Chiral Integrable Boundary States in the SU(4) Alternating Spin Chain Yang Liu, Jun-Bao Wu Abstract Previously identified integrable boundary states in ABJM theory are exclusively achiral. This paper presents the first chiral integrable boundary states in the S ​ U ​ ( 4 ) SU(4) alternating spin chain from the planar two-loop dilatation operator in the scalar sector. Utilizing a sufficient condition for the untwisted integrable condition, we identify specific two-site and four-site basis boundary states as chiral integrable states. Numerical evidence indicates that other basis states are unlikely to be chiral integrable. Furthermore, we compute the overlaps between these chiral integrable basis states and on-shell Bethe eigenstates. 1 Introduction Integrable boundary states, more precisely understood as integrable initial states in their physical interpretation, were first studied in the context of two-dimensional integrable quantum field theories with boundaries [ 1 ] . These states are characterized by the fact that they are annihilated by all odd conserved charges. In 2017, Piroli, Pozsgay, and Vernier proposed an analogous definition for integrable boundary states \left|\mathcal{B}\right\rangle in lattice models [ 2 ] : \displaystyle Q_{2k+1}\left|\mathcal{B}\right\rangle=0,k\geq 1, (1) where the conserved charges Q_{n+1},n\geq 1 are generated from the transfer matrix \tau\left(u\right) via \displaystyle Q_{n+1}\propto\left.\frac{\partial^{n}}{\partial u^{n}}\ln\tau\left(u\right)\right|_{u=u_{\ast}}, (2) with a suitable choice of u_{\ast} such that Q_{n+1} can be expressed as a sum over operators with finite range. Since the odd conserved charges are generated by the transfer matrix, a sufficient condition for a boundary state to be integrable can be formulated directly in terms of transfer matrix \tau(u) : \displaystyle\Pi\tau\left(u\right)\Pi\left|\mathcal{B}\right\rangle=\tau\left(u\right)\left|\mathcal{B}\right\rangle, (3) with \Pi being the reflection operator \displaystyle\Pi\left|A_{1}A_{2}\cdots A_{L}\right\rangle=\left|A_{L}\cdots A_{2}A_{1}\right\rangle. (4) Motivated by significant applications in both quench dynamics of spin chains and gauge/string duality, there has been considerable interest in deriving explicit overlap formulas between integrable boundary states and eigenstates (on-shell Bethe states) of the transfer matrix. These overlaps exhibit several universal features. Certain selection rules emerge, determining when the overlap can be non-vanishing. One crucial selection rule demands the Bethe roots characterizing a given eigenstate must exhibit a specific pairing structure. This is because integrable states preserve infinitely many conserved charges, as shown in ( 1 ). In higher-rank spin chains, Bethe roots can be paired in different ways [ 3 ] . An overlap is referred to as chiral if the pairing occurs among Bethe roots of the same type, and achiral if the pairing involves roots of different