Absolutely Maximally Entangled States

• Absolutely Maximally Entangled States are multipartite quantum states where every bipartition yields a maximally mixed reduced state, ensuring extreme entanglement.
• They are constructed via combinatorial designs, graph-state methods, and tensor-network approaches, which facilitate practical quantum error correction and secret sharing protocols.
• Their unique structure enables optimal teleportation, robust holographic codes, and advances in high-dimensional quantum communication applications.

Absolutely maximally entangled (AME) states are multipartite pure quantum states that exhibit maximal entanglement across every possible bipartition: for any subset of up to half the parties, the reduced density matrix is exactly maximally mixed. This property renders AME states as the most extremal representatives of multipartite entanglement, with profound connections to quantum error correction, secret sharing, teleportation, and holographic codes. The mathematical structure of AME states reveals deep relationships with algebraic combinatorics, coding theory, and matrix/tensor analysis.

1. Formal Definition and Structural Properties

Let $|\Psi\rangle \in (\mathbb{C}^d)^{\otimes n}$ be a pure state on $n$ systems of local dimension $d$. For any subset $A \subset \{1, \dots, n\}$ with $|A| \leq \lfloor n/2 \rfloor$, let $$\rho_A = \operatorname{Tr}_{\bar{A}}(|\Psi\rangle\langle\Psi|)$$. The state is called absolutely maximally entangled (AME$(n,d)$) if

$\rho_A = \frac{1}{d^{|A|}} \mathbb{I}_{d^{|A|}}$

for all such $A$. Equivalently, the von Neumann entropy of every reduced state for $|A| \leq \lfloor n/2 \rfloor$ is $|A| \log d$ (Helwig et al., 2013, Goyeneche et al., 2015, Burchardt et al., 2020). AME is thus the extremal case of $k$-uniformity with $k = \lfloor n/2 \rfloor$.

Key structural features include:

• Genuine multipartite entanglement: No bipartition is separable; all bipartitions demonstrate maximal Schmidt rank and entropy.
• Perfect tensor property: The defining tensor is an isometry under any flattening that maps up to $k$ indices to output and the rest to input (perfect tensors) (Mazurek et al., 2019).
• Complete symmetry among parties: All marginal spectra on equal-size subsystems are identical.

2. Existence, Minimal Support, and Classification

The existence of AME$(n,d)$ states is highly parameter-dependent.

• Qubits ($d=2$): AME$(n,2)$ states exist if and only if $n \in \{2,3,5,6\}$ (Huber et al., 26 Jun 2025, Huber et al., 2016, Zhang et al., 2024). For $n=4$ or $n \geq 7$, no qubit AME states exist (Huber et al., 2016, Huber et al., 26 Jun 2025).
• Qudits ($d>2$): By increasing $d$ (especially to prime powers), further AME states are possible; e.g., AME$(4,3)$ exists, with support on 9 basis states (Helwig, 2013, Goyeneche et al., 2015). A necessary condition for minimal-support AME$(n,d)$ states is $d \geq \frac{n}{2}+1$ (Bernal, 2018); if $d$ is a prime power, Reed-Solomon codes provide constructions with $n=d+1$ (Bernal, 2018).
• Minimal support: A minimal-support AME has exactly $d^{\lfloor n/2\rfloor}$ nonzero coefficients in the computational basis, realized using MDS codes and mutually orthogonal Latin hypercubes (Bernal, 2018, Goyeneche et al., 2015, Helwig, 2013).

Classification by equivalence under local unitaries (LU) and SLOCC (stochastic local operations and classical communication) reveals sharp dichotomies:

• For $n=3$ and $d=2$, GHZ is unique up to LU; for $n=4, d=3$, there is again a single LU class (Rather et al., 2022, Ramadas et al., 2024).
• For larger $d$ and $n\geq5$, typically infinitely many LU-inequivalent AME states exist, parameterized by continuous families of phases associated with irredundant orthogonal arrays (Ramadas et al., 2024, Rather et al., 2022, Burchardt et al., 2020).
• For $n\geq6$ and minimal-support AME states, there are necessarily infinitely many LU and SLOCC classes (Burchardt et al., 2020).

3. Construction Methods

Combinatorial and Coding Constructions

Many AME states correspond directly to combinatorial objects:

• Classical MDS codes: The codewords yield the basis states of the AME via $$\ket{\mathrm{AME}} = \frac{1}{\sqrt{d^k}} \sum_{x \in \mathbb{F}_d^k} |Gx\rangle$$, where $G$ is the generator matrix (Helwig et al., 2013, Helwig, 2013, Bernal, 2018).
• Orthogonal arrays (OA) and irredundant OA: The existence of an irredundant OA with suitable parameters (rows $r > Nd - (N-1)$) guarantees infinitely many LU-inequivalent AME$(N,d)$ states (Ramadas et al., 2024).

