The square-lattice dimer model on a strip with Uq(sl(2))U_q(sl(2))-invariant boundary conditions is exactly solvable by Yang–Baxter methods: the double-row transfer matrix satisfies inversion identities, finite-size spectra yield a logarithmic conformal field theory with central charge c=−2c=-2, and the associated quantum Hamiltonian — the logarithmic derivative of the transfer matrix, coinciding with the non-Hermitian Uq(sl(2))U_q(sl(2))-invariant XX chain — exhibits rank-2 Jordan blocks at every examined finite width in odd sectors. What is not established is the continuum limit itself: whether these lattice Jordan blocks provably survive as rank-2 Jordan cells of the Virasoro dilatation operator L0L_0, what the resulting Virasoro modules Ds\mathcal{D}_s are (the proposed Loewy diagrams contain arrows whose presence the lattice analysis cannot determine, leaving the modules indecomposable-or-decomposable unresolved), and whether the empirical selection rules that fix the eigenvalue degeneracies hold for all widths. The problem is to convert this strongly supported physical picture into a mathematically controlled derivation — or to refute it.

Background

Close-packed dimers on a square lattice — perfect matchings covering every vertex by exactly one domino — form one of the classical exactly solved systems of statistical mechanics. Pearce and Vittorini-Orgeas showed that the dimer model is Yang–Baxter integrable: viewing it as the six-vertex model at its free-fermion point (crossing parameter λ=π/2\lambda=\pi/2, with the dimers rotated by 45∘45^\circ), the model enters the standard machinery of integrable lattice models, in which a solution of the Yang–Baxter equation yields a commuting family of transfer matrices. On a strip of width NN, one builds a double-row transfer matrix D(u)\mathbf{D}(u) that incorporates left and right boundary weights through boundary Yang–Baxter (reflection) equations; choosing boundary weights compatible with Uq(sl(2))U_q(sl(2)) invariance (the quantum-group deformed symmetry of the six-vertex model, with boundary parameter x=ix=i) makes the model exactly solvable at arbitrary finite width. In a suitable gauge the local face operators represent the Temperley–Lieb algebra — the diagram algebra with generators eje_j obeying ej2=βeje_j^2=\beta e_j and ejej±1ej=eje_j e_{j\pm1} e_j=e_j, where the loop fugacity β\beta is the weight of a closed loop, here β=2cos⁡λ=0\beta=2\cos\lambda=0 — and a particle (fermion) representation makes the free-fermion structure explicit.

Pearce, Rasmussen and Vittorini-Orgeas solved the strip model through inversion identities: functional equations D(u)D(u+λ)=(scalar)⋅I\mathbf{D}(u)\mathbf{D}(u+\lambda)=\text{(scalar)}\cdot\mathbb{I} that, combined with analyticity, determine the transfer-matrix eigenvalues. Because the dimer model and critical dense polymers (the logarithmic minimal model LM(1,2)) share the underlying Temperley–Lieb algebra, the two models have the same eigenvalues, but their degeneracies and characters differ, so the two boundary theories appear to lie in different universality classes. Finite-size corrections to the eigenvalues yield central charge c=−2c=-2 and, in sectors labelled s=∣Sz∣+1s=|S_z|+1 (with SzS_z the magnetization), conformal weights Δs=((2−s)2−1)/8\Delta_s=((2-s)^2-1)/8; the finitized characters — finite-width qq-series counting lattice states in a sector, truncating the continuum conformal characters — decompose into qq-Narayana numbers, equivalently skew qq-binomials. The eigenvalue degeneracies are not fixed by the inversion identities alone: they are fixed by empirical selection rules, namely a classification of admissible patterns of zeros (the zero structure of the eigenvalue polynomials) and degeneracy matrices A(s)A^{(s)}, B(s)B^{(s)} (special upper-triangular Toeplitz matrices), verified by direct inspection of the spectra out to width N=N+w≤8\mathcal{N}=N+w\le 8.

