Dark states in an integrable X ​ Y XY central spin model Jaco van Tonder, Jon Links Abstract Eigenstates of central spin models in which the central spin is unentangled with the environment are known as dark states. They have recently been observed in a class of integrable X ​ X XX models. Here we find that dark states are present in X ​ Y XY models, but only for particular configurations of the central spin magnetic field. We show this via an explicit construction of the states. 1 Introduction Central spin models have gained renewed attention due to their possible applications in modern quantum technologies focused on quantum sensing and metrology [ 16 ] . This is due to their integrability allowing high-fidelity control of these systems at a mesoscopic scale, where the exponentially increasing size of the Hilbert space would in general be prohibitive [ 6 ] . The central spin allows the dynamics of the spin bath to be monitored, and for feedback to be used to steer the dynamics of the bath in a desired direction. There are several physical or engineered systems for which central spin models provide a theoretical model. These include nitrogen vacancy (NV) centres in diamonds [ 23 ] , room temperature quantum memory storage [ 18 , 21 , 22 ] , quantum batteries [ 11 ] and quantum dots [ 2 , 19 ] – an active area of research which gained public attention through the 2023 Nobel prize for Chemistry [ 13 ] . On the theoretical side, these models have been known for their integrability since Gaudin’s seminal paper [ 8 ] . Integrability allows for the analytic solution of the eigenstates and eigenspectrum of the models through the Bethe Ansatz. This makes these models well suited for studies of their equilibrium and dynamical behaviour; in particular, tests of the Eigenstate Thermalisation Hypothesis (ETH) and investigation of many-body localisation [ 1 , 12 , 17 , 22 ] . Integrability is also essential for perturbative solutions of physical models. For example, integrability provides a wavefunction Ansatz in variational eigensolvers to model strong electron correlation in quantum chemistry [ 7 ] . Integrability is also of great interest in mathematical physics. There is ongoing work on extending or modifying known integrable models to obtain new ones. This has recently led to establishing that the XX model, which models certain resonant dipolar spin systems, is integrable [ 5 , 21 ] . Integrability was subsequently extended to the XY model [ 20 ] using conserved charges discussed in [ 14 ] , and the same charges with self-interaction obtained by Skrypnyk [ 15 ] . These charges were diagonalised using several modifications of the standard Algebraic Bethe Ansatz for arbitrary spin [ 15 ] . These expressions for the eigenvalues can also be obtained through the use of functional relations as described in [ 20 ] . Here we derive the XY central spin model Hamiltonian and its charges through a limit of the charges for the XYZ Richardson-Gaudin model in [ 15 ] . We employ a reparametrisation of the results of [ 15 ] to obtain regularised eigenvalue and eigenstate results for the central spin model which reproduce the diagonalisation results in the literature [ 3 , 21 ] in the appropriate limits. It is found that dark states, states in which the central spin is not entangled with the environment, are present for special configurations of the magnetic field \vec{B}=(B^{x},B^{y},B^{z}) . This is reminiscent of the XX model for which these occur for an out-of-plane magnetic field, B^{x}=B^{y}=0 [ 5 , 21 ] . In Sect. 2 we give a summary of the results found in this work. In the following Sect. 3 we show how the XY central spin Hamiltonian and its conserved charges for spin- 1/2 central spin can be derived from the XYZ Gaudin model charges. This allows the eigenstate results of [ 15 ] to be used to find the eigenstates for the XY central spin model in Sect. 4 . In Sec. 5 we show that for special magnetic field configurations dark states occur, in analogy to those that were seen in the out-of-plane XX model [ 21 ] . Concluding remarks are given in Sect. 6 . In the Appendix 7 we take the isotropic limit to confirm that these recover the results for the XX model. This is supplemented with some heuristic arguments and numerical results about how the dark states emerge from the states with generic magnetic field configurations. 2 Overview of main results Consider a set of L+1 spin operators \{S^{x}_{j},\,S_{j}^{y},\,S^{z}_{j}:j=0,1,\dots,L\} satisfying the standard canonical commutation relations \displaystyle[S^{\theta}_{j},\,S^{\eta}_{k}]=i\delta_{jk}\sum_{\kappa\in\{x,y,z\}}\varepsilon^{\theta\eta\kappa}S^{\kappa}_{j}, (1) where i=\sqrt{-1} and \varepsilon^{\theta\eta\kappa} is the Levi-Civita symbol. The main focus of this work is the central spin Hamiltonian [ 20 ] \displaystyle H \displaystyle=B^{x}S^{x}_{0}+B^{y}S^{y}_{0}+B^{z}S^{z}_{0}+2\sum_{j=1}^{L}(X_{j}S^{x}_{0}S^{x}_{j}+Y_{j}S^{y}_{0}S^{y}_{j}) (2) describing a central spin- 1/2 particle interacting with a bath of L surrounding spins through an inhomogeneous XY interaction. The central spin is also subject to an external magnetic field \vec{B}=\{B^{x},B^{y},B^{z}\} . In [ 20 ] it was shown that the model is integrable when Y_{j}^{2}-X_{j}^{2}=2\beta , with the constant \beta\in\mathbb{R} being free to be chosen through rescalings of the parameters and magnetic field components. The spins s_{j} of the surrounding bath particles are arbitrary. Along with \beta we introduce a set of distinct parameters \{\epsilon_{j}:j=1,\dots,L\} such that \epsilon_{j}\pm\beta>0 for all j=1,\dots,L . We also define f_{j}^{\pm}=\sqrt{\epsilon_{j}\pm\beta} . In terms of these it is convenient to use the parametrisation \displaystyle X_{j}=\sqrt{\epsilon_{j}-\beta},\quad Y_{j}=\sqrt{\epsilon_{j}+\beta}. The Hamiltonian ( 2 ) possesses L conserved charges [ 20 ] \displaystyle Q_{i} \displaystyle=-2S_{0}^{z}S_{i}^{z}+\frac{B^{x}}{f_{i}^{+}}S_{i}^{x}+\frac{B^{y}}{f_{i}^{-}}S^{y}_{i}+\frac{f_{i}^{-}}{f_{i}^{+}}S_{i}^{x}S_{i}^{x}+\frac{f_{i}^{+}}{f_{i}^{-}}S_{i}^{y}S_{i}^{y} \displaystyle\quad+2\sum_{\mathrel{{\mathop{j=1}\limits}}}^{L}\frac{1}{\epsilon_{i}-\epsilon_{j}}(f_{i}^{+}f_{j}^{-}S^{x}_{i}S^{x}_{j}+f_{i}^{-}f_{j}^{+}S^{y}_{i}S^{y}_{j}+f_{j}^{+}f_{j}^{-}S^{z}_{i}S^{z}_{j}). To describe its eigenstates we note that we can also bring the Hamiltonian into the form \displaystyle H \displaystyle=B^{z}S_{0}^{z}+S_{0}^{-}A^{+}+S_{0}^{+}A^{-}, where A^{\pm}=(B^{x}\pm iB^{y})I/2+\sum_{j=1}^{L}(f_{j}^{-}S^{x}_{j}\pm