The Bethe ansatz equations of the periodic spin-12\frac{1}{2} isotropic Heisenberg (XXX) chain admit singular solutions containing the pair {i2,−i2}\{\tfrac{\mathrm{i}}{2},-\tfrac{\mathrm{i}}{2}\}, for which the naive Bethe eigenvector vanishes and the energy diverges; after regularization, only those singular solutions that satisfy an additional consistency (physicality) condition correspond to genuine eigenvectors. Giri and Deguchi (2015) showed that for even chain length NN the sign-symmetric singular solutions reduce the Bethe equations to (M−2)/2(M-2)/2 equations for even MM and (M−3)/2(M-3)/2 for odd MM, constructed the explicit M=2M=2 and M=3M=3 singular eigenstates, and proved that for odd NN physical singular solutions in the M=3M=3 sector exist exactly when N=3(2k+1)N=3(2k+1), with λ3=±32\lambda_3=\pm\tfrac{\sqrt{3}}{2}; they also excluded M=4M=4 solutions at N=15N=15 graphically. However, no complete classification is known: for even NN the sign-symmetry (flip-invariance) of all physical singular solutions is supported by numerics up to N=14N=14 and by rigged-configuration conjectures but is unproven, and for odd NN the M≥4M\ge4 sectors are analytically unexplored (numerical searches up to N=13N=13 found no physical singular solutions, and a graphical argument excluded N=15N=15, M=4M=4). The problem asks for this complete classification: prove or refute even-NN sign-symmetry, and determine existence and structure of the physical singular solutions for odd NN and M≥4M\ge4.

Background

The spin-12\frac{1}{2} isotropic Heisenberg chain (XXX chain) on a ring of NN sites with periodic boundary conditions is the archetypal exactly solved quantum many-body model. In the algebraic Bethe ansatz, each eigenstate in the sector of MM down spins (the number of spins flipped relative to the fully polarized reference state, also called the magnon number) is labelled by MM complex rapidities, called Bethe roots, {λ1,…,λM}\{\lambda_1,\ldots,\lambda_M\}, which solve the Bethe ansatz equations (BAE)

(λα+i2λα−i2)N=∏β≠αMλα−λβ+iλα−λβ−i,α=1,…,M.\left(\frac{\lambda_\alpha+\tfrac{\mathrm{i}}{2}}{\lambda_\alpha-\tfrac{\mathrm{i}}{2}}\right)^{N}=\prod_{\beta\neq\alpha}^{M}\frac{\lambda_\alpha-\lambda_\beta+\mathrm{i}}{\lambda_\alpha-\lambda_\beta-\mathrm{i}},\qquad \alpha=1,\ldots,M .

Most solutions of the BAE, the regular solutions with pairwise distinct roots, produce nonzero Bethe eigenvectors. However, the BAE also admit singular solutions: solutions containing the pair {λ1=i2,λ2=−i2}\{\lambda_1=\tfrac{\mathrm{i}}{2},\lambda_2=-\tfrac{\mathrm{i}}{2}\}, for which the algebraic Bethe ansatz expression for the eigenvector vanishes and the energy diverges. Singular solutions cannot simply be discarded, because without them the Bethe ansatz solutions do not complete the spectrum; for example, at N=6N=6, M=3M=3 the BAE have 5 singular solutions of which exactly one is physical (see below), and at N=8N=8, M=4M=4 there are 21 singular solutions of which 3 are physical.

Nepomechie and Wang (2013) regularized such solutions by shifting the singular pair and taking a limit. The outcome is a practical criterion: a singular solution is physical, i.e. corresponds to a genuine eigenvector of the transfer matrix (the generating operator of the chain's commuting conserved quantities, whose logarithmic derivative gives the Hamiltonian) and of the Hamiltonian, if and only if the remaining rapidities λ3,…,λM\lambda_3,\ldots,\lambda_M (which must be pairwise distinct and different from ±i2\pm\tfrac{\mathrm{i}}{2}) satisfy the pole-free form of the BAE,

(λk+i2λk−i2)N−1λk−3i2λk+3i2=∏j=3, j≠kMλk−λj+iλk−λj−i,k=3,…,M,\left(\frac{\lambda_k+\tfrac{\mathrm{i}}{2}}{\lambda_k-\tfrac{\mathrm{i}}{2}}\right)^{N-1}\frac{\lambda_k-\tfrac{3\mathrm{i}}{2}}{\lambda_k+\tfrac{3\mathrm{i}}{2}} =\prod_{j=3,\, j\neq k}^{M}\frac{\lambda_k-\lambda_j+\mathrm{i}}{\lambda_k-\lambda_j-\mathrm{i}},\qquad k=3,\ldots,M ,

together with the consistency condition, called the physicality condition,

(−∏j=3Mλj+i2λj−i2)N=1.\left(-\prod_{j=3}^{M}\frac{\lambda_j+\tfrac{\mathrm{i}}{2}}{\lambda_j-\tfrac{\mathrm{i}}{2}}\right)^{N}=1 .

