Completing the exact solution of the one-dimensional $J_1$–$J_2$ frustrated Potts chain: subleading transfer-matrix spectrum, correlation-level status of the field-driven mapping, and noninteger $q$

In 2025 the classical one-dimensional $q$-state Potts chain with competing nearest-neighbor coupling $J_1$ and next-nearest-neighbor coupling $J_2$ was exactly solved for arbitrary integer $q$ by a maximally symmetric subspace (MSS) reduction of the transfer matrix: the dominant (Perron) eigenvalue of a two-dimensional symmetry block yields the full thermodynamics, and the dominant eigenvalue is shown to coincide with that of a simpler field-driven Potts chain, establishing an exact thermodynamic mapping. The exact solution of this same model is, however, incomplete in three linked ways, which the source paper leaves open and no later work has addressed: (A) closed-form subleading eigenvalues and finite-size spectral gaps — hence exact correlation lengths and two-point correlation functions at general separation — are not known; (B) whether the mapping to the field-driven chain holds beyond the dominant eigenvalue, at the level of correlation functions and microscopic configuration statistics, is unanalyzed; and (C) the reduction and the mapping are confined to integer $q$, with the continuously varying (noninteger) $q$ regime in the Fortuin–Kasteleyn random-cluster formulation unaddressed. This problem asks for the completion of the exact solution of this one model along these three axes.

Background

The $q$-state Potts model assigns to each site $i$ of a lattice a spin $\sigma_i \in \{1,\dots,q\}$ and couples pairs of sites with an energy favoring equal spin values, $-J\,\delta(\sigma_i,\sigma_j)$, where $\delta$ is the Kronecker delta; $q=2$ is the Ising model. In one dimension with short-range interactions, such models are exactly solvable by the transfer-matrix method: with periodic boundary conditions the partition function of a chain of $N$ sites is $Z = \mathrm{tr}\, T^N$, where $T$ is a finite matrix whose entries are Boltzmann weights connecting neighboring spin configurations, and in the thermodynamic limit the free energy per site is $f = -k_B T \ln \lambda_{\max}$, with $\lambda_{\max}$ the Perron (dominant) eigenvalue of $T$.

When a next-nearest-neighbor coupling $J_2$ competes with the nearest-neighbor coupling $J_1$ (for example ferromagnetic $J_1 > 0$ with antiferromagnetic $J_2 < 0$: making neighboring spins equal, as $J_1 > 0$ prefers, forces next-nearest neighbors to be equal as well and thus violates the antiferromagnetic $J_2$ bonds), the chain is frustrated — no configuration minimizes all bonds at once. (For $q = 2$ the antiferromagnetic combination $J_1 < 0$ with $J_2 < 0$ is frustrated too; for $q \ge 3$ it is not, because a repeating three-coloring such as $1,2,3,1,2,3,\dots$ keeps both neighbors and next-nearest neighbors distinct, so for $q \ge 3$ the frustration resides in the $J_1 > 0$, $J_2 < 0$ quadrant, as the source paper's zero-temperature phase diagram shows.) This is the Potts analogue of the axial next-nearest-neighbor Ising (ANNNI) chain, studied exactly for $q = 2$ by Dobson (1969) and Stephenson (1970). Notably, the interesting physics of the frustrated Ising chain — anisotropic (elliptic) decay of two-point correlations, disorder points — is governed not by the dominant eigenvalue but by the subleading eigenvalues of its transfer matrix. For $q \ge 3$ the frustrated ("ANNNP") chain had no exact solution before 2025; earlier one-dimensional Potts studies with competing or long-range interactions, such as Glumac and Uzelac (1993), were numerical.

