Detecting multipartite entanglement

Andrew C. Doherty, Pablo A. Parrilo, Federico M. SpedalieriPhysical 2005 · DOI · arXiv

Abstract

Determining whether the state of several quantum mechanical subsystems is entangled remains a fundamental challenge in quantum information theory. This paper introduces a systematic procedure for checking separability based on finding state extensions with specific symmetric properties, implemented via semidefinite programming (SDP). We demonstrate that this approach forms a complete hierarchy: if a multipartite state is entangled, the procedure is guaranteed to detect it at some level of the hierarchy. Furthermore, the method provides a constructive means of generating entanglement witnesses for experimental verification.

Introduction

In quantum mechanics, a state is considered separable if it can be prepared using only local operations and classical communication (LOCC). Conversely, any state that cannot be so prepared is entangled. While entanglement is a cornerstone of quantum technologies, identifying it is notoriously difficult. For bipartite systems, the Positive Partial Transpose (PPT) criterion serves as a necessary condition for separability, but it is not sufficient; "bound entangled" states exist which are entangled yet satisfy the PPT condition.

The complexity increases significantly in multipartite systems (three or more parties). The general separability problem is NP-hard, meaning no single efficient algorithm can solve it for all cases. This work addresses the need for a reliable, step-by-step detection framework that is guaranteed to converge for any number of parties.

Methodology: Locally Symmetric Extensions

The core of the proposed method is the concept of Locally Symmetric Extensions (LSE), which leverages the "monogamy of entanglement." If a state is separable, one should be able to mathematically construct a larger system where each party possesses multiple identical copies, maintaining global consistency and symmetry.

Semidefinite Programming and Duality

The search for these extensions is formulated as a Semidefinite Program (SDP). SDPs are a class of convex optimization problems that can be solved efficiently on modern computers. A key advantage of this formulation lies in its dual problem. By analyzing the dual, the algorithm automatically generates an entanglement witness—a specific observable operator $Z$. For a state $\rho$, if the expectation value $\text{Tr}[Z\rho] < 0$, the state is definitively entangled. This provides experimentalists with a "smoking gun" to verify entanglement in the lab.

The Quantum de Finetti Theorem

To prove that this hierarchy is complete, the authors utilize a strengthened version of the Quantum de Finetti Theorem. They demonstrate that as the number of "imaginary copies" in the extension increases, the test becomes increasingly sensitive. If a state possesses symmetric extensions for all possible levels $k$, it must be separable. This inductive proof ensures that no entangled state can indefinitely hide from the hierarchy.

Key Theoretical Insights

  • Completeness via Induction: The most significant theoretical contribution is the proof that this hierarchy is complete for any number of parties $N$.
  • The "One-Party" Shortcut: A significant finding (Corollary 1) reveals that it is not necessary to add copies for every party. Adding copies to $N-1$ parties is sufficient to eventually detect any entanglement, which substantially reduces the computational overhead required for the tests.
  • Physical Interpretation: The existence of these extensions is linked to local hidden variable theories. States that possess such extensions behave "classically" in specific types of Bell-like experiments.

Validation: Detecting Bound Entanglement

The effectiveness of the hierarchy was tested against the Bennett et al. Unextendible Product Basis (UPB) state. This 2x2x2 tripartite state is a classic "hard" case: it is entangled but remains positive under partial transposition across any bipartition, rendering it invisible to the standard PPT test.

The authors successfully detected the entanglement in the UPB state using a $(2, 1, 1)$ extension test. They explicitly calculated an $8 \times 8$ witness matrix $Z$ where $\text{Tr}[Z\rho] = -3/8$, while simultaneously proving that $Z$ remains non-negative for all product states, confirming its validity as a witness.

Scaling and Limitations

For a fixed number of parties and a fixed level in the hierarchy, the computational requirements scale polynomially with the dimension of the Hilbert space. However, the number of variables in the SDP grows rapidly as one moves to higher levels or increases the number of parties $N$. The authors note that the number of inequivalent partial transposes increases very quickly with $N$.

Furthermore, some open questions remain:

  • LOCC Invariance: It is not yet known if the property of having a symmetric extension is preserved under all LOCC operations.
  • Constructive Decompositions: While the method is excellent at detecting entanglement, it does not currently provide a "recipe" (a product state decomposition) for states that are found to be separable. Developing a procedure to construct these decompositions is a priority for future work.

Conclusion

This work provides a powerful, theoretically grounded framework for the multipartite entanglement problem. By transforming a hard existence problem into a structured hierarchy of semidefinite programs, it offers both a proof of entanglement and a practical tool for experimentalists through the construction of entanglement witnesses.

Summary updated

Abstract

We discuss the problem of determining whether the state of several quantum mechanical subsystems is entangled. As in previous work on two subsystems we introduce a procedure for checking separability that is based on finding state extensions with appropriate properties and may be implemented as a semidefinite program. The main result of this work is to show that there is a series of tests of this kind such that if a multiparty state is entangled this will eventually be detected by one of the tests. The procedure also provides a means of constructing entanglement witnesses that could in principle be measured in order to demonstrate that the state is entangled.
BibTeX
@inproceedings{doherty2005detecting,
  author = {Andrew C. Doherty and Pablo A. Parrilo and Federico M. Spedalieri},
  title = {{Detecting multipartite entanglement}},
  booktitle = {Physical},
  year = {2005},
  doi = {10.1103/physreva.71.032333},
}
Figure