Off-shell Bethe vectors beyond the classical Lie series: intrinsic meaning of the uniformizing parameters and extension to exceptional and quantum affine algebras

A 2026 paper of Liashyk, Pakuliak and Ragoucy (arXiv:2601.00713) gave a uniform, intrinsic definition of off-shell Bethe vectors for generic $\mathfrak g$ -invariant integrable models for all classical Lie series $\mathfrak{gl}_n$ , $\mathfrak{o}_{2n+1}$ , $\mathfrak{sp}_{2n}$ , $\mathfrak{o}_{2n}$ , based on two defining relations in the RTT presentation and on two discrete uniformizing parameters $\epsilon_{\mathfrak g}$ and $\xi_{\mathfrak g}$ whose algebraic origin is unexplained, and posed three open directions. The direction concerning a detailed proof of the correspondence with the projection (current) construction in the Yangian double was subsequently resolved for the classical series by the authors' own follow-up papers arXiv:2606.31624 and arXiv:2607.00594. The open problem consists of the following three linked parts: (1) give $\epsilon_{\mathfrak g}$ and $\xi_{\mathfrak g}$ an intrinsic algebraic interpretation, for instance through root and weight data; (2) extend the uniform RTT-based definition and its derived machinery (on-shell property, action formulas, rectangular recurrence relations, coproduct formula) to the exceptional Lie algebras $G_2$ , $F_4$ , $E_6$ , $E_7$ , $E_8$ ; and (3) extend it to the quantum affine (trigonometric) deformations. As of August 2026 only a first $G_2$ step exists (the equivalence of the fifth-order Serre relation in $DY(\mathfrak g_2)$ with a quadratic composed-current relation), and the parameter-interpretation question is untouched.