Graph and Stabilizer States

• Graph-state formalism: For prime $d$, stabilizer (graph) states correspond to adjacency matrices satisfying a linear-independence criterion in every $\lfloor n/2\rfloor$-subset, efficiently allowing the identification of AME states (Helwig, 2013).
• Multi-unitary/multi-isometry tensors: AME$(2k,d)$ states correspond to $k$-unitary (multi-unitary) tensors; for odd or heterogeneous systems, the requirement generalizes to multi-isometry tensors (Goyeneche et al., 2015, Shen et al., 2020), unifying approaches for homogeneous and heterogeneous dimensions.

Circuit and Tensor-Network Approaches

• Quantum circuits: Efficient gate decompositions are known for several small AME$(n,d)$, minimizing circuit depth and entangling gate count; graph-state and code-based AME circuit implementations are prominent (Cervera-Lierta et al., 2019, Casas et al., 7 Apr 2025). For high $d$, non-stabilizer AME states with explicit multi-unitary diagonal gates have been constructed for $d=4,6,8$ (Casas et al., 7 Apr 2025).
• Tensor network decompositions: AME$(6,D)$ and AME$(8,D)$ for $D \geq 5,7$ can be constructed from a minimal network of perfect 4-leg tensors (perfect tensors), dramatically reducing gate counts compared to full graph-state approaches (Pozsgay et al., 2023).

Heterogeneous and Irreducible AME States

• Heterogeneous local dimensions: In $$\mathbb{C}^{l}\otimes\mathbb{C}^{m}\otimes\mathbb{C}^{n}$$ with $3 \leq l < m < n \leq m+l-1$, existence is governed by the feasibility of a "magic solution array" (MSA): a nonnegative matrix satisfying three combinatorial constraints, giving rise to AME states in heterogeneous systems (Shen et al., 2020).
• Irreducibility criteria: If a tripartite AME has one prime local dimension and the remaining two are coprime, the AME is irreducible; for multipartite systems, at least three primes among the local dimensions enforce irreducibility (Shen et al., 2020).

4. Applications and Connections

Quantum Error Correction

• Quantum MDS codes: There is a one-to-one correspondence between AME$(n,d)$ and pure [[n,0,d/2+1]] quantum MDS codes, with AME(5,2) being the 5-qubit perfect code and AME(6,2) its code concatenation (Mazurek et al., 2019, Helwig et al., 2013, Goyeneche et al., 2015).
• Code concatenation: Entanglement swapping among AME states recapitulates code concatenation, as in the construction of holographic error-correcting codes (e.g., pentagon code) (Mazurek et al., 2019).

Quantum Secret Sharing

• Threshold and ramp schemes: Even-party AME$(2m,d)$ states implement threshold $(m,2m-1)$ QSS schemes, while ramp QSS protocols generalize this using $(m,L,2m-L)$ access structures with the same AME resource (Helwig et al., 2013, Helwig, 2013).
• Equivalence: Every even-party AME state corresponds to a unique pure-state threshold QSS, and vice versa (Helwig et al., 2013, Helwig et al., 2012).

Multi-User Teleportation and Parallel Communication

• Open-destination teleportation: An AME$(n,d)$ enables the perfect teleportation of up to $\lfloor n/2\rfloor$ unknown $d$-level states to any subset of parties, with the flexibility of choosing senders and recipients after state distribution (Helwig et al., 2012, Mazurek et al., 2019).
• Parallel transfer: Both joint and local operations are possible, providing maximal transfer rates compatible with the quantum marginal properties of AME states.

Holography and Perfect Tensors

• Holographic codes: AME states, particularly perfect tensors, are central to toy models of AdS/CFT and holography, saturating corrected Ryu–Takayanagi entropy bounds and ensuring maximal boundary entanglement in tensor-network representations (Mazurek et al., 2019, Goyeneche et al., 2015).
• Generalized swapping: Concatenation of perfect tensors realizes bulk-to-boundary mappings with precise control of entanglement structure.

Quantum Steering and Nonlocality

• Assisted entanglement distillation and steering: Heterogeneous AME states constructed via multi-isometry matrices enable one-sided device-independent steering protocols, with projective measurements on one system collapsing the remainder into maximally entangled states (Shen et al., 2020).

5. Invariants, Classification Results, and Open Problems

LU/SLOCC Equivalence and Parameter Counting

• LU classes: For small $n, d$, especially $n=3$ or $d=2,3$, AME states are unique up to LU. For higher $d$ and $n$ ($n \geq 5$), the existence of irredundant OAs implies that there are usually infinitely many LU-inequivalent AME states, parameterized by continuous phase degrees of freedom associated with the OA rows (Rather et al., 2022, Ramadas et al., 2024, Burchardt et al., 2020).
• SLOCC inequivalence: For $n\ge6$, SLOCC-inequivalent AME states abound, as phase insertions in minimal-support constructions persist under invertible local operations (Burchardt et al., 2020).