types. Furthermore, the non-vanishing overlaps involving integrable boundary states and eigenstates are relatively simple, compared to non-integrable cases, often admitting compact expressions in terms of ratios of Gaudin-like determinants multiplied by scalar factors which are functions of Bethe roots. In the context of the AdS_{5}/CFT_{4} correspondence, both types of pairing structures have appeared in the literature, notably in studies of correlation functions in the presence of defects and three-point functions [ 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 ] . In contrast, within the AdS_{4}/CFT_{3} framework, only the achiral pairing structure with pairing condition {\mathbf{u}}={-\mathbf{v}} and {\mathbf{w}}={-\mathbf{w}} imposed on the sets of Bethe roots \mathbf{u},\mathbf{w},\mathbf{v} has been observed [ 20 , 21 , 22 , 23 , 24 , 25 , 26 ] 1 1 1 On the string theory side, the integrability of open strings ending on a D4 -brane dual to a domain wall was revealed in [ 27 ] , based on earlier studies on dynamical reflection matrices [ 28 ] . See [ 29 ] for a very nice review on this subject. . This paper aims to fill this gap by identifying special two-site and four-site states which are chiral integrable states among the basis states. We also have numerical evidence which supports that the other basis states are not chiral integrable boundary states. The planar two-loop dilatation operator acting on the scalar sector of ABJM theory [ 30 ] gives an integrable Hamiltonian of an alternating spin chain [ 31 , 32 ] . States on odd/even sites of this chain are in the fundamental/anti-fundamental representation. In this case, there are two transfer matrices \tau(u) , \bar{\tau}(u) which commute for arbitrary spectral parameters \displaystyle[\tau(u),\tau(v)]=[\bar{\tau}(u),\bar{\tau}(v)]=[\tau(u),\bar{\tau}(v)]=0. (5) These transfer matrices generate two sets of conserved charges Q_{l},\bar{Q}_{l} . Two types of integrable boundary states are anticipated. They are chiral integrable states satisfying the untwisted integrable condition \displaystyle\Pi\tau(u)\Pi\left|\mathcal{B}\right\rangle=\tau(u)\left|\mathcal{B}\right\rangle, (6) and achiral integrable states satisfying the twisted integrable condition \displaystyle\Pi\tau(u)\Pi\left|\mathcal{B}\right\rangle=\bar{\tau}(u)\left|\mathcal{B}\right\rangle. (7) In this paper, we focus on the chiral integrable states. Instead of dealing with integrable conditions in terms of conserved charges Q_{l} and \bar{Q}_{l} , we found that taking ( 6 ) as a starting point turns out to be more tractable. The remainder of this paper is organized as follows. In section 2 , we review the fundamental setup of the SU(4) alternating spin chain. Then section 3 is devoted to identifying chiral integrable boundary states through a detailed derivation. Overlap formulas between these boundary states and Bethe states are presented in section 4 . Finally, we give a conclusion in section 5 . Several Bethe roots used in our numerical checks are listed in Appendix A . 