Geomagic Square Representation of a Real AME$(4, 6)$ State

AI summary

Overview

Research area: Quantum information theory — specifically absolutely maximally entangled (AME) states of four qudits, combined with combinatorial design and recreational geometry (geomagic squares).

Technical level: Advanced. The paper assumes familiarity with Pauli matrices, unitary flattenings, tensor reshuffling and partial transposition, though its central object — the picture — is described in enough detail to be followed by a patient non-specialist.

Scope (one sentence): The paper constructs a real AME(4, 6) state from a table of signed Pauli matrices and shows that a locally equivalent representative can be drawn as a genuine geomagic square whose 36 pieces, colors and hatching encode the full tensor.

What This Paper Is About

An AME(4, d) state is a four-party state so entangled that discarding any two parties leaves the remaining two in a maximally mixed state. For d = 6 no "permutation" solution exists, because orthogonal Latin squares of order six do not exist — Euler's 36 officers problem — so every known AME(4, 6) representative had complex coefficients, which are hard to draw faithfully. The paper asks whether a real representative exists whose coefficients can be pictured exactly, and then builds such a picture: a hexagon-tiling geomagic square in which areas give magnitudes and hatching gives signs.

Key Contributions

1. An independent construction of a real AME(4, 6) state. A table of signed Pauli matrices (Eq. 2) plus twelve unit entries on odd permutations of (0, 1, 2, 3) defines a tensor with 60 nonzero coefficients: 12 equal to 1 and 48 equal to ±1/√2 (broken down in Appendix A as 12 ones, 34 at 1/√2 and 14 at −1/√2).

2. A short, explicit proof of two-unitarity. All three flattenings are shown to be orthogonal by reducing each mixed block condition to P Rᵀ + Q Sᵀ = 0 and verifying it for every x in all three cuts, giving two-party reductions equal to 𝕀₃₆/36.

3. A genuine geomagic square carrying the state. After applying local orthogonal maps built from the Hadamard matrix H = (X + Z)/√2, the resulting tensor T^g has 152 nonzero entries with squared values 1/2, 1/4 or 1/8, and yields a square in which every row, column and both main diagonals tile a common hexagon with all 36 pieces pairwise non-congruent.

4. Full encode/decode rules. Colors, sector keys, cell labels and hatching allow the tensor to be read back off the picture, and a combinatorial test proves non-congruence for any allowed positive interval split.

Main Findings

• Sparsity and normalization: The two-unitary matrix U has 60 nonzero entries equal to 1 or ±1/√2, and the sum of squared coefficients is 12 + 48/2 = 36, so Eq. (1) is correctly normalized.
• Block structure of every cut: In each of the three flattenings, row and column permutations split the matrix into 𝕀₁₂, two 4×4 Bell-basis blocks and four 4×4 mixed blocks. The identity block comes from the odd permutations, and the Bell blocks are orthonormal because I, X, Z and iY are orthogonal in the Hilbert–Schmidt inner product.
• Verified orthogonality identity: The condition P Rᵀ + Q Sᵀ = 0 holds for every x in all three cuts; the paper's worked example is K₃, giving −(iY)(iY)ᵀ + X Xᵀ = 0.
• Locally equivalent picture tensor: T^g has 152 nonzero entries with squares 1/2, 1/4 or 1/8, versus 60 nonzero entries in the sparse U.
• Diagonal-friendly ordering: Displaying rows in the order (0, 2, 4, 5, 3, 1) and columns in (2, 4, 0, 1, 5, 3) places the cell sets of D = (2, 3, 4, 5, 0, 1) and D′ = (3, 2, 5, 4, 1, 0) on the two geometric diagonals; this works for 48 row orders, those where rows paired by D⁻¹D′ = (0 1)(2 3)(4 5) occupy mirrored positions.
• Transversal decomposition: Table 1 gives 32 records yielding 48 weighted transversals, with weight 1/4 for k = 0 to 3 and 1/16 for k = 4 and 5, which lets each squared coefficient be split into triangles placed in specific cells.
• Non-congruence proved combinatorially: Pieces contain 3 to 12 lobes and there are 32 distinct edge words up to rotation and reflection. Only four candidate pairs survive, distinguished either by lobe counts (8 versus 6, and 10 versus 11) or, for the shared words 123400 and 004420, by occupancy at the start of sector 4's outer edge.
• Metric cross-check: In the unequal split used for Figure 1, the 36 perimeters are pairwise different, an additional check of non-congruence.
• Two-unitarity is not sufficient: The paper stresses that two-unitarity alone fixes areas but does not prevent overlaps, constrains no diagonal and does not exclude congruent pieces; the diagonal and non-congruence properties come from the specific transversal structure.
• Placement robustness: All 720 sector placements also give pairwise non-congruent pieces, for both the canonical and the illustrated split.
• Picture is faithful but not unique: The picture determines the full real tensor, but the converse fails — other interval splits, sector placements, display orders or matching orders decode to the same T^g.

