[Paper Review] Quantum entanglement without eigenvalue spectra:multipartite case

Hao Chen|ArXiv.org|Sep 13, 2001

Quantum Information and Cryptography2 references3 citations

TL;DR

This paper introduces eigenvalue-free algebraic set invariants for multipartite quantum mixed states, based on the geometric configuration of eigenvectors in product complex projective spaces. These invariants yield a new separability criterion independent of eigenvalues, proving that separable states must have algebraic sets that are sums of products of linear subspaces, and demonstrating that entanglement cannot be fully characterized by partial trace spectra—especially in tripartite systems with LOCC-incomparable states sharing identical spectra.

ABSTRACT

We introduce algebriac sets in the products of complex projective spaces for multipartite mixed states, which are independent of their eigenvalues and only measure the "position" of their eigenvectors, as their non-local invariants (ie. remaining invariant after local untary transformations). These invariants are naturally arised from the physical consideration of checking multipartite mixed states by measuring them with multipartite separable pure states. The algebraic sets have to be the sum of linear subspaces if the multipartite mixed state is separable, and thus we give a new separability criterion of multipartite mixed states. A continuous family of 4-party mixed states, whose members are separable for any 2:2 cut and entangled for any 1:3 cut (thus bound entanglement if 4 parties are isolated), is constructed and studied from our invariants and separability criterion. Examples of LOCC-incomparable entangled tripartite pure states are given to show it is hopeless to characterize the entanglement properties of tripartite pure states by only using the eigenvalue speactra of their partial traces. We also prove that at least $n^2+n-1$ terms of separable pure states, which are "orthogonal" in some sence, are needed to write a generic pure state in $H_A^{n^2} \otimes H_B^{n^2} \otimes H_C^{n^2}$ as a linear combination of them.

Motivation & Objective

• To develop entanglement invariants for multipartite mixed states that are independent of eigenvalue spectra.
• To address the gap in understanding multipartite entanglement, especially for mixed states and pure states beyond bipartite systems.
• To construct a new separability criterion based on geometric configurations of eigenvectors rather than eigenvalues.
• To demonstrate that entanglement properties of multipartite pure states cannot be captured by partial trace eigenvalue spectra alone.
• To generalize bound entangled states (e.g., Smolin-type) and analyze their structure via the proposed invariants.

Proposed method

• Define algebraic sets as the degenerating locus of Hermitian bilinear forms arising from measuring a mixed state with multipartite separable pure states.
• Construct invariants in the product of complex projective spaces that are invariant under local unitary transformations and depend only on the 'position' of eigenvectors.
• Use algebraic geometry to characterize the structure of these invariants, including dimensions, irreducible components, and differential geometric properties like volume and curvature.
• Prove a new separability criterion: a multipartite mixed state is separable only if its associated algebraic set is a sum of products of linear subspaces.
• Apply the criterion to construct a continuous family of 4-party bound entangled states, generalizing Smolin’s construction.
• Analyze tripartite pure states via the invariants to show LOCC-incomparability despite identical partial trace spectra.

Experimental results

Research questions

• RQ1 Can multipartite entanglement be characterized without relying on eigenvalue spectra of density matrices or their partial traces?
• RQ2 What geometric invariants of eigenvectors remain invariant under local unitary transformations and can serve as entanglement measures?
• RQ3 Why do certain tripartite pure states with identical partial trace spectra remain LOCC-incomparable, and how can this be detected algebraically?
• RQ4 Can a continuous family of bound entangled 4-party mixed states be constructed and characterized using eigenvector-based invariants?
• RQ5 How many orthogonal separable pure states are required to decompose a generic tripartite pure state in $H_{A}^{n^2} \otimes H_{B}^{n^2} \otimes H_{C}^{n^2}$?

Key findings

• A new eigenvalue-free separability criterion is established: a multipartite mixed state is separable only if its associated algebraic set is a sum of products of linear subspaces.
• The criterion detects entire continuous families of entangled states—any state with the same eigenvectors and arbitrary eigenvalues is entangled if the criterion is violated.
• A continuous family of 4-party bound entangled mixed states is constructed, generalizing Smolin’s state, each separable under any 2:2 partition but entangled under any 1:3 partition.
• Examples of LOCC-incomparable tripartite pure states are constructed that have identical eigenvalue spectra for all partial traces, proving that spectra alone cannot characterize multipartite entanglement.
• For a generic pure state in $H_{A}^{n^2} \otimes H_{B}^{n^2} \otimes H_{C}^{n^2}$, at least one of the $n^2$ components in a Schmidt-like decomposition must have Schmidt rank at least $n$, showing a fundamental difference from bipartite systems.
• It is proven that at least $n^2 + n - 1$ orthogonal separable pure states are required to express a generic pure state in $H_{A}^{n^2} \otimes H_{B}^{n^2} \otimes H_{C}^{n^2}$ as a linear combination.