if_{j}^{+}S^{y}_{j}). (3) Along with the eigenstates \ket{\pm}_{0} of S^{z}_{0} , define reference states from which to build the eigenstates via the Bethe Ansatz \displaystyle v^{\pm}_{B} \displaystyle=v_{1}^{\pm}\otimes\cdots\otimes v_{L}^{\pm}. Here the v_{k}^{\pm} satisfy \displaystyle T_{k}^{z}v_{k}^{+}=-s_{k}v_{k}^{+},\;T_{k}^{-}v_{k}^{+}=0;\quad T_{k}^{z}v_{k}^{-}=s_{k}v_{k}^{-},\;T_{k}^{+}v_{k}^{-}=0 for the operators \displaystyle T_{k}^{z} \displaystyle=-i\frac{\sqrt{2\beta}}{f_{k}^{-}}S_{k}^{x}-\frac{f_{k}^{+}}{f_{k}^{-}}S_{k}^{z},\quad T_{k}^{\pm} \displaystyle=\pm iS_{k}^{y}-\frac{f_{k}^{+}}{f_{k}^{-}}S_{k}^{x}+i\frac{\sqrt{2\beta}}{f_{k}^{-}}S_{k}^{z}. In terms of the operators \displaystyle S^{\pm}_{k}(u) \displaystyle=\pm(2k-1)i\sqrt{2\beta}+B^{x}\pm iB^{y}-i\zeta\sqrt{2\beta} \displaystyle\quad+2\sum_{j=1}^{L}\frac{f^{+}_{j}(u-\beta)}{\epsilon_{j}-u}T_{j}^{\mp}-2i\sqrt{2\beta}\sum_{j=1}^{L}T_{j}^{z}, the eigenstates generically take the form \displaystyle\ket{\kappa;u_{1},\dots,u_{M}} \displaystyle=\left.\left[A^{+}-\kappa S_{0}^{+}\right]\ket{-}_{0}\otimes S^{+}_{1}(u_{1})S^{+}_{2}(u_{2})\cdots S^{+}_{M}(u_{M})v^{-}_{B}\right|_{\zeta=-1} for M=N_{s}\equiv\sum_{j=1}^{L}2s_{j} and have energy E=-B^{z}/2-\kappa . They depend on a parameter \kappa satisfying \displaystyle\kappa^{2}+B^{z}\kappa+iB^{x}\sqrt{2\beta}-2\beta-2\sum_{j=1}^{L}(\epsilon_{j}-\beta)s_{j}+2\sum_{m=1}^{M}(u_{m}-\beta) \displaystyle=-\frac{1}{4}\left(\left((M+2)\sqrt{2\beta}-iB^{x}\right)^{2}-(B^{y})^{2}\right)\frac{\prod_{m=1}^{M}(u_{m}-\beta)}{\prod_{j=1}^{L}(\epsilon_{j}-\beta)^{2s_{j}}}. and a set of Bethe roots \{u_{m}\}_{m=1}^{M} satisfying the Bethe Ansatz equations \displaystyle 2u_{l}+4\beta-iB^{x}\sqrt{2\beta}+2\sum_{j=1}^{L}\frac{(\epsilon_{j}-\beta)(u_{l}+\beta)}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{M}\frac{(u_{m}-\beta)(u_{l}+\beta)}{u_{l}-u_{m}} \displaystyle\;=\frac{1}{4}\left(\left((M+2)\sqrt{2\beta}-iB^{x}\right)^{2}-(B^{y})^{2}\right)\prod_{j=1}^{L}\left(\frac{\epsilon_{j}-u_{l}}{\epsilon_{j}-\beta}\right)^{2s_{j}}\prod_{m\neq l}^{M}\frac{u_{m}-\beta}{u_{m}-u_{l}}. These states are generically bright states, in other words entangled states of the bath spins and the central spin. However, if the magnetic field components satisfy B^{x}=0,\quad\frac{B^{y}}{2\sqrt{2\beta}}=M^{\prime}-N_{s}/2 (4) for M^{\prime}\in\{0,1,\dots,N_{s}\} , then some of these states become dark states in which the eigenstates are product states of the central spin and bath spins. These are the eigenstates \displaystyle\left.\ket{+}_{0}\otimes S^{-}_{1}(u_{1})\cdots S^{-}_{M^{\prime}}(u_{M^{\prime}})\,v^{+}_{B}\right|_{\zeta=-1}, \displaystyle\left.\ket{-}_{0}\otimes S^{+}_{1}(u_{1})\cdots S^{+}_{M^{\prime}}(u_{M^{\prime}})\,v^{-}_{B}\right|_{\zeta=+1}. These have respective energy E=\pm B^{z}/2 and Bethe roots satisfying the Bethe Ansatz equations \displaystyle 2\beta+2\sum_{j=1}^{L}\frac{(u_{l}+\beta)(\epsilon_{j}-\beta)}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{M^{\prime}}\frac{(u_{l}+\beta)(u_{m}-\beta)}{u_{l}-u_{m}}=0. 3 Derivation of integrable XY central spin Hamiltonians A class of integrable central spin models, with spin-1/2 central spin and arbitrary bath spins, is derived as follows. Introducing an additional parameter \epsilon_{0} we consider the L+1 conserved charges 1 1 1 We use \zeta instead of B^{z} to prevent confusion with B^{z} of the XY Hamiltonian ( 6 ) below. [ 15 , 20 ] \displaystyle\mathcal{Q}_{i}=\zeta S^{z}_{i}+\frac{B^{x}}{f_{i}^{+}}S^{x}_{i}+\frac{B^{y}}{f_{i}^{-}}S^{y}_{i}+\frac{f_{i}^{-}}{f_{i}^{+}}S^{x}_{i}S^{x}_{i}+\frac{f_{i}^{+}}{f_{i}^{-}}S^{y}_{i}S^{y}_{i} \displaystyle\quad\quad\quad\quad+2\sum_{j\neq i}^{L}\frac{1}{\epsilon_{i}-\epsilon_{j}}(f_{i}^{+}f_{j}^{-}S^{x}_{i}S^{x}_{j}+f_{i}^{-}f_{j}^{+}S^{y}_{i}S^{y}_{j}+f_{j}^{+}f_{j}^{-}S^{z}_{i}S^{z}_{j}), (5) \displaystyle[\mathcal{Q}_{i},\mathcal{Q}_{j}]=0,\quad\forall{i,j}\in\{0,1,\dots,L\}. We take \{S^{x}_{0},\,S_{0}^{y},\,S^{z}_{0}\} to be the spin- 1/2 operators, identify the zero subscript with the central spin and the other subscripts with the bath spins. The zeroth charge \mathcal{Q}_{0} will then become the central spin Hamiltonian and \{\mathcal{Q}_{j}:j=1,\dots,L\} its conserved charges. Noting that for spin-1/2 operators we have \displaystyle\left(S^{x}_{0}\right)^{2}=\left(S^{y}_{0}\right)^{2}=\frac{1}{4}I, it is convenient for the zeroth charge to subtract off the squared spin operator terms. After making the change of variables \{\zeta\mapsto B^{z}/\sqrt{\epsilon_{0}}\} for the magnetic field, we obtain the XY central spin model through the limit as \epsilon_{0}\to\infty with the Hamiltonian being \displaystyle H \displaystyle=\lim\limits_{\epsilon_{0}\to\infty}\left[\sqrt{\epsilon_{0}}\mathcal{Q}_{0}-\frac{\sqrt{\epsilon_{0}}}{4}\left(\frac{f^{-}_{0}}{f^{+}_{0}}+\frac{f^{+}_{0}}{f^{-}_{0}}\right)\right] \displaystyle=B^{x}S^{x}_{0}+B^{y}S^{y}_{0}+B^{z}S^{z}_{0}+2\sum_{j=1}^{L}(f_{j}^{-}S^{x}_{0}S^{x}_{j}+f_{j}^{+}S^{y}_{0}S^{y}_{j}), (6) and the charges \displaystyle Q_{i} \displaystyle=\lim\limits_{\epsilon_{0}\to\infty}\mathcal{Q}_{i} \displaystyle=-2S_{0}^{z}S_{i}^{z}+\frac{B^{x}}{f_{i}^{+}}S_{i}^{x}+\frac{B^{y}}{f_{i}^{-}}S^{y}_{i}+\frac{f_{i}^{-}}{f_{i}^{+}}S_{i}^{x}S_{i}^{x}+\frac{f_{i}^{+}}{f_{i}^{-}}S_{i}^{y}S_{i}^{y} \displaystyle\quad+2\sum_{\mathrel{{\mathop{j=1}\limits}}}^{L}\frac{1}{\epsilon_{i}-\epsilon_{j}}(f_{i}^{+}f_{j}^{-}S^{x}_{i}S^{x}_{j}+f_{i}^{-}f_{j}^{+}S^{y}_{i}S^{y}_{j}+f_{j}^{+}f_{j}^{-}S^{z}_{i}S^{z}_{j}), (7) reproducing those introduced in [ 20 ] . 4 The Bethe Ansatz The eigenstates of the charges \mathcal{Q}_{j} , as well as their eigenvalues, have been found in [ 15 ] . Here we recall the algebraic setup and the results required in order to obtain the eigenstates and eigenvalues of H and its conserved charges Q_{j} . 