Existence and Non-Existence

• No-go theorems:
  • AME$(4,2)$: No four-qubit AME exists; proof via polynomial invariants (Luque–Thibon), quantum code propagation, and anti-commutator arguments in Pauli expansion (Huber et al., 26 Jun 2025).
  • AME$(7,2)$: Nonexistence proved using Bloch-representation and weight parity rules (Huber et al., 2016, Huber et al., 26 Jun 2025).
• Construction failures and exceptional solutions: Euler’s 36-officers case ($d=6$): no pair of classical orthogonal Latin squares, but a quantum (nonclassical) solution via entangled orthogonal squares yields AME$(4,6)$ (Rather et al., 2022).
• Open existence: For large $n$ and small $d$, minimal-support AME existence is essentially governed by code-theoretic and combinatorial constraints (Singleton bound, MDS conjecture) (Bernal, 2018). For composite $d$ not a prime power, explicit construction remains open.

Technical Tools and Criteria

• Bloch representation: Provides characterization via vanishing of weight-$k$ correlation tensors for $k \leq \lfloor n/2 \rfloor$ and explicit criteria for nonexistence based on sign of trace norms (Li et al., 2018).
• Parity rule: In Pauli expansions, the parity of operator weight determines algebraic constraints for possible AME states (Huber et al., 2016, Zhang et al., 2024).
• Maximum-concurrence criterion: The purity of every marginal must attain its minimal value, equating to maximal bipartite I-concurrence in all cuts (Bag et al., 26 Jan 2025).

6. Advanced Generalizations: Heterogeneous and Multipartite Regimes

• Heterogeneous systems: The generalization of $k$-uniformity and AME to systems with varying local dimensions leads to new combinatorial existence problems, often settled by magic solution arrays – arrays of nonnegative entries meeting row/column and modular sum constraints (Shen et al., 2020). Existence and irreducibility in such contexts are classified by number-theoretic relations among subsystem dimensions.
• Irreducibility: The criterion that AME states with three or more prime local dimensions among an odd number of parties are irreducible unless they factor into smaller AMEs shows the atomicity of such states in the multipartite entanglement hierarchy (Shen et al., 2020).

7. Open Questions and Outlook

• Classification for higher $n$ and composite $d$: Many existence questions reduce to combinatorial and coding-theoretic problems; for $d$ not a prime power, minimal-support constructions are incomplete (Bernal, 2018, Rather et al., 2022).
• Non-stabilizer AME states and tensor structures: Most applications and explicit circuits are based on stabilizer or graph-state AMEs; broader families may be accessible via designs based on biunimodular arrays, OAs, or quantum Latin squares (Casas et al., 7 Apr 2025, Rather et al., 2022).
• Noise robustness and application in realistic devices: The response of AME states to local noise, especially Pauli and dephasing channels, illustrates both their invariance (under depolarizing channels) and possible symmetry breaking—relevant for quantum error correction and benchmarking (Stawska et al., 10 May 2025).

---

References:

• (Helwig et al., 2013) Absolutely Maximally Entangled States: Existence and Applications
• (Goyeneche et al., 2015) Absolutely Maximally Entangled states, combinatorial designs and multi-unitary matrices
• (Helwig, 2013) Absolutely Maximally Entangled Qudit Graph States
• (Bernal, 2018) On the Existence of Absolutely Maximally Entangled States of Minimal Support II
• (Mazurek et al., 2019) Quantum error correction codes and absolutely maximally entangled states
• (Rather et al., 2022) Absolutely maximally entangled state equivalence and the construction of infinite quantum solutions to the problem of 36 officers of Euler
• (Ramadas et al., 2024) Local unitary equivalence of absolutely maximally entangled states constructed from orthogonal arrays
• (Zhang et al., 2024) Extremal Maximal Entanglement
• (Huber et al., 2016) Absolutely maximally entangled states of seven qubits do not exist
• (Burchardt et al., 2020) Stochastic Local Operations with Classical Communication of Absolutely Maximally Entangled States
• (Huber et al., 26 Jun 2025) On two maximally entangled couples
• (Bag et al., 26 Jan 2025) A maximum concurrence criterion to investigate absolutely maximally entangled states
• (Casas et al., 7 Apr 2025) Quantum Circuits for High-Dimensional Absolutely Maximally Entangled States
• (Stawska et al., 10 May 2025) Sending absolutely maximally entangled states through noisy quantum channels
• (Pozsgay et al., 2023) Tensor network decompositions for absolutely maximally entangled states
• (Shen et al., 2020) Absolutely maximally entangled states in tripartite heterogeneous systems