For M=2M=2 the product is empty and the condition reduces to (−1)N=1(-1)^N=1, so the singular pair {i2,−i2}\{\tfrac{\mathrm{i}}{2},-\tfrac{\mathrm{i}}{2}\} is physical exactly when NN is even. An equivalent form of the consistency condition (equivalent given the pole-free equations above), convenient for sector-by-sector analysis, is ∏α=3M(λα+i2)(λα+3i2)(λα−i2)(λα−3i2)=(−1)N\prod_{\alpha=3}^{M}\frac{(\lambda_\alpha+\tfrac{\mathrm{i}}{2})(\lambda_\alpha+\tfrac{3\mathrm{i}}{2})}{(\lambda_\alpha-\tfrac{\mathrm{i}}{2})(\lambda_\alpha-\tfrac{3\mathrm{i}}{2})}=(-1)^N.

Kirillov and Sakamoto (2014) used rigged configurations — combinatorial objects (Young-diagram partitions carrying integer labels called riggings), introduced earlier in the theory of integrable systems to count Bethe states — and conjectured (their Conjectures 13 and 14) that for even NN the physical solutions, regular plus physical singular, are in bijection with rigged configurations, with the flip map {λα}↦{−λα}\{\lambda_\alpha\}\mapsto\{-\lambda_\alpha\} acting as a flip of the riggings; in particular, physical singular solutions of even-length chains are predicted to be sign-symmetric as multisets, {λα}={−λα}\{\lambda_\alpha\}=\{-\lambda_\alpha\}, which implies the vanishing of the sum of the rapidities; the converse implication is not known in general and is itself part of the open question. These conjectures agree with the numerical data up to N=14N=14 but remain unproven.

Giri and Deguchi (2015) analyzed the physical singular solutions sector by sector. For even NN they showed that sign-symmetric singular solutions reduce the BAE to (M−2)/2(M-2)/2 equations for even MM and (M−3)/2(M-3)/2 equations for odd MM; they derived sufficient conditions implying sign-symmetry (the remaining rapidities pairwise sign-opposite, together with λ3=0\lambda_3=0 when MM is odd); they constructed the explicit regularized singular eigenstates for M=2M=2 and M=3M=3; and they reported that, numerically, no other even-NN singular solutions appear up to N=14N=14. For odd NN they proved that no M=2M=2 physical singular solutions exist, and that in the M=3M=3 sector the consistency condition forces λ3=±32\lambda_3=\pm\tfrac{\sqrt{3}}{2} while the BAE then hold exactly for N=3(2k+1)N=3(2k+1), k=1,2,3,…k=1,2,3,\ldots; for M=4M=4 they gave a graphical search method and excluded physical singular solutions at N=15N=15.

More recent systematic frameworks settle completeness of the spectrum in generalized formulations, but not the classification of the standard BAE. Tarasov (2018) proved completeness of the Bethe ansatz for the periodic isotropic Heisenberg model by describing a system of algebraic equations, in bijection with transfer-matrix eigenvalues, for pairs of polynomials with prescribed discrete Wronskian (Casorati determinant, the discrete analogue of a Wronskian) and divisibility conditions; this system reduces to the standard BAE when the smaller-degree polynomial is coprime with the Wronskian, which is precisely the condition that fails for singular solutions. The rational Q-system of Hou, Jiang and Zhu (2024), a system of functional equations whose polynomial solutions give all and only the physical solutions of the BAE (completeness proved building on Tarasov's theorem), reproduces the physicality condition for singular solutions at spin 12\frac{1}{2}; He, Hou, Liu and Tan (2025) extended the framework to twisted (periodic chains whose Bethe equations carry a nontrivial boundary phase) and open boundaries, deriving physicality conditions for singular solutions (recovering the Nepomechie-Wang condition in the twistless limit) and observing numerically that a generic twist eliminates physical singular solutions altogether. None of these works classifies which (N,M)(N,M) sectors of the periodic spin-12\frac{1}{2} chain contain physical singular solutions, or determines their root structure in general.