4.1 Relations for the Algebraic Bethe Ansatz Introduce the Lax algebra generators \displaystyle S^{-}(u) \displaystyle=B^{x}-iB^{y}-i\zeta\sqrt{2\beta}+2\sum_{j=0}^{L}\frac{f^{+}_{j}(u-\beta)}{\epsilon_{j}-u}T_{j}^{+}-2i\sqrt{2\beta}\sum_{j=0}^{L}T_{j}^{z}, \displaystyle S^{+}(u) \displaystyle=B^{x}+iB^{y}-i\zeta\sqrt{2\beta}+2\sum_{j=0}^{L}\frac{f^{+}_{j}(u-\beta)}{\epsilon_{j}-u}T_{j}^{-}-2i\sqrt{2\beta}\sum_{j=0}^{L}T_{j}^{z} \displaystyle S^{z}(u) \displaystyle=\sqrt{2\beta}\frac{iB^{x}}{u-\beta}+\zeta\frac{u+\beta}{u-\beta}+2\sum_{j=0}^{L}\frac{(u+\beta)(\epsilon_{j}-\beta)}{(u-\beta)(\epsilon_{j}-u)}T_{j}^{z} where the operators \{T_{k}^{+},T_{k}^{z},T_{k}^{-}\} satisfy the \mathfrak{su}(2) commutation relations \displaystyle[T_{k}^{+},T_{l}^{-}]=2\delta_{k,l}T_{k}^{z},\quad[T_{k}^{z},T_{l}^{\pm}]=\pm\delta_{k,l}T_{k}^{+}, and in terms of the spin operators take the form \displaystyle T_{k}^{z} \displaystyle=-i\frac{\sqrt{2\beta}}{f_{k}^{-}}S_{k}^{x}-\frac{f_{k}^{+}}{f_{k}^{-}}S_{k}^{z},\quad \displaystyle T_{k}^{-} \displaystyle=-iS_{k}^{y}-\frac{f_{k}^{+}}{f_{k}^{-}}S_{k}^{x}+i\frac{\sqrt{2\beta}}{f_{k}^{-}}S_{k}^{z},\quad T_{k}^{+}=iS_{k}^{y}-\frac{f_{k}^{+}}{f_{k}^{-}}S_{k}^{x}+i\frac{\sqrt{2\beta}}{f_{k}^{-}}S_{k}^{z}. The Lax algebra generators 2 2 2 Specifically these are a reparametrisation of those in [ 15 ] in order to regularise the eigenstates in the isotropic \beta\to 0 limit recovering the XX model. See Appendix 7.1 for the details. satisfy the relations \displaystyle{[}S^{\pm}(u),S^{\pm}(v){]} \displaystyle=\pm 2i\sqrt{2\beta}\left(S^{\pm}(v)-S^{\pm}(u)\right), (8) \displaystyle{[}S^{z}(u),S^{\pm}(v){]} \displaystyle=\pm 2\frac{(u+\beta)(v-\beta)}{(u-v)(u-\beta)}\left(S^{\pm}(v)-S^{\pm}(u)\right), \displaystyle{[}S^{z}(u),S^{z}(v){]} \displaystyle=0, \displaystyle{[}S^{+}(u),S^{-}(v){]} \displaystyle=\frac{(u-\beta)(v-\beta)}{u-v}\left(S^{z}(v)-S^{z}(u)\right) (9) \displaystyle\quad\quad\quad-2i\sqrt{2\beta}S^{-}(v)-2i\sqrt{2\beta}S^{+}(u). We also define operators shifted by a constant term, necessary to build the appropriate Bethe states, \displaystyle S^{\pm}_{k}(u)=S^{\pm}(u)\pm(2k-1)i\sqrt{2\beta}I,\quad k\in\mathbb{Z}_{\geq 0} and with these define \displaystyle\mathcal{S}^{\pm}(u_{1},u_{2},\dots,u_{M})=S^{\pm}_{1}(u_{1})S^{\pm}_{2}(u_{2})\cdots S^{\pm}_{M}(u_{M}). (10) Following the notation of [ 15 ] we also introduce the reference states v^{\pm} defined as \displaystyle v^{\pm} \displaystyle=v_{0}^{\pm}\otimes v_{1}^{\pm}\otimes\cdots\otimes v_{L}^{\pm} (11) where the v_{k}^{\pm} are respectively lowest and highest weight states satisfying \displaystyle T_{k}^{z}v_{k}^{+}=-s_{k}v_{k}^{+},\; \displaystyle T_{k}^{-}v_{k}^{+}=0, \displaystyle T_{k}^{z}v_{k}^{-}=s_{k}v_{k}^{-},\; \displaystyle T_{k}^{+}v_{k}^{-}=0 for half non-negative integers s_{k} , i.e. s_{k}\in\{0,1/2,1,3/2,\dots\} . For the spin labelled by the 0 subscript, for which s_{0}=1/2 , we use the following notational convention \ket{+}_{0}=v^{-}_{0},\quad\ket{-}_{0}=v^{+}_{0}. For conciseness we will also use the notation v_{B}^{\pm}=v_{1}^{\pm}\otimes\cdots\otimes v_{L}^{\pm}. 4.2 Generic eigenstates of the charges \mathcal{Q}_{j} To describe the eigenstates of the XY model for generic magnetic field configurations we will use the corresponding results of [ 15 ] for the charges \mathcal{Q}_{j} . As was shown in [ 15 ] , the charges \displaystyle\mathcal{Q}_{j}=\zeta S^{z}_{j}+\frac{B^{x}}{f_{j}^{+}}S^{x}_{j}+\frac{B^{y}}{f_{j}^{-}}S^{y}_{j}+\frac{f_{j}^{-}}{f_{j}^{+}}S^{x}_{j}S^{x}_{j}+\frac{f_{j}^{+}}{f_{j}^{-}}S^{y}_{j}S^{y}_{j} \displaystyle\quad\quad\quad\quad+2\sum_{k\neq j}^{L}\frac{1}{\epsilon_{j}-\epsilon_{k}}(f_{j}^{+}f_{k}^{-}S^{x}_{j}S^{x}_{k}+f_{j}^{-}f_{k}^{+}S^{y}_{j}S^{y}_{k}+f_{k}^{+}f_{k}^{-}S^{z}_{j}S^{z}_{k}), have the eigenstates constructed from raising operators \displaystyle\ket{u_{1},\dots,u_{M}} \displaystyle=\mathcal{S}^{+}(u_{1},\dots,u_{M})v^{-}, (12) with respective eigenvalues \displaystyle\mathfrak{q}_{j} \displaystyle=\frac{f_{j}^{+}}{f_{j}^{-}}s_{j}(s_{j}+1)-\frac{\epsilon_{j}+3\beta}{f_{j}^{+}f_{j}^{-}}s_{j}^{2}-\left(iB^{x}\frac{\sqrt{2\beta}}{f_{j}^{+}f_{j}^{-}}+\zeta\frac{f_{j}^{+}}{f_{j}^{-}}\right)s_{j} \displaystyle\quad+2\frac{f_{j}^{+}}{f_{j}^{-}}\sum_{m=1}^{M}\frac{u_{m}-\beta}{u_{m}-\epsilon_{j}}s_{j}+2\frac{f_{j}^{+}}{f_{j}^{-}}\sum_{k\neq j}^{L}\frac{\epsilon_{k}-\beta}{\epsilon_{j}-\epsilon_{k}}s_{j}s_{k}. (13) The Bethe roots \{{u_{m}}\}_{m=1}^{M} , with M=1+N_{s} where N_{s}=\sum_{j=1}^{L}2s_{j} , are the solutions to the set of Bethe Ansatz equations \displaystyle u_{l}+3\beta-iB^{x}\sqrt{2\beta}-\zeta(u_{l}+\beta) \displaystyle+2\sum_{j=0}^{L}\frac{(\epsilon_{j}-\beta)(u_{l}+\beta)}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{M}\frac{(u_{m}-\beta)(u_{l}+\beta)}{u_{l}-u_{m}} \displaystyle=\frac{1}{4}\left(\left((M+1)\sqrt{2\beta}-(iB^{x}+\zeta\sqrt{2\beta})\right)^{2}-(B^{y})^{2}\right) \displaystyle\quad\times\prod_{j=0}^{L}\left(\frac{\epsilon_{j}-u_{l}}{\epsilon_{j}-\beta}\right)^{2s_{j}}\prod_{m\neq l}^{M}\frac{u_{m}-\beta}{u_{m}-u_{l}}. (14) It was also found in [ 15 ] that some of the eigenstates reduce to \displaystyle\ket{u_{1},\dots,u_{M^{\prime}}} \displaystyle=\mathcal{S}^{+}(u_{1},\dots,u_{M^{\prime}})v^{-}, (15) with the M^{\prime}\in\{0,1,\dots,N_{s}\} Bethe roots satisfying “standard” Bethe Ansatz equations \displaystyle u_{l}+3\beta-iB^{x}\sqrt{2\beta}-\zeta(u_{l}+\beta) \displaystyle+2\sum_{j=0}^{L}\frac{(\epsilon_{j}-\beta)(u_{l}+\beta)}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{M^{\prime}}\frac{(u_{m}-\beta)(u_{l}+\beta)}{u_{l}-u_{m}}=0 (16) when the magnetic field satisfies \displaystyle\frac{-iB^{x}-B^{y}}{\sqrt{2\beta}}-\zeta \displaystyle=N_{s}-2M^{\prime}. (17) In other words, one can see ( 15 ) as a generic Bethe state ( 12 ) in which N_{s}-M^{\prime} of the roots become equal to \beta . Assuming that B^{x},B^{y},\zeta are real valued, as required for the charges to be Hermitian, imposes B^{x}=0 . Before moving onto the eigenstates of the central spin model, we note that for generic model parameters the states can be constructed by either applying the lowering operators to the highest weight state v^{+} , or the raising operators to the lowest weight state v^{-} , with each separately constituting a complete set of eigenstates. Completeness follows from quadratic identities satisfied by the charges \mathcal{Q}_{j} similar to those seen in [ 20 ] , and extended to the eigenvalues. From these the Bethe Ansatz equations are seen to be the consistency equations for the parametrisation in terms of the roots \{u_{m}\} . Completeness is then argued similarly as in [ 9 , 10 ] by assuming regularisability and the one-dimensionality of the simultaneous eigenspace for generic parameters. 