2 ABJM spin chain We begin by reviewing the SU(4) alternating spin chain from ABJM theory. Gauge invariance requires that single-trace operators be represented as states in an alternating SU(4) spin chain. States on odd sites of this chain transform in the fundamental representation of the R-symmetry group SU(4) , and those on even sites transform in the anti-fundamental representation. The planar two-loop dilatation operator acting on the scalar sector of ABJM theory can be described by the following spin chain Hamiltonian [ 31 , 32 ] \displaystyle H=\frac{\lambda^{2}}{2}\sum_{l=1}^{2L}\left(2-2P_{l,l+2}+P_{l,l+2}K_{l,l+1}+K_{l,l+1}P_{l,l+2}\right), (8) where P_{ij} and K_{ij} are permutation and trace operators, respectively. Periodic boundary condition is understood. The orthonormal bases of the local Hilbert spaces are \left|A\right\rangle ( A=1,\cdots,4 ) for the fundamental representation sites and \left|\bar{A}\right\rangle ( \bar{A}=\bar{1},\cdots,\bar{4} ) for the anti-fundamental representation ones. The permutation operator P and trace operator act as \displaystyle P\left|A\right\rangle\otimes\left|B\right\rangle \displaystyle=\left|B\right\rangle\otimes\left|A\right\rangle,\quad P\left|\bar{A}\right\rangle\otimes\left|\bar{B}\right\rangle=\left|\bar{B}\right\rangle\otimes\left|\bar{A}\right\rangle; (9) \displaystyle K\left|A\right\rangle\otimes\left|\bar{B}\right\rangle \displaystyle=\delta_{AB}\sum_{C=1}^{4}\left|C\right\rangle\otimes\left|\bar{C}\right\rangle,\quad K\left|\bar{A}\right\rangle\otimes\left|B\right\rangle=\delta_{AB}\sum_{C=1}^{4}\left|\bar{C}\right\rangle\otimes\left|C\right\rangle. (10) It is also convenient to introduce the index notation for these operators. For example, \displaystyle I_{a_{1}a_{2}}^{b_{1}b_{2}}=\delta_{a_{1}}^{b_{1}}\delta_{a_{2}}^{b_{2}},\quad P_{a_{1}a_{2}}^{b_{1}b_{2}}=\delta_{a_{1}}^{b_{2}}\delta_{a_{2}}^{b_{1}},\quad K_{a_{1}\bar{a}_{2}}^{b_{1}\bar{b}_{2}}=\delta^{\bar{b}_{1}\bar{b}_{2}}\delta_{a_{1}a_{2}}, (11) P_{ij} and K_{ij} satisfy the following relations which will be used extensively in subsequent derivations in subsection 3.2 \displaystyle P_{ij}P_{ij} \displaystyle=I_{ij}, (12) \displaystyle P_{ij}P_{ik} \displaystyle=P_{ik}P_{kj}=P_{jk}P_{ij}, (13) \displaystyle K_{ij}K_{ij} \displaystyle=4K_{ij}, (14) \displaystyle K_{ij}K_{ik} \displaystyle=K_{ij}P_{jk}=P_{jk}K_{ik}, (15) \displaystyle\textbf{tr}_{i}P_{ij} \displaystyle=I_{j},\quad\textbf{tr}_{i}K_{ij}=I_{j}. (16) The SU(4) alternating spin chain has four R matrices, \displaystyle R_{0j}(u) \displaystyle=uI+P_{0j}, \displaystyle R_{0\bar{j}}(u) \displaystyle=-(u+2)I+K_{0\bar{j}}, \displaystyle R_{\bar{0}j}(u) \displaystyle=-(u+2)I+K_{\bar{0}j}, \displaystyle R_{\bar{0}\bar{j}}(u) \displaystyle=uI+P_{\bar{0}\bar{j}}, (17) in which 0 and \bar{0} denote the auxiliary space in the fundamental and anti-fundamental representations of the R symmetry, respectively. 