ABJM Spin Chain: Integrability and Fusion

• The ABJM spin chain is an integrable quantum spin chain emerging from planar ABJM theory with alternating SU(4) representations.
• It employs the algebraic Bethe ansatz and fusion procedures to overcome non-regular R-matrix challenges, leading to nearest-neighbor Hamiltonians and higher conserved charges.
• Chiral integrable boundary states and matrix product states are constructed via fused reflection equations, informing defect CFT observables in the AdS/CFT framework.

The Aharony–Bergman–Jafferis–Maldacena (ABJM) spin chain refers to an integrable quantum spin chain arising in the planar limit of ABJM theory, a three-dimensional superconformal Chern–Simons-matter theory with OSp$(6|4)$ symmetry. The theory's two-loop planar dilatation operator in the scalar SU(4) subsector is mapped to an alternating spin chain of length $2L$, with odd sites in the fundamental $\mathbf{4}$ and even sites in the anti-fundamental $\bar{\mathbf{4}}$ of SU(4). The spin chain possesses intricate integrable properties, with distinct physical and mathematical structures appearing in its fundamental alternating and fused (nearest-neighbor) incarnations, and supports the explicit construction of chiral integrable boundary states relevant for defect CFT observables in the AdS/CFT correspondence (Bai et al., 2024, Liu et al., 2 Feb 2026).

1. Algebraic Bethe Ansatz and the Alternating ABJM Spin Chain

The planar ABJM “alternating” spin chain is constructed via the algebraic Bethe ansatz, with the quantum space

$$\mathcal H = \bigotimes_{l=1}^{2L} V_l, \qquad V_{2j-1}\cong\mathbf{4}, \quad V_{2j}\cong\bar{\mathbf{4}},$$

where odd sites transform under the SU(4) fundamental and even sites under the anti-fundamental. Integrability is encoded through R-matrices that interchange or trace SU(4) indices: $$R_{ab}(u) = u\,\mathbf{1}_{ab} + P_{ab}, \qquad R_{a\bar b}(u) = -(u+2)\,\mathbf{1}_{a\bar b} + K_{a\bar b},$$ where $P$ is the permutation and $K$ the trace operator. Two monodromy matrices,

$$T_0(u) = R_{0,1}(u)\cdots R_{0,2L}(u),\quad \widehat{T}_{\bar0}(u) = R_{\bar0,1}(u)\cdots R_{\bar0,2L}(u),$$

lead to transfer matrices $\tau(u)$ and $\widehat\tau(u)$ whose commutativity encodes conserved quantities arising from Yang–Baxter integrability (Bai et al., 2024, Liu et al., 2 Feb 2026).

2. Fusion Procedure and the Fused Model

The non-regularity of the fundamental R-matrix prevents the construction of a genuine nearest-neighbor Hamiltonian. To circumvent this, the fusion procedure is introduced: every pair of adjacent sites $(2j-1,2j)$ is merged to form a 16-dimensional vector space $$V_{2j-1}\otimes V_{2j}\cong\mathbf{4}\otimes\bar{\mathbf{4}}$$. The fused R-matrix,

$$R^{\rm (fus)}_{(aa),(bb)}(u) = R_{a,b}(u)\,R_{a,\bar b}(u)\,R_{\bar a,b}(u)\,R_{\bar a,\bar b}(u),$$

acts between fused quantum or auxiliary spaces and satisfies a regularity condition,

$R^{\rm (fus)}_{(aa),(bb)}(0) = 4\,P_{13}\,P_{24},$

yielding a nearest-neighbor model via a fused transfer matrix $t^{\rm(fus)}(u) = \Tr_{(00)} \mathcal{T}_{(00)}(u) = \tau(u)\widehat{\tau}(u)$ (Bai et al., 2024).

3. Hamiltonian, Boost Operator, and Local Charges

The Hamiltonian of the fused ABJM chain is obtained from the logarithmic derivative of the transfer matrix at the regular point $u=0$: $$H = i\,\frac{d}{du}\ln t^{\rm (fus)}(u)\Big|_{u=0} = \sum_{j=1}^L H_{(2j-1,2j),(2j+1,2j+2)},$$ with the two-site local density expressed as

$$H_{(2j-1,2j),(2j+1,2j+2)} = h_{2j-1,2j,2j+1} + h_{2j,2j+1,2j+2},$$

where

$$h_{j,j+1,j+2} = P_{j,j+2} - K_{j+1,j+2} P_{j,j+2} - P_{j,j+2} K_{j,j+1}.$$

The fused model, being regular, allows the definition of a boost operator,

$$B = -\sum_{k=1}^{L-1} k\,H_{(2k-1,2k),(2k+1,2k+2)},$$

which generates higher local charges recursively: $$Q_{n+1} = [B, Q_n],\qquad Q_2 \equiv -iH, \quad Q_1 = -i\ln t^{\rm(fus)}(0).$$ Such a boost operator is absent in the original alternating chain, highlighting the structural advantage of the fused construction (Bai et al., 2024).