The central finding concerns Jordan structure. The quantum Hamiltonian Hw=−12 dlog⁡D(u)/du∣u=0\mathcal{H}_w=-\tfrac12\,\mathrm{d}\log\mathbf{D}(u)/\mathrm{d}u|_{u=0} (the logarithmic derivative of the transfer matrix; for w=0w=0 it coincides with the standard non-Hermitian Uq(sl(2))U_q(sl(2))-invariant XX spin chain, whose eigenvalues are nevertheless all real) is non-diagonalizable: for ss odd (N\mathcal{N} even) its Jordan normal form exhibits rank-2 Jordan blocks (ε10ε)\begin{pmatrix}\varepsilon&1\\0&\varepsilon\end{pmatrix}, empirically (N−2d−1)\binom{N-2}{d-1} of them in sectors with dd down arrows (1≤d≤N−11\le d\le N-1), while for ss even no non-trivial blocks occur. The same blocks had been observed for the quantum-group-invariant XX chain by Gainutdinov, Hao, Nepomechie and Sommese. Non-diagonalizability of the dilatation operator is the hallmark of a logarithmic conformal field theory (LCFT): in the continuum the Virasoro generator L0L_0 acts as L0∣χ⟩=h∣χ⟩+∣ψ⟩L_0|\chi\rangle=h|\chi\rangle+|\psi\rangle on pairs of states, which produces logarithmic terms in correlation functions and forces indecomposable representations. Arguing that the lattice Jordan blocks are stable as NN grows, the authors conclude that the strip dimer model realizes a boundary LCFT with c=−2c=-2, minimal conformal weight Δmin⁡=−1/8\Delta_{\min}=-1/8 and effective central charge ceff=1c_{\mathrm{eff}}=1, in which the familiar c=−2c=-2 staggered modules (rank-2 L0L_0 Jordan cells built on symplectic fermions, as studied by Gaberdiel and Kausch) appear as submodules of new modules Ds\mathcal{D}_s, ss odd. The proposed Loewy diagrams — diagrams encoding the socle (maximal completely reducible submodule) and head (maximal completely reducible quotient) filtration of a module — however contain left-pointing horizontal arrows whose presence the lattice analysis cannot determine: as drawn the modules are indecomposable; with those arrows removed they decompose.

Three gaps separate this physical picture from a theorem. First, the persistence of the lattice Jordan blocks as rank-2 Jordan cells of L0L_0 in the N→∞N\to\infty scaling limit is argued, not proven: a proof must control generalized eigenvectors — the Hamiltonian is non-Hermitian, and its Jordan blocks arise where single-particle eigenvectors coalesce at exceptional points (parameter values at which a non-Hermitian matrix becomes non-diagonalizable) — and must identify the limit operator with L0L_0 including its off-diagonal (logarithmic) matrix elements. Second, the module structure of Ds\mathcal{D}_s is unsettled: the lattice analysis cannot determine the left-pointing arrows, and the lattice W-algebra analysis of Gainutdinov, Saleur and Tipunin for related XXZ chains supports the decomposable option. Third, the selection rules underpinning the state counting are empirical; the analogous statement for critical dense polymers was proven by Morin-Duchesne only in separate work. Numerical work by Vasseur, Jacobsen and Saleur on indecomposability parameters of loop models, and the recent construction by Io, Huang and Hsieh of a PT-symmetric free-fermion lattice model whose exceptional-point criticality realizes a c=−2c=-2 LCFT with L0L_0 Jordan blocks, both support the plausibility of persistence, but neither constitutes a proof for the dimer strip. The problem is to close these gaps for the pinned model.

Problem Statement

Establish, or refute, the logarithmic conformal field theory content of the Yang–Baxter integrable dimer model on a strip with Uq(sl(2))U_q(sl(2))-invariant boundary conditions, starting from its exact finite-size solution. Concretely, for the model defined by the free-fermion six-vertex RR-matrix at crossing parameter λ=π/2\lambda=\pi/2, boundary weights KL(b∣a)=x1−2aδa,bK^L(b|a)=x^{1-2a}\delta_{a,b}, KR(b∣a)=δa,bK^R(b|a)=\delta_{a,b} with x=ix=i, and double-row transfer matrix D(u)\mathbf{D}(u) on strips of width NN (with Hamiltonian Hw=−12 dlog⁡D(u)/du∣u=0\mathcal{H}_w=-\tfrac12\,\mathrm{d}\log\mathbf{D}(u)/\mathrm{d}u|_{u=0}, boundary parameter w=0,1w=0,1), the problem comprises three linked deliverables that jointly amount to a controlled derivation of the continuum theory:

(1) Persistence of Jordan cells. Prove, in a stated precise sense of convergence of the N→∞N\to\infty scaling limit, that the rank-2 Jordan blocks of the finite-NN transfer matrices and Hamiltonians (sectors with ss odd, i.e. N=N+w\mathcal{N}=N+w even, where the sector label is s=∣Sz∣+1s=|S_z|+1 with SzS_z the total magnetization) converge to rank-2 Jordan cells of the Virasoro dilatation operator L0L_0 acting on the c=−2c=-2 continuum theory with conformal weights Δr,1=((2r−1)2−1)/8\Delta_{r,1}=((2r-1)^2-1)/8 for r=1,2,3,…r=1,2,3,\ldots (these are the weights Δ2r+1\Delta_{2r+1} among the sector conformal weights Δs=((2−s)2−1)/8\Delta_s=((2-s)^2-1)/8) — or rigorously refute persistence, e.g. by proving that the limiting action of L0L_0 is diagonalizable on the affected spectrum.

(2) Module structure. Determine the Virasoro modules Ds\mathcal{D}_s, ss odd, generated in the scaling limit: in particular, settle whether the left-pointing horizontal arrows in the proposed Loewy diagrams are present (modules indecomposable) or absent (modules decomposable), consistently with the rank-2 Jordan-block data, the finitized characters and their qq-Narayana (skew qq-binomial) decompositions, and the appearance of the standard c=−2c=-2 staggered modules as submodules.