Background

A quantum integrable model in one dimension is encoded in a monodromy matrix $T(u)$ , a matrix-valued generating series in a spectral parameter $u$ whose entries are operators on a Hilbert space. The entries obey quadratic commutation relations of RTT type (so named after the order $RTT$ of the factors), $R(u,v)T_1(u)T_2(v)=T_2(v)T_1(u)R(u,v)$ , where $R(u,v)$ is a solution of the Yang-Baxter equation (the consistency condition ensuring that the factorized scattering of three auxiliary particles does not depend on their order) acting on the tensor product of two auxiliary spaces, and $T_1$ , $T_2$ denote $T$ acting in the first and second factor. When $R$ is rational and invariant under a finite-dimensional simple Lie algebra $\mathfrak g$ (a $\mathfrak g$ -invariant model), these relations define the Yangian $Y(\mathfrak g)$ , a Hopf-algebra deformation of the enveloping algebra of the current algebra $\mathfrak g[t]$ , in its original matrix (RTT) realization. The trace of $T(u)$ , the transfer matrix, then generates a commuting family of conserved quantities, and the central computational problem is to diagonalize it. The algebraic Bethe ansatz (ABA) does so by constructing eigenvectors of the transfer matrix as polynomials in the creation entries of $T(u)$ acting on a pseudo-vacuum state (a reference state annihilated by the entries of $T(u)$ below the diagonal, on which the diagonal entries act by scalar functions). The eigenvalue property holds only when the free parameters (the Bethe parameters) satisfy a coupled system of algebraic Bethe equations; vectors built the same way with unconstrained parameters are called off-shell Bethe vectors, and they become on-shell Bethe vectors (true eigenvectors) when the Bethe equations hold. Off-shell Bethe vectors are the workhorses of the theory of scalar products, form factors and correlation functions, so their precise definition matters: the loose characterization by the on-shell limit does not determine them, since any term vanishing on the Bethe equations could be added. For $\mathfrak{gl}_n$ , and partially for the orthogonal and symplectic series, off-shell Bethe vectors were previously constructed by the nested ABA (the ansatz applied iteratively, one simple root at a time, to build excitations carrying several colors), by trace and combinatorial formulas, and by the projection method. In the projection method one uses Drinfeld's new (current) realization of the Yangian double $DY(\mathfrak g)$ , the quantum double of the Yangian generated by half-currents attached to the simple roots (the standard basis of the root system of $\mathfrak g$ ), and obtains off-shell Bethe vectors as projections of normally ordered products of simple-root currents onto intersections of opposite Borel subalgebras (the two conjugate maximal triangular subalgebras of the double); these projections are the universal weight functions. The bridge between the RTT and current realizations goes back to Ding-Frenkel isomorphisms and, for types B, C and D, to the isomorphisms of Jing, Liu and Molev and to the Gauss-coordinates-versus-currents analysis of Liashyk and Pakuliak. The source paper of this candidate (Liashyk, Pakuliak, Ragoucy, arXiv:2601.00713, v1 January 2026, v2 June 2026) reversed the logic: it defines off-shell Bethe vectors intrinsically in the RTT presentation and derives the familiar properties as theorems. Uniformly for $\mathfrak g=\mathfrak{gl}_n,\mathfrak{o}_{2n+1},\mathfrak{sp}_{2n},\mathfrak{o}_{2n}$ , the off-shell Bethe vectors $\mathbb B(\bar t)$ , where $\bar t$ is a collection of sets $\bar t^a$ of Bethe parameters of color $a$ , one color per simple root, are fixed up to normalization by the vacuum condition $\mathbb B(\varnothing)=|0\rangle$ and two defining relations: the action of the highest monodromy entry $T_{n',n}(z)$ (the upper-right corner entry of the monodromy matrix), and the action of the zero modes $\mathbb T_a$ associated with the simple roots of $\mathfrak g$ (constant terms of monodromy-matrix entries, located in the lower triangular part of the monodromy matrix), given by sums over partitions of the sets $\bar t^a$ . From these relations alone the paper proves, uniformly for all classical $\mathfrak g$ : that on-shell vectors are transfer-matrix eigenvectors; the action formulas of arbitrary monodromy entries on the Bethe vectors; the rectangular recurrence relations, which express Bethe vectors whose parameters fill a rectangular array through vectors with fewer parameters (generalizing the $\mathfrak{gl}_n$ and $\mathfrak{o}_{2n+1}$ recursions of arXiv:2412.05224 and the earlier ABA for $\mathfrak{o}_{2n+1}$ in arXiv:2008.03664); and the coproduct formula, which controls how Bethe vectors of a chain composed of two subchains factor. The uniformity over the four classical series is achieved through two discrete parameters whose algebraic origin is unexplained. The integer $\xi_{\mathfrak g}$ fixes the dimension $N=2n+1-\xi_{\mathfrak g}$ of the defining representation ( $\xi_{\mathfrak g}=n+1,0,1,1$ for $\mathfrak{gl}_n,\mathfrak{o}_{2n+1},\mathfrak{sp}_{2n},\mathfrak{o}_{2n}$ ) and defines the index involution $i\mapsto i'=\xi_{\mathfrak g}-i$ on matrix indices, which enters the twisted transposition of the monodromy matrix (the transposition of a matrix combined with this index reversal and sign factors). The sign $\epsilon_{\mathfrak g}$ , equal to $0,+1,-1$ for $\mathfrak{gl}_n,\mathfrak{o}_N,\mathfrak{sp}_{2n}$ , enters the uniform formula for the $\mathfrak g$ -invariant $R$ -matrix, in which the usual permutation operator (the operator swapping the two tensor factors) is accompanied by its image $\mathbf Q$ under the twisted transposition, an operator that vanishes identically for $\mathfrak{gl}_n$ ; a sign of this kind had already appeared, as a mere $\pm$ , in the twisted Yangians of Molev, Nazarov and Olshanskii (subalgebras of the Yangian $Y(\mathfrak{gl}_N)$ attached to a symmetric or skew-symmetric bilinear form). The conclusion of the paper asks whether these parameters admit a more intrinsic algebraic interpretation, for instance in terms of the roots and weights of $\mathfrak g$ (the roots being the nonzero weights of the adjoint representation of $\mathfrak g$ , and weights the eigenvalues of the Cartan generators that label the irreducible representations); announces a forthcoming complete proof of the correspondence with the projection construction in $DY(\mathfrak g)$ ; and poses the extension of the framework beyond the classical series, namely to the exceptional Lie algebras $G_2$ , $F_4$ , $E_{6,7,8}$ and to the quantum affine (trigonometric) deformations $U_q(\widehat{\mathfrak g})$ — the $q$ -deformations of the enveloping algebra of the loop algebra of $\mathfrak g$ , whose $R$ -matrices are trigonometric functions of multiplicative spectral parameters — where analogous questions remain largely unexplored. Two follow-up papers by the same authors, posted in June and July 2026, settle the correspondence announcement: arXiv:2606.31624 reformulates the Serre relations (the current-algebra relations encoding the Cartan matrix, generalizing $[E_i,[E_i,E_j]]$ -type identities) as quadratic commutation relations of composed currents, that is, normal-ordered products of simple-root currents attached to positive roots (the roots expressible as nonnegative combinations of the simple roots), within the category of highest-weight representations (representations generated from a vector annihilated by all raising operators); and arXiv:2607.00594 proves that the projection-method Bethe vectors satisfy the two defining relations of arXiv:2601.00713 for all classical series, with a normalization cancelling the zeros induced by the Serre relations. The exceptional direction, by contrast, has only received a first step, for $G_2$ : the equivalence of the fifth-order Serre relation in $DY(\mathfrak g_2)$ with a quadratic relation involving a composed current, while the classification of composed currents for the exceptional types is explicitly deferred to future publications.