Problem Statement

Determine completely, for the homogeneous periodic spin-12\frac{1}{2} XXX chain of NN sites, the physical singular solutions of the Bethe ansatz equations in every down-spin sector MM (the number of spins flipped relative to the fully polarized reference state); that is, for each pair (N,M)(N,M), decide whether physical singular solutions exist, and describe all of them explicitly. Here the Bethe ansatz equations (BAE) for the sector of MM down spins are

(λα+i2λα−i2)N=∏β≠αMλα−λβ+iλα−λβ−i,α=1,…,M;\left(\frac{\lambda_\alpha+\tfrac{\mathrm{i}}{2}}{\lambda_\alpha-\tfrac{\mathrm{i}}{2}}\right)^{N}=\prod_{\beta\neq\alpha}^{M}\frac{\lambda_\alpha-\lambda_\beta+\mathrm{i}}{\lambda_\alpha-\lambda_\beta-\mathrm{i}},\qquad \alpha=1,\ldots,M ;

a singular solution is a solution of the BAE containing the pair {i2,−i2}\{\tfrac{\mathrm{i}}{2},-\tfrac{\mathrm{i}}{2}\}, with the remaining rapidities λ3,…,λM\lambda_3,\ldots,\lambda_M pairwise distinct and different from ±i2\pm\tfrac{\mathrm{i}}{2}; and it is physical if those remaining rapidities satisfy both the pole-free Bethe equations

(λk+i2λk−i2)N−1λk−3i2λk+3i2=∏j=3, j≠kMλk−λj+iλk−λj−i,k=3,…,M,\left(\frac{\lambda_k+\tfrac{\mathrm{i}}{2}}{\lambda_k-\tfrac{\mathrm{i}}{2}}\right)^{N-1}\frac{\lambda_k-\tfrac{3\mathrm{i}}{2}}{\lambda_k+\tfrac{3\mathrm{i}}{2}} =\prod_{j=3,\, j\neq k}^{M}\frac{\lambda_k-\lambda_j+\mathrm{i}}{\lambda_k-\lambda_j-\mathrm{i}},\qquad k=3,\ldots,M ,

and the consistency condition (physicality condition)

(−∏j=3Mλj+i2λj−i2)N=1.\left(-\prod_{j=3}^{M}\frac{\lambda_j+\tfrac{\mathrm{i}}{2}}{\lambda_j-\tfrac{\mathrm{i}}{2}}\right)^{N}=1 .

The completion boundary consists of the following linked components, left open by Giri and Deguchi (2015):

  1. (Even NN.) Prove or disprove that every physical singular solution of an even-length chain is sign-symmetric, i.e. invariant as a multiset under negation, {λα}={−λα}\{\lambda_\alpha\}=\{-\lambda_\alpha\} (this implies ∑α=1Mλα=0\sum_{\alpha=1}^{M}\lambda_\alpha=0; the converse is not known in general and must not be assumed). The known sufficient conditions, namely that the remaining rapidities be pairwise sign-opposite together with λ3=0\lambda_3=0 when MM is odd, are not known to be necessary, and the evidence beyond them is numerical, up to N=14N=14. A valid answer either proves sign-symmetry for all even NN and all MM (thereby establishing the flip-invariance predicted by the rigged-configuration conjectures), or exhibits a physical singular solution of an even-length chain that is not sign-symmetric.

  2. (Odd NN, M≥4M\ge4.) The sectors M=2M=2 (excluded) and M=3M=3 (solutions exist exactly for N=3(2k+1)N=3(2k+1), k=1,2,3,…k=1,2,3,\ldots, with λ3=±32\lambda_3=\pm\tfrac{\sqrt{3}}{2}) are completely classified. Determine, for every M≥4M\ge4 and every odd NN, whether physical singular solutions exist, and classify all of them; the published results in this range are the numerical tables of Hao, Nepomechie and Sommese, which found no physical singular solutions for the odd lengths N≤13N\le13 in all computed sectors M≤6M\le6 (while those authors expected that such exceptions do occur for higher values of MM), and the graphical exclusion at N=15N=15, M=4M=4.