4.3 Generic eigenstates of the XY model We now proceed to obtain the eigenstates of the XY model with the limits defined in Sect. 3 . Recall from ( 6 ) that we obtain the Hamiltonian by taking the limit of \mathcal{Q}_{0} as \epsilon_{0}\to\infty after applying a rescaling and a constant shift. Hence the eigenstates are found from the eigenstates of the charge \mathcal{Q}_{0} in the above limit. Due to the possibility that the Bethe roots \{u_{m}\} diverge, we analyse the behaviour as two cases. Finite roots in the limit. Assuming the roots stay finite in the limit, the raising operators become \displaystyle\tilde{S}^{+}(u) \displaystyle=B^{x}+iB^{y}+2i\sqrt{2\beta}S_{0}^{z} \displaystyle\quad+2\sum_{j=1}^{L}\frac{f_{j}^{+}(u-\beta)}{\epsilon_{j}-u}\left(-iS_{j}^{y}-\frac{f_{j}^{+}}{f_{j}^{-}}S_{j}^{x}+i\frac{\sqrt{2\beta}}{f_{j}^{-}}S_{j}^{z}\right) \displaystyle\quad+2\sqrt{2\beta}\sum_{j=1}^{L}\left(-\frac{\sqrt{2\beta}}{f_{j}^{-}}S_{j}^{x}+i\frac{f_{j}^{+}}{f_{j}^{-}}S_{j}^{z}\right) Hence the eigenstates take the form of product states of the central spin and bath spins \displaystyle\ket{u_{1},\dots,u_{M}} \displaystyle=\left.\ket{-}_{0}\otimes\mathcal{S}^{+}(u_{1},u_{2},\dots,u_{M})v_{B}^{-}\right|_{\zeta=+1}, namely dark states. However, we will show in Sec. 5.2.1 that dark states only occur when ( 4 ) holds. In this case the Bethe Ansatz has to be modified to that of Sec. 5 . One of the roots diverges. Assuming that u_{M}\to\infty we have \displaystyle\lim_{\epsilon_{0},u_{M}\to\infty}S^{+}(u_{M}) \displaystyle=B^{x}+iB^{y}-2\kappa S_{0}^{-}+2i\sqrt{2\beta}S_{0}^{z}+2\sum_{j=1}^{L}\left(f_{j}^{-}S_{j}^{x}+if_{j}^{+}S_{j}^{y}\right) where \kappa=\lim_{\epsilon_{0},u_{M}\to\infty}\left[\frac{u_{M}}{\sqrt{\epsilon_{0}}}\right]. By using the expression ( 3 ) for A^{+} and noting that \displaystyle S^{+}_{k+1}(u)|_{\zeta={+1}}=S_{k}^{+}(u)|_{\zeta={-1}}, (18) the eigenstates become \displaystyle\ket{\kappa;u_{1},\dots,u_{M-1}} \displaystyle=\left.2\left[A^{+}-\kappa S_{0}^{+}\right]\ket{-}_{0}\otimes\mathcal{S}^{+}(u_{1},\dots,u_{M-1})v_{B}^{-}\right|_{\zeta=-1} (19) for M-1=N_{s} . Applying the same limits to the Bethe Ansatz equations ( 14 ) it follows that the roots satisfy \displaystyle 2u_{l}+4\beta-iB^{x}\sqrt{2\beta}+2\sum_{j=1}^{L}\frac{(\epsilon_{j}-\beta)(u_{l}+\beta)}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{M-1}\frac{(u_{m}-\beta)(u_{l}+\beta)}{u_{l}-u_{m}} \displaystyle= \displaystyle\frac{1}{4}\left(\left((M+1)\sqrt{2\beta}-iB^{x}\right)^{2}-(B^{y})^{2}\right)\prod_{j=1}^{L}\left(\frac{\epsilon_{j}-u_{l}}{\epsilon_{j}-\beta}\right)^{2s_{j}}\prod_{m\neq l}^{M-1}\frac{u_{m}-\beta}{u_{m}-u_{l}} (20) and \kappa satisfies \displaystyle\kappa^{2}+B^{z}\kappa+iB^{x}\sqrt{2\beta}-2\beta-2\sum_{j=1}^{L}(\epsilon_{j}-\beta)s_{j}+2\sum_{m=1}^{M-1}(u_{m}-\beta) \displaystyle=-\frac{1}{4}\left(\left((M+1)\sqrt{2\beta}-iB^{x}\right)^{2}-(B^{y})^{2}\right)\frac{\prod_{m=1}^{M-1}(u_{m}-\beta)}{\prod_{j=1}^{L}(\epsilon_{j}-\beta)^{2s_{j}}}. The energy is \displaystyle E \displaystyle=-B^{z}/2-\kappa (21) and the eigenvalues \{q_{i}\} of the charges \{Q_{i}\} are \displaystyle q_{i}= \displaystyle-\frac{2\beta}{f_{i}^{+}f_{i}^{-}}s_{i}^{2}+\left(-iB^{x}\frac{\sqrt{2\beta}}{f_{i}^{+}f_{i}^{-}}+2\frac{f_{i}^{+}}{f_{i}^{-}}\right)s_{i} \displaystyle+2\frac{f_{i}^{+}}{f_{i}^{-}}\sum_{m=1}^{M}\frac{u_{m}-\beta}{u_{m}-\epsilon_{i}}s_{i}+2\frac{f_{i}^{+}}{f_{i}^{-}}\sum_{j\neq i}^{L}\frac{\epsilon_{j}-\beta}{\epsilon_{i}-\epsilon_{j}}s_{i}s_{j}. 5 The dark states There is an alternative way of constructing the eigenstates more analogous to that employed for the XX model in [ 21 ] , namely exploiting the connection of the model with a supersymmetric Hamiltonian. This method is more conducive to obtaining the dark states and the conditions admitting their existence and so it will be followed with the construction and characterisation of the dark states. We remark that it is convenient to construct the dark states from both the raising S^{+}(u) and lowering S^{-}(u) operators. For this reason we will give expressions for both concurrently. We use the notation \Theta to denote either an element of the set \{+,-\} or \{+1,-1\} . 5.1 Construction from supersymmetry Define the spin raising and lowering operators S_{0}^{\pm}=S_{0}^{x}\pm iS_{0}^{y} , and supercharges \mathcal{A}^{\pm}=S_{0}^{\mp}A^{\pm} , where A^{\pm} is given by ( 3 ). The supercharges satisfy (\mathcal{A}^{\pm})^{2}=0 and S_{0}^{z}\mathcal{A}^{\pm}=-\mathcal{A}^{\pm}S_{0}^{z} . Recall from [ 20 ] that the Hamiltonian ( 6 ) can be written as \displaystyle H \displaystyle=B^{z}S_{0}^{z}+\mathcal{A}^{+}+\mathcal{A}^{-}. (22) With this one sees that H^{2} is related to a supersymmetric Hamiltonian through \displaystyle H^{2} \displaystyle=(B^{z})^{2}I/4+\mathcal{A}^{+}\mathcal{A}^{-}+\mathcal{A}^{-}\mathcal{A}^{+} (23) \displaystyle=((B^{x})^{2}+(B^{y})^{2}+(B^{z})^{2})I/4+\sum_{j=1}^{L}f_{j}^{+}f_{j}^{-}Q_{j}. Expressed in the eigenbasis of S^{z}_{0} yields \displaystyle H^{2} \displaystyle=(B^{z})^{2}/4+(I/2-S^{z}_{0})A^{+}A^{-}+(I/2+S^{z}_{0})A^{-}A^{+} \displaystyle=\begin{pmatrix}(B^{z})^{2}/4+A^{-}A^{+}&0\cr 0&(B^{z})^{2}/4+A^{+}A^{-}\end{pmatrix} \displaystyle=\frac{|\vec{B}|^{2}}{4}I+\begin{pmatrix}\sum_{j=1}^{L}f_{j}^{+}f_{j}^{-}\mathcal{Q}^{-}_{j}&0\cr 0&\sum_{j=1}^{L}f_{j}^{+}f_{j}^{-}\mathcal{Q}^{+}_{j}\end{pmatrix}, where \mathcal{Q}_{j}^{\pm}=\mathcal{Q}_{j}|_{\zeta=\pm 1} . This shows that H^{2} can be diagonalised with eigenstates of the form \ket{\Psi_{\Theta}}=\ket{\Theta}_{0}\otimes\ket{\psi_{\Theta}}_{B},\quad\Theta\in\{+,-\} where the subscript 0 denotes a central spin state and B a bath-spin state. Observe that the states \ket{\psi_{\Theta}}_{B} are respectively eigenstates of the charges \mathcal{Q}_{j}^{-\Theta} . Due to the supersymmetry, the eigenstates of H^{2} are either singlets or doublets. The singlets \ket{\Psi_{\Theta}} , \mathcal{A}^{+}\ket{\Psi_{\Theta}}=\mathcal{A}^{-}\ket{\Psi_{\Theta}}=0, are also eigenstates of H . Whereas for the doublets \ket{\Psi_{\Theta}} , \displaystyle\mathcal{A}^{+}\ket{\Psi_{+}}=\Lambda\ket{\Psi_{-}},\quad\mathcal{A}^{-}\ket{\Psi_{-}}=\Lambda\ket{\Psi_{+}},\quad\Lambda\in\mathbb{R} we can diagonalise H to obtain the eigenstates \displaystyle\ket{E_{\pm}} \displaystyle=\Lambda\ket{\Psi_{+}}+\left(E_{\pm}-B^{z}/2\right)\ket{\Psi_{-}} (24) \displaystyle=\left(\mathcal{A}^{-}+\left(E_{\pm}-B^{z}/2\right)\right)\ket{\Psi_{-}} \displaystyle=\left(A^{-}+\left(E_{\pm}-B^{z}/2\right)S_{0}^{-}\right)\ket{+}_{0}\otimes\ket{\psi_{-}}_{B} where E_{\pm}=\pm\sqrt{(B^{z})^{2}/4+\Lambda^{2}} . From the above construction it is clear that we can obtain the eigenstates of the XY model from those of \mathcal{Q}_{j}^{-\Theta} or equivalently, due to simultaneous diagonalisability, respectively from A^{-\Theta}A^{\Theta} . 5.2 Dark states The dark states can be constructed from the singlets via the Bethe Ansatz. Doing so one finds these are the states \displaystyle\left.