Problem Statement

The problem is to complete the uniform theory of off-shell Bethe vectors founded in arXiv:2601.00713 along the directions that remain unresolved as of August 2026. (The third direction posed in that paper's conclusion, the detailed proof of the correspondence between the RTT-based definition and the projection construction in the Yangian double $DY(\mathfrak g)$ , has been discharged for all classical series by the follow-up papers arXiv:2606.31624 and arXiv:2607.00594 and is therefore not asked here.) The open problem has three linked parts, sharing one scientific objective: a single, intrinsically grounded, uniform construction of off-shell Bethe vectors valid for every simple Lie algebra in the rational case and for its trigonometric deformation. All three parts are required for a complete resolution of this objective; settling only part (1), only part (2), or only part (3) is a self-standing partial result and does not complete the objective. (1) Intrinsic interpretation of the uniformizing parameters. Give the two discrete parameters $\epsilon_{\mathfrak g}$ and $\xi_{\mathfrak g}$ , or a common generalization, an intrinsic algebraic or representation-theoretic characterization, for instance in terms of the root and weight data of $\mathfrak g$ , the action of the Weyl group (the finite reflection group generated by the simple-root reflections) on the defining module, or the invariant bilinear form of $\mathfrak g$ . The characterization must reproduce the known values $\epsilon_{\mathfrak g}=0,+1,-1$ for $\mathfrak{gl}_n,\mathfrak{o}_N,\mathfrak{sp}_{2n}$ and $\xi_{\mathfrak g}=n+1,0,1,1$ for $\mathfrak{gl}_n,\mathfrak{o}_{2n+1},\mathfrak{sp}_{2n},\mathfrak{o}_{2n}$ , explain their role in the uniform defining relations, in the $\mathfrak g$ -invariant $R$ -matrix and in the twisted transposition, and be more than a relabeling of the case list: the characterization must be a single rule in the intrinsic root and weight data of $\mathfrak g$ that reproduces the listed classical values and is defined for every simple $\mathfrak g$ , so that it also determines $\epsilon_{\mathfrak g}$ and $\xi_{\mathfrak g}$ for the exceptional types. (2) Extension to the exceptional Lie algebras. Extend the RTT-based definition of off-shell Bethe vectors — that is, the vacuum condition $\mathbb B(\varnothing)=|0\rangle$ plus the two defining relations of Definition 2.1 of arXiv:2601.00713 (the action of the highest monodromy entry $T_{n',n}(z)$ and the action of the simple-root zero modes $\mathbb T_a$ ) — to generic $\mathfrak g$ -invariant integrable models with exceptional $\mathfrak g=G_2,F_4,E_6,E_7,E_8$ . This entails establishing the needed RTT realization with the corresponding $\mathfrak g$ -invariant $R$ -matrices, and proving from the defining relations the same derived structures as in the classical case: the on-shell (transfer-matrix eigenvalue) property under the Bethe equations, the action formulas of the monodromy entries, the rectangular recurrence relations, and the coproduct formula. (3) Trigonometric (quantum affine) extension. Extend the same uniform RTT-based definition and its derived machinery to the quantum affine (trigonometric) deformations, where the rational Yangian doubles are replaced by the quantum affine doubles $U_q(\widehat{\mathfrak g})$ , and verify that the expected action, recursion and coproduct structures continue to follow.