(3) Selection rules. Determine whether the empirical selection rules hold for all widths NN and both boundary parameters w=0,1w=0,1 (proving or refuting them): the classification of admissible patterns of zeros — encoded as double-column diagrams, i.e. pairs of occupation columns S=(L,R)S=(L,R) with L⪯RL\preceq R componentwise — and the degeneracy matrices A(s)A^{(s)}, B(s)B^{(s)}, which the inversion identities alone do not determine.

Deliverable (3) is the lattice-combinatorial foundation for the state counting used in (1) and (2); the three together close the gap, left open by the exact solution, between the finite-NN Jordan structure and the conjectured logarithmic c=−2c=-2 boundary conformal field theory. An answer that resolves the central scaling-limit question (1) in the negative (a diagonalizable limiting L0L_0) closes the problem by refutation, provided the diagonalizability claim is established with the same rigour. In general the task is closed by establishing all three deliverables or by rigorously refuting any one of them, at the same standard of rigour; concretely, refuting (1) means a diagonalizable limiting L0L_0, refuting (2) means decomposable modules Ds\mathcal{D}_s with the left-pointing arrows absent, and refuting (3) means a failure of the selection rules at some width NN or boundary parameter ww; if none is refuted, all of (1)-(3) must be established affirmatively.

References

  1. P. A. Pearce, J. Rasmussen, A. Vittorini-Orgeas, 'Yang–Baxter integrable dimers on a strip', J. Stat. Mech. (2020) 013107, doi:10.1088/1742-5468/ab54bd, arXiv:1907.07610, https://arxiv.org/abs/1907.07610
  2. P. A. Pearce, A. Vittorini-Orgeas, 'Yang–Baxter solution of dimers as a free-fermion six-vertex model', J. Phys. A 50 (2017) 434001, doi:10.1088/1751-8121/aa86bc, arXiv:1612.09477, https://arxiv.org/abs/1612.09477
  3. A. Morin-Duchesne, A. Klümper, P. A. Pearce, 'Critical site percolation on the triangular lattice: from integrability to conformal partition functions', J. Stat. Mech. (2023) 043103, doi:10.1088/1742-5468/acc4b3, arXiv:2211.12379, https://arxiv.org/abs/2211.12379
  4. X. Poncini, J. Rasmussen, 'Integrability of planar-algebraic models', J. Stat. Mech. (2023) 073101, doi:10.1088/1742-5468/acdce7, arXiv:2206.14462, https://arxiv.org/abs/2206.14462
  5. X. Poncini, J. Rasmussen, 'Integrable models from singly generated planar algebras', Nucl. Phys. B 994 (2023) 116308, doi:10.1016/j.nuclphysb.2023.116308, arXiv:2302.11712, https://arxiv.org/abs/2302.11712
  6. I.-F. Io, F.-H. Huang, C.-T. Hsieh, 'Non-Hermitian free-fermion critical systems and logarithmic conformal field theory', arXiv:2602.02649 (2026), https://arxiv.org/abs/2602.02649
  7. A. Morin-Duchesne, Y. Saint-Aubin, 'The Jordan structure of two-dimensional loop models', J. Stat. Mech. (2011) P04007, doi:10.1088/1742-5468/2011/04/P04007, arXiv:1101.2885, https://arxiv.org/abs/1101.2885
  8. A. Morin-Duchesne, Y. Saint-Aubin, 'Jordan cells of periodic loop models', J. Phys. A 46 (2013) 494013, doi:10.1088/1751-8113/46/49/494013, arXiv:1302.5483, https://arxiv.org/abs/1302.5483
  9. R. Vasseur, J. L. Jacobsen, H. Saleur, 'Indecomposability parameters in chiral logarithmic conformal field theory', Nucl. Phys. B 851 (2011) 314–345, doi:10.1016/j.nuclphysb.2011.05.018, arXiv:1103.3134, https://arxiv.org/abs/1103.3134
  10. A. M. Gainutdinov, H. Saleur, I. Yu. Tipunin, 'Lattice W-algebras and logarithmic CFTs', J. Phys. A 47 (2014) 495401, doi:10.1088/1751-8113/47/49/495401, arXiv:1212.1378, https://arxiv.org/abs/1212.1378
  11. A. Morin-Duchesne, 'A proof of selection rules for critical dense polymers', J. Phys. A 44 (2011) 495003, doi:10.1088/1751-8113/44/49/495003, arXiv:1109.6397, https://arxiv.org/abs/1109.6397
  12. A. M. Gainutdinov, W. Hao, R. I. Nepomechie, A. J. Sommese, 'Counting solutions of the Bethe equations of the quantum group invariant open XXZ chain at roots of unity', J. Phys. A 48 (2015) 494003, doi:10.1088/1751-8113/48/49/494003, arXiv:1505.02104, https://arxiv.org/abs/1505.02104
  13. M. R. Gaberdiel, H. G. Kausch, 'Indecomposable fusion products', Nucl. Phys. B 477 (1996) 293–318, doi:10.1016/0550-3213(96)00364-1, arXiv:hep-th/9604026, https://arxiv.org/abs/hep-th/9604026