\ket{\Theta}_{0}\otimes\mathcal{S}^{-\Theta}(u_{1},\dots,u_{M})\,v^{\Theta}_{B}\right|_{\zeta=-\Theta} in which the M\in\{0,1,\dots,N_{s}\} roots satisfy the Bethe Ansatz equations \displaystyle 2\beta+2\sum_{j=1}^{L}\frac{(u_{l}+\beta)(\epsilon_{j}-\beta)}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{M}\frac{(u_{l}+\beta)(u_{m}-\beta)}{u_{l}-u_{m}}=0. (25) These are eigenstates of H with energy E=\Theta B^{z}/2 if the magnetic field components satisfy \displaystyle B^{x}=0,\quad B^{y}=(2M-N_{s})\sqrt{2\beta}. (26) Note that here one requires the solutions for both \Theta=+1 and \Theta=-1 to obtain a complete set of eigenstates. Recalling the replacement \zeta\to B^{z}/\sqrt{\epsilon_{0}} made in the derivation of the XY central spin model, we remark that ( 26 ) is consistent with ( 17 ). These states can however be seen as coming from the generic eigenstates of Sec. 4 when the magnetic field configuration takes the form ( 26 ). 3 3 3 See also the appendix 7.3 for numerical results confirming this and a heuristic discussion in 7.2.1 To see this note that due to the eigenstates ( 19 ) being also eigenstates of the charges \{Q_{j}\} the bath states are eigenstates of \mathcal{Q}^{\pm}_{j} . Specifically, the bath state of \displaystyle\left.\ket{-}_{0}\otimes A^{+}\mathcal{S}^{+}(u_{1},\dots,u_{M-1})v_{B}^{-}\right|_{\zeta=-1} is an eigenstate of \mathcal{Q}^{+}_{j} and so is equivalent to an eigenstate of the form ( 15 ) and so has Bethe roots being solutions of Bethe Ansatz equations of the form in ( 25 ). There is however a simpler way of constructing the dark states which circumvents the Bethe Ansatz. It also allows a clear characterisation of which magnetic field configurations admit dark states. 5.2.1 Characterisation and construction. We first show that the eigenstates given in the previous section, namely the singlet states, are the only dark states. As seen in the construction in Sec. 5.1 , states constructed from singlets are evidently dark states. Examining then the eigenstates obtained from the doublets, namely \ket{E}=\Lambda\ket{-}_{0}\otimes\ket{\psi_{-}}_{B}+\left(E+B^{z}/2\right)\ket{+}_{0}\otimes\ket{\psi_{+}}_{B}, we see that for these to be dark states requires (E+B^{z}/2)\ket{\psi_{+}}_{B}\propto\ket{\psi_{-}}_{B}. If we assume that neither of these is zero this implies that \ket{\psi_{\Theta}}_{B},\,\Theta\in\{+,-\} are eigenstates of both charges \mathcal{Q}_{j}^{\pm} since \ket{\psi_{\Theta}}_{B} are respectively eigenstates of the charges \mathcal{Q}_{j}^{-\Theta} due to A^{+}\mathcal{Q}_{j}^{-}=\mathcal{Q}_{j}^{+}A^{+}. Hence they are also eigenstates of \mathcal{Q}_{j}^{+}-\mathcal{Q}_{j}^{-}=2S_{j}^{z} for all j=1,\dots,L . This determines \ket{\psi_{\Theta}}_{B},\,\Theta\in\{+,-\} and would lead to a contradiction if we took e.g. B^{x}\gg 1 since this holds for all B^{x} . This also means that dark states can only occur if \Lambda\ket{\psi_{-}}_{B}=0 or (E+B^{z}/2)\ket{\psi_{+}}_{B}=0 in which case the central spin state is respectively \ket{\Theta}_{0} . Furthermore, by the supersymmetry this means that \ket{\Theta}_{0}\otimes\ket{\psi_{\Theta}}_{B} is a singlet (as these were originally doublets). Next we construct the dark states. From the above argument these are the singlet states \ket{\Phi_{\Theta}}=\ket{\Theta}_{0}\otimes\ket{\phi_{\Theta}}_{B},\quad\Theta\in\{+,-\} with A^{\Theta}\ket{\phi_{\Theta}}=0. Set \mathfrak{f}_{j}^{\pm}:=(f_{j}^{-}\pm f_{j}^{+})/2 , so that we can write the local operators of A^{\pm} in terms of S_{j}^{\pm} as \displaystyle f^{-}_{j}S_{j}^{x}\pm if_{j}^{+}S_{j}^{y} \displaystyle=\mathfrak{f}_{j}^{\pm}S_{j}^{+}+\mathfrak{f}_{j}^{\mp}S_{j}^{-}. We define operators \displaystyle E_{j}^{\pm} \displaystyle=-\frac{\sqrt{\mathfrak{f}_{j}^{\pm}}}{2\sqrt{\mathfrak{f}_{j}^{\mp}}}S_{j}^{+}+S_{j}^{z}+\frac{\sqrt{\mathfrak{f}_{j}^{\mp}}}{2\sqrt{\mathfrak{f}_{j}^{\pm}}}S_{j}^{-},\;H_{j}^{\pm}=\frac{\sqrt{\mathfrak{f}_{j}^{\pm}}}{\sqrt{\mathfrak{f}_{j}^{\mp}}}S_{j}^{+}+\frac{\sqrt{\mathfrak{f}_{j}^{\mp}}}{\sqrt{\mathfrak{f}_{j}^{\pm}}}S_{j}^{-}, \displaystyle F_{j}^{\pm} \displaystyle=\frac{\sqrt{\mathfrak{f}_{j}^{\pm}}}{2\sqrt{\mathfrak{f}_{j}^{\mp}}}S_{j}^{+}+S_{j}^{z}-\frac{\sqrt{\mathfrak{f}_{j}^{\mp}}}{2\sqrt{\mathfrak{f}_{j}^{\pm}}}S_{j}^{-}, that satisfy the \mathfrak{sl}(2) relations \displaystyle[H_{j}^{\pm},E_{j}^{\pm}]=2E_{j}^{\pm},\quad[E_{j}^{\pm},F_{j}^{\pm}]=H_{j}^{\pm},\quad[H_{j}^{\pm},F_{j}^{\pm}]=-2F_{j}^{\pm}. Noting that \displaystyle A^{\pm} \displaystyle=(B^{x}\pm iB^{y})I/2+\sum_{j=1}^{L}\sqrt{\mathfrak{f}^{+}_{j}\mathfrak{f}^{-}_{j}}\,H_{j}^{\pm} \displaystyle=(B^{x}\pm iB^{y})I/2+i\sum_{j=1}^{L}\sqrt{\beta/2}\,H_{j}^{\pm} we can respectively diagonalise A^{\pm} with the respective eigenvectors \displaystyle\ket{n_{1},\dots,n_{L}}^{+}=\prod_{j=1}^{L}(E_{j}^{+})^{n_{j}}\chi^{+}, \displaystyle\ket{n_{1},\dots,n_{L}}^{-}=\prod_{j=1}^{L}(F_{j}^{-})^{n_{j}}\chi^{-}, where the respectively lowest and highest weight vector is defined as \displaystyle\chi^{\pm}=\chi_{1}^{\pm}\otimes\cdots\chi_{L}^{\pm}, \displaystyle F_{j}^{+}\chi_{j}^{+}=0,\;H_{j}^{+}\chi_{j}^{+}=-2s_{j}\chi_{j}^{+};\;-2s_{j}\leq n_{j}\leq 2s_{j}, \displaystyle E_{j}^{-}\chi_{j}^{-}=0,\;H_{j}^{-}\chi_{j}^{-}=2s_{j}\chi_{j}^{-};\;-2s_{j}\leq n_{j}\leq 2s_{j}. It is found that A^{\pm} has respective eigenvalues \displaystyle(B^{x}\pm iB^{y})/2\pm\left(-i\sqrt{2\beta}\sum_{j=1}^{L}n_{j}+\frac{N_{s}}{2}i\sqrt{2\beta}\right),\quad-2s_{j}\leq n_{j}\leq 2s_{j}. We see that these are zero, i.e. the corresponding states \ket{\Phi_{\Theta}}=\ket{\Theta}_{0}\otimes\ket{n_{1},\dots,n_{L}}^{\Theta} are dark states, when \displaystyle\frac{-iB^{x}+\Theta B^{y}}{\sqrt{2\beta}} \displaystyle=\Theta\left(2\sum_{j=1}^{L}n_{j}-N_{s}\right)\in\{-N_{s}/2,\dots,N_{s}/2\} holds. Due to the Hamiltonian being Hermitian this requires B^{x}=0 and we obtain ( 26 ) after making the identification M=\sum_{j=1}^{L}n_{j} . We briefly note that for the XX model case the dark states are zero eigenvalue eigenstates of A^{\pm}=(B^{x}\pm iB^{y})I/2+\sum_{j=1}^{L}\sqrt{\epsilon_{j}}S_{j}^{\pm}. Hence, it follows that A^{\pm} is strictly upper or lower triangular if and only if B^{x}=B^{y}=0 , and so dark states can only occur if B^{x}=B^{y}=0 , namely that treated in [ 21 ] . 