References

A. Liashyk, S. Pakuliak, E. Ragoucy, Bethe Vectors in Quantum Integrable Models with Classical Symmetries, Commun. Math. Phys. 407, 218 (2026), DOI:10.1007/s00220-026-05724-z, https://doi.org/10.1007/s00220-026-05724-z; preprint arXiv:2601.00713, https://arxiv.org/abs/2601.00713
A. Liashyk, S. Pakuliak, E. Ragoucy, Serre Relations in Yangian Doubles, arXiv:2606.31624 (2026), https://arxiv.org/abs/2606.31624
A. Liashyk, S. Pakuliak, E. Ragoucy, Yangian Doubles and off-Shell Bethe Vectors, arXiv:2607.00594 (2026), https://arxiv.org/abs/2607.00594
A. Molev, M. Nazarov, G. Olshanskii, Yangians and Classical Lie Algebras, Russian Math. Surveys 51 (1996) 205-282, DOI:10.1070/RM1996V051N02ABEH002772, https://doi.org/10.1070/RM1996V051N02ABEH002772; preprint arXiv:hep-th/9409025, https://arxiv.org/abs/hep-th/9409025
A. Liashyk, S. Z. Pakuliak, Recurrence relations for off-shell Bethe vectors in trigonometric integrable models, J. Phys. A 55 (2022) 075201, DOI:10.1088/1751-8121/ac491b, https://doi.org/10.1088/1751-8121/ac491b; preprint arXiv:2109.07528, https://arxiv.org/abs/2109.07528
A. Gerrard, V. Regelskis, Nested algebraic Bethe ansatz for deformed orthogonal and symplectic spin chains, Nucl. Phys. B 956 (2020) 115021, DOI:10.1016/j.nuclphysb.2020.115021, https://doi.org/10.1016/j.nuclphysb.2020.115021; preprint arXiv:1912.11497, https://arxiv.org/abs/1912.11497
A. Liashyk, S. Z. Pakuliak, Gauss Coordinates vs Currents for the Yangian Doubles of the Classical Types, SIGMA 16 (2020) 120, DOI:10.3842/SIGMA.2020.120, https://doi.org/10.3842/SIGMA.2020.120; preprint arXiv:2006.01579, https://arxiv.org/abs/2006.01579
A. Liashyk, S. Pakuliak, E. Ragoucy, Rectangular Recurrence Relations in $\mathfrak{gl}_n$ and $\mathfrak{o}_{2n+1}$ Invariant Integrable Models, SIGMA 21 (2025) 078, DOI:10.3842/SIGMA.2025.078, https://doi.org/10.3842/SIGMA.2025.078; preprint arXiv:2412.05224, https://arxiv.org/abs/2412.05224
A. Liashyk, S. Z. Pakuliak, Algebraic Bethe ansatz for $\mathfrak{o}_{2n+1}$ -invariant integrable models, Theoret. Math. Phys. 206 (2021) 19-39, DOI:10.1134/S0040577921010025, https://doi.org/10.1134/S0040577921010025; preprint arXiv:2008.03664, https://arxiv.org/abs/2008.03664
N. Zare Harofteh, R. I. Nepomechie, Preparing multi-qudit states in a definite-weight subspace, arXiv:2606.24659 (2026), https://arxiv.org/abs/2606.24659
J. Ding, I. Frenkel, Isomorphism of two realizations of quantum affine algebra $U_q(\widehat{\mathfrak{gl}}_n)$ , Commun. Math. Phys. 156 (1993) 277-300, DOI:10.1007/BF02098484, https://doi.org/10.1007/BF02098484
N. Jing, M. Liu, A. Molev, Isomorphism between the R-matrix and Drinfeld presentations of Yangian in types B, C and D, Commun. Math. Phys. 361 (2018) 827-872, DOI:10.1007/s00220-018-3185-x, https://doi.org/10.1007/s00220-018-3185-x; preprint arXiv:1705.08155, https://arxiv.org/abs/1705.08155