These components jointly deliver one scientific objective: a complete finite-NN classification of the physical singular sector of the XXX Bethe ansatz, closing the last case-by-case gap in the description of the solution set required for completeness of the Bethe ansatz.

References

  1. P. R. Giri and T. Deguchi, 'Singular eigenstates in the even(odd) length Heisenberg spin chain', J. Phys. A: Math. Theor. 48 (2015) 175207, arXiv:1411.5839, doi:10.1088/1751-8113/48/17/175207, https://doi.org/10.1088/1751-8113/48/17/175207.
  2. R. I. Nepomechie and C. Wang, 'Algebraic Bethe ansatz for singular solutions', J. Phys. A: Math. Theor. 46 (2013) 325002, arXiv:1304.7978, doi:10.1088/1751-8113/46/32/325002, https://doi.org/10.1088/1751-8113/46/32/325002.
  3. W. Hao, R. I. Nepomechie and A. J. Sommese, 'Completeness of solutions of Bethe's equations', Phys. Rev. E 88 (2013) 052113, arXiv:1308.4645, doi:10.1103/PhysRevE.88.052113, https://doi.org/10.1103/PhysRevE.88.052113.
  4. W. Hao, R. I. Nepomechie and A. J. Sommese, 'Singular solutions, repeated roots and completeness for higher-spin chains', J. Stat. Mech. (2014) P03024, arXiv:1312.2982, doi:10.1088/1742-5468/2014/03/P03024, https://doi.org/10.1088/1742-5468/2014/03/P03024.
  5. R. I. Nepomechie and C. Wang, 'Twisting singular solutions of Bethe's equations', J. Phys. A: Math. Theor. 47 (2014) 505004, arXiv:1409.7382, doi:10.1088/1751-8113/47/50/505004, https://doi.org/10.1088/1751-8113/47/50/505004.
  6. A. N. Kirillov and R. Sakamoto, 'Singular solutions to the Bethe ansatz equations and rigged configurations', J. Phys. A: Math. Theor. 47 (2014) 205207, arXiv:1402.0651, doi:10.1088/1751-8113/47/20/205207, https://doi.org/10.1088/1751-8113/47/20/205207.
  7. T. Deguchi and P. R. Giri, 'Non self-conjugate strings, singular strings and rigged configurations in the Heisenberg model', J. Stat. Mech. (2015) P02004, arXiv:1408.7030, doi:10.1088/1742-5468/2015/02/P02004, https://doi.org/10.1088/1742-5468/2015/02/P02004.
  8. R. Sakamoto, 'Rigged configurations approach for the spin-1/2 isotropic Heisenberg model', J. Phys. A: Math. Theor. 48 (2015) 165201, arXiv:1501.00532, doi:10.1088/1751-8113/48/16/165201, https://doi.org/10.1088/1751-8113/48/16/165201.
  9. V. Tarasov, 'Completeness of the Bethe ansatz for the periodic isotropic Heisenberg model', Rev. Math. Phys. 30 (2018) 1840018, doi:10.1142/S0129055X18400184, https://doi.org/10.1142/S0129055X18400184.
  10. J. Hou, Y. Jiang and R.-D. Zhu, 'Spin-s rational Q-system', SciPost Phys. 16 (2024) 113, arXiv:2303.07640, doi:10.21468/SciPostPhys.16.4.113, https://doi.org/10.21468/SciPostPhys.16.4.113.
  11. Y.-J. He, J. Hou, Y.-C. Liu and Z.-X. Tan, 'Spin-s Q-systems: twist and open boundaries', SciPost Phys. Core 8 (2025) 057, arXiv:2502.15636, doi:10.21468/SciPostPhysCore.8.3.057, https://doi.org/10.21468/SciPostPhysCore.8.3.057.
  12. G. Aupetit-Diallo, G. Pecci, A. G. Volosniev, M. Albert, A. Minguzzi and P. Vignolo, 'Necklace ansatz for strongly repulsive spin mixtures on a ring', SciPost Phys. Core 8 (2025) 022, arXiv:2407.16618, doi:10.21468/SciPostPhysCore.8.1.022, https://doi.org/10.21468/SciPostPhysCore.8.1.022.
  13. J. Links, 'Integrable 3-site, tilted, extended Bose-Hubbard model with nearest-neighbour interactions', SIGMA 21 (2025) 077, arXiv:2506.21789, doi:10.3842/SIGMA.2025.077, https://doi.org/10.3842/SIGMA.2025.077.