5.2.2 Quadratic relations. We mention that some of the roots becoming equal to \beta is consistent with the quadratic relations of the charges \mathcal{Q}_{j} as seen in [ 20 ] . Restricting to spin- 1/2 for all of the spins, the charges of the XYZ Richardson-Gaudin model \mathcal{Q}_{j} 4 4 4 With a constant shift introduced to make the relations cleaner. \displaystyle\mathcal{Q}_{j} \displaystyle=\zeta S_{j}^{z}+\frac{B^{x}}{f_{j}^{+}}S_{j}^{x}+\frac{B^{y}}{f_{j}^{-}}S_{j}^{y} \displaystyle\quad+2\sum_{k\neq j}^{L}\frac{1}{\epsilon_{j}-\epsilon_{k}}\left(f_{j}^{+}f_{k}^{-}S_{j}^{x}S_{k}^{x}+f_{j}^{-}f_{k}^{+}S_{j}^{y}S_{k}^{y}+f_{k}^{+}f_{k}^{-}\left(S_{j}^{z}S_{k}^{z}-\frac{1}{4}\right)\right) satisfy the following quadratic relations \displaystyle\mathcal{Q}_{j}^{2}= \displaystyle\frac{1}{4}\left(\zeta^{2}+\frac{(B^{x})^{2}}{(f^{+}_{j})^{2}}+\frac{(B^{y})^{2}}{(f^{-}_{j})^{2}}\right)-\sum_{k\neq j}^{L}\frac{f^{+}_{j}f^{-}_{j}}{\epsilon_{j}-\epsilon_{k}}\left(\mathcal{Q}_{j}-\mathcal{Q}_{k}\right) \displaystyle\quad+\frac{1}{4}\sum_{k\neq j}^{L}\frac{(f^{+}_{j}f^{-}_{k}-f^{-}_{j}f^{+}_{k})^{2}}{(\epsilon_{j}-\epsilon_{k})^{2}} (27) with these also holding true for their eigenvalues \displaystyle\mathfrak{q}_{j}^{2}= \displaystyle\frac{1}{4}\left(\zeta^{2}+\frac{(B^{x})^{2}}{(f^{+}_{j})^{2}}+\frac{(B^{y})^{2}}{(f^{-}_{j})^{2}}\right)-\sum_{k\neq j}^{L}\frac{f^{+}_{j}f^{-}_{j}}{\epsilon_{j}-\epsilon_{k}}\left(\mathfrak{q}_{j}-\mathfrak{q}_{k}\right) \displaystyle\quad+\frac{1}{4}\sum_{k\neq j}^{L}\frac{(f^{+}_{j}f^{-}_{k}-f^{-}_{j}f^{+}_{k})^{2}}{(\epsilon_{j}-\epsilon_{k})^{2}}. With a slightly more general ansatz than that of [ 20 ] , for a parametrisation of the eigenvalues in terms of the parameters \{u_{m}\}_{m=1}^{M} \displaystyle\mathfrak{q}_{j} \displaystyle=C\frac{f_{j}^{+}}{f_{j}^{-}}+D\frac{f_{j}^{-}}{f_{j}^{+}}-\sum_{m=1}^{M}\frac{f_{j}^{+}f_{j}^{-}}{\epsilon_{j}-u_{m}}+\frac{1}{2}\sum_{k\neq j}^{L}\frac{f_{j}^{+}f_{j}^{-}-f_{k}^{+}f_{k}^{-}}{\epsilon_{j}-\epsilon_{k}} (28) \displaystyle=\alpha_{j}(\epsilon_{j})\frac{\epsilon_{j}}{f_{j}^{+}f_{j}^{-}}-\sum_{m=1}^{M}\frac{f_{j}^{+}f_{j}^{-}}{\epsilon_{j}-u_{m}}+\frac{1}{2}\sum_{k\neq j}^{L}\frac{f_{j}^{+}f_{j}^{-}-f_{k}^{+}f_{k}^{-}}{\epsilon_{j}-\epsilon_{k}} where \displaystyle\alpha(\epsilon_{j}) \displaystyle=\alpha_{0}+\frac{\alpha_{1}}{\epsilon_{j}}=(C+D)+(C-D)\frac{\beta}{\epsilon_{j}}, the quadratic relations give consistency equations for the parameters \{u_{m}\}_{m=1}^{M} . These are the Bethe Ansatz equations ( 14 ) for \displaystyle C=\frac{1}{2}\left(2M-L+1-\frac{iB^{x}}{\sqrt{2\beta}}-\zeta\right),\quad D=\frac{iB^{x}}{2\sqrt{2\beta}}. From this we see that when some of the M=L roots become equal to \beta the eigenvalues and Bethe Ansatz equations reduce to ( 16 ) when the magnetic field satisfies ( 17 ). 6 Conclusion We have derived the XY central spin model Hamiltonian H_{XY} ( 6 ) and charges given in [ 20 ] , as a limit of the XYZ Richardson-Gaudin model charges with self-interaction \mathcal{Q}_{i} obtained by Skrypnyk [ 15 ] . This allowed the diagonalisation results found in [ 15 ] for the charges \mathcal{Q}_{i} via a modified Algebraic Bethe Ansatz to be used to diagonalise H_{XY} . However, in order for these results to connect with those found for the XX model we regularised the charges \mathcal{Q}_{i} via a reparametrisation. Lastly, we showed that the dark states of the XX model without an in-plane magnetic field [ 5 , 21 ] can be seen as a special case of dark states of the XY model for special configurations of the magnetic field. These results open the way for finding form factors and correlation functions for the XY central spin model as well as probing the dynamics for large systems. The characterisation of the magnetic field configurations supporting dark states and their emergence via special limits of the parameter \kappa can provide experimental criteria for preparing these states with high-fidelity and determining how far a bright state is from being a product state. Another extension of this work would be to the case of a spin-1 central spin. Here it has been shown that the XX central spin model is integrable for an out-of-plane magnetic field [ 17 ] and more recently for an arbitrarily oriented magnetic field [ 4 ] . Integrability for the XY model can also be shown through a minor modification of the charges found in [ 4 ] . In the latter work it was numerically demonstrated that dark states can emerge at asymptotically large coupling, just as seen in the central spin-1/2 model for an arbitrarily oriented magnetic field [ 5 ] . This raises the possibility of the existence of these states for certain special magnetic field configurations of the XX and XY spin-1 central spin models. However, determining whether these dark states actually exist is complicated by the fact that the relationship between the model and the known class of Richardson-Gaudin models is not yet understood. 7 Appendix 7.1 Reparametrisation. In the derivation of the expressions for the eigenstates in Sec. 4 from those of [ 15 ] we employed a reparametrisation of the model parameters. Write the charges \{\mathcal{Q}_{j}\} ( 5 ) in terms of the parameters \{\phi_{j}\} and magnetic field components \{\tilde{B^{x}},\tilde{\zeta},\tilde{B^{z}}\} \displaystyle\tilde{\mathcal{Q}_{j}} \displaystyle=\tilde{B^{z}}S_{j}^{z}+\frac{\tilde{B^{x}}}{h_{j}^{+}}S_{j}^{x}+\frac{\tilde{\zeta}}{h_{j}^{-}}S_{j}^{y}+\frac{h_{j}^{-}}{h_{j}^{+}}S_{j}^{x}S_{j}^{x}+\frac{h_{j}^{+}}{h_{j}^{-}}S_{j}^{y}S_{j}^{y} \displaystyle\quad+2\sum_{k\neq j}^{L}\frac{1}{\phi_{j}-\phi_{k}}(h_{j}^{+}h_{k}^{-}S_{j}^{x}S_{k}^{x}+h_{j}^{-}h_{k}^{+}S_{j}^{y}S_{k}^{y}+h_{k}^{+}h_{k}^{-}S_{j}^{z}S_{k}^{z}), where h^{\pm}_{j}=\sqrt{\phi_{j}\pm\beta} . Then the reparametrisation that was applied is the following \displaystyle\epsilon_{j}=\beta+\frac{4\beta^{2}}{\phi_{j}-\beta}, \displaystyle\tilde{B^{x}}=-B^{x},\quad\tilde{\zeta}=-\zeta\sqrt{2\beta},\quad\tilde{B^{z}}=B^{y}/\sqrt{2\beta}. (29) The spectral parameter u in the argument of the Lax generators was similarly reparametrised as \quad u=\beta+\frac{4\beta^{2}}{w-\beta}, where the Lax generators in [ 15 ] correspond to those of Sec. 8 via \displaystyle S^{z}(w)=2\beta S^{-}(u(w)),\;S^{+}(w)=\frac{1}{\sqrt{2\beta}}S^{+}(u(w)),\;S^{-}(w)=\frac{1}{\sqrt{2\beta}}S^{-}(u(w)), and these are identified with \{A(w),B(w),C(w),D(w)\} as \displaystyle S^{z}(w)=\frac{A(w)-D(w)}{2},\quad S^{+}(w)=C(w),\quad S^{-}(w)=D(w). The reparametrisation is required to regularise the eigenstates in the isotropic limit \beta\to 0 in order to recover the eigenstates of the XX model as seen in [ 21 ] . It is enabled by the observation that under this reparametrisation the charges ( 5 ) take the form (after making the rotation S^{y}\to S^{z},\;S^{z}\to-S^{y} and adding \mathbf{S}_{i}^{2} ) \displaystyle\tilde{\mathcal{Q}}_{i} \displaystyle=-\frac{f_{i}^{-}}{\sqrt{2\beta}}\bigg\{-\frac{f_{i}^{+}}{f_{i}^{-}}s_{i}(s_{i}+1)-\frac{\tilde{\zeta}}{\sqrt{2\beta}}S^{z}_{i}-\frac{\tilde{B^{x}}}{f^{+}_{i}}S^{x}_{i}+\tilde{B^{z}}\frac{\sqrt{2\beta}}{f^{-}_{i}}S^{y}_{i}+\frac{f_{i}^{-}}{f_{i}^{+}}S_{i}^{x}S_{i}^{x} \displaystyle\quad+\frac{f_{i}^{+}}{f_{i}^{-}}S_{i}^{y}S_{i}^{y}+2\sum_{j\neq i}^{L}\frac{1}{\epsilon_{i}-\epsilon_{j}}\left(f^{+}_{i}f^{-}_{j}S^{x}_{i}S^{x}_{j}+f^{-}_{i}f^{+}_{j}S^{y}_{i}S^{y}_{j}+f^{+}_{j}f^{-}_{j}S^{z}_{i}S^{z}_{j}\right)\bigg\} \displaystyle=-\frac{f_{i}^{-}}{\sqrt{2\beta}}\left(\mathcal{Q}_{i}-\frac{f_{i}^{+}}{f_{i}^{-}}s_{i}(s_{i}+1)\right) (30) where \{\mathcal{Q}_{i}\} are simply the charges ( 5 ). 