2L sites of the alternating spin chain are labeled as 1,\bar{1},2,\bar{2},\cdots,L,\bar{L} . Based on these, two monodromy matrices are defined as \displaystyle T_{0}(u) \displaystyle=R_{01}(u)R_{0\bar{1}}(u)\cdots R_{0L}(u)R_{0\bar{L}}(u), (18) \displaystyle T_{\bar{0}}(u) \displaystyle=R_{\bar{0}1}(u)R_{\bar{0}\bar{1}}(u)\cdots R_{\bar{0}L}(u)R_{\bar{0}\bar{L}}(u). (19) The associated transfer matrices are obtained by tracing over the auxiliary space \displaystyle\tau(u) \displaystyle=\text{tr}_{0}T_{0}(u),\quad\bar{\tau}(u)=\text{tr}_{\bar{0}}T_{\bar{0}}(u). (20) A general on-shell Bethe state is characterized by three sets of Bethe roots, \mathbf{u} , \mathbf{w} and \mathbf{v} , which satisfy a set of Bethe equations \displaystyle\begin{gathered}\begin{aligned} 1=\left(\frac{u_{j}+\frac{i}{2}}{u_{j}-\frac{i}{2}}\right)^{L}\prod_{\begin{array}[]{c}k=1\\ k\neq j\end{array}}^{N_{\mathbf{u}}}S(u_{j},u_{k})\prod_{k=1}^{N_{\mathbf{w}}}\tilde{S}(u_{j},w_{k})\end{aligned}\\ \begin{aligned} 1=\prod_{\begin{array}[]{c}k=1\\ k\neq j\end{array}}^{N_{\mathbf{w}}}S(w_{j},w_{k})\prod_{k=1}^{N_{\mathbf{u}}}\tilde{S}(w_{j},u_{k})\prod_{k=1}^{N_{\mathbf{v}}}\tilde{S}(w_{j},v_{k})\end{aligned}\\ \begin{aligned} 1=\left(\frac{v_{j}+\frac{i}{2}}{v_{j}-\frac{i}{2}}\right)^{L}\prod_{\begin{array}[]{c}k=1\\ k\neq j\end{array}}^{N_{\mathbf{v}}}S(v_{j},v_{k})\prod_{k=1}^{N_{\mathbf{w}}}\tilde{S}(v_{j},w_{k}),\end{aligned}\end{gathered} where N_{u},N_{w},N_{v} denote the number of rapidities in the sets \mathbf{u}=\{u_{1},\dots,u_{N_{u}}\} , \mathbf{w}=\{w_{1},\dots,w_{N_{w}}\} , \mathbf{v}=\{v_{1},\dots,v_{N_{v}}\} , and \displaystyle S(u,v) \displaystyle=\frac{u-v-i}{u-v+i},\quad\tilde{S}(u,v)=\frac{u-v+\frac{i}{2}}{u-v-\frac{i}{2}}. (24) Finally, the eigenvalues of the transfer matrices are expressed as \displaystyle\Lambda(u)= \displaystyle(-u-2)^{L}(u+1)^{L}\prod_{i=1}^{K_{u}}\frac{u-iu_{i}-1/2}{u-iu_{i}+1/2} \displaystyle+(-u)^{L}(u+1)^{L}\prod_{i=1}^{K_{v}}\frac{u-iv_{i}+5/2}{u-iv_{i}+3/2} \displaystyle+(-u)^{L}(u+2)^{L}\prod_{i=1}^{K_{u}}\frac{u-iu_{i}+3/2}{u-iu_{i}+1/2}\prod_{i=1}^{K_{w}}\frac{u-iw_{i}}{u-iw_{i}+1} \displaystyle+(-u)^{L}(u+2)^{L}\prod_{i=1}^{K_{v}}\frac{u-iv_{i}+1/2}{u-iv_{i}+3/2}\prod_{i=1}^{K_{w}}\frac{u-iw_{i}+2}{u-iw_{i}+1}, (25) \displaystyle\bar{\Lambda}(u)= \displaystyle(-u)^{L}(u+1)^{L}\prod_{i=1}^{K_{u}}\frac{u-iu_{i}+5/2}{u-iu_{i}+3/2} \displaystyle+(-u-2)^{L}(u+1)^{L}\prod_{i=1}^{K_{v}}\frac{u-iv_{i}-1/2}{u-iv_{i}+1/2} \displaystyle+(-u)^{L}(u+2)^{L}\prod_{i=1}^{K_{u}}\frac{u-iu_{i}+1/2}{u-iu_{i}+3/2}\prod_{i=1}^{K_{w}}\frac{u-iw_{i}+2}{u-iw_{i}+1} \displaystyle+(-u)^{L}(u+2)^{L}\prod_{i=1}^{K_{v}}\frac{u-iv_{i}+3/2}{u-iv_{i}+1/2}\prod_{i=1}^{K_{w}}\frac{u-iw_{i}}{u-iw_{i}+1}. (26) The corresponding Bethe states \left|\bf{u},\bf{v},\bf{w}\right\rangle can be constructed using nested coordinate Bethe ansatz [ 20 ] where the vacuum was chosen to be \left|1\bar{4}\right\rangle^{\otimes L} . 