In 2025, Weiguo Yin (arXiv:2503.23758, published as Physical Review B 112, 094424) exactly solved the classical one-dimensional $J_1$–$J_2$ $q$-state Potts chain for arbitrary integer $q$. The key step is a maximally symmetric subspace (MSS) reduction: because the Hamiltonian is invariant under global permutations of the $q$ spin values, the transfer matrix — taken to act on overlapping pairs of neighboring spins, as a $q^2 \times q^2$ matrix — commutes with the permutation group $S_q$, and its sector of fully symmetric states is only two-dimensional, since $S_q$ has exactly two orbits on ordered spin pairs (equal pairs and unequal pairs). The Perron eigenvalue lies in this $2 \times 2$ block; diagonalizing it gives the exact free energy for all $J_1, J_2, T$, and differentiating gives the exact nearest- and next-nearest-neighbor observables $\langle \delta(\sigma_i,\sigma_{i+1})\rangle$ and $\langle \delta(\sigma_i,\sigma_{i+2})\rangle$. The paper further shows that this dominant eigenvalue coincides with that of a simpler field-driven Potts chain (a $q$-state chain whose nearest-neighbor coupling is $J_2$ and which sits in an effective magnetic field of strength $J_1$, so that it has no next-nearest-neighbor interaction), proving an exact thermodynamic mapping between the two Hamiltonians, and uses it to explain a "dome" mechanism for ultranarrow thermal crossovers proposed for stacked ordered layers in materials such as $\mathrm{La}_{2-x}\mathrm{Sr}_x\mathrm{NiO}_4$, $1T$-$\mathrm{TaS}_2$, and EuAuSb.

The 2025 solution is thus a dominant-eigenvalue solution, and the exact solution of this same model is incomplete in three ways. First, no other eigenvalue of the transfer matrix is obtained: correlation lengths, finite-size spectral gaps, and genuine two-point functions $\langle \delta(\sigma_0,\sigma_r)\rangle$ at separation $r \ge 3$ require the subleading spectrum, and for $q \ge 3$ the subleading eigenvalues generally live outside the two-dimensional symmetric block, in symmetry sectors whose dimensions grow with $q$ — the tractability that Stephenson exploited at $q = 2$ does not directly extend. Second, the mapping to the field-driven chain is established only at the level of the dominant eigenvalue and the thermodynamics derived from it; whether the two Hamiltonians also share correlation functions and microscopic configuration weights is not analyzed. Third, the solution is confined to integer $q$ in the transfer-matrix formalism, whereas continuously varying (noninteger) $q$ is meaningful through the Fortuin–Kasteleyn random-cluster representation, in which the partition function is reorganized as a sum over bond configurations and becomes a polynomial in $q$; whether the reduction and the mapping survive in that formulation is unaddressed.

Problem Statement

For the classical one-dimensional $q$-state Potts chain with nearest-neighbor coupling $J_1$ and next-nearest-neighbor coupling $J_2$ (arbitrary real $J_1, J_2$ and temperature $T$, periodic boundary conditions), complete the exact solution begun by the dominant-eigenvalue result of Yin (2025), by resolving the following — components (A) and (B) for arbitrary integer $q \ge 2$, and component (C) for continuously varying (noninteger) $q$:

(A) Subleading spectrum and correlations. Determine the subleading eigenvalues of the transfer matrix — the full set of eigenvalues entering the $r$-dependence of the two-point function, which includes the eigenvalue of second-largest modulus together with the identification of which symmetry sector of the permutation group $S_q$ carries it at each point of the $(J_1, J_2, T)$ parameter space — in exact form (closed form where it exists, otherwise exact roots of an explicitly given finite-degree polynomial), and derive from them the exact correlation length, the finite-size spectral gap, and the two-point correlation functions $\langle \delta(\sigma_0, \sigma_r)\rangle$ at general separation $r$.

