Amol Aggarwal, Matthew Nicoletti and Leonid Petrov, "Colored interacting particle systems on the ring: stationary measures from Yang-Baxter equations", Compositio Mathematica 161 (2025) 1855-1922, DOI 10.1112/s0010437x25007195, https://arxiv.org/abs/2309.11865
13. Rouven Frassek, "Steady state representations for the harmonic process", J. Phys. A: Math. Theor. 59 (2026) 085202, DOI 10.1088/1751-8121/ae4799, https://arxiv.org/abs/2507.20436
14. Jeffrey Kuan, "An algebraic construction of duality functions for the stochastic $U_q(A_n^{(1)})$ vertex model and its degenerations", Commun. Math. Phys. 359 (2018) 121-187, DOI 10.1007/s00220-018-3108-x, https://arxiv.org/abs/1701.04468