7.2 Eigenstates of the XX central spin model Due to the reparametrised expressions for the eigenstates and eigenvalues of the XYZ Richardson-Gaudin model charges all being non-singular in the limit \beta\to 0 , we can rederive the diagonalisation results for the XX model. This will also give us an idea of how the dark states emerge in the limit of the special magnetic field configurations ( 26 ) of Sect. 5 . Taking the limit as \beta\to 0 , the eigenstates of the charges \displaystyle\mathcal{Q}_{i}=\zeta S^{z}_{i}+\frac{B^{x}}{\sqrt{\epsilon_{i}}}S^{x}_{i}+\frac{B^{y}}{\sqrt{\epsilon_{i}}}S^{y}_{i}+S_{i}^{x}S_{i}^{x}+S_{i}^{y}S_{i}^{y} \displaystyle\qquad+2\sum_{j\neq i}^{L}\frac{1}{\epsilon_{i}-\epsilon_{j}}\left(\sqrt{\epsilon_{i}}\sqrt{\epsilon_{j}}S^{x}_{i}S^{x}_{j}+\sqrt{\epsilon_{i}}\sqrt{\epsilon_{j}}S^{y}_{i}S^{y}_{j}+\epsilon_{j}S^{z}_{i}S^{z}_{j}\right) become \displaystyle\ket{u_{1},\dots,u_{M}}_{-\Theta} \displaystyle=S^{-\Theta}(u_{1})S^{-\Theta}(u_{2})\cdots S^{-\Theta}(u_{M})v^{\Theta}, with \Theta\in\{+,-\} . Here \displaystyle S^{-\Theta}(u) \displaystyle=B^{x}-\Theta iB^{y}+2\sum_{j=1}^{L}\frac{u\sqrt{\epsilon_{j}}}{u-\epsilon_{j}}S_{j}^{-\Theta} and the reference states v^{\Theta}=v_{0}^{\Theta}\otimes v_{1}^{\Theta}\otimes\cdots\otimes v_{L}^{\Theta} now satisfy \displaystyle S_{k}^{z}v_{k}^{+}=s_{k}v_{k},\;S_{k}^{+}v_{k}^{+}=0;\quad S_{k}^{z}v_{k}^{-}=-s_{k}v_{k},\;S_{k}^{-}v_{k}^{-}=0, or in other words v^{\Theta}=\ket{\Theta s_{1},\dots,\Theta s_{L}}. The Bethe roots in turn now satisfy \displaystyle 1+\Theta\zeta+ \displaystyle 2\sum_{j=1}^{L}\frac{\epsilon_{j}}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{M}\frac{u_{m}}{u_{l}-u_{m}} \displaystyle=-\frac{(B^{x})^{2}+(B^{y})^{2}}{4}\frac{\prod_{j=1}^{L}\left(u_{l}^{-1}-\epsilon_{j}^{-1}\right)^{2s_{j}}}{\prod_{m\neq l}^{M}(u_{l}^{-1}-u_{m}^{-1})} (31) with M=\sum_{j=1}^{L}2s_{j} , while the eigenvalues \{\mathfrak{q}^{-\Theta}_{j}\} of the charges \{\mathcal{Q}_{j}\} are \displaystyle\mathfrak{q}^{-\Theta}_{i} \displaystyle=\left(1+\Theta\zeta\right)s_{i}+2\sum_{m=1}^{M}\frac{u_{m}}{u_{m}-\epsilon_{i}}s_{i}+2\sum_{j\neq i}^{L}\frac{\epsilon_{j}}{\epsilon_{i}-\epsilon_{j}}s_{i}s_{j}. To obtain the eigenstates of the XX central spin model we can simply use the results for the XY model in the limit \beta\to 0 . For a magnetic field with a non-zero in-plane component the eigenstates are entangled states of the central spin and bath, the bright states of [ 21 ] , taking the form \displaystyle\ket{\kappa;u_{1},\dots,u_{M-1}}_{\Theta}= \displaystyle\left[B^{x}-\Theta iB^{y}+2\sum_{j=1}^{L}\sqrt{\epsilon_{j}}S_{j}^{-\Theta}-2\kappa S_{0}^{-\Theta}\right]\ket{\Theta}_{0} \displaystyle\otimes\left.\mathcal{S}^{-\Theta}(u_{1},u_{2},\dots,u_{M-1})\ket{\Theta s_{1},\dots,\Theta s_{L}}\right|_{\zeta=\Theta} with the M-1=\sum_{j=1}^{L}2s_{j} roots satisfying the Bethe Ansatz equations \displaystyle 2+2\sum_{j=1}^{L}\frac{\epsilon_{j}}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{M-1}\frac{u_{m}}{u_{l}-u_{m}} \displaystyle=-\frac{(B^{x})^{2}+(B^{y})^{2}}{4}\frac{\prod_{j=1}^{L}(u_{l}^{-1}-\epsilon_{j}^{-1})^{2s_{j}}}{\prod_{m\neq l}^{M-1}(u_{l}^{-1}-u_{m}^{-1})} and the equation for \kappa coming from the Bethe Ansatz equation for u_{M} , \displaystyle\kappa^{2}-\Theta B^{z}\kappa-2\sum_{j=1}^{L}\epsilon_{j}s_{j}+2\sum_{m=1}^{M-1}u_{m} \displaystyle=-\frac{(B^{x})^{2}+(B^{y})^{2}}{4}\frac{\prod_{m=1}^{M-1}u_{m}}{\prod_{j=1}^{L}(\epsilon_{j})^{2s_{j}}}. (32) The energy is \displaystyle E \displaystyle=\Theta B^{z}/2-\kappa while the eigenvalues \{q_{i}\} of the charges \{Q_{i}\} are \displaystyle q_{i} \displaystyle=2s_{i}-2\sum_{m=1}^{M-1}\frac{u_{m}}{\epsilon_{i}-u_{m}}s_{i}+2\sum_{j\neq i}^{L}\frac{\epsilon_{j}}{\epsilon_{i}-\epsilon_{j}}s_{i}s_{j}. Observe here that if \kappa is a solution to the first equation of (\ref{eq: kappa XX}) then \kappa^{\prime}=\kappa-B^{z} is a solution to the second (respectively, if \kappa is a solution to the second equation of (\ref{eq: kappa XX}) then \kappa^{\prime}=\kappa+B^{z} is a solution to the first equation), due to the Bethe equations for the roots being the same for both. These have identical energy and charge eigenvalues. Therefore, based on the argument of the correspondence between eigenstates constructed using raising or lowering operators, they are proportional to the same eigenstate. We note these expressions match those of [ 3 ] and [ 21 ] , where in the former s_{j}=1/2 and in the latter B^{x}=B^{y}=0 . 7.2.1 Recovering the dark states for B^{x}=B^{y}=0 . As argued at the end of Sect. 5.2.1 , in the case of a non-zero in-plane magnetic field, (B^{x})^{2}+(B^{y})^{2}>0 , there are no dark states. We find numerically (see also Sect. 7.3 ) that the dark states of the XX model with no in-plane field are recovered in the limit as (B^{x})^{2}+(B^{y})^{2}\to 0 via \kappa\to\Theta B^{z} or \kappa\to 0 for some of the eigenstates. For simplicity we will only discuss the case where for all the bath spins s_{j}=1/2 . In the case where \kappa\to\Theta B^{z} , the states become \ket{\psi_{\Theta}}=2\left[\sum_{j=1}^{L}\sqrt{\epsilon_{j}}S_{j}^{-\Theta}+B^{z}S_{0}^{-\Theta}\right]\ket{\Theta}_{0}\otimes\left.