3 Chiral Integrable Boundary States By explicitly examining the expressions for \Lambda(u)\eqref{lmd} and \bar{\Lambda}(u)\eqref{lmdb} , we observe that when \left\{\mathbf{u}\right\}=\left\{-\mathbf{u}\right\},\left\{\mathbf{w}\right\}=\left\{-\mathbf{w}\right\},\left\{\mathbf{v}\right\}=\left\{-\mathbf{v}\right\} , the relation \Lambda(u)=\bar{\Lambda}(-u-2) holds. This motivates the expectation that the integrability condition \displaystyle\tau(u)\left|\mathcal{B}\right\rangle=\bar{\tau}(-u-2)\left|\mathcal{B}\right\rangle (27) leads to the chiral pairing structure \displaystyle\left\{\mathbf{u}\right\}=\left\{-\mathbf{u}\right\},\left\{\mathbf{w}\right\}=\left\{-\mathbf{w}\right\},\left\{\mathbf{v}\right\}=\left\{-\mathbf{v}\right\} (28) for the overlap \left\langle\mathcal{B}|\mathbf{u},\mathbf{w},\mathbf{v}\right\rangle to be non-vanishing. In fact, ( 27 ) can be derived from the untwisted integrable condition [ 3 ] \displaystyle\Pi\tau(u)\Pi\left|\mathcal{B}\right\rangle \displaystyle=\tau(u)\left|\mathcal{B}\right\rangle, (29) where \Pi denotes the spatial reflection operator defined by \displaystyle\Pi\left|B^{1}\bar{B}_{1}B^{2}\bar{B}_{2}\cdots B^{L}\bar{B}_{L}\right\rangle=\left|\bar{B}_{L}B^{L}\cdots\bar{B}_{2}B^{2}\bar{B}_{1}B^{1}\right\rangle. (30) The transformation \Pi\tau(u)\Pi reverses the order of R-matrix products \displaystyle\Pi\tau(u)\Pi=\text{tr}_{0}R_{0\bar{L}}(u)R_{0L}(u)\cdots R_{0\bar{1}}(u)R_{01}(u). (31) Using crossing symmetry relations for R-matrices \displaystyle R_{0\bar{j}}^{t_{a}}(u)=R_{\bar{0}\bar{j}}(-u-2),R_{0j}^{t_{a}}(u)=R_{\bar{0}j}(-u-2), (32) where t_{a} denotes transposition with respect to auxiliary space, we deduce that \displaystyle\Pi\tau(u)\Pi \displaystyle=\text{tr}_{0}\left(R_{\bar{0}1}(-u-2)R_{\bar{0}\bar{1}}(-u-2)\cdots R_{\bar{0}L}(-u-2)R_{\bar{0}\bar{L}}(-u-2)\right)^{t_{a}} (33) \displaystyle=\bar{\tau}\left(-u-2\right). Thus gives rise to ( 27 ) . We now aim to identify the states that satisfy the integrability condition ( 27 ) . In general, such a state is a specific and intricate linear combination of the orthonormal basis states spanning the full Hilbert space of the spin chain. For simplicity, however, we restrict our attention to basis states \left|B\right\rangle=\left|B^{1}\bar{B}_{1}B^{2}\bar{B}_{2}\cdots B^{L}\bar{B}_{L}\right\rangle , with B^{i}=1,2,3,4,\ i=1,\cdots,L and \bar{B}_{j}=\bar{1},\bar{2},\bar{3},\bar{4},\ j=1,\cdots,L , and ask which of these are integrable. To further facilitate the analysis, we expand the transfer matrix in a specific form and derive a sufficient condition for the integrability condition ( 27 ) , as done for achiral integrable boundary states in the ABJM spin chain [ 23 , 25 ] . Starting from ( 20 ) , we write \displaystyle\tau(u) \displaystyle=\sum_{m,n=0}^{L}u^{L-m}\left(-(u+2)\right)^{L-n}\mathcal{O}_{m,n}\ , (34) where, omit all identity operators I , \displaystyle\mathcal{O}_{m,n}=\sum_{\begin{subarray}{c}i_{1}<\cdotsk\\ \sigma_{j}<\sigma_{k}\end{subarray}}S\left(u_{\sigma_{j}},u_{\sigma_{k}}\right) (98) \displaystyle= \displaystyle\left(\prod_{j>k}f^{-1}\left(u_{j},u_{k}\right)\right)\sum_{\sigma\in S_{\frac{L}{2}}}\prod_{j=1}^{\frac{L}{2}}\left(l_{u_{\sigma_{j}}}\right)^{2j-1}\prod_{j>k}f\left(u_{\sigma_{j}},u_{\sigma_{k}}\right). (99) When the Bethe state is on-shell and the rapidities are parity-invariant, this expression simplifies to a compact formula involving the Gaudin determinant [ 37 ] \displaystyle\left\langle B_{2\bar{4}1\bar{4},\frac{L}{2}}|\mathbf{u}\right\rangle= \displaystyle\prod_{j>k=1}^{N_{u}}f^{-1}\left(u_{j},u_{k}\right)\times (100) \displaystyle\begin{cases}\prod_{i=1}^{\frac{N_{u}}{2}}\frac{\left(u_{i}^{2}+\frac{1}{4}\right)^{2}}{4u_{i}^{2}}\prod_{j>k=1}^{\frac{N_{u}}{2}}F\left(u_{j},u_{k}\right)\times\det G_{+}^{even}&N_{u}=\frac{L}{2}\ \text{even},\\ \frac{1}{8}\prod_{i=1}^{\left\lfloor\frac{N_{u}}{2}\right\rfloor}\frac{\left(u_{i}^{2}+\frac{1}{4}\right)^{2}}{4u_{i}^{2}}\frac{u_{i}^{2}+1}{u_{i}^{2}}\prod_{j>k=1}^{\left\lfloor\frac{N_{u}}{2}\right\rfloor}F\left(u_{j},u_{k}\right)\times\det G_{+}^{odd}&N_{u}=\frac{L}{2}\ \text{odd},\end{cases} where \displaystyle F\left(u_{j},u_{k}\right)=f\left(u_{j},u_{k}\right)f\left(u_{j},-u_{k}\right)f\left(u_{k},u_{j}\right)f\left(-u_{k},u_{j}\right). (101) Thus, the overlap between the boundary state and the normalized Bethe state is given by \displaystyle\frac{\left\langle B_{2\bar{4}1\bar{4},\frac{L}{2}}|\mathbf{u}\right\rangle}{\sqrt{\left\langle\mathbf{u}|\mathbf{u}\right\rangle}}= \displaystyle\prod_{j>k=1}^{N_{u}}f^{-1}\left(u_{j},u_{k}\right)\times (102) \displaystyle\begin{cases}\prod_{i=1}^{\frac{N_{u}}{2}}\frac{\left(u_{i}^{2}+\frac{1}{4}\right)}{4u_{i}^{2}}\prod_{j>k=1}^{\frac{N_{u}}{2}}F\left(u_{j},u_{k}\right)\times\sqrt{\frac{\det G_{+}^{even}}{\det G_{-}^{even}}}&N_{u}=\frac{L}{2}\ \text{even},\\ \frac{\sqrt{{p}^{\prime}\left(0\right)}}{8}\prod_{i=1}^{\left\lfloor\frac{N_{u}}{2}\right\rfloor}\frac{u_{i}^{2}+\frac{1}{4}}{4u_{i}^{2}}\frac{u_{i}^{2}+1}{u_{i}^{2}}\prod_{j>k=1}^{\left\lfloor\frac{N_{u}}{2}\right\rfloor}F\left(u_{j},u_{k}\right)\times\sqrt{\frac{\det G_{+}^{odd}}{\det G_{-}^{odd}}}&N_{u}=\frac{L}{2}\ \text{odd}.\end{cases} For \left\langle B\right|=\left\langle 1\bar{4}2\bar{4}\right|^{\otimes\frac{L}{2}} , the results are basically the same as \left\langle B\right|=\left\langle 2\bar{4}1\bar{4}\right|^{\otimes\frac{L}{2}} . \displaystyle\left\langle B_{1\bar{4}2\bar{4},\frac{L}{2}}|\mathbf{u}\right\rangle= \displaystyle\sum_{\sigma\in S_{\frac{L}{2}}}\prod_{j=1}^{\frac{L}{2}}\left(l_{u_{\sigma_{j}}}\right)^{2j}\prod_{\begin{subarray}{c}j>k\\ \sigma_{j}<\sigma_{k}\end{subarray}}S\left(u_{\sigma_{j}},u_{\sigma_{k}}\right), (103) \displaystyle= \displaystyle\left(\prod_{j>k}f^{-1}\left(u_{j},u_{k}\right)\right)\sum_{\sigma\in S_{\frac{L}{2}}}\prod_{j=1}^{\frac{L}{2}}\left(l_{u_{\sigma_{j}}}\right)^{2j}\prod_{j>k}f\left(u_{\sigma_{j}},u_{\sigma_{k}}\right). (104) By