The quantum crystal-melting Hamiltonian introduced by Dijkgraaf, Orlando and Reffert describes the stochastic-style growth and erosion of a crystal corner whose configurations are plane partitions (stacks of unit cubes, i.e. three-dimensional Young diagrams). Its one- and two-dimensional reductions are exactly solvable — the two-dimensional problem is the Heisenberg XXZ ferromagnet — and both are gapped with the same mass gap, which led to the conjecture that the full three-dimensional problem is also integrable with the same gap. This question asks whether the three-dimensional Hamiltonian admits a Yang–Baxter, vertex-model-type integrable formulation: explicit, position-independent local Boltzmann weights on the Kagome/hexagonal-lattice description of plane-partition growth that assemble into Lax matrices and transfer matrices satisfying RLL/RTT (Yang–Baxter) relations with an explicit R-matrix, so that the commuting transfer-matrix family reproduces the melting Hamiltonian and unlocks transfer-matrix/Bethe-ansatz access to the three-dimensional spectrum, including the mass-gap conjecture. The problem remains open: the most direct attempt (Araujo, 2021) produced a free-fermion/spin-chain reformulation and a classical vertex-like model whose local weights yield two commuting six-vertex-descendant subsystems, but the full row-to-row transfer matrices fail to commute, and no later work has constructed the integrable structure, diagonalized the Hamiltonian, or established an obstruction.

Background

A plane partition is a two-dimensional array of nonnegative integers that is weakly decreasing along both rows and columns; visually it is a stack of unit cubes in the corner of a room (a three-dimensional Young diagram). Plane partitions with the so-called MacMahon weight, where a configuration with nn cubes carries weight qnq^{n} (so the ground-state amplitude is qn/2q^{n/2}), are the classical statistical-mechanics description of a melting crystal corner, and their partition function is ∏k≥1(1−qk)−k\prod_{k\geq 1}(1-q^{k})^{-k} — the MacMahon function ∏k≥1(1−Qk)−k\prod_{k\geq 1}(1-Q^k)^{-k} evaluated at Q=qQ=q; this combinatorics counts equivariant D-brane BPS states and Donaldson–Thomas-type invariants on the simplest toric Calabi-Yau spaces, which is one reason the model is of physical interest. Dijkgraaf, Orlando and Reffert (Nucl. Phys. B 811 (2009) 463) quantized this growth process: their crystal-melting Hamiltonian H=−J∑(∣■⟩⟨□∣+∣□⟩⟨■∣)+V∑(q ∣□⟩⟨□∣+1q ∣■⟩⟨■∣)H=-J\sum\big(|\blacksquare\rangle\langle\square|+|\square\rangle\langle\blacksquare|\big)+V\sum\big(\sqrt{q}\,|\square\rangle\langle\square|+\tfrac{1}{\sqrt{q}}\,|\blacksquare\rangle\langle\blacksquare|\big) has kinetic terms that create or annihilate a cube at an allowed corner position, and diagonal terms that count the places where a cube can consistently be added (∣□⟩|\square\rangle) or removed (∣■⟩|\blacksquare\rangle). Restricted to growth along one axis (the 1D problem) or in a quadrant (the 2D problem, i.e. growth of ordinary partitions) the model is exactly solvable: the 2D problem maps to the Heisenberg XXZ spin chain — a prototypical integrable quantum spin chain whose spectrum is found by the Bethe ansatz, a method that solves interacting many-body systems through parametrized scattering data — and the spectrum under domain-wall boundary conditions (fixed, oppositely oriented spin configurations imposed on the two halves of the chain boundary — the setting relevant for the melting problem) was derived explicitly by Orlando, Reffert and Reshetikhin. The 1D and 2D problems have the same mass gap (the energy difference between the ground state and the first excitation), which led Dijkgraaf, Orlando and Reffert to conjecture that the full 3D problem — genuine plane-partition dynamics in the first octant, equivalent to a system of coupled XXZ chains — is also integrable and shares the same gap. Later work recast the 3D problem geometrically: plane partitions are equivalent to dimer coverings of the hexagonal lattice (a dimer covering, or perfect matching, is a selection of lattice edges such that every vertex is touched by exactly one selected edge), and via a dimer–particle duality to an occupation problem on the Kagome lattice (the trihexagonal lattice of corner-sharing triangles), where Araujo, Orlando and Reffert (Phys. Rev. D 103 (2021) 026020) constructed a fermion–boson duality for plane partitions and conjectured that the growth operators are represented in the affine Yangian of gl^(1)\widehat{\mathfrak{gl}}(1) — an infinite-dimensional quantum algebra (a Yangian-type deformation of the loop algebra of gl(1)\mathfrak{gl}(1)) whose MacMahon representation organizes plane-partition states. This is an algebraic hint of an underlying integrable structure, but it does not by itself produce a solvable formulation of the Hamiltonian. The standard constructive route to exact solvability for a lattice model is the Yang–Baxter framework: one assigns local Boltzmann weights to the allowed local configurations of a two-dimensional classical vertex model (a model whose degrees of freedom live on edges meeting at vertices, the 6-vertex model being the archetype), arranges them into a Lax matrix L(u)L(u), and builds a row-to-row transfer matrix T(u)=tr L1(u)⋯LN(u)T(u)=\mathrm{tr}\,L_1(u)\cdots L_N(u). If the weights satisfy the Yang–Baxter (RLL) relation R(u,v)L(u)L(v)=L(v)L(u)R(u,v)R(u,v)L(u)L(v)=L(v)L(u)R(u,v) with an explicit R-matrix R(u,v)R(u,v) — equivalently the transfer matrices satisfy RTT-type commutation relations — then the transfer matrices commute, [T(u),T(v)]=0[T(u),T(v)]=0, giving infinitely many conserved charges, a commuting family from which the Hamiltonian emerges, and Bethe-ansatz access to the spectrum. Araujo (Nucl. Phys. B 963 (2021) 115287) carried this program as far as it has gone: rewriting the Kagome system through two families of free fermions and spin chains, he classified the allowed local hexagon configurations of the Kagome lattice (18 local states), defined a classical statistical model with Boltzmann weights and a Lax operator, and showed that a reduction gives two Yang–Baxter integrable subsystems descending from the 6-vertex model; however, the two natural row-to-row transfer matrices built from the full weight set fail to commute with one another, so the Yang–Baxter equations for the weights of the full plane-partition problem could not even be imposed, and the exact solvability of the 3D problem — and the mass-gap conjecture — remained open.