(B) Correlation-level status of the mapping. Determine whether the exact mapping between the frustrated $J_1$–$J_2$ chain and the field-driven Potts chain extends beyond the dominant eigenvalue, along two properties to be settled separately: do the two Hamiltonians yield identical correlation functions at arbitrary separation, and separately do they yield identical microscopic configuration statistics, or does at least one of these two properties fail beyond the thermodynamic level — the free energy and the observables derivable from it, such as the bond observables $\langle\delta(\sigma_i,\sigma_{i+1})\rangle$ and $\langle\delta(\sigma_i,\sigma_{i+2})\rangle$ — with correlations at separations $r \ge 3$ or configuration probabilities differing? A proof or disproof of equality settles each of the two properties, and this component is complete once both are settled; a demonstration that a property fails must exhibit a parameter point at which the two models' correlation functions $\langle\delta(\sigma_0,\sigma_r)\rangle$ differ at some separation $r \ge 3$, or at which their microscopic configuration probabilities differ, with the two chains compared under the mapping's parameter correspondence (the frustrated chain at $(J_1, J_2, T)$ against the field-driven chain with nearest-neighbor coupling $J_2$ in the effective field of strength $J_1$) and its observable correspondence (the frustrated chain's nearest-neighbor bond $\langle\delta(\sigma_i,\sigma_{i+1})\rangle$ corresponds to the field-driven chain's magnetization $\langle\delta(\sigma_i,1)\rangle$, and its next-nearest-neighbor bond $\langle\delta(\sigma_i,\sigma_{i+2})\rangle$ corresponds to the field-driven chain's nearest-neighbor bond $\langle\delta(\sigma_i,\sigma_{i+1})\rangle$), so that the separations $r \ge 3$ at which the correlation functions are compared lie beyond these two thermodynamic-level bond observables.

(C) Noninteger $q$. Extend the maximally symmetric subspace reduction and the exact mapping to continuously varying (noninteger) $q$, formulated in the Fortuin–Kasteleyn random-cluster representation, and state which parts of the integer-$q$ solution — free energy, dome mechanism, mapping — persist and which are modified.

These three components jointly complete the exact solution of this one model beyond its dominant eigenvalue and beyond integer $q$; they are one problem, not three. Extensions to quantum Potts chains, to higher-dimensional lattices, and the experimental fitting of layered materials are explicitly outside this problem and remain open directions.

References

1. Weiguo Yin, "Exact solution of the frustrated Potts model with next-nearest-neighbor interactions in one dimension via AI bootstrapping", Physical Review B 112, 094424 (2025), doi:10.1103/y5vc-3t6q, https://arxiv.org/abs/2503.23758 (source paper)
2. J. F. Dobson, "Many-Neighbored Ising Chain", Journal of Mathematical Physics 10, 40–45 (1969), doi:10.1063/1.1664757, https://doi.org/10.1063/1.1664757
3. John Stephenson, "Two one-dimensional Ising models with disorder points", Canadian Journal of Physics 48, 1724–1734 (1970), doi:10.1139/p70-217, https://doi.org/10.1139/p70-217
4. Z. Glumac and K. Uzelac, "Critical behaviour of the 1D q-state Potts model with long-range interactions", Journal of Physics A: Mathematical and General 26, 5267–5278 (1993), doi:10.1088/0305-4470/26/20/013, https://doi.org/10.1088/0305-4470/26/20/013
5. Weiguo Yin, "Half-ice, half-fire-driven ultranarrow phase crossover in one-dimensional decorated q-state Potts ferrimagnets: An AI-co-led exploration", APS Open Science 1, 000047 (2026), doi:10.1103/425p-bt1x, https://arxiv.org/abs/2505.02303
6. Weiguo Yin, "Site-decorated model for unconventional frustrated magnets: Ultranarrow phase crossover and two-dimensional spin reversal transition", Physical Review B 113, 104418 (2026), doi:10.1103/v5qd-5n7z, https://arxiv.org/abs/2502.11270
7. Weiguo Yin and A. M. Tsvelik, "Metamagnetic Transition in Low-Dimensional Site-Decorated Quantum Heisenberg Ferrimagnets", arXiv:2511.06442 (2025), https://arxiv.org/abs/2511.06442
8. Balázs Pozsgay and István Vona, "Bethe Ansatz with a Large Language Model", arXiv:2603.29932 (2026), https://arxiv.org/abs/2603.29932
9. Jurriaan Wouters, Fabian Hassler, Hosho Katsura, and Dirk Schuricht, "Phase diagram of an extended parafermion chain", SciPost Physics Core 5, 008 (2022), doi:10.21468/SciPostPhysCore.5.1.008, https://arxiv.org/abs/2106.15823