Abstract
This paper identifies six inequivalent $4\times4$ unextendible product bases (UPBs) of size eight derived from 4-qubit product vectors. These results are applied to construct rank-nine Positive-Partial-Transpose (PPT) entangled states. Notably, one UPB is orthogonal to an "almost" genuinely entangled space, which lacks $4\times4$ product vectors across any bipartition. Furthermore, a structural theorem is established relating multipartite UPBs to $n \times n \times n$ tripartite systems, addressing an open problem in quantum information theory.
The Challenge of Bound Entanglement
In quantum information theory, Unextendible Product Bases (UPBs) are critical for understanding the boundary between separable and entangled states. A UPB is a set of orthogonal product vectors whose orthogonal complement contains no product vectors. This property makes them ideal for constructing Positive-Partial-Transpose (PPT) entangled states—states that are entangled but cannot be distilled into pure entanglement, often referred to as "bound entanglement."
Constructing Six Inequivalent UPBs
The researchers utilized the framework of Unextendible Orthogonal Matrices (UOMs) to represent product vectors. By analyzing the algebraic properties of these matrices, they identified six distinct $4 \times 4$ UPBs, labeled $F_1$ through $F_6$.
As shown in the table above, the "maximum number of independent variables" serves as a mathematical signature for each matrix. Because these signatures differ, the six bases cannot be transformed into one another through local unitaries or qubit permutations, proving they are mathematically distinct objects.
Orthogonality and "Almost" Genuinely Entangled Spaces
A key contribution of this work is the exploration of Genuinely Entangled (GE) spaces—subspaces where every state is entangled across every possible bipartition. While finding a UPB orthogonal to a true GE space remains an open challenge, the authors discovered that the basis $F_6$ is orthogonal to an "almost" GE space.
The orthogonality structure depicted in Figure 1 highlights how these vectors are constrained. The basis $F_6$ is unique because it acts as a bipartite UPB across all three major $4 \times 4$ partitions of a 4-qubit system (AB:CD, AC:BD, and AD:BC). This symmetry ensures that its orthogonal complement contains no $4 \times 4$ product vectors in any of these bipartitions, representing a significant step toward finding a UPB orthogonal to a true GE space.
Implications and the $n \times n \times n$ Theorem
Beyond the specific $4 \times 4$ constructions, the paper proves a significant structural theorem: a multipartite UPB orthogonal to a GE space exists if and only if a corresponding $n \times n \times n$ tripartite UPB exists. This reduction simplifies the search for GE-orthogonal UPBs by focusing on tripartite systems.
Furthermore, the construction of rank-nine PPT states from subsets of these UPBs challenges the traditional view that the entire range must be devoid of product states to ensure entanglement. These findings provide a new toolkit for investigating the complex landscape of multipartite entanglement and bound entangled states.
Summary updated