comparison with ( 98 ) , we deduce that the final result is identical to ( 100 ) and ( 102 ) , up to an overall multiplicative factor \displaystyle\prod_{j=1}^{N_{u}}l_{u_{j}}=(-1)^{N_{u}}=(-1)^{\frac{L}{2}}. (105) Here the pair condition on the Bethe roots \mathbf{u} has been used. The remaining results are as follows. For the quantum numbers (N_{u},N_{w},N_{v})=\left(0,0,\frac{L}{2}\right) , the expression corresponding to \left\langle B\right|=\left\langle 1\bar{3}1\bar{4}\right|^{\otimes\frac{L}{2}} can be obtained from that of \left\langle B\right|=\left\langle 2\bar{4}1\bar{4}\right|^{\otimes\frac{L}{2}} by replacing \mathbf{u} with \mathbf{v} . Likewise, the result for \left\langle B\right|=\left\langle 1\bar{4}1\bar{3}\right|^{\otimes\frac{L}{2}} is derived from that of \left\langle B\right|=\left\langle 1\bar{4}2\bar{4}\right|^{\otimes\frac{L}{2}} under the same substitution \mathbf{u}\rightarrow\mathbf{v} . In the SU(2)\times SU(2) sector characterized by the quantum numbers (N_{u},N_{w},N_{v})=\left(\frac{L}{2},0,\frac{L}{2}\right) , the corresponding results are given below: \displaystyle\left\langle B_{2\bar{3}1\bar{4},\frac{L}{2}}|\mathbf{u},\mathbf{v}\right\rangle \displaystyle=\left\langle B_{2\bar{4}1\bar{4},\frac{L}{2}}|\mathbf{u}\right\rangle\times\left\langle B_{1\bar{3}1\bar{4},\frac{L}{2}}|\mathbf{v}\right\rangle, (106) \displaystyle\left\langle B_{2\bar{4}1\bar{3},\frac{L}{2}}|\mathbf{u},\mathbf{v}\right\rangle \displaystyle=\left\langle B_{2\bar{4}1\bar{4},\frac{L}{2}}|\mathbf{u}\right\rangle\times\left\langle B_{1\bar{4}1\bar{3},\frac{L}{2}}|\mathbf{v}\right\rangle, (107) \displaystyle\left\langle B_{1\bar{3}2\bar{4},\frac{L}{2}}|\mathbf{u},\mathbf{v}\right\rangle \displaystyle=\left\langle B_{1\bar{4}2\bar{4},\frac{L}{2}}|\mathbf{u}\right\rangle\times\left\langle B_{1\bar{3}1\bar{4},\frac{L}{2}}|\mathbf{v}\right\rangle, (108) \displaystyle\left\langle B_{1\bar{4}2\bar{3},\frac{L}{2}}|\mathbf{u},\mathbf{v}\right\rangle \displaystyle=\left\langle B_{1\bar{4}2\bar{4},\frac{L}{2}}|\mathbf{u}\right\rangle\times\left\langle B_{1\bar{4}1\bar{3},\frac{L}{2}}|\mathbf{v}\right\rangle. (109) 5 Conclusion In this paper, we studied chiral integrable boundary states in the ABJM spin chain. These states had not appeared in previous investigations of integrable boundary states within the AdS_{4}/CFT_{3} correspondence. Among the basis states, we identified specific two-site or four-site states satisfy the untwisted integrable condition and exhibit this chiral property. We have some numerical evidences that other basis states are not chiral integrable. We further computed the overlap between these chiral integrable states and general Bethe states. A key step in our analysis was the use of ( 38 ) which is a sufficient condition for a state to be chiral integrable. Crucially, this sufficient condition is simpler to solve in practice than ( 41 )—which is equivalent to the untwisted integrable condition—or than