Methodology in Plain English

The approach has three layers. First, a compact recipe: for each pair of subsystems (AB, AC, AD, BC, BD, CD) and each label x from 0 to 3, a small table names one of the four real Pauli-type matrices I, X, Z, iY. Setting two indices to x and the remaining two to 4 + u and 4 + v, with u, v in {0, 1}, fills in entries scaled by 1/√2. Twelve more entries, set to 1, come from the odd permutations of (0, 1, 2, 3). Everything else vanishes.

Second, a check: the author reorganizes the same entries into the three two-party splittings — the original arrangement, a reshuffling, and a partial transpose — and shows that each reorganised matrix splits into an identity block, Bell blocks and small mixed blocks that are all orthogonal. Since the block identity reduces to a single algebraic relation, verifying it case by case for every x finishes the proof.

Third, the picture: applying local orthogonal transformations built from the Hadamard matrix spreads the sparse 60-entry tensor into a 152-entry tensor whose squared magnitudes are 1/2, 1/4 or 1/8 — convenient rational areas. A table of "colored transversals" then specifies how to cut each hexagonal sector into intervals of prescribed total length and place the resulting triangles in the cells of a 6×6 array, so that each interval appears exactly once in every row, column and diagonal, and each color covers the hexagon once overall. Signs are recorded by hatching, so the drawing keeps the phase information needed for interference. A finite combinatorial check of lobe counts on the six outer edges rules out congruent pieces. Notably, the author acknowledges using ChatGPT (OpenAI, GPT-6 Astra Pro) for the pictorial search and exact computational checks and Claude (Anthropic, Claude Opus 5.5) for reproduction code and figures.

Why This Matters

Impact on research. The real-existence question for AME(4, 6) states was explicitly raised in earlier work on complex constructions; this paper supplies a real representative without claiming priority, since independent projects arrived at the same answer with different motivations and visibly different matrices. Beyond the state itself, the paper proposes a dictionary between normalization conditions of the three flattenings and geometric properties — areas, tilings, color balance, and hatching for orthogonality — and raises a tensor analogue of unistochasticity: given the areas of a geomagic square, do signed or phased square roots yield a two-unitary matrix?

Real-world applications. The paper reports no applications; the following are plausible directions rather than claims made in the text.

• Perfect tensors and quantum error-correcting codes, where AME states underlie code constructions.
• Quantum secret sharing and multipartite protocols that rely on maximal entanglement across every bipartition.
• Verification and certification tooling: a pictorial representation could serve as a human-checkable certificate of two-unitarity.
• Visualization and teaching of high-dimensional entanglement, where complex phases are otherwise invisible.

Industry relevance. The work is foundational rather than applied. Its relevance is indirect: to groups working on quantum error correction, entanglement-based cryptography, and to quantum software teams interested in compact, exact representations of states that are otherwise unwieldy — the author collects exact coordinates, source files and reproduction code in a public GitHub repository.

Future Directions

1. Make the cancellations visible. A drawing in which the orthogonality between different rows appears as a geometric cancellation would complete the dictionary between algebra and picture.
2. Test the unistochasticity analogue. Ask whether the areas of arbitrary geomagic squares admit signs or phases whose square roots give two-unitary matrices, using the squares of Ref. [13] as natural test cases.
3. Optimize and extend the drawing. Other locally equivalent representatives, sector placements or interval splits might reduce the 480 triangular fragments of Figure 1, connect the interiors of pieces, or add symmetry; encoding phases via a special hatching direction could extend faithful pictures to complex AME states such as those of Refs. [10, 2].
4. Generalize the construction. Local transformations of the permutation solutions at d = 3 and d = 4 might yield genuine geomagic squares, states of more parties would require higher-dimensional arrays, and a companion note in the repository discusses self-similar geomagic representations of real AME(4, 6ⁿ) states generated from the present square.

Target Audience

Quantum information researchers working on AME states, perfect tensors and multipartite entanglement; mathematicians interested in combinatorial designs, orthogonal Latin squares and geomagic squares; and scientifically literate readers drawn to exact pictorial representations of abstract algebraic objects. Readers seeking experimental protocols or complexity analysis will not find them here — the paper is a construction, a proof and a picture.

Authors’ abstract

Seeking a pictorial representation of four-party absolutely maximally entangled (AME) states of local size six, we independently construct a real AME$(4,6)$ state from a table of signed Pauli matrices. Its two-unitary matrix has $60$ nonzero entries, equal to $1$ or $\pm 1/\sqrt{2}$. A locally equivalent representative yields a genuine geomagic square, where every row, column and main diagonal tiles a hexagon and all $36$ pieces are pairwise non-congruent. Colors, shapes and hatching retain the full tensor, allowing its reconstruction from the picture.