Problem Statement

Determine whether the three-dimensional quantum crystal-melting Hamiltonian of Dijkgraaf–Orlando–Reffert — the Hamiltonian H=−J∑(∣■⟩⟨□∣+∣□⟩⟨■∣)+V∑(q ∣□⟩⟨□∣+1q ∣■⟩⟨■∣)H=-J\sum\big(|\blacksquare\rangle\langle\square|+|\square\rangle\langle\blacksquare|\big)+V\sum\big(\sqrt{q}\,|\square\rangle\langle\square|+\tfrac{1}{\sqrt{q}}\,|\blacksquare\rangle\langle\blacksquare|\big) (equation (1.1) of their paper) acting on the Hilbert space spanned by plane partitions, with real couplings JJ and VV and deformation parameter 0<q<10<q<1, where the kinetic terms create or annihilate a cube at an admissible corner position of the plane partition and the diagonal growth terms count, respectively, the positions where a cube can consistently be added (projector ∣□⟩⟨□∣|\square\rangle\langle\square|, weight q\sqrt{q}) or removed (projector ∣■⟩⟨■∣|\blacksquare\rangle\langle\blacksquare|, weight 1/q1/\sqrt{q}) — admits a Yang–Baxter, vertex-model-type integrable formulation. Concretely: on the Kagome/hexagonal-lattice description of plane-partition growth (where plane partitions map to dimers on the hexagonal lattice and, by dimer–particle duality, to an occupation problem on the dual Kagome lattice), can one assign explicit, position-independent local Boltzmann weights to the allowed local hexagon configurations — the 18 local states of Araujo's classical model, or an equivalent local vertex formulation of the same problem (equivalent in that its commuting transfer-matrix family reproduces the same 3D melting Hamiltonian on the plane-partition Hilbert space, in the sense specified below) — that assemble into Lax matrices L(u)L(u) and transfer matrices T(u)T(u) satisfying the Yang–Baxter/RLL/RTT relations with an explicit R-matrix, such that the resulting commuting transfer-matrix family reproduces the 3D melting Hamiltonian and its spectrum on the plane-partition Hilbert space? Here "reproduces" is understood in the standard transfer-matrix sense: the plane-partition Hilbert space arises as an invariant sector of the commuting family (the larger classical configuration space being restricted to the plane-partition sector), and HH belongs to the commuting charge algebra generated by T(u)T(u) — for instance as a logarithmic derivative of T(u)T(u) at a suitable spectral point, up to an additive constant and a similarity transformation — so that the spectrum of HH follows from that of the commuting family. A positive answer is such a construction, supplemented by the nondegeneracy conditions that the R-matrix is a genuine solution of the Yang–Baxter equation on an auxiliary space of dimension greater than one (not proportional to the identity) and that the commuting transfer-matrix family generated by T(u)T(u) carries infinitely many independent conserved charges rather than merely functions of HH alone; connecting it to the observed algebraic structures (the MacMahon-weighted ground state, the affine Yangian of gl^(1)\widehat{\mathfrak{gl}}(1) acting on plane-partition states) and using transfer-matrix and Bethe-ansatz tools to extract the 3D spectrum, including the Dijkgraaf–Orlando–Reffert conjecture that the 3D model is integrable and has the same mass gap as its 1D and 2D reductions, are the intended payoffs of the construction rather than additional acceptance conditions. A negative answer would be a rigorous obstruction result of matching generality: it must exclude the natural ansatz class of position-independent local Kagome weights that includes the Lax operators of the source program, together with the equivalent local vertex formulations admitted above; a no-go confined to a single one of these local ansatz classes would count only as a partial negative result; nonlocal transfer objects, tetrahedron-equation-type structures, and affine-Yangian routes may remain open. The question is posed at its natural generality: no restriction to particular reductions, subsystems, parameter windows, or answer forms is intended, and the two known six-vertex-descendant integrable subsystems do not by themselves answer it, since the question is whether the full plane-partition dynamics is Yang–Baxter integrable.