the formulation involving local higher charges generated from the transfer matrices \tau(u) and \bar{\tau}(u) . Since numerical results support that other basis states are not chiral integrable, we conjecture that this sufficient condition is also necessary. If one can proof this, then the determination of all chiral integrable states among a certain set of states will be greatly simplified. We also anticipate discovering more complex chiral integrable states, such as matrix product states (MPS), based on solutions to the corresponding KT relation [ 3 ] . An alternative approach involves generalizing the necessary integrability condition for MPS [ 45 ] from SO(N) spin chains to the ABJM setting. This condition allows us to systematically identify candidate states. Then we can try to determine genuine integrable states through analytical or numerical methods. It would also be valuable to explore how these chiral integrable states might be realized as operators or defects in ABJM theory and to seek their interpretation in the dual string/M-theory. Notably, known integrable boundary states in AdS_{4}/CFT_{3} correspondence arising from BPS operators or defects are exclusively achiral. This presents a stark contrast to {\cal N}=4 SYM theory, where both chiral and achiral integrable states in the SO(6) sector emerge in computations of specific correlators. For example, chiral states appear in heavy-heavy-light (HHL) three-point functions [ 12 , 13 ] , while achiral states feature in domain-wall one-point functions [ 4 ] . Understanding the deeper underlying reasons for this difference between the two theories merits further investigation. Appendix A Bethe roots Solving rational Q-system [ 33 , 34 ] of given quantum numbers (L,N_{u},N_{w},N_{v}) yields \left\{Q_{1}(u),Q_{2}(w),Q_{3}(v)\right\} whose zeros are Bethe roots. SU(2) \left(N_{u},N_{w},N_{v}\right)=\left(\frac{L}{2},0,0\right) L 2 4 6 Q_{1}(u) u u^{2}-\frac{1}{12} u^{3}+\frac{u}{12}-\frac{1}{4\sqrt{3}} u^{3}+\frac{u}{12}+\frac{1}{4\sqrt{3}} u^{3}+\frac{1}{12}\left(5-2\sqrt{13}\right)u u^{3}+\frac{1}{12}\left(2\sqrt{13}+5\right)u SU(2)\times SU(2) \left(N_{u},N_{w},N_{v}\right)=\left(\frac{L}{2},0,\frac{L}{2}\right) L 2 4 \left\{Q_{1}(u),Q_{3}(v)\right\} \left\{u,v\right\} \left\{u^{2}-\frac{1}{12},v^{2}-\frac{1}{12}\right\} Table 2: Bethe roots Several Bethe roots don‘t exhibit chiral pair structure are also used in our numerical check, which are: • L=4n-2\,(n=1,2,\cdots),\,N_{u}=N_{w}=N_{v}=1,\,{\bf u}={\bf v}={\bf w}=\{\pm 1/2\} , • L=3n-1,(n=1,2,\cdots),\,N_{u}=N_{w}=N_{v}=1,\,{\bf u}=-{\bf v}=\{1/(2\sqrt{3})\},\,{\bf w}=\{0\} , • L=3,\,N_{u}=N_{w}=1,\,N_{v}=2,\,{\bf u}={\bf w}=\{\pm\frac{\sqrt{3}}{2}\},{\bf v}=\{\frac{1}{8}(\pm\sqrt{3}+\sqrt{11}),\frac{1}{8}(\pm\sqrt{3}-\sqrt{11})\} , 3 3 3 This set of Bethe roots was first found numerically in [ 24 ] . We now find the analytic expression. • Bethe roots listed in the appendixes of [ 20 , 24 ] besides the above ones.