4. Open Boundary Conditions and Chiral Integrable States

Integrable open boundary conditions in the ABJM spin chain are introduced via boundary K-matrices $K^\pm(u)$ acting on the fused auxiliary space. The open-chain double-row transfer matrix,

$t_{\rm open}(u) = \Tr_{(00)}\left[K^+_{(00)}(u)\, \mathcal T_{(00)}(u)\, K^-_{(00)}(u)\, \mathcal T_{(00)}^{-1}(-u)\right],$

yields, via logarithmic differentiation,

$$H_{\rm open} = H_{\rm bulk} + H_{\rm left} + H_{\rm right},$$

where the boundary terms factorize into Kronecker-delta structures up to model-dependent coefficients. The boundary K-matrices satisfy (fused) reflection equations, enabling a systematic construction of chiral integrable boundary states (Bai et al., 2024, Liu et al., 2 Feb 2026).

Chiral boundary states $|\mathcal{B}\rangle$ within the ABJM chain are defined by annihilation under all odd conserved charges. These “chiral” states are constructed by solving the condition

$$\tau(u)\,|\mathcal B\rangle = \Pi\,\tau(u)\,\Pi\,|\mathcal B\rangle,$$

with $\Pi$ the site-reversal permutation, or equivalently for $\widehat\tau(u)$ (Liu et al., 2 Feb 2026).

5. Reflection Equations, Fusion of Boundary States, and Matrix Product States

To generate higher-rank chiral integrable boundary states, the standard fusion procedure is applied to the K-matrices: $$K_{(12)}(u) = \tilde K_2(u)\,R_{12}(2u)\,\tilde K_1(u),$$ and by repeating this process $n$ times, one constructs $n$-fused K-matrices $K^{(n)}(u)$ solving fused reflection equations. Each solution yields a $2n$-site block state,

$$|\Phi^{(n)}(u)\rangle = [K^{(n)}(u)]^{i_1\ldots i_n}_{j_1\ldots j_n} |i_n,\ldots,i_1, j_n,\ldots,j_1\rangle \in V^{\otimes 2n},$$

forming translationally invariant matrix product states (MPS) under repetition across the chain. All such states satisfy the chiral integrability condition imposed by the transfer matrices (Liu et al., 2 Feb 2026).

6. Overlaps, Bethe States, and Gaudin Determinants

For the special case $n=2$, the four-site block MPS $$|\Psi_4\rangle = |\Phi^{(2)}(-1)\rangle^{\otimes L/2}$$ possesses explicitly computable overlaps with Bethe eigenstates characterized by three rapidity sets: $$\langle\Psi_4|\mathbf u,\mathbf w,\mathbf v\rangle = (-2)^{L/2} \prod_{i=1}^{N_u/2} u_i(u_i-\tfrac{i}{2}) \prod_{j=1}^{N_v/2} v_j(v_j-\tfrac{i}{2}) \prod_{k=1}^{N_w/2} \frac{w_k}{w_k+\tfrac{i}{2}} \times \det G_+,$$ where $G_+$ denotes a sector of the Gaudin matrix arising under chiral pairing of rapidities, and the construction extends to various block sizes and dressing procedures. These formulas underpin the exact computation of one-point functions in certain defect CFT contexts (Liu et al., 2 Feb 2026).

7. Classification and Applications of Chiral Integrable Subspaces

Numerical investigation of the space of chiral integrable boundary states reveals its rapid growth with system size: for $L=2$, the space has dimension $196$, and for $L=3$, $616$. Simple product ansätze or four-site MPS together span only subspaces of the full space, implying the existence of yet-undiscovered integrable states beyond those generated by standard fusion procedures. This structure underpins applications in AdS/CFT, including one-point observables in defect CFTs such as Wilson lines or domain walls, and provides a foundation for the study of quantum quenches and boundary correlation functions in nested SU(4) integrable systems (Liu et al., 2 Feb 2026).

---

For a comprehensive algebraic and analytic treatment, refer to "The Fused Model of Alternating Spin Chain from ABJM Theory" (Bai et al., 2024) and "Chiral Integrable Boundary States of ABJM Spin Chain from Reflection Equations" (Liu et al., 2 Feb 2026).