References

  1. Robbert Dijkgraaf, Domenico Orlando, Susanne Reffert, "Quantum Crystals and Spin Chains", Nuclear Physics B 811 (2009) 463–490. DOI: 10.1016/j.nuclphysb.2008.11.027; arXiv:0803.1927. https://arxiv.org/abs/0803.1927
  2. Domenico Orlando, Susanne Reffert, Nicolai Reshetikhin, "On domain wall boundary conditions for the XXZ spin Hamiltonian", arXiv:0912.0348 (2009). https://arxiv.org/abs/0912.0348
  3. Thiago Araujo, Domenico Orlando, Susanne Reffert, "Quantum crystals, Kagome lattice and plane partitions fermion-boson duality", Physical Review D 103 (2021) 026020. DOI: 10.1103/PhysRevD.103.026020; arXiv:2005.09103. https://arxiv.org/abs/2005.09103
  4. Thiago Araujo, "Towards Yang-Baxter integrability of quantum crystal melting: From Kagome lattice to vertex models", Nuclear Physics B 963 (2021) 115287. DOI: 10.1016/j.nuclphysb.2020.115287; arXiv:2009.06812. https://arxiv.org/abs/2009.06812
  5. Na Wang, "3D fermions and affine Yangian", Nuclear Physics B 969 (2021) 115461. DOI: 10.1016/j.nuclphysb.2021.115461. https://doi.org/10.1016/j.nuclphysb.2021.115461
  6. Andrea Appel, Sachin Gautam, Curtis Wendlandt, "The R-matrix of the affine Yangian", arXiv:2309.02377 (2023). https://arxiv.org/abs/2309.02377
  7. Dmitry Kolyaskin, Alexey Litvinov, Arkady Zhukov, "R-matrix formulation of affine Yangian of gl^(1∣1)\hat{\mathfrak{gl}}(1|1)", arXiv:2206.01636 (2022). https://arxiv.org/abs/2206.01636
  8. Alexey Litvinov, Ilya Vilkoviskiy, "Liouville reflection operator, affine Yangian and Bethe ansatz", arXiv:2007.00535 (2020). https://arxiv.org/abs/2007.00535
  9. Federico Balducci, Andrea Gambassi, Alessio Lerose, Antonello Scardicchio, Carlo Vanoni, "Interface dynamics in the two-dimensional quantum Ising model", Physical Review B 107 (2023) 024306. DOI: 10.1103/PhysRevB.107.024306; arXiv:2209.08992. https://arxiv.org/abs/2209.08992
  10. Federico Balducci, Antonello Scardicchio, Carlo Vanoni, "Slow melting of a disordered quantum crystal", Physical Review B 107 (2023) 024201. DOI: 10.1103/PhysRevB.107.024201; arXiv:2207.11246. https://arxiv.org/abs/2207.11246
  11. Dmitry Galakhov, Wei Li, "Charging solid partitions", Journal of High Energy Physics 01 (2024) 043. DOI: 10.1007/JHEP01(2024)043; arXiv:2311.02751. https://arxiv.org/abs/2311.02751
  12. Jonah Guse, David Jiang, David Keating, "Colored Vertex Models and Interacting Reverse Plane Partitions", arXiv:2505.13806 (2025). https://arxiv.org/abs/2505.13806
  13. Ajesh Kumar, Andrew C. Potter, "Symmetry-enforced fractonicity and two-dimensional quantum crystal melting", Physical Review B 100 (2019) 045119. DOI: 10.1103/PhysRevB.100.045119. https://doi.org/10.1103/PhysRevB.100.045119
  14. Abhishodh Prakash, Jaydev Singh Rao, Siddharth A. Parameswaran, Alessio Lerose, "Interface phases and dynamics in two-dimensional quantum magnets: A "holographic" approach from universality to quantum simulation", arXiv:2608.12312 (2026). https://arxiv.org/abs/2608.12312