\mathcal{S}^{-\Theta}(u_{1},u_{2},\dots,u_{\mathfrak{M}-1})\ket{\Theta s_{1},\dots,\Theta s_{L}}\right|_{\zeta=\Theta} where, from ( 31 ), the roots satisfy \displaystyle\underbrace{1+\Theta\left(\Theta\right)}_{=2}+2\sum_{j=1}^{L}\frac{\epsilon_{j}}{u_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{\mathfrak{M}-1}\frac{u_{m}}{u_{l}-u_{m}}=0, with \mathfrak{M}\leq M due to some of the roots becoming zero and their corresponding raising and lowering operators being asymptotically S^{-\Theta}(u)\propto c(B^{x},B^{y})I . This follows by noting that the Bethe states will become eigenstates of the total S^{z} operator, as the raising and lowering operators are proportional to products of total spin raising or lowering operators in the zero in-plane field limit, and since the XX model in this limit has a global U(1) symmetry. Using the correspondence coming from the symmetry of the charges, \{S^{z}\to-S^{z},\;i\to-i,\;\zeta\to-\zeta\} , this can also be written as \ket{\psi_{\Theta}}\propto 2\left[\sum_{j=1}^{L}\sqrt{\epsilon_{j}}S_{j}^{-\Theta}+B^{z}S_{0}^{-\Theta}\right]\ket{\Theta}_{0}\otimes\left.\mathcal{S}^{\Theta}(\tilde{u}_{1},\tilde{u}_{2},\dots,\tilde{u}_{\mathcal{M}-1})\ket{-\Theta s_{1},\dots,-\Theta s_{L}}\right|_{\zeta=-\Theta} with the roots \{\tilde{u}_{m}\}_{m=1}^{\mathcal{M}-1} satisfying \displaystyle\underbrace{1+\Theta\left(-\Theta\right)}_{=0}+2\sum_{j=1}^{L}\frac{\epsilon_{j}}{\tilde{u}_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{\mathcal{M}-1}\frac{\tilde{u}_{m}}{\tilde{u}_{l}-\tilde{u}_{m}}=0. Note that the bath component of the state now satisfies (see for example [ 21 ] ) \displaystyle\underbrace{\left(\sum_{j=1}^{L}\sqrt{\epsilon_{j}}S_{j}^{-\Theta}\right)}_{A^{-\Theta}}\left.\mathcal{S}^{\Theta}(\tilde{u}_{1},\tilde{u}_{2},\dots,\tilde{u}_{\mathcal{M}-1})\ket{-\Theta s_{1},\dots,-\Theta s_{L}}\right|_{\zeta=-\Theta}=0, giving the dark state \displaystyle\ket{\psi_{\Theta}} \displaystyle\propto\ket{-\Theta}_{0}\otimes\left.\mathcal{S}^{\Theta}(\tilde{u}_{1},\tilde{u}_{2},\dots,\tilde{u}_{\mathcal{M}-1})\ket{-\Theta s_{1},\dots,-\Theta s_{L}}\right|_{\zeta=-\Theta}. As explained in Sect. 7.2 , the states with \kappa^{\prime}=\kappa+B^{z} constructed with lowering operators correspond to states with \kappa constructed with raising operators (and vice versa for \kappa^{\prime}=\kappa-B^{z} and \kappa ). So the states constructed with raising operators for which \kappa\to 0 are the same dark states as those constructed with lowering operators with the behaviour \kappa\to+B^{z} (and vice versa for \kappa\to 0 and \kappa\to-B^{z} ) that we just found. However, for completeness we will sketch here the behaviour in this limit. For conciseness let \rho_{-\Theta}=(B^{x}-\Theta iB^{y})/2 so that the state asymptotically becomes \displaystyle\ket{\psi_{\Theta}}\sim\ket{\Theta}_{0}\otimes \displaystyle\left(\rho_{-\Theta}+\sum_{j=1}^{L}\sqrt{\epsilon_{j}}S_{j}^{-\Theta}\right)\left(\rho_{-\Theta}+\sum_{j=1}^{L}\frac{u_{1}\sqrt{\epsilon_{j}}}{u_{1}-\epsilon_{j}}S_{j}^{\Theta}\right)\times \displaystyle\cdots\left(\rho_{-\Theta}+\sum_{j=1}^{L}\frac{u_{\mathfrak{M}-1}\sqrt{\epsilon_{j}}}{u_{\mathfrak{M}-1}-\epsilon_{j}}S_{j}^{\Theta}\right)\ket{\Theta s_{1},\dots,\Theta s_{L}} where again some of the roots \{u_{i}\}_{i=1}^{M-1} go to zero. Expanding this out, the state has the form (for s_{i}=1/2 ) \displaystyle\ket{\psi_{\Theta}}\sim\ket{\Theta}_{0}\otimes \displaystyle\left(\rho_{-\Theta}^{\mathfrak{M}}+\rho_{-\Theta}^{\mathfrak{M}-1}\sum_{i_{1}}c_{i_{1}}S_{i_{1}}^{\mp}+\cdots+\rho_{-\Theta}^{\mathfrak{M}-K}\sum_{i_{1}<\cdots\mathcal{M} are zero while the leading order coefficients are c_{i_{1}\cdots i_{\mathcal{M}}}=\sum_{\sigma\in S_{\mathcal{M}}}\left(\frac{\tilde{u}_{1}\sqrt{\epsilon_{i_{\sigma(1)}}}}{\tilde{u}_{1}-\epsilon_{i_{\sigma(1)}}}\cdots\frac{\tilde{u}_{\mathcal{M}}\sqrt{\epsilon_{i_{\sigma(\mathcal{M})}}}}{\tilde{u}_{\mathcal{M}}-\epsilon_{i_{\sigma(\mathcal{M})}}}\right). The state in the limit then becomes equal to the dark state \displaystyle\ket{\psi_{\Theta}}\propto\ket{\Theta}_{0}\otimes\left(\sum_{j=1}^{L}\frac{\tilde{u}_{1}\sqrt{\epsilon_{j}}}{\tilde{u}_{1}-\epsilon_{j}}S_{j}^{\Theta}\right)\cdots\left(\sum_{j=1}^{L}\frac{\tilde{u}_{\mathcal{M}}\sqrt{\epsilon_{j}}}{\tilde{u}_{\mathcal{M}}-\epsilon_{j}}S_{j}^{\Theta}\right)\ket{\Theta s_{1},\dots,\Theta s_{L}} with \mathcal{M}\leq\mathfrak{M} and where the roots satisfy 2\sum_{j=1}^{L}\frac{\epsilon_{j}}{\tilde{u}_{l}-\epsilon_{j}}s_{j}-2\sum_{m\neq l}^{\mathcal{M}}\frac{\tilde{u}_{m}}{\tilde{u}_{l}-\tilde{u}_{m}}=0. 7.3 Emergence of dark states In the following we show numerical examples of dark states emerging due to \kappa\to 0,\Theta B^{z} when condition ( 26 ) is met. We set the model parameters as L=3,\beta=0.1,B^{z}=1 , and 5 5 5 This \underline{\epsilon} was generated in Mathematica with SeedRandom 1324 using RandomReal in the range 0.4 to 1.3. \displaystyle\underline{\epsilon}=\{0.5869853530170386,1.270553831408777,1.2426482643150194\}, and only look at the case where all spins are s_{i}=1/2 . Then for different B^{y} we change B^{x}\to 0 so that we approach conditions where dark states emerge. As we use states constructed from v^{+} with the lowering operators S^{-} , we should observe for dark states that \kappa\to 0,+B^{z} . We label states by the eigenvalue q_{1} of Q_{1} . 7.3.1 Dark states for B^{y}=L\sqrt{2\beta} . In this case there are only two dark states in the limit, both with the same eigenvalues for the charges but with different energies. B^{x}/2 \kappa_{1} \kappa_{2} q_{1} 10^{0} -5.72938\times 10^{-2} 1+5.72938\times 10^{-2} -2.57221 10^{-1} -6.70952\times 10^{-5} 1+6.70952\times 10^{-5} -2.14310 10^{-2} -6.33204\times 10^{-7} 1+6.33204\times 10^{-7} -2.13838 10^{-3} -6.32830\times 10^{-9} 1+6.32830\times 10^{-9} -2.13833 10^{-4} -6.32829\times 10^{-11} 1+6.32829\times 10^{-11} -2.13833 7.3.2 Dark states for B^{y}=\left(L-2\right)\sqrt{2\beta} . In this case there are six dark states in the limit. B^{x}/2 \kappa_{1} \kappa_{2} q_{1} 10^{-1} -1.03242\times 10^{-5} 1+1.03242\times 10^{-5} -1.90365 10^{-2} -9.33203\times 10^{-8} 1+9.33203\times 10^{-8} -1.89852 10^{-3} -9.32225\times 10^{-10} 1+9.32225\times 10^{-10} -1.89846 10^{-4} -9.30916\times 10^{-12} 1+9.30916\times 10^{-12} -1.89846 B^{x}/2 \kappa_{1} \kappa_{2} q_{1} 10^{-1} -2.97399\times 10^{-3} 1+2.97399\times 10^{-5} -9.88279\times 10^{-2} 10^{-2} -2.91182\times 10^{-5} 1+2.91182\times 10^{-5} -8.68202\times 10^{-2} 10^{-3} -2.91118\times 10^{-7} 1+2.91118\times 10^{-7} -8.66989\times 10^{-2} 10^{-4} -2.91117\times 10^{-9} 1+2.91117\times 10^{-9} -8.66977\times 10^{-2} B^{x}/2 \kappa_{1} \kappa_{2} q_{1} 10^{-1} -1.07577\times 10^{-3} 1+1.07577\times 10^{-3} 3.78262 10^{-2} -1.03663\times 10^{-5} 1+1.03663\times 10^{-5} 3.77776 10^{-3} -1.03624\times 10^{-7} 1+1.03624\times 10^{-7} 3.77771 10^{-4} -1.03623\times 10^{9} 1